Diplo

Health Policy and
Systems Research
A Methodology Reader
Edited by Lucy Gilson
Health Policy and
Systems Research
A Methodology Reader
Edited by Lucy Gilson
WHO Library Cataloguing-in-Publication Data
Health policy and systems research: a methodology reader / edited by Lucy Gilson.
1.Health policy. 2.Health services research. 3.Delivery of health care. 4.Research. 5.Review literature. I.Gilson, Lucy.
II.Alliance for Health Policy and Systems Research. III.World Health Organization.
ISBN 978 92 4 150313 6 (NLM classification: W 84.3)
© World Health Organization 2012
All rights reserved. Publications of the World Health Organization are available on the WHO web site (www.who.int) or can be purchased
from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857;
e-mail: bookorders@who.int).
Requests for permission to reproduce or translate WHO publications – whether for sale or for noncommercial distribution – should be addressed to
WHO Press through the WHO web site (http://www.who.int/about/licensing/copyright_form/en/index.html).
The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part
of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of
its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement.
The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health
Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are
distinguished by initial capital letters.
All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the
published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of
the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use.
The named authors alone are responsible for the views expressed in this publication.
Printed by the WHO Document Production Services, Geneva, Switzerland

Designed by Capria
o te t
Contents
About this Reader ............................................................................................................................................... 11
Part 1
Introduction to Health Policy and Systems Research ....................................................................... 19
1. What is Health Policy and Systems Research? ..................................................................................... 21
Key characteristics ............................................................................................................................................... 21
Key areas of HPSR .............................................................................................................................................. 22
2. Health systems ........................................................................................................................................... 23
Goals .................................................................................................................................................................. 23
Elements and characteristics ............................................................................................................................... 23
Multi-levels of operation ..................................................................................................................................... 24
Interactions and interrelationships ....................................................................................................................... 25
3. Health system development or strengthening ..................................................................................... 26
4. Health policy ............................................................................................................................................... 28
5. Health policy analysis ................................................................................................................................ 28
Policy actors ........................................................................................................................................................ 29
The focus and forms of policy analysis ................................................................................................................ 29
6. The boundaries of HPSR ........................................................................................................................... 30
What HPSR is ...................................................................................................................................................... 31
What HPSR is not ............................................................................................................................................... 32
The distinction between HPSR and service delivery/disease programme research ............................................... 32
Fuzzy boundaries ................................................................................................................................................ 33
7. Understanding the nature of social and political reality ................................................................... 34
Positivism ............................................................................................................................................................ 35
Relativism ........................................................................................................................................................... 35
Critical realism .................................................................................................................................................... 36
HPSR perspectives on causality, generalizability and learning ............................................................................. 36
References ..........................................................................................................................................................38
Contents 5
Part 2
Doing HPSR: Key steps in the process ...................................................................................................... 41
Step 1: Identify the research focus and questions .................................................................................. 43
Networking and creative thinking ....................................................................................................................... 44
Literature search ................................................................................................................................................. 45
Key challenges .................................................................................................................................................... 46
Identifying the purpose of the research .............................................................................................................. 47
Taking account of multidisciplinarity ................................................................................................................... 51
Finalizing research questions ............................................................................................................................... 51
Step 2: Design the study ............................................................................................................................... 52
Using theory and conceptual frameworks to inform the study ............................................................................ 54
Step 3: Ensure research quality and rigour ............................................................................................... 55
Step 4: Apply ethical principles ................................................................................................................... 58
References ..........................................................................................................................................................59
Part 3
Understanding Health Policy and Systems ............................................................................................. 61
Health system frameworks ........................................................................................................................... 63
Bloom G, Standing H, Lloyd R (2008). Markets, information asymmetry and health care: Towards new
social contracts. ................................................................................................................................................... 63
de Savigny D et al. (2009). Systems thinking: What it is and what it means for health systems.
In: de Savigny D, Adam T, eds. Systems thinking for health systems strengthening. ............................................ 63
Conceptual frameworks for HPSR ............................................................................................................... 64
Atun R et al. (2010). Integration of targeted health interventions into health systems: a conceptual
framework for analysis. ....................................................................................................................................... 64
Bossert T (1998). Analyzing the decentralization of health systems in developing countries: decision
space, innovation and performance. .................................................................................................................... 64
Brinkerhoff D (2004). Accountability and health systems: toward conceptual clarity and policy relevance. ......... 64
Franco LM, Bennett S, Kanfer R (2002). Health sector reform and public sector health worker
motivation: a conceptual framework. .................................................................................................................. 64
Gilson L (2003). Trust and health care as a social institution. ............................................................................. 64
Kutzin J (2001). A descriptive framework for country-level analysis of health care financing arrangements. ....... 64
Vian T (2007). Review of corruption in the health sector: theory, methods and interventions. ............................ 64
Walt G, Gilson L (1994). Reforming the health sector in developing countries: the central role
of policy analysis. ................................................................................................................................................ 64
6 Gilson L, ed. (2012). Health Policy and Systems Research: A Methodology Reader
Alliance for Health Policy and Systems Research, World Health Organization
Part 4
Empirical papers .................................................................................................................................................. 65
Overview: research strategies and papers ................................................................................................. 67
1. Cross-sectional perspectives .................................................................................................................... 72
Blaauw D et al. (2010). Policy interventions that attract nurses to rural areas: a multicountry
discrete choice experiment. ................................................................................................................................. 75
Glassman A et al. (1999). Political analysis of health reform in the Dominican Republic. ................................... 83
Morrow M et al. (2009) Pathways to malaria persistence in remote central Vietnam: a mixed-method
study of health care and the community. ............................................................................................................ 95
Ramanadhan et al. (2010). Network-based social capital and capacity-building programs: an example
from Ethiopia ...................................................................................................................................................... 105
Ranson MK, Jayaswal R, Mills AJ (2011). Strategies for coping with the costs of inpatient care: a mixed
methods study of urban and rural poor in Vadodara District, Gujarat, India. ...................................................... 116
Riewpaiboon et al. (2005). Private obstetric practice in a public hospital: mythical trust in obstetric care. ......... 129
Rwashana AS, Williams DW, Neema S (2009). System dynamics approach to immunization healthcare
issues in developing countries: a case study of Uganda. .................................................................................... 139
Sheikh K, Porter J (2010). Discursive gaps in the implementation of public health policy guidelines
in India: The case of HIV testing. ........................................................................................................................ 152
2. The case study approach .......................................................................................................................... 161

Atkinson S et al. (2000). Going down to local: incorporating social organisation and political culture
into assessments of decentralised health care. .................................................................................................... 166
Murray SF, Elston MA (2005). The promotion of private health insurance and its implications for the social
organisation of health care: a case study of private sector obstetric practice in Chile. (Reproduced in the print version only)
Mutemwa RI (2005). HMIS and decision-making in Zambia: re-thinking information solutions for
district health management in decentralised health systems. .............................................................................. 184
Rolfe B et al. (2008). The crisis in human resources for health care and the potential of a ‘retired’
workforce: case study of the independent midwifery sector in Tanzania. ............................................................ 197
Russell S, Gilson L (2006). Are health services protecting the livelihoods of the urban poor in Sri Lanka?
Findings from two low-income areas of Colombo. .............................................................................................. 210
Shiffman J et al. (2004). The emergence of political priority for safe motherhood in Honduras. ......................... 223
3. The ethnographic lens ............................................................................................................................... 235

Aitken JM (1994). Voices from the inside: Managing district health services in Nepal. (Reproduced in the print version only)
Behague DT, Storeng KT (2008). Collapsing the vertical-horizontal divide: an ethnographic study
of evidence-based policymaking in maternal health. ........................................................................................... 239
George A (2009). By papers and pens, you can only do so much: views about accountability and human
resource management from Indian government health administrators and workers. (Reproduced in the print version only)
Lewin S, Green J (2009). Ritual and the organisation of care in primary care clinics in Cape Town,
South Africa. ....................................................................................................................................................... 245
Contents 7
Part 4 (continued)
4. Advances in impact evaluation ............................................................................................................... 253
Björkman M, Svensson J (2009). Power to the people: evidence from a randomized field experiment
on community-based monitoring in Uganda. ...................................................................................................... 257
Macinko J et al. (2007). Going to scale with community-based primary care: an analysis of the family
health programme and infant mortality in Brazil. ................................................................................................ 292
Marchal B, Dedzo M, Kegels G (2010). A realist evaluation of the management of a well-performing
regional hospital in Ghana. ................................................................................................................................. 303
Wang H et al. (2009). The impact of rural mutual health care on health status: evaluation of a social
experiment in rural China. (Reproduced in the print version only)
5. Investigating policy and system change over time ............................................................................. 317

Brown TM, Cueto M, Fee E (2006). The World Health Organization and the transition from
‘international’ to ‘global’ public health. .............................................................................................................. 320
Crichton J (2008). Changing fortunes: analysis of fluctuating policy space for family planning in Kenya. ........... 331
Masanja H et al. (2008).Child survival gains in Tanzania: analysis of data from demographic and
health surveys. .................................................................................................................................................... 343
Van Ginneken N, Lewin S, Berridge S. (2010). The emergence of community health worker
programmes in the late apartheid era in South Africa: an historical analysis. ..................................................... 351
6. Cross-national analysis .............................................................................................................................. 361

Bryce J et al. (2005). Programmatic pathways to child survival: results of a multi-country evaluation
of Integrated Management of Childhood Illness. ................................................................................................ 363
Gilson L et al. (2001). Strategies for promoting equity: experience with community financing
in three African countries. ................................................................................................................................... 369
Lee K et al. (1998). Family planning policies and programmes in eight low-income countries:
A comparative policy analysis. ............................................................................................................................ 400
O’Donnell O et al. (2007). The incidence of public spending on healthcare: comparative evidence
from Asia. ........................................................................................................................................................... 411
7. Action research ........................................................................................................................................... 443

Khresheh R, Barclay L (2007). Practice-research engagement (PRE): Jordanian experience
in three Ministry of Health hospitals. .................................................................................................................. 445
Khresheh R, Barclay L (2008). Implementation of a new birth record in three hospitals in Jordan:
a study of health system improvement. .............................................................................................................. 461
Part 5
Reflections on Health Policy and Systems Research ......................................................................... 469
de Savigny D et al. (2009). Systems thinking: Applying a systems perspective to design and evaluate
health systems interventions. In: de Savigny D, Adam T, eds. Systems thinking for health systems
strengthening. ..................................................................................................................................................... 471
English M et al. (2008). Health systems research in a low-income country: easier said than done. .................... 471
Erasmus E, Gilson L (2008). How to start thinking about investigating power in the organizational
settings of policy implementation. ....................................................................................................................... 471
Hanson K et al. (2008). Vouchers for scaling up insecticide-treated nets in Tanzania: methods for
monitoring and evaluation of a national health system intervention. ................................................................. 471
Hyder A et al. (2010). Stakeholder analysis for health research: case studies from low- and
middle-income countries. .................................................................................................................................... 472
Molyneux CS et al. (2009). Conducting health-related social science research in low income settings:
ethical dilemmas faced in Kenya and South Africa. ............................................................................................. 472
Ridde V (2008). Equity and heath policy in Africa: using concept mapping in Moore (Burkina Faso). ................ 472
Sakyi EK (2008). A retrospective content analysis of studies of factors constraining the implementation
of health sector reform in Ghana. ....................................................................................................................... 472
Schneider H, Palmer N (2002). Getting to the truth? Researching user views of primary health care. ................ 472
van der Geest S, Sarkodie S (1998). The fake patient: a research experiment in a Ghanaian hospital. ............... 472
Walt G et al. (2008). ‘Doing’ health policy analysis: methodological and conceptual reflections
and challenges. ................................................................................................................................................... 472
Contents 9
List of Figures
Figure 1 The different levels of health systems ..................................................................................................... 24
Figure 2 The interconnections among the health system building blocks ............................................................. 25
Figure 3 The terrain of HPSR ................................................................................................................................ 30
Figure 4 Multiple research purposes ..................................................................................................................... 50
List of Tables
Table 1 Typical system constraints and possible disease-specific and health-system responses ........................... 27
Table 2 Key elements of knowledge paradigms as applied in HPSR .................................................................... 35
Table 3 Priority research questions in three health policy and systems areas,
results of international priority-setting processes .................................................................................... 44
Table 4 Examples of HPSR questions .................................................................................................................. 48
Table 5 A summary of broad study designs ........................................................................................................ 52
Table 6 Key features of fixed and flexible research strategies ............................................................................. 53
Table 7 Criteria and questions for assessing research quality .............................................................................. 56
Table 8 Processes for ensuring rigour in case study and qualitative data collection and analysis ....................... 56
Table 9 Overview of papers presented ................................................................................................................ 68
Table 10 Procedures to ensure trustworthiness in case study research ................................................................. 162
List of Boxes
Box 1 Suggested topics for health systems research ......................................................................................... 31
Box 2 Topics addressed by existing empirical HPSR studies .............................................................................. 32
Box 3 Broad research questions of interest to national policy-makers .............................................................. 45
Box 4 The HPSR questions of different health policy and systems actors ......................................................... 46
Box 5 The purpose of different types of research ............................................................................................. 49
Box 6 Links between purpose and broad forms of research questions ............................................................. 50
Box 7 Eight ethical principles for clinical research in low- and middle-income countries .................................. 59
A o t th Reader
About this Reader
What does this Reader offer?
Health Policy and Systems Research (HPSR) is often criticized for lacking rigour, providing a weak basis for generalization of
its findings and, therefore, offering limited value for policy-makers. This Reader aims to address these concerns through
supporting action to strengthen the quality of HPSR.
The Reader is primarily for researchers and research users, teachers and students, particularly those working in low- and
middle-income countries (LMICs). It provides guidance on the defining features of HPSR and the critical steps in conducting
research in this field. It showcases the diverse range of research strategies and methods encompassed by HPSR, and it
provides examples of good quality and innovative HPSR papers.
The production of the Reader was commissioned by the Alliance for Health Policy and Systems Research (the Alliance) and
it will complement its other investments in methodology development and postgraduate training.
Why is the Reader needed?

Health systems are widely recognized to be vital elements of the social fabric of every society. They are not only critical for
the treatment and prevention of ill-health but are central strategies for addressing health inequity and wider social injustice
(Commission on the Social Determinants of Health, 2008). Health systems also provide the platform from which to launch
dedicated efforts to address major diseases and health conditions that burden low-income populations, such as HIV/AIDS,
tuberculosis and malaria. Given these roles, the early 2000s saw a significant expansion of international and national
interest in health systems as one component of sustainable development in LMICs. Health system strengthening is now
seen to be essential for the achievement of the Millennium Development Goals (Travis et al., 2004).
However, the knowledge base to support health system strengthening and policy change in LMICs is surprisingly weak
(World Health Organization, 2009). The body of available work is quite limited compared to other areas of health research
and suffers from various weaknesses. Thus, HPSR is criticized as being unclear in its scope and nature, lacking rigour in the
methods it employs and presenting difficulties in generalizing conclusions from one country context to another (Mills,
2012). Review of health policy analysis work, in particular, also shows that research in this area is often weakly
contextualized and quite descriptive, and offers relatively limited insights into its core questions of how and why policies
are developed and implemented effectively over time (Gilson & Raphaely, 2008). As HPSR remains a ‘cinderella’, or
marginal, field in health research these weaknesses are not particularly surprising. Within LMICs there are very few
national researchers working on health policy and systems issues, and there is a lack of relevant training courses (Bennett
et al., 2011). Yet the need is clear – as Julio Frenk, Dean of the Harvard School of Public Health, stated at the First Global
Symposium on Health Systems Research held in Montreux, Switzerland, in 2010:
we need to mobilise the power of ideas in order to influence the ideas of power, that is to say, the ideas of those
with the power to make decisions.
About this Reader 13

What does the Reader aim to do?
This Reader aims to support the development of the field of HPSR, particularly in LMICs. It complements the range of
relevant texts that are already available (see examples at the end of this section) by providing a particular focus on
methodological issues for primary empirical health policy and systems research.
More specifically, the Reader aims to support the practice of, and training in, HPSR by:
encouraging researchers to value a multidisciplinary approach, recognizing its importance in addressing
n
the complexity of health policy and systems challenges;
stimulating wider discussion about the field and relevant research questions;
n
demonstrating the breadth of the field in terms of study approaches, disciplinary perspectives, analytical
n
approaches and methods;
highlighting newer or relatively little-used methods and approaches that could be further developed.
n
The Reader is mainly for use by:

researchers and health system managers who wish to understand and apply the multidisciplinary approaches
n
of HPSR in order to identify comprehensive strategies that address the complex challenges of health system
development;
teachers and facilitators involved in HPSR training;
n
students, from any discipline or background, who are new to the field of HPSR.
n
How is the Reader structured?

There are four main sections in the Reader:
Part 1 provides an overview of the field of HPSR in LMICs and some of the key challenges of this kind of research.
Part 2 outlines key steps to follow when conducting HPSR studies.
Part 3 presents some key references of papers which provide overarching conceptual frameworks for understanding health
policy and health systems.
Part 4 is the main body of the Reader and presents a set of empirical papers drawn exclusively from LMICs. The papers
were selected because they:
together demonstrate the breadth and scope of HPSR work

n
provide good examples of different forms of research strategy relevant to HPSR
n
are high quality and innovative.
n
Part 5 presents a set of references for papers that reflect on specific concepts or methods relevant to HPSR as well as
some of the particular challenges of working in this field.
Doing HPSR: from research questions to reseach strategy
The defining feature of primary HPSR is that it is problem- or question-driven, rather than, as with epidemiology, method-
driven. Therefore, as outlined in Part 2, the first step in doing rigorous and good quality research is to clarify the purpose of
the research, what the study is trying to achieve, and to identify and develop relevant and well-framed research questions.
Good quality work then demands an understanding of the research strategy that is appropriate to the questions of focus.
The strategy is neither primarily a study design nor a method, but instead represents an overarching approach to
conducting the research; it considers the most appropriate methods of data collection and sampling procedure in terms of
the research purpose and questions. The art of study design in HPSR, as with all ‘real world research’, is about turning
research questions into valid, feasible and useful projects (Robson, 2002).
The papers in Part 4 are grouped by research strategy in order to encourage critical and creative thinking about the nature
and approach of HPSR, and to stimulate research that goes beyond the often quite descriptive cross-sectional analyses that
form the bulk of currently published work in the field. The research strategies were chosen to demonstrate the breadth of
HPSR work, covering both dominant and emerging approaches in the field.
They are:
1. Cross-sectional perspectives
2. The case study approach
3. The ethnographic lens
4. Advances in impact evaluation
5. Investigating policy and system change over time
6. Cross-national analysis
7. Action research
Each of the sections in Part 4 includes: a brief overview of the relevance of the research strategy to HPSR; critical elements
of the strategy that must be considered in conducting rigorous work; and an introduction to the selected papers.
We note that secondary research or synthesis methods are not addressed here, and readers interested in that particular
research area are encouraged to use relevant supporting materials. These include, for example, a Handbook developed
with the Alliance support and downloadable from:
http://www.who.int/alliance-hpsr/projects/alliancehpsr_handbook systematicreviewschile.pdf
Three broader texts of use to those doing HPSR are:

Fulop N et al., eds (2001). Issues in studying the organisation and delivery of health services. In: Fulop N et al., eds.
Studying the organisation and delivery of health services: research methods. London, Routledge.
Robson C (2011). Real world research: a resource for social scientists and practitioner-researchers, 3rd ed. Oxford,
Blackwell Publishing.
Thomas A, Chataway J, Wuyts M, eds (1998). Finding out fast: investigative skills for policy and development. London, Sage
Publications.

How was the Reader developed?
The Reader was developed through a process of five steps:
1. engagement with relevant researchers across the world to identify potential papers for inclusion and comment
on an initial draft of Part 2;
2. development and teaching of a new course, “intoduction to Health Systems Research and Evaluation” as part of
the University of Cape Town’s Master’s in Public Health (health systems) degree programme;
3. review of papers and selection of an initial “long list” for possible inclusion in the Reader;
4. presentation and discussion of the initial ideas for the Reader and the long list of papers, at the 2010 Montreux,
First Global Symposium on Health Systems Research;
5. final selection of papers and finalization of the section introductions.
The team
A multidisciplinary group of researchers, with a range of relevant experience and organizational bases, supported the
Reader’s development process. The team was led by:
Lucy Gilson (health policy/health economics, South Africa/United Kingdom of Great Britain and Northern Ireland)
n
and included:
Sara Bennett (health policy/health economics, United States of America)
n
Kara Hanson (health economics, United Kingdom of Great Britain and Northern Ireland)
n
Karina Kielmann (medical anthropology, United Kingdom of Great Britain and Northern Ireland)
n
Marsha Orgill (health policy/health systems, South Africa)
n
Helen Schneider (public health/health policy, South Africa).
n
Irene Agyepong (public health manager/health policy, Ghana), Kabir Sheikh (health policy/public health, India) and Freddie
Ssengooba (health systems/health policy, Uganda), also contributed greatly to conceptualizing Part 2, in part through
their collaboration with Sara Bennett, Lucy Gilson and Kara Hanson in a set of parallel papers published in PLoS Medicine
(Bennett et al., 2011; Gilson et al., 2011; Sheikh et al., 2011).
A range of inputs or comments on the Reader’s development were also received from a broader group of colleagues who
deserve a special note of thanks (see below).
Ultimately, however, the selection of papers in this Reader reflects the particular perspectives of those most closely involved
in its development – both on the nature of the field and on what constitutes a good quality or unusual study and paper.
The Reader is, therefore, a starting point for reflection on HPSR, not an end point. It must be seen as a living document
that will develop over time.
Please note that this Reader is mostly available online at: http://www.who.int/alliance-hpsr/resources/reader/en.
For their comments and ideas, particular thanks to:
Virginia Berridge Prasanta Mahapatra Steven Russell
Thomas Bossert Martin Mckee William Savedoff
Susan Cleary Barbara McPake Jeremy Shiffman
Don de Savigny Anne Mills Sameen Siddiqi
Mike English Sassy Molyneux Barbara Starfield
Abdul Ghaffar Susan Fairley Murray Viroj Tangcharoensathien
Miguel Angel González Block Piroska Östlin Sally Theobold
Felicia Knaul Weerasak Putthasri Stephen Tollman
Mylene Lagarde Kent Ranson Wim Van Damme
Asa Cristina Laurell Michael Reich Frank Wafula
John Lavis Valéry Ridde Gill Walt
Bruno Marchal John-Arne Röttingen
Thanks also to all those who participated in the special session to discuss ideas about the Reader that was held as part of
the First Global Symposium on Health Systems Research, Montreux, November 2010.
Lydia Bendib managed the overall production of the Reader.
Publishers acknowledgement
Grateful acknowledgement is made to the various publishers for permission to reproduce in this Reader, the full text articles
included in Part 4.

References
Bennett S et al. (2011). Building the field of health policy and systems research: an agenda for action. PLoS Medicine
8(8):e1001081.
Commission on the Social Determinants of Health (CSDH) (2008). Closing the gap in a generation: health equity through
action on the social determinants of health. Geneva, World Health Organization.
Gilson L, Raphaely N (2008). The terrain of health policy analysis in low and middle income countries: a review of
published literature 1994–2007. Health Policy and Planning, 23(5):294–307.
Gilson L et al. (2011). Building the field of health policy and systems research: social science matters. PLoS Medicine
8(8):e1001079.
Mills A (2012). Health policy and systems research: defining the terrain; identifying the methods. Health Policy and
Planning 27(1):1-7.
Robson C (2002). Real world research: a resource for social scientists and practitioner-researchers, 2nd ed. Oxford,
Blackwell Publishing.
Sheikh K et al. (2011). Building the field of health policy and systems research: framing the questions. PLoS Medicine
8(8):e1001073.
Travis P et al. (2004). Overcoming health-systems constraints to achieve the Millennium Development Goals. Lancet,
364:900–906.
World Health Organization (2009). Scaling up research and learning for health systems: now is the time. Report of a High
Level Task Force, presented and endorsed at the Global Ministerial Forum on Research for Health 2008, Bamako, Mali.
Geneva, World Health Organization.
Part
trod t o to ealth ol y
a d y te Re ear h
Lucy Gilson
ni e sity o e o n out ic nd
London c ool o y iene nd o ic l edicine nited in do o Ge t it in
nd o t e n el nd
1. What is Health Policy and Systems Research?
his part of the Reader provides an overview of Health Policy and Systems Research HPSR and the key
elements and issues with which it is concerned. t includes an outline of the main knowledge paradigms that are
encompassed within this field of research.
Key points from this section
Features that define HPSR are:

• the types of issues it addresses
• the fact that it seeks to address real-world situations and issues
• it is multidisciplinary, drawing on methods and perspectives from a range of disciplines.
HPSR investigates issues such as: how health care is financed, organized, delivered and used; how health
policies are prioritized, developed and implemented; and how and why health systems do or do not generate
health and wider social goals.
It brings together health policy and health systems work into one integrated field. This combined focus on
health policy and health systems issues provides a strong basis for identifying what can be done to:
1. strengthen health systems so they can better achieve their health and broader social goals; and
2. ensure that the related research is applied research that has the potential to support the implementation
of health policies and health system development.
Key characteristics of HPSR

Health policy and systems research (HPSR) is defined Health Policy and Systems Research:
as a field: is a multidisciplinary research field, distinguished by
n
the issues and questions addressed through the
that seeks to understand and improve how
research rather than by a particular disciplinary base
societies organi e themselves in achieving collective or set of methods;
health goals, and how different actors interact in
includes research that focuses on health services as
n
the policy and implementation processes to well as on the promotion of health in general;
contribute to policy outcomes. y nature, it is inter- includes concern for global and international issues
n
disciplinary, a blend of economics, sociology, as well as national and sub-national issues, as global
anthropology, political science, public health and forces and agencies have important influences over
epidemiology that together draw a comprehensive health systems in low- and middle-income countries;
picture of how health systems respond and adapt encompasses research on or of policy, which means
n
to health policies, and how health policies can shape that it is concerned with how policies are developed
and implemented and the influence that policy actors
and be shaped by health systems and the
have over policy outcomes – it addresses the politics
broader determinants of health. lliance for Health of health systems and health system strengthening;
Policy and Systems Research, . promotes work that explicitly seeks to influence
n
This definition also highlights its key characteristics policy, that is, research for policy.
(Alliance for Health Policy and Systems Research, 2007;
Mills, 2012).
Part 1 ntroduction to Health Policy and Systems Research 21

An integrated approach 4. A specific focus on policy implementation allows
for and requires a better understanding of the
Importantly, HPSR brings together health policy and
organizational dynamics of health systems, which is a
health systems work into one research field, as there are
critical and often overlooked element of health
four linkages between these apparently separate areas of
system functioning.
work, as listed below.
In practice, therefore, the two apparently different areas
1. Health policies can be seen as the purposeful and
of work – health policy and health systems – overlap.
deliberate actions through which efforts are made to
Together they provide the knowledge base relevant to
strengthen health systems in order to promote
strengthen health systems whilst also showing how
population health.
knowledge and other forms of power together influence
2. Health policy actions must not only be informed by
policy decision-making. In these ways, HPSR work always
an understanding of the current dynamics of health
seeks to be policy relevant.
system functioning and performance, but are also
sustained, or undermined, by whether and how they
find expression in the health system. Key areas of HPSR
3. A better understanding of the politics of health policy Each of the four central elements in HPSR are considered
change, the actors and interests driving the processes in the following sections. Some key definitions, concepts
through which policies are developed and implemen- and frameworks are discussed. These provide a foun-
ted, contributes to understanding how to influence dation for thinking about issues related to HPSR, defining
policy and take action to strengthen health systems. appropriate research questions and analysing the
findings of such research.
Key points from the following four sections
Four central elements in HPSR are:

• health systems • health policy
• health system development or strengthening • health policy analysis.
The issues related to each of these elements can be understood through a range of definitions, concepts and
frameworks, which also help to generate relevant and appropriately framed research questions. Such
frameworks allow us to understand the various elements, characteristics and dimensions of a health system;
and to identify the different connections and interrelationships within a health system that need to be
considered in order to strengthen them.
New health policies represent efforts to introduce deliberate and purposeful change within health systems.
Ideas and concepts related to policy and the analysis of such policy are an important part of HPSR. In seeking
to support better policy implementation, it is critical that we understand the factors that influence policy
outcomes. Through understanding the nature of policy and the processes of policy change, we gain new
insights that help to explain how health system actors, and the relationships of power and trust among them,
influence health system performance.
2. Health systems Elements and characteristics
Health systems can be defined either by what they seek In terms of the elements they comprise, health systems
to do and achieve, or by the elements of which they are can be understood as:
comprised. 1. Encompassing the population the system serves, as
well as the supply or delivery of services, interven-
tions and activities intended to promote health and
oals wider value. Members of the population play five
critical health-related roles. They are:
The defining goal of health systems is generally seen as
health improvement – achieved not only through the patients with health needs requiring care
n
provision of curative and preventive health services but consumers with expectations of how they will be
n
also through the protection and promotion of public treated
health, emergency preparedness and intersectoral action taxpayers who provide the main source of
n
financing for the system
(Mackintosh & Koivusalo, 2005).
citizens who may have access to health care as a
n
However, health systems are also part of the social fabric right
in any country, offering value beyond health (Gilson, co-producers of health through their healthsee-
n
2003; Mackintosh, 2001). Their wider goals include king and health-promoting behaviours (Frenk,
equity, or fairness, in the distribution of health and the 2010).
costs of financing the health system as well as protection 2. A set of six functions, or building blocks, some of
for households from the catastrophic costs associated which are clearly represented in the goals outlined
with disease; responsiveness to the expectations of the above (World Health Organization, 2007):
population; and the promotion of respect for the dignity service delivery
n
of persons (World Health Organization, 2007). These last health workforce
n
two goals specifically require: information
n
ethical integrity, citizen’s rights, participation and
n medical products, vaccines and technologies
n
involvement of health system users in policy financing
n
development, planning and accountability and
leadership/governance.
n
respect of confidentiality as well as dignity in service
provision (Mackintosh & Koivsualo, 2005); 3. Incorporating, within the service delivery function
(Van Damme et al., 2010):
building and maintaining the social relations that
n
general curative and preventive health services
n
support sustained resource redistribution, through
and services aimed at specific health problems,
strategies and activities that include, rather than
including specific disease control programmes
exclude, socially marginalized population groups
and personal and population-based services;
within all decision-making activities (Freedman et al.,
2005). a range of modes or channels of service delivery
n
including various levels of facility, other outlets for
Therefore, health systems, through both their service health goods (such as pharmacies or shops) and
provision role and their influence over societal relations, other strategies (such as community-based health
are a critical field of action to address the social deter- workers and activities);
minants of health and the related health inequities a complex mixture of service providers – public
n
(Commission on the Social Determinants of Health, 2008; and private, for profit and not-for-profit, formal
and informal, professional or non-professional,
Gilson et al., 2008).
allopathic or traditional, remunerated and volun-
tary – the pluralistic health care system (Bloom,
Standing & Lloyd, 2008).

ulti-levels of operation There has, however, been growing realization of the
strong influence of the broader global context over
Health systems operate at, and across, the macro, meso population health and health care (Smith & Hanson,
and micro levels (Fulop et al., 2001; Van Damme et al., 2011). Critical influences include international trade,
2010). This is illustrated in Figure 1. international aid and global changes, such as economic
trends or climate change. There are also a range of very
As Figure 1 suggests, the macro level has traditionally
influential global organizations and actors, including
focused mainly on the national, or domestic, health system
multilateral and bilateral organizations, and global
whilst recognizing that this system is also influenced by a
public-private initiatives. Therefore, the domestic health
wider national and international context. Key system roles
system must be understood as an open system within
at the national level include:
the global context, influenced by and influencing global
balancing policies, strategies, resource allocation and
n
forces.
health worker reward systems in line with overall
system goals; The meso level comprises both the local health system,
coordination across functions and service delivery
n often called the district health system, and the organiza-
activities and interventions; tional level, such as hospitals. System roles at this level
the development of policy and regulations;
n include:
engaging with health system actors, including
n responding to local needs and circumstances, in
n
citizens; terms of provision of health services and wider health
interactions with other national agencies that
n promoting activities;
influence health as well as international agencies and coordination among local actors;
n
processes.
management of health services, activities and health
n
workers;
Figure 1 The different levels of health systems
MACRO LEVEL
lo al at o al o te t
o e t ealth y te
Policy
lites
MICRO LEVEL:
Individuals
Health
Managers
Citizens
Patients
Pro iders
MESO LEVEL:
Organisation & Local level
supervision and training of service providers;
n Interactions and
adaptation of national policy and guidelines to local
n
circumstances. interrelationships
Finally, the micro-level is the level of the individuals in Health systems encompass not only various elements but
the system. It includes providers and patients as well as also the interactions and interrelationships between
citizens, managers and policy elites – and the interac- those elements and between the various individuals
tions between them. Critical roles of individuals at this within the system (Frenk, 1994). These relationships
level include: not only support service delivery towards health
the search for care, compliance with health advice
n improvement but are also central to the wider social
and broader health behaviours; value generated by the health system (Gilson, 2003).
the provision of health care and health promoting
n
he building blocks do not alone constitute a system,
activities;
any more than a pile of bricks constitutes a func-
the development of new forms of provider–patient
n
tioning building. t is the multiple relationships and
interaction, such as the use of patient information for
follow-up; interactions among the blocks how one affects and
influences the others, and is in turn affected by
the development of broader local relationships
n
between health system agents and the population; them that converts these blocks into a system
managerial decision-making and leadership across
n de Savigny dam, : see Figure .
the health system. The relationships are, moreover, shaped and influ-
enced by both the hardware and the software of the
health system and, in turn, influence levels of system
performance.
Figure 2 The interconnections among the health system building blocks

Source: de Savigny dam, : .
G E E
E E nd
E L GE
E LE
E E G
E E
EL E

Health system hardware includes the particular organiza-
tional, policy, legal and financing frameworks that struct-
3. Health system
ure any health system, as well as its clinical and service development or
delivery requirements. The software encompasses the
institutions (norms, traditions, values, roles and proce-
strengthening
dures) embedded within the system. As explained in the previous section, health systems are
These two health system dimensions are often tied shaped by both structural (hardware) components and
together. For example, financing mechanisms not only social (software) elements. Therefore, in order to iden-
influence the level of funding available for the health tify actions to develop or strengthen health systems,
system, but also indicate what is valued by that system. researchers need to consider:
Here is an example: the taxation-based elements of changes in the structures of the system that are likely
n
the system signal the extent to which society is prepared to generate performance gains; as well as
to take collective action to support redistribution; whilst what can be done to influence the behaviour and
n
the level of fee for service within the system signals practices of health system agents; and
the extent to which society values choice, allowing those how to implement both sets of changes in ways that
n
are most likely to secure intended effects (Roberts et
who can afford to, to pay for health care to buy more or
al., 2008).
better services. The set of financing mechanisms, more-
over, influence relationships between the state and its These system-level interventions sometimes focus on
citizens as well as between providers and patients, and more than one of the building blocks, such as pay-for-
has a direct influence over levels and patterns of health performance systems that together address human
care utilization, the extent to which the health system resource and financing issues.
offers financial protection in times of health crisis and Alternatively, through the governance or information
the contribution of the heath system to generating social building blocks, such intervention can encompass
solidarity (Gilson et al., 2008). processes and strategies that bring about change across
The recent attention on systems thinking, therefore, the system as a whole – that is, across system building
encourages a focus on the nature of health system blocks, levels and/or dimensions (de Savigny & Adam,
relationships and the synergies emerging from them, 2009). The hardware and software dimensions of health
recognizing that the sum of the whole is more than systems may together be addressed by, for example, new
the sum of the parts (de Savigny & Adam, 2009). accountability mechanisms, or processes, and monitoring
and evaluation strategies. There is also potential for new
leadership and management approaches to focus on the
deliberate development of the institutional and relational
nature of the health system (Gilson, 2012).
Some disease or programme-specific interventions also

have system-wide effects, such as scaling up anti-
retroviral therapy or integrating vouchers for malaria-
preventing bednets into ante natal care (de Savigny &
Adam, 2009). However, most disease programme or
service-specific strategies are unlikely by themselves to
bring about improvements across the health system.
Such strategies suffer one or more of the following
weaknesses (Travis et al., 2004).
They may: nachieve short-term goals but prevent the develop-
ment of long-term strategies to sustain those goals
limit the policy options considered in system impro-
n
(as when donor-funded financial incentives encou-
vement by focusing more on actions at the micro
rage performance gains in one programme that
level (individual) rather than meso level (local and
cannot be sustained over time or do not benefit all
organizational) or macro levels (national and global);
services due to resource constraints).
crowd-out routine activities (as when a number of
n
training activities occurs at the expense of service In contrast, Table 1 shows that system-level responses
delivery); to the common constraints that particular services or
programmes may face are broad in focus and aim
to tackle the root causes of the problems. However, such
responses generally take longer to have effect and their
implementation is likely to be more difficult to manage.
Table 1 Typical system constraints and possible disease-specific and health-system responses
(Source: Travis et al., 2004)
Constraint Disease-specific response Health-system response
Financial inaccessibility: Exemptions/reduced prices Development of risk-pooling

inability to pay, informal fees for focal diseases strategies
Physical inaccessibility: Outreach for focal diseases Reconsideration of long-term

distance to facility plan for capital investment and
siting of facilities
Inappropriately skilled staff Continuous education and Review of basic medical and
training workshops to develop nursing training curricula to
skills in focal diseases ensure that appropriate skills
included in basic training
Poorly motivated staff Financial incentives to reward Institution of proper performance

delivery of particular priority review systems, creating greater
services clarity of roles and expectations
regarding performance of roles,
review of salary structures and
promotion procedures
Weak planning and management Continuous education and training Restructuring ministries of health,
workshops to develop skills in recruitment and development of
planning and management cadre of dedicated managers
Lack of inter sectoral action and Creation of special disease-focused Building systems of local gover-
partnership cross-sectoral committees and task nment that incorporate represen-
forces at national level tatives from health, education,
agriculture, and promote accoun-
tability of local governance
structures to the people
Poor quality care amongst private Training for private sector Development of accreditation and
sector providers providers regulation systems

4. Health policy 5. Health policy analysis
Health policy can be understood as the: Health policy analysis is a central strand of HPSR. It is
sometimes understood as the technical work that
...courses of action and inaction that affect the sets
underpins the development of new policies or the central
of institutions, organi ations, services and funding
element of their evaluation. It includes, for example,
arrangements of the health system. t includes policy
epidemiological analysis that identifies risk factors for
made in the public sector by government as well as
particular diseases and the important targets for health
policies in the private sector. ut because health is
interventions; or cost-effectiveness analysis that identifies
influenced by many determinants outside the health
which of several possible interventions to address a
system, health policy analysts are also interested in
particular health problem provides the best value for
the actions and intended actions of organi ations
money. However, a more political and organizational
e ternal to the health system which has an impact
approach to policy analysis sees policy itself as a
on health for e ample, the food, tobacco or pharma-
process, the process of decision-making, rather than
ceutical industries use, Mays alt, : .
focusing only on policy as the output of that process or
Commonly, health policies are understood as the formal, as a management input (Harrison, 2001; Thomas, 1998).
written documents, rules and guidelines that present
Technical analysts often conceive of policy analysis as
policy-makers’ decisions about what actions are deemed
including several stages, such as getting a problem or
legitimate and necessary to strengthen the health system
issue prioritized for policy action, defining what the
and improve health. However, these formal documents
problem is and what objectives would represent an
are translated through the decision-making of policy
improvement to it, identifying the causes of the problem
actors (such as middle managers, health workers,
and how they are inter linked, identifying possible
patients and citizens) into their daily practices (for
interventions that would address the factors causing
example, management, service delivery, interactions with
the problem, considering options for intervention,
others). Ultimately, these daily practices become health
implementing selected options, evaluation and feedback
policy as it is experienced, which may differ from the
(Harrison, 2001).
intentions of the formal documents (Lipksy, 1980).
Therefore, policy can be seen not only as the formal However, analysts adopting a political and organizational
statements of intent but also as the informal, unwritten approach to policy analysis do not assume that these
practices (Buse, Mays & Walt, 2005). stages are sequential or that they always occur in every
decision-making process. Indeed, these policy analysts
often describe the policy process as a mess, a set of
incremental decisions:
not only is policy designed to change a given

situation but the situation is changing anyway and
giving rise to changing pressures for changes in
policy. he fact that policy is constantly developing in
this way makes it useful to think of policy itself as a
process. homas, : .
The focus of this form of policy analysis goes beyond the
content of particular policies and gives greater attention
The focus and forms
to the behaviour of health policy actors: their processes of policy analysis
of decision-making and the actions they take; their lack
Policy analysis specifically considers: (a) the roles of
of action and unintended actions; the influence of actors who influence policy change at different levels –
content on those actions; and the context that influences from individual, organizational, national to global – and
and is influenced by these behaviours (Walt & Gilson,
their interests; (b) the influence of power relations,
1994). Such analysis offers insights that can be well institutions (the rules, laws, norms and customs that
combined with those of systems thinking (Gilson, 2012). shape human behaviour) and ideas (arguments and
For some, health policy is “synonymous with politics evidence), over health system operations and policy
and deals explicitly with who influences policy-making, change within them; and (c) global political economy
how they exercise that influence, and under what issues. It also seeks to understand the forces influencing
conditions” (Buse, Mays & Walt, 2005:6). why and how policies are initiated, developed or formu-
lated, negotiated, communicated, implemented and
Policy actors evaluated, including how researchers influence policy-

making (Overseas Development Institute, 2007). The
Within national settings, policy actors include those who: latter includes considering whether and why routine
have specific responsibility for developing formal
n practices differ from, and may even contradict, formal
policies in the public or private sectors, including policies, and generate an implementation gap between
those outside the health sector working on health- policy intentions and routine practice.
influencing policies, and international agencies and
organizations; Finally, although policy analysis may be conducted retro-
spectively, to understand past experience, it can also be
influence how policies are translated into practice
n
(such as middle managers, health workers, patients used prospectively to support health policy change and
and citizens); health system strengthening. Prospective policy analysis
has been proposed as an important support for advocacy
seek to influence the formal policy process (such as
n
civil society groups or interest groups at national and efforts (Buse, 2008) and as a key component of health
international levels). system leadership and governance activities (Gilson,
2012).
At global level, policy actors include the range of multi-
lateral and bilateral organizations engaged in activities A new approach to health system development, global
that are likely to influence health, as well as the newly health diplomacy (Smith & Hanson, 2011), also reco-
powerful global public–private initiatives (such as the gnizes that health policy actors must increasingly nego-
Gates Foundation), and transnational civil society tiate and engage with a range of actors at national and
movements. international levels, and outside the national health
system. Examples of global health diplomacy include
action to influence the global tobacco trade or to
develop the World Health Organization Code on the
Ethical Recruitment of Health Personnel; and, at national
level, efforts to secure increased health budgets in
African countries – in line with the Abuja target of 15
of total government budget.

6. The boundaries of HPSR
his section focuses on the types of issues addressed through HPSR. s HPSR is a new and emerging field,
the issues it addresses and how it differs from other related areas of health research are not always understood.
he four elements outlined in the previous section health systems and their development, health policy
and policy analysis provide the basis for the ideas presented in this section. Figure illustrates key elements
of the field of HPSR.
HPSR is an emerging area of health research. It focuses on health policies and health systems – what they
are; how policies are implemented; how health systems work; and what can be done to improve policy
implementation and the functioning of health systems.
Issues relevant to HPSR are wide ranging, include a variety of actors, and may be studied at local, national
and global levels.
HPSR can be distinguished from research focused on specific health programmes, for example those relating
to malaria or HIV/AIDS, by its focus on the broader setting in which such programmes are implemented. HPSR
includes, for example, work on the financing, human resource or governance elements of the health system
that underpin all service provision.
However, HPSR has fuzzy boundaries – it has overlaps with health services research and operational research,
and there are some grey areas between HPSR and aspects of management and some discipline-specific
research.
Figure 3 The terrain of HPSR
lo al at o al or e
Health Systems Health Policy
Hardware: Content &

Structure; Instruments
Organization; System
Technology; functioning Actors, Power
Resourcing & & Politics
Policy
Software: Institutions, Interests
Values;
Change & Ideas
Norms;
Actors & Relationships
What HPSR is the processes and institutional arrangements within
n
which policy change is developed and implemented
(meso level analysis);
HPSR encompasses research on the policies, organiza-
tions, programmes and people that make up health the impact of specific people on policy change and its
n
impacts (micro level analysis) – the balance of struc-
systems, as well as how the interactions amongst these
ture (institutional influences) and agency (autonomy)
elements, and the broader influences over decision- that shapes such actions (Hudson & Lowe, 2004).
making practices within the health system, influence
HPSR considers the full range of policy actors, not only
system performance.
those with formal policy influence, or in formal policy-
HPSR seeks to understand: making positions at the top or centre of the system. As
what health systems are and how they operate
n important are the patients, citizens, front line providers
what needs to be done to strengthen health systems
n and managers at the bottom or periphery of the system.
in order to improve performance in terms of health Their actions and interactions represent the practices that
gain and wider social value are ultimately experienced not only as health policy but
how to influence policy agendas to embrace actions
n also as the health system (see, for example, Ssengooba
to strengthen health systems
et al., 2007; Walker & Gilson, 2004), and through which
how to develop and implement such actions in ways
n health improvement and wider social value is achieved.
that enhance their chances of achieving performance
gains HPSR may also be undertaken through studies imple-
mented at national or sub-national levels, and through
The scope of HPSR covers work implemented across
studies implemented in multiple countries.
the various elements and dimensions of the health
system (see Figures 1 and 3). An HPSR study may involve The variety of issues that are relevant for HPSR is shown
considering one or more of the following aspects: in Boxes 1 and 2.
the wider arena in which policy is made (macro level
n
analysis);
Box 1: Suggested topics for health systems research
Financial and human resources:

• Community-based financing and national health insurance
• Human resources for health at the district level and below
• Human resources for health at the national level
Organization and delivery of health services:
• Community involvement
• Equitable, effective, and efficient health care
• Approaches to the organization of health services
• Drug and diagnostic policies
Governance, stewardship, and knowledge management:
• Governance and accountability
• Health information systems
• Priority-setting and evidence-informed policy-making
• Effective approaches for inter-sectoral engagement in health
Global influences:
• Effects of global initiatives and policies (including trade, donors, and international agencies) on health systems
Source: Sanders Haines,

Box 2: Topics addressed by existing empirical HPSR studies
HPSR has been undertaken to investigate a wide range of health policy and system issues, such as:
• describing and assessing particular system building blocks (such as decentralization; health financing);
• describing particular experiences of policy change in particular settings;
• explaining how multinational corporations influence transnational and national policies
(for example tobacco companies);
• explaining the influences over aspects of particular policy actors’ decision-making
(such as health-seeking behaviour studies; health worker motivation studies);
• assessing whether new interventions generate performance gains, and of what level
(conditional grant assessments), as well as the cost-effectiveness of alternative interventions;
• understanding stakeholder power and positions around specific new policies or actions, and assessing
the likely implications for the acceptability of new policies or interventions;
• understanding particular experiences of policy implementation, or explaining variations between settings
in the experience of implementing a particular policy;
• explaining overall health system performance impacts and their variation across health systems
(for example cross-national analysis of catastrophic health expenditure levels).
What HPSR is not

Falling outside the definition of HPSR are more tradi- upstream, from particular health conditions, services or
tional medical and public health research issues, such as: programmes to consider their health system and policy
basic scientific research on new pharmaceutical
n context. This context has critical influence over sustained
products or medical technologies; action to tackle particular health conditions and sus-
assessing the clinical efficacy and effectiveness of
n tained delivery of particular services or programmes
particular treatments or technologies; (Travis et al., 2004).
the measurement of population health profiles and
n HPSR, therefore, addresses the full range of health
patterns.
system building blocks rather than being primarily
concerned with aspects of the service delivery block.
The distinction between HPSR has particular concern for the horizontal dimen-
HPSR and service delivery sions of the health system (for example, planning,
disease programme research management, organizational functioning). Nonetheless,
it may involve research within certain programme areas
HPSR is concerned with the system-level factors and (which are often called the vertical elements of the
forces that cut across actions dedicated to tackling system) in order to understand the systemic challenges
particular health problems, as well as those that of responding to different health conditions and of sus-
underpin and shape the performance of health pro- taining different types of health programmes. In HPSR,
grammes that target specific health conditions. From a the health problems or programmes of focus are selected
service delivery perspective this includes, for example, because they have system-wide demands (as with anti-
assessing new organizational models of care or new retroviral therapy for HIV/AIDS) or because they serve as
roles for different types of health-care providers. tracers for understanding and/or influencing health policy
However, much HPSR broadens the focus, or goes and system dynamics (Alliance for Health Policy and
Systems Research, 2007).
Fuzzy boundaries and organisational behaviour.
http: www.implementationscience.com about ,
A range of terms are used by different groups of people accessed anuary
to address slightly different aspects of HPSR. In contrast, HPSR adopts a broader approach to
The older term ‘health services research’ is perhaps more implementation research that is rooted in the decades-
commonly used in higher income countries, and its old and rich body of policy implementation theory (Hill &
starting point is the service delivery function of health Hupe, 2009), among other research traditions. It sees
systems, sometimes in relation to other functions. Health research on implementation as being both central to the
services research may, for example, study the patient study of governance in health systems and focused on
–provider relationship and interventions to improve understanding how change is driven or shaped. Asking
uptake of clinical guidelines by health-care practitioners. ‘What actually happens and why?’ rather than ‘Why is
The term ‘health policy and systems research’ was intro- there an implementation gap?’, this approach sees
duced by the Alliance for Health Policy and Systems implementation as an organizational, social and political
Research to cover a broader terrain of work, and process to be enabled rather than as a centrally con-
although the Alliance has particular concern for work in trolled and almost mechanical process. It considers,
low- and middle-income countries, the term HPSR is therefore, the practices of management and communi-
now being more widely embraced. HPSR may start from cation that support the scale-up of a new idea or inter-
any of the health system building blocks, and includes vention within a health system, rather than focusing
concern for the policy process as well as global influ- more exclusively on, for example, new ways of shaping
ences. Other areas of research related to HPSR include provider behaviours. It also acknowledges the practices
implementation and operational research – and there is of power or relationships of trust that shape imple-
some degree of overlap between these particular forms mentation experience.
of research and management activities. Rather than As HPSR draws insights from a range of disciplines, a
trying to establish explicit and clear boundaries between second set of fuzzy boundaries are those between more
these different areas of work it might be better to see specialist disciplinary work and HPSR. For example, most
most of them as, essentially, sets of overlapping areas epidemiological work would not fall within HPSR, but
with fuzzy boundaries. those analyses which shed light on health system
However, the differences between HPSR and the performance and change over time are relevant (see
emerging field of implementation science illuminate Masanja et al., 2008, in Part 4 of this Reader). Similarly,
some key differences in perspective (Sheikh et al., 2011). the anthropological work that sheds light on health
system functioning and performance includes, for
As currently discussed in international health debates,
example, research focused on relationships among health
implementation science can be seen to be primarily
system actors (George, 2009, later) or on policy itself
concerned with improving the delivery of particular
(Behague & Storeng, 2008, see Part 4). More classical
services or treatment interventions that have already
anthropological work, perhaps addressing lay perspec-
been proven to be clinically effective. For example:
tives around particular health programmes, is less directly
mplementation research is the scientific study of relevant to HPSR. Political science and sociology also
methods to promote the systematic uptake of clinical have much to offer HPSR (for example, Shiffman et al.,
research findings and other evidence-based practices 2004 and Murray & Elston, 2005, see Part 4), although
into routine practice, and hence to improve the not all work from these disciplinary perspectives would
quality and effectiveness of health care. t includes fall squarely into the field of HPSR.
the study of influences on health-care professional

Finally, whilst health economics is a central discipline of Ultimately, by definition, studies falling within the field of
HPSR, the analyses most centrally falling within HPSR HPSR must address health policy and systems issues, as
include work focussed on financing (for example defined here, and offer insights that have fairly clear
O’Donnell et al., 2007, see Part 4), and human resource policy relevance.
issues (for example Blaauw et al., 2010, see Part 4),
rather than, for example, cost-effectiveness analysis of
specific disease technologies.
7. Understanding the nature of social and

political reality
his section outlines different ways of understanding researchers views of the world they investigate, views
which influence the type of research they choose to do. iscussion of these issues is a common feature of wider
social and development research but is more rare in health research.
All research is influenced by the researcher’s understanding of what reality and knowledge mean.
As a researcher, it is always important to acknowledge the way you understand the world – as this influences
the types of question you ask, and the types of research strategy you choose.
Positivism, relativism and critical realism are terms describing three key ways of looking at the world and
finding out about it.
Because HPSR draws on a range of disciplinary perspectives it embraces a wider range of understandings of
social and political reality than most health research. This also influences the understandings of causality,
generalizability and learning accepted within the field. More specifically, HPSR seeks to investigate complex
causality; draws on comparative analysis to generate conclusions that are relevant in various settings; and
takes a fairly engaged approach to promoting learning from research.
A fundamental difference between HPSR and wider questions and approaches (see Table 2). The differences
health research lies in their different understandings of between these paradigms underlie some of the common
the nature of reality, what is out there to know, and how criticisms of HPSR, as well as the different research
to gather knowledge about that ‘reality’. Biomedical and strategies used compared to biomedical and epidemio-
clinical research, and some epidemiological and ecological research (Gilson et al., 2011). The following brief
nomics research, is founded on the same positivist overview of these differences draws particularly on Grix,
understandings as natural and physical sciences. 2004; Harrison, 2001; Robson, 2002.
However, unlike the dominant health research traditions

HPSR draws strongly on social science perspectives,
embracing not only the critical realist but also the relativist
paradigm of knowledge – and related sets of research
Table 2 Key elements of knowledge paradigms as applied in HPSR
Knowledge
Relativism
paradigm Positivism Critical Realism (interpretivism /
social constructionism)
Types of questions Is the policy or intervention What works for whom under How do actors experience and
addressed (cost)-effective? which conditions? understand different types of
interventions or policies?
What are the social processes,
including power relations,
influencing actors’ understan-
dings and experiences?
Related disciplinary Epidemiology Policy analysis Anthropology

perspectives Welfare economics Organizational studies Sociology
Political science Political science
(rational choice theory) (sociological institutionalism)
Key research Deductive: Hypothesis driven Deductive and inductive Inductive

approaches and Measurement through (theory testing and building) (maybe theory building and/or
methods surveys, use of archival and Multiple data collection testing)
other data records methods including review Multiple data collection methods
Statistical analysis of documents, range of including in-depth interviewing
interviewing methods, (individuals and groups),
Qualitative data collected observation documentary review but also
through, for example, semi- participant observation or life
structured interviews and histories, for example.
interviewing procedures
HPSR articles Björkman & Svensson, 2009 Marchal, Dedzo & Kegels, 2010 Riewpaiboon et al., 2005
that illustrate Shiffman, 2009
the paradigm Sheikh & Porter, 2010
(see Part 4)
Positivism
Positivist research, such as biomedical or epidemiological evidence. HPSR rooted in this paradigm has a central
research, starts from the position that the phenomena or focus on identifying what interventions work best and
issues of investigation exist independently of how they have most impact.
are understood and seen by people. Research in this
tradition works with the understanding that these
Relativism
phenomena comprise a set of facts that can be observed
and measured by the researcher, without disturbing The social sciences, however, encompass the under-
them, and that there are patterns and regularities within standing that the phenomena being investigated (such
them, causes and consequences, that can be identified as health policies and systems) are produced through
through empirical research. Indeed, the central task of interaction among social actors. Such phenomena do
such research is considered to be to detect the laws of not, therefore, exist independently of these actors but
cause and effect that operate in reality and that remain are, in essence, constructed through the way the actors
‘true’ in different contexts and times, by describing them interpret or make meaning of their experience, and these
and testing hypotheses (or predictions) against the interpretations change over time.

From this perspective, facts are not clearly distinct from
the values people hold, and searching for laws of cause
HPSR perspectives on
and effect is an almost irrelevant task. Instead, research causality, generalizability
grounded in this tradition focuses on people’s intentions, and learning
beliefs, values, reasons and how they make meaning.
The broader understandings of knowledge and social
It acknowledges that the researcher also constructs
reality incorporated within HPSR, as compared to
knowledge through how they interpret what they hear
positivist research, underpin its recognition of the socio-
and observe. The central task of HPSR in this tradition
political and ideological influences over health policies
is, thus, not to explain but rather to understand the
and health systems. It also leads to important differences
meanings given by actors to social phenomena, including
in perspectives on causality, generalizability and learning
the language used to construct reality.
between these research fields.
Critical realism Causality

A third perspective, critical realism, can be seen as placed HPSR embraces complex causality – the understanding
somewhere between the other two perspectives. Like that an effect is not linked by a linear and predictable
positivism, this perspective understands social reality to path to a cause, but that there are multiple-interacting
exist independently of social actors, although it accepts causes generating a set of often unpredictable effects.
that actors’ interpretations of that reality have influence Such complex causality can be seen as a result of the
over the nature of social change. The pre-existing influence of actors and their interpretations over how
structures and processes of society therefore affect, and problems are defined, which form interventions or
are affected by, actors; and human action is influenced policies take in implementation, how health systems
by a range of individual, group, organizational and work and how interventions or policies play out through
societal processes and structures. health systems (Pawson & Tilley, 1997).
Like positivists, critical realists seek to identify the causal Complex causality also results from the open nature
mechanisms underpinning social phenomena (such as of health systems – there are multiple, interacting
health policies and systems), but they also adopt influences over them and embedded in them. Therefore,
an interpretive understanding. In other words, they do interventions and policies often do not generate the
not accept that cause and effect mechanisms hold same impacts over time and in different places (de
across context and times, but believe that there are Savigny & Adam, 2009). In addition, research takes place
a range of mechanisms mediating between cause within the health system, even as it changes in ways that
and effect, including those linked to actors and to may have nothing to do with the particular focus of
contexts. inquiry (Robson, 2002). HPSR must therefore adopt
For critical realists, therefore, the task of research and research strategies that allow investigation of complex
evaluation is to generate theories that explain the social causality. In particular, systems thinking is increasingly
world and, in particular, to identify the mechanisms that seen to offer insights and perspectives of relevance
explain the outcomes of interventions. The dominant to HPSR (Atun & Menabde, 2008; de Savigny & Adam,
HPSR question from this perspective is ‘What works for 2009).
whom in which conditions?’ (Pawson & Tilley, 1997).
eneralization Knowledge generation and learning
HPSR recognizes various approaches to generalization. Finally, HPSR embraces different understandings of
Research from the positivist tradition looks for conclu- knowledge generation and learning to that of biomedical
sions that have external validity and that can be stati- and epidemiological work. Research in the positivist
stically generalized beyond the initial study setting tradition tends to see learning as an act of engineering –
and population. In evaluation work, randomized control the transfer of knowledge from one setting to another
trials have become the gold standard study design – whereas the relativist perspective of social science sees
because they allow such generalizations. However, HPSR learning as an integral part of the process of policy
also embraces analytic or theoretical generalizability, development and implementation (Freeman, 2006). As
as commonly applied in case study research. General Rose (2005), for example, has argued, policy lessons are
insights derived from one or a few experiences, or cases, not just direct copies of interventions implemented in
through a careful process of analysis, are judged to hold one setting. Instead, they are ideas drawn from obser-
a sufficient degree of universality to be projected to vations of interventions in other settings, observations
other settings (Robson, 2002). that are abstracted, generalized and then and re-
contextualized in a new setting.
The process of analysis involves the development of
conclusions from detailed findings about context, In the positivist tradition, the researcher’s job is to
processes and outcomes in one or more settings; con- identify the causal mechanisms that can be transplanted
clusions that are lifted to a sufficient level of abstraction from one setting to another. In the relativist tradition the
or generality to have resonance in a different context. researcher’s job is to assist in the process of unders-
Comparisons across similar cases also allow such middle tanding and promoting change – including through
range theory (”ideas about how the world works, understanding how social actors interpret and make
comprising concepts derived from analysis and ideas meaning of their realities and through helping policy
about how these concepts are linked together”, Gilson et actors to negotiate mutually acceptable solutions to
al., 2011:2) to be tested and revised in repeated cycles problems (Harrison, 2001).
of theory-building and theory-testing.
In comparative case study analysis, generalization is not

grounded in the representativeness of the population
sampled but instead in a process of abstracting from the
specifics of one case to ideas that encompass several
cases. There is, therefore, growing interest in comparative
case study analysis among health policy and systems
researchers interested in explanation (Gilson & Raphaely,
2008; Marchal, Dedzo and Kegels, 2010). Nonetheless, it
should also be noted that HPSR encompasses research
that does not seek to generalize (for example about
actors and their meaning-making) but works instead with
the particular and specific, aiming to illuminate and
understand these experiences (for example, George,
2009; Sheikh & Porter, 2010).

References Commission on the Social Determinants of Health
(2008). Closing the gap in a generation: health equity
Alliance for Health Policy and Systems Research (AHPSR) through action on the social determinants of health.
(2007). hat is Health Policy and Systems Research and Geneva, World Health Organization.
why does it matter AHPSR Briefing Note 1. Geneva, de Savigny D, Adam T, eds (2009). Systems thinking for
World Health Organization. (http://www.who.int/alliance- health systems strengthening. Geneva, World Health
hpsr/en/, accessed 22 August 2011). Organization (http://www.who.int/alliance-hpsr/resources
Alliance for Health Policy and Systems Research (2011). /9789241563895/en/index.html, accessed 10 July 2010).
hat is HPSR verview. Geneva, World Health Organi- Freedman L et al. (2005). Who’s got the power? rans-
zation (http://www.who.int/alliance-hpsr/about/hpsr/ forming health systems for women and children.UN
en/index.html, accessed 18 May 2011). Millennium Project Task Force on Child Health and Maternal
Atun R, Menabde N (2008). Health systems and Health. London, Earthscan (http://www.unmillennium
systems thinking. In: Coker R, Atun R, McKee M, eds. project.org/documents/TF4Childandmaternalhealth.pdf,
Health systems and the challenge of communicable accessed 15 September 2011).
diseases: e periences from urope and Latin merica. Freeman R (2006). Learning in public policy. In: Moran
Maidenhead, Open University Press. (European Observa- M, Rein M, Goodin R, eds. he ford handbook of
tory on Health Systems and Policies Series):121–140. public policy. Oxford, Oxford University Press:367–388.
Behague DT, Storeng KT (2008). Collapsing the vertical– Frenk J (1994). Dimensions of health system reform.
horizontal divide: an ethnographic study of evidence- Health Policy, 27:19–34.
based policy-making in maternal health. merican
ournal of Public Health, 98(4):644–649. Frenk J (2010). The global health system: strengthening
national health systems as the next step for global
Björkman M, Svensson J (2009). Power to the people: progress. PLoS Medicine, 7(1):1–3.
evidence from a randomized field experiment on
community-based monitoring in Uganda. he uarterly Fulop N et al. eds (2001). Issues in studying the organi-
ournal of conomics, 124(2):735–769. sation and delivery of health services. In: Fulop N et al.,
eds. Studying the organisation and delivery of health
Blaauw D et al. (2010). Policy interventions that attract services: research methods. London, Routledge:1–23.
nurses to rural areas: a multicountry discrete choice
experiment. ulletin of the orld Health rgani ation, George A. (2009). By papers and pens, you can only
88:350–356. do so much: views about accountability and human
resource management from Indian government health
Bloom, G, Standing H, Lloyd R (2008). Markets, informa- administrators and workers. nternational ournal of
tion asymmetry and health care: Towards new social Health Planning and Management, 24(3):205-224.
contracts. Social Science Medicine, 66(10):2076–2087.
Gilson L (2003). Trust and the development of health
Buse K (2008). Addressing the theoretical, practical and care as a social institution. Social Science Medicine,
ethical challenges inherent in prospective policy analysis. 56(7):1453–1468.
Health Policy and Planning 23(5):351–360.
Gilson L (2012). Health systems and institutions. In:
Buse K, Mays N, Walt G (2005). Making health policy. Smith R, Hanson K, eds. Health systems in low- and
Milton Keynes, Open University Press. middle-income countries: an economic and policy
perspective. Oxford, Oxford University Press. pp21-45.
Gilson L, Raphaely N (2008). The terrain of health policy Masanja H et al. (2008). Child survival gains in Tanzania:
analysis in low- and middle-income countries: a review analysis of data from demographic and health surveys.
of published literature 1994–2007. Health Policy and Lancet, 371:1276–1283.
Planning, 23(5):294–307.
Mills A (2012). Health policy and systems research:
Gilson L et al. (2008). Challenging inequity through defining the terrain; identifying the methods. Health
health systems. Final report of the Knowledge Net- Policy and Planning, 27(1): 1-7.
work on Health Systems, World Health Organization
Commission on the Social Determinants of Health. Murray SF, Elston MA (2005). The promotion of private
Johannesburg, Centre for Health Policy, E UINET, London health insurance and its implications for the social orga-
School of Hygiene and Tropical Medicine. nisation of health care: a case study of private sector
obstetric practice in Chile. Sociology of Health and llness
Gilson L et al. (2011). Building the field of Health Policy
27(6):701–721.
and Systems Research: Social Science Matters. PLoS
Medicine (8): e1001079. O’Donnell O et al. (2007). The incidence of public spen-
Grix J (2004). he foundations of research. Houndsmill, ding on healthcare: comparative evidence from Asia.
Basingstoke, Palgrave Macmillan. orld ank conomic Review, 21:93–123.
Harrison S (2001). Policy analysis. In: Fulop et al., eds. Overseas Development Institute (ODI) (2007). How can
Studying the organisation and delivery of health services: analysis of power and process in policy-making improve
research methods. London, Routledge:90–106. health outcomes Moving the agenda forward. ODI
briefing paper 25 (http://www.odi.org.uk/resources/
Hill M, Hupe P (2009). mplementing public policy, 2nd
download/343.pdf, accessed 22 August 2011).
ed. London, Sage Publications.
Hudson J, Lowe S (2004). nderstanding the policy Pawson R, Tilley N (1997). Realistic Evaluation. London,
process: analysing welfare policy and practice. Bristol, Sage Publications.
The Policy Press.
Riewpaiboon W et al. (2005). Private obstetric practice in
Lipsky M (1980). Street-level bureaucracy: dilemmas of a public hospital: mythical trust in obstetric care. Social
the individual in public services. New York, Russell Sage Science Medicine, 61(7):1408–1417.
Foundation.
Roberts M et al. (2008). etting health reform right:
Mackintosh, M. (2001). Do health care systems contri- guide to improving performance and equity. New York,
bute to inequalities? In: Leon D, Walt G, eds. Poverty, Oxford University Press.
inequality and health: an international perspective.
Oxford, Oxford University Press:175–193. Robson C (2002). Real world research: a resource for
social scientists and practitioner-researchers, 2nd ed.
Mackintosh M, Koivusalo M (2005). Health systems
Oxford, Blackwell Publishing.
and commercialization: in search of good sense. In:
Mackintosh M, Koivusalo M, eds. Commerciali ation of Rose R (2005). Learning from comparative public policy:
health care. Houndsmill, Basingstoke, Palgrave a practical guide. Abingdon, Routledge.
Macmillan:3–21.
Sanders D, Haines A (2006). Implementation research is
Marchal B, Dedzo M, Kegels G (2010). A realist
needed to achieve international health goals. PLoS
evaluation of the management of a well-performing
Medicine, 3(6): e186.
regional hospital in Ghana. MC Health Services
Research, 10:24.

Sheikh K, Porter J (2010). Discursive gaps in the Walt G, Gilson L (1994). Reforming the health sector in
implementation of public health policy guidelines developing countries: the central role of policy analysis.
in India: The case of HIV testing. Social Science Health Policy and Planning, 9(4):353–370.
Medicine, 71(11):2005–2013.
World Health Organization (2007). verybody s business:
Sheikh K et al. (2011) Building the field of health policy Strengthening health systems to improve health out-
and systems research: framing the questions. PLoS comes: H s framework for action. Geneva, World Health
Medicine, 8(8):e1001073. Organization (http://www.who.int/entity/healthsystems/
strategy/everybodys_business.pdf, accessed 16 September
Shiffman J (2009). A social explanation for the rise and
2011).
fall of global health issues. ulletin of the orld Health
rgani ation. 87(8):608–613.
Shiffman J et al. (2004). The emergence of political

priority for safe motherhood in Honduras. Health Policy
and Planning, 19(6): 380–390.
Smith R, Hanson K (2011). What is a health system.

In: Smith R, Hanson K, eds. Health systems in low- and
middle-income countries: an economic and policy
perspective. Oxford, Oxford University Press: Chapter 1.
Ssengooba F et al. (2007). Health sector reforms and

human resources for health in Uganda and Bangladesh:
mechanisms of effect. Human Resources for Health, 5:3.
Thomas A (1998). Introduction. In: Thomas A, Chataway J,

Wuyts M, eds. Finding out fast: investigative skills for
policy and development. London, Sage Publications:1–18.
Travis P et al. (2004). Overcoming health-systems con-

straints to achieve the Millennium Development Goals.
Lancet, 364:900–06.
Van Damme W et al. (2010). How can disease control

programmes contribute to health systems strengthening
in Sub-Saharan Africa? Studies in Health Services Organi-
sation & Policy. Working Paper Series. Working Paper
no1. Antwerp, Institute of Tropical Medicine
(http://www.itg.be/itg/GeneralSite/default.aspx?WPID
756&L e&miid, accessed 22 August 2011).
Walker L, Gilson L (2004). We are bitter but we are

satisfied: nurses as street level bureaucrats in South
Africa. Social Science Medicine, 59(6):1251–1261.
Part
o g ealth ol y a d
y te Re ear h ey te
the ro e
Lucy Gilson
ni e sity o e o n out ic nd
London c ool o y iene nd o ic l edicine nited in do o Ge t it in
nd o t e n el nd
This part of the Reader outlines the four critical steps in due to the influence of other health researchers or specific
developing a primary Health Policy and Systems Research interest groups. National research priority-setting pro-
(HPSR) study that should be addressed by all researchers: cesses are, therefore, important as a means of ensuring
1. identify the research focus and questions dialogue and engagement between researchers and
2. design the study health policy-makers and managers. The aims of such
engagement are to turn health system and policy pro-
3. ensure research quality and rigour
blems into researchable questions, identify priorities
4. apply ethical principles.
among them and, ultimately, support the uptake of
When assessing the quality of empirical HPSR work, research findings. Greater national funding for HPSR may
it is important to consider all steps, not only Step 3. be a further consequence (Green & Bennett, 2007).
Examples of international research priority-setting

Step 1: Identify the processes include those convened by the Alliance for
research focus and

Health Policy and Systems Research in 2007-2008 which
identified priority topics for research in human resources,
questions financing and the role of the non-state sector (see
Table 3). At national level, the Essential National Health
The process of developing an HPSR study begins with Research approach has provided a framework for
identifying the topic of focus – the issue or problem you priority-setting that has been applied in various countries
want to investigate – and the related questions. There (Green & Bennett, 2007; Alliance for Health Policy
are two main reasons for this: and Systems Research, 2009). See also the work of
1. HPSR is defined by the topics and questions it the Council on Health Research for Development at
addresses rather than the disciplinary perspective http://www.cohred.org.
or the particular approach to data collection and Beyond networking with policy actors and other
analysis it adopts. researchers, identifying an HPSR topic and related
2. HPSR always aims to be policy relevant and to inform research questions should involve:
the decisions taken by those who influence how
thinking creatively, for example to identify new areas
n
health systems evolve and perform – the policy of work or different approaches to an investigation;
actors, from household to global levels. (Note that
exploring theory and conceptual understandings
n
policy relevance is a key criterion used to assess the relevant to HPSR generally, and the topic of focus;
ethical value of HPSR research, Henning, 2004). conducting a literature search to identify relevant
n
As policy relevance is always important to HPSR, those publications and research studies.
working in the field have paid increasing attention to the Finally, pragmatism is important when identifying
process of setting research priorities. A particular concern a research question. The research needs to be feasible,
has been the influence of global actors (conventional for example, the scope and size of the study must be
multilateral and bilateral research funders as well as considered relative to the resources and time available
global public–private initiatives) over priority-setting (Robson, 2002; Varkevisser, Pathmanathan & Brownlee,
within low- and middle-income countries. The priorities of 2003).
these global actors have often emphasized upstream
health research or commodity procurement, rather than
systems strengthening questions and initiatives. Even
amongst national research communities, HPSR questions
may receive less attention than other research questions
Part 2 oing Health Policy and Systems Research: ey ste s in the rocess 43
Table 3 Priority research questions in three health policy and systems areas, results
of international priority-setting processes (Source: Alliance for Health Policy and
Systems Research, 2009)
Human resources for health Health system financing on-state sector
st To what extent do financial and How do we develop and implement How can the government create a
non-financial incentives work in universal financial protection? better environment to foster non-
attracting and retaining qualified state providers in the achievement
health workers to under-serviced of health systems outcomes?
areas?
nd What is the impact of dual practice What are the pros and cons of the What is the quality and/or coverage
(i.e. practice by a single health care different ways of identifying the of health care services provided by
worker in both the public and poor? the non-state sector for the poor?
the private sectors) and multiple
employment?
Are regulations on dual practice
required, and if so, how should they
be designed and implemented?
rd How can financial and non-financial To what extent do health benefits What types of regulation can
incentives be used to optimize reach the poor? improve health systems outcomes,
efficiency and quality of health and under what conditions?
care?
4th What is the optimal mix of financial, What are the pros and cons of How best to capture data and
regulatory and non-financial policies implementing demand-side trends about private sector
to improve distribution and subsidies? providers on a routine basis?
retention of health workers?
th What are the extent and effects of What is the equity impact of social What are the costs and affordability
the out-migration of health workers health insurance and how can it be of the non-state sector goods and
and what can be done to mitigate improved? services relative to the state sector?
problems of out-migration? And to whom?
etworking and creative

thinking
Engaging with policy actors and other researchers helps Networking can also help to stimulate creative thinking.
to ensure that the topic and research questions are In addition, exploring conceptual understandings and
policy relevant. Both groups, through their experience in theory can highlight new areas of work rarely considered
different settings, will have insights into the challenges in the past, or new ways of understanding how to
and opportunities that face health systems. The types of investigate a topic on which there is already some
questions that may interest national policy-makers are research.
shown in Box 3. Such questions focus on both policy
content and policy processes.
Box : Broad research uestions of interest to national
policy ma ers
Policy formulation
• What is the nature and extent of problem ?
• What happened before in response to problem , and what were the consequences? What were
the unexpected consequences?
• What are cost-effective responses to the problem?
• How long will it be before the impacts of response are seen? How can popular and political support
be sustained until the impacts are seen?
Policy implementation
• What happens in practice when policy is implemented, and why?
• Do policy implementors have the same understanding of the problem that the policy aims to address, and
the same policy goals, as the policy-makers? If not, how does that difference affect policy implementation?
• Is the organizational response adequate/sustained?
Policy evaluation
• Were the policy, or programme, objectives met?
• What were the unexpected outcomes?
• Did the policy objectives remain the same over time?
• Did the condition being addressed change over time?
• Was the programme or policy? implemented effectively?
Source: Rist
Literature search
It is important to find out what relevant research has or developing new ideas on topics that have already
already been conducted in order to avoid unnecessary been considered.
duplication and to build on existing research.
The growth of interest in systematic reviews and
Although researchers can draw on their own knowledge syntheses of existing research reflects, in part, the
of a particular setting, it is always important that they concern that existing primary research is frequently
conduct more formal literature reviews of research not used as a basis for changing policy and practice,
previously conducted in other settings and not only in or for developing new research work. An important
the area with which they are familiar. While there is value resource for health policy and systems researchers is,
in replication studies (deliberately replicating work therefore, the Health Systems Evidence web site at
previously conducted in one setting in a new setting to http://www.healthsystemsevidence.org .
generate new insights, for example (Robson, 2002)), This is a continuously updated and searchable repository
the duplication of a research study simply because of of syntheses of research evidence about governance,
limited knowledge about existing research is a waste of financial and delivery arrangements within health
resources and so unethical (Emanuel et al., 2004). systems, and about implementation strategies that
can support change in health systems.
New studies must always offer value, that is they must
build on existing work, for example by addressing a
question not previously considered in a particular setting,
Key challenges
Two key challenges related to identifying appropriate conducted elsewhere. Similarly, managers of a
research questions are discussed below. particular health programme, be it HIV/AIDS, nutri-
1. Framing policy relevant and valuable HPSR questions tion or school health, tend to be most interested in
through networking with research users. research about how to strengthen their particular
programme and less interested in the systemic
A challenge of generating new research ideas
support needs across programmes. Yet, as discussed
through networking with policy actors is that the
earlier (see Part 1, Section 6), HPSR focuses on such
types of topics and questions identified as important
systemic needs rather than on programme-specific
will vary between policy actors, depending on their
needs.
roles and responsibilities within the health system
(as illustrated in Box 4). For example, policy actors Therefore, health policy and systems researchers need
working at lower levels of the health system have to think carefully about the fuzzy boundary between
particular operational needs which, while important, HPSR and management (see Part 1, Section 6) and
might limit the wider application of the work if other seek either to support managers to conduct their
policy actors do not see its relevance to them or own operational research, or to identify the wider
if it requires the duplication of research already value of the particular research question.
Box : The HPSR uestions of different health policy

and systems actors
National policy-makers might ask:

• How can we prevent the HIV/AIDS programme from draining resources (time and staff) from other equally
important programmes?
• How can HIV/AIDS resources be used in ways that strengthen other areas of the health system?
• Should antiretrovirals be prescribed only by doctors or is prescription by nurses more cost-effective?
District managers might ask:
• Why are there more patient complaints about facility than others in my district?
• Why are patient waiting times at clinics still very long, although we have already tried to reorganize
services to address the problem?
• How can we develop an integrated HIV/AIDS and tuberculosis service, in line with national policy?
Hospital managers might ask:
• How can we decrease the pharmacy waiting time?
• How can we reduce the average length of stay for chronically ill patients?
• Are ambulatory services available and adequate?
• Are patients coming late for treatment and why?
Patient groups might ask:
• Why do we have to wait so long to get care?
• Why do health workers treat us so rudely?
2. dentifying research questions that are relevant to a challenges. Therefore, the health policy and systems
range of policy actors and that add to the e isting researcher can see the particular programme issue as
knowledge base. a case study of policy implementation.
The challenge for the health policy and systems All these approaches show how research around one
researcher is to identify policy-relevant and valuable programme can represent a tracer for understanding
research questions that not only directly address the and/or influencing health policy and system dynamics,
concerns of the main group of policy actors with as discussed earlier.
whom they seek to work and influence, but also have Overall, compared with research focused on a parti-
relevance to a wider range of actors and add to the cular disease programme, service area or clinical
existing HPSR knowledge base. treatment, HPSR requires the researcher to consider
For example, how can work on reducing a particular the system within which the specific service or
hospital’s pharmacy waiting time have relevance to treatment is nested. This means thinking:
other hospitals or to national managers concerned broad – beyond the disease or treatment of
n
with supporting all hospitals to reduce waiting times? focus;
Similarly, how can research linked to a particular up – above the programme or service to the
n
disease programme be undertaken in ways that offer facility, district, province etc.; and
policy and systems lessons that benefit other about the cross-cutting functions that underlie
n
service and programme delivery – the system
programmes as well? In both cases, it is important to
building blocks and interactions among them
see the specific focus of the research as an entry (Schneider, 2011).
point for considering an issue of relevance to a
broader range of actors and settings.
In terms of pharmacy waiting times, there could be
Identifying the purpose of
value in seeing the work in one hospital as a case the research
study of how to tackle such a problem. The case study
could generate ideas on processes and strategies that In developing research questions that will be policy
can initially be tested in other hospitals. Then, drawing relevant and valuable, it is also important to think about
on several experiences, this can become the basis for the overall purpose of the research, in particular:
compiling general insights into ways to address the What is the research trying to achieve? or Why is it
n
common problem of waiting times. This is an example being done?
of the process of analytic generalization and it To whom will it be useful?
n
provides the basis for the sort of policy learning in How will it be useful?
n
implementation discussed in Part 1, Section 7. How will it add to the existing knowledge base?
n
Another approach would be to see how work in
Thinking about such questions will also inform the
a particular programmatic area offers insights into
design of the research study (see Step 2).
a broader systems’ question of relevance across
programmes. For example, work on task shifting As research questions are developed, four dimensions
within an HIV/AIDS programme offers insights on can be considered:
the types of human resource development and 1. Whole field or specific policy: Will the research seek
management needs that must be addressed in any to focus on the field as a whole, and so expand
new policy initiative that involves an expansion of the knowledge of the nature and functioning of the key
scope of work of lower-level cadres; it also highlights elements of health policy and systems, or will it
the possible challenges to the political feasibility of seek to focus on a particular policy and support its
such an initiative and ways of managing those implementation?
2. Normative/evaluative or descriptive/explanatory can add to our general knowledge of policy
research questions: Will the research address norma- development and implementation (analysis of
tive or evaluative questions (which may involve value policy)? If so, this will generally demand longer
time frames, with a focus on the broader research
judgements) or descriptive or explanatory questions?
questions through which the complex and
Table 4 provides some examples of HPSR questions dynamic trajectories of policy experience, for
across dimensions 1 and 2. example, are more amenable to investigation.
3. Analysis ‘for’ or ‘of’ policy (Parsons, 1995) – where 4. Primary research purpose: Will the research primarily
focused on a particular policy: seek to explore an issue or phenomena in order to
describe it or to explain it? Or will it adopt a more
Will the research aim to support policy implemen-
n
tation in real time (analysis for policy)? (Whether critical stance in generating understanding, perhaps
considering the technical content of the policy or working with other people to bring about change
experience of the actors and processes engaged rather than focusing only on generating knowledge?
in its implementation.) If so, this may demand Although these research purposes often overlap in
shorter time frames and is likely to be focused on
practice, Robson (2002) identifies their different aims,
narrower research questions; or
see Box 5.
Will the research aim to generate, from that
n
policy’s experience, a broader understanding that
Table 4 Examples of HPSR questions (Source: Adapted from Potter and Subrahmanian, 1998)
ormative/evaluative questions Descriptive/e planatory questions
uestions a out Cell 1 Cell 3

the policy itself Should this policy be adopted? Which agencies are stakeholders in this policy,
How does policy impact on health seeking what positions do they take on the policy and
behaviour? why?
Which actor management strategies are How did policy come about?
likely to be most useful in supporting Is there capacity to implement policy ?
implementation of policy x? How do front line providers understand policy ?
uestions a out Cell 2 Cell 4

the field Which type of health system performs best? What are patterns of health seeking behaviour
What are the different approaches to actor and what influences that behaviour?
management that can be considered by those How is the health system organized at present?
seeking to manage policy change? What if a new provider was available, how
would health seeking behaviour change and
how would it affect the performance of the
system overall?
What influences how front line providers under-
stand policies, and how does their understanding
influence their implementation of the policy?
Note: The questions in Cell 1 are asked by those responsible for policy implementation, and essentially demand judgements,
at least some of which are likely to be informed by work addressing the questions proposed in Cell . The questions in Cell
2, meanwhile, address what people should do, and may be informed by the what if’ questions included in Cell . Finally,
questions in Cell encompass the areas of interest in health policy analysis, as outlined earlier: the context, history,
interests and organizations that shape a particular policy.
Box : The purpose of different types of research
ploratory research seeks to:
• find out what is happening, especially in little-understood situations
• generate new insights and ask questions
• assess phenomena in new light
• generate ideas and hypotheses for future research
escriptive research seeks to:
• give an accurate profile of people, events, situations
planatory research seeks to:
• explain a situation or problem, traditionally, but not necessarily, in the form of a causal relationship
(evaluative research)
• explain patterns relating to the phenomenon being researched
• identify and explain relationships between aspects of phenomenon/phenomena
mancipatory research seeks to:
• create opportunities and the will to engage in social action
Critical research: Focuses on the lives and experiences of those traditionally marginalized, analysing how
and why inequities are reflected in power imbalances and examining how research into inequities leads
to political and social action
ction research: seeks improvements in practices, understandings of practice and situations of practice,
and is undertaken by and with those who will take action
Source: Robson,
The purpose of the research should reflect the current Relativists, however, are more likely to conduct forms of
state of knowledge about the topic. Exploratory work, for exploratory, descriptive and/or explanatory research that
example, is important when little is known about a topic aim to deepen our understanding of the phenomena
or when theory suggests a new way of examining of focus and the complex relationships among aspects
and understanding it; but descriptive research requires of those phenomena. Sheikh et al. (2011:5) have speci-
extensive knowledge of the situation in order to identify fically suggested that more HPSR work needs to adopt
what is useful to investigate. However, in empirical work this perspective and address the “fundamental, explora-
researchers often pursue more than one purpose at the tory and explanatory questions” that shape policy and
same time (see Figure 4). provide a platform for further research. For relativists,
emancipatory research also represents an important form
The purpose of the research will also reflect the resear-
of research – analysis for policy.
cher’s understanding of social and political reality (see
Part 1, Section 7). Positivists and critical realists tend to Box 6 shows how the different purposes of research
focus on evaluating causal relationships, based on translate into different basic forms of research questions.
particular forms of descriptive work. For them, therefore,
Finally, across these different research purposes, research
explanatory questions are the same, more or less, as the
might address one or more of the different levels of the
evaluative questions outlined in Table 4, Cell 1; perhaps
system (from micro, meso or macro level) and work with
also entailing forms of descriptive work and preceded by
different (conceptual) units of analysis such as individual
exploratory pilot studies, or accompanied by exploratory
behaviour, patient–provider relationships, the primary
work to support explanation.
health care system, the district hospital, etc.
Figure 4 ultiple research purposes
A to art atory re ear h
Exploratory: Descriptive:
hat ho, what, where,
new insights how many, how much
Explanatory (Evaluative):
Impact
hy and how
Box : in s between purpose and broad forms

of research uestions
Exploratory/descriptive questions
hat or how many much , or who or where questions
• What is the experience of patients with new programme x?
• What is the experience of health workers in training programme x?
• What is the understanding of patient groups or health workers about a problem or a new programme?
• To what extent are family members involved in the programme?
• Who is exposed to condition x or health risk y?
Explanatory questions
valuation questions
• Does programme x lead to reduced health problems from the condition addressed?
• Is programme x more effective than programme y in treating this condition?
• For which group of patients is programme x most effective?
How and why questions:
• How does programme x generate these impacts?
• Why is programme x more effective than programme y?
• Why do health workers act unexpectedly when implementing the programme?
• How do policy actors’ values and beliefs influence their decision-making practices?
• Who supports and opposes new policy x, and why and how?
Taking account of health economics to understand the impacts achieved
n
by a particular form of community accountability; and
multidisciplinarity an historical perspective to track the changing roles
n
of international organizations within global health
Within HPSR, different disciplinary perspectives generate policy.
different research questions on the same topic and
so generate varied policy-relevant insights on the issue
of focus. Therefore, on the one hand, it is important
Finalizing research
to consider the disciplinary perspective that you as a questions
researcher bring to the topic and the type of research
Ultimately, good research questions (Robson, 2002), i.e.
questions you are likely to consider. On the other hand, it
those that will drive valuable and sound research, are:
may be useful to think about how to draw on other
disciplinary perspectives that address the same topic. clear – unambiguous and easily understood;
n
specific – sufficiently specific to be clear about what
n
Work on human resources, for example, may draw on constitutes an answer;
economics and sociological perspectives to understand
answerable – clearly indicate what type of data are
n
motivation; alternatively political science or organiza- needed to answer the question and how the data
tional management perspectives may be applied to will be collected;
understand the decision-making of front line providers; interconnected – a set of questions are related in a
n
or the work may draw on clinical insights to understand meaningful way and form a coherent whole;
skills needs. All have policy relevance. substantively relevant – worthwhile, non-trivial
n
questions, worthy of the effort to be expanded in the
Therefore, Part 4 of this Reader includes papers that
research.
address particular health system functions, or building
blocks, from different disciplinary perspectives (see Part 4:
Table 8). Financing issues, for example, are examined
using:
policy analysis and sociological perspectives in order
n
to understand what influences why and how parti-
cular financing policies are prioritised, developed and
implemented;
the health economics lens in order to understand
n
what cost burdens households experience in acces-
sing care and how they cope with these costs, and
what is the impact on health of community-based
health insurance.
The papers addressing leadership and governance issues

draw, moreover, on:
policy analysis to understand the influences over
n
various experiences of policy change;
anthropology to generate in-depth insights about
n
decentralisation experience and explore global
discourses around maternal health care provision;
management sciences to understand the use of
n
information in district decision-making;
Step 2: Design the study
Once you have the research question/s, the next step is The research purpose and question/s shape the research
to develop the overarching design of the study: to turn strategy. Table 5 provides examples of the different over-
the questions into a project. The overarching study design arching designs that are relevant for different purposes
is not just a set of data collection methods. The design is across the dominant paradigms of knowledge.
comprised of the:
purpose of the study (see Step 1)
n
particular questions to be addressed (see Step 1)
n
strategy for data collection and analysis
n
sampling strategy
n
theory to be used within the study (Robson, 2002).
n
Table 5 A summary of broad study designs (Source: Adapted from Klopper, 2008;
Potter and Subrahmanian, 1998; Yin, 2009.)
Paradigm of Research strategy

Purpose Descriptive/e planatory questions
nowledge Collection of new data Analysis of existing data
Positivist Explanatory Experimental and quasi-experimental Simple and multiple-variable

design including, for example, before modelling
and after studies
Descriptive Survey designs: questionnaires, interviews Secondary data analysis (census

and indirect observation; data, record data)
Repeated surveys to allow trend analysis Quantitative content analysis
over time (newspaper reports, speeches, etc.)
Exploratory Survey designs (pilot studies)
Relativist Explanatory Case study (theory building, longitudinal) Qualitative content analysis
Grounded theory (theory building) Discourse analysis
Historical analysis
Descriptive Case study

Ethnographic designs with the focus
on unstructured direct and indirect
observations, for example narrative
inquiry, critical ethnography
Exploratory Field designs or ethnographic designs with

the emphasis on the use of informants,
for example autho-ethnography,
autobiography, life histories
Case study (such as generating
categorizations)
Qualitative interviews and panels
Research strategies can also be grouped into two main knowledge paradigms to which they are mostly linked.
sets: fixed designs that are established before data It also highlights examples of common data collection
collection and flexible designs that evolve during the study methods, key principles of sampling and the primary
(Robson, 2002). Table 6 summarizes the key chara- characteristics of analysis. Note that within either strategy
cteristics and forms of these two sets of strategies, set, multiple methods may be used in which qualitative
and links them both to the standard forms of research and quantitative data collection approaches are combined.
questions for which they are appropriate and the
Table 6 Key features of fixed and flexible research strategies (Source: adapted from Robson, 2002)
i ed strategy le i le strategy
Characteristics Calls for tight pre-specification before data Design evolves during data collection
collection Data often non-numerical
Data generally numbers Often called qualitative
Often called quantitative Quantitative data may also be collected
Rarely collect qualitative data (multi-method study)
Dominant Positivist Critical realist

nowledge Interpretivist/Social constructivist
paradigm
verarching Primary data collection methods Primary data collection methods

study design Experimental Case study
types Quasi-experimental Grounded Theory
Non-experimental (for example cross- Ethnography
sectional, before and after studies, trend Life histories
analyses) Phenomenological research (qualitative interviewing)
Secondary data analysis Secondary data analysis
Modelling Historical analysis
Archive analysis
Discourse analysis
orms of What is impact of x? How and why? (where investigator has little
research How and why? (where investigator has control over events, or limited knowledge
question control over events, and existing knowledge about mechanism involved)
about mechanisms involved) What (what is going on here)?
What (how many, how much, who, where?)
amples of Structured and semi-structured interviews Qualitative individual interviews

dominant data (including open-ended questions) Focus group discussions
collection Routine record review Observation
methods Document review
Key sampling Representive of sample population Purposive sampling guided by theory, to ensure
principles maximum variability across relevant units
Characteristics Statistical analysis following predetermined Iterative

of data analysis rules Interpretative
There is also a third category of research strategy: mixed-
method studies, which deliberately combine elements
Using theory and
of fixed and flexible design “to expand the scope of, conceptual frameworks
and deepen the insights from, their studies” (Sandelowski, to inform the study
2000). This strategy is not linked to a particular know-
ledge paradigm or set of methods, nor does it reflect a Given the complexity of the phenomena addressed by
mix of paradigms. Instead it purposefully combines HPSR, theory should play an important role within every
different methods of inquiry in order to capture different study design and within both fixed and flexible research
strategies. In evaluation work, for example, there is
dimensions of the central phenomenon of focus. Mixed-
increasing acknowledgement of the importance of
method studies, thus, entail various combinations of
theory-driven inquiry in adequately addressing complex
sampling and/or data collection and/or data analysis
causality (de Savigny & Adam, 2009) – in both experi-
techniques in order to:
mental or quasi experimental designs and the case study
allow triangulation across data sets;
n
work linked to critical realist evaluation (see Part 4,
enable the elaboration of results, through complemen-
n ‘Advances in impact evaluation’). However, currently,
tary data and analyses;
theory is too rarely used in HPSR and as a result policy
guide the development of an inquiry by identifying
n
analysis work, for example, is often quite descriptive.
additional sampling, data collection and analysis
Opportunities for the theory-building and explanatory
needs.
work that would better inform policy-making and imple-
Within a study different methods may be used sequen- mentation are ignored (Gilson & Raphaely, 2008; Walt et
tially (at different times) or concurrently (at the same al., 2008).
time). Examples of what a mixed-method study could
entail in practice are given below. In broad terms, theory provides a language for describing
and explaining the social world being studied and
The research could entail an initial small-scale
n
represents a general explanation of what is going on in a
intensive study using qualitative methods to develop
detailed understanding of a phenomenon. This would situation. It offers the basis for generating hypotheses
be followed by a larger-scale structured survey under- (predicted answers that can be statistically tested in fixed
taken to generate more extensive understanding designs), as well as looser propositions of how different
of the same phenomenon, and that uses a tool deve- dimensions of a phenomena may be linked, which can
loped with the detailed understanding generated
be explored or considered in analysis (flexible designs).
from the initial study.
The ‘middle range theory’ represented by the latter can
An initial structured survey, using a random sampling
n
be captured in the form of a conceptual framework (a
approach to gather knowledge around a pheno-
menon within one population of respondents, could set of concepts and their inter-linkages) that may offer
provide the basis for purposeful sampling of respon- explanations or predictions of behaviour, or outcomes,
dents within the same population to allow more but may also simply identify relevant elements and
detailed inquiry and gain a deeper understanding of relationships.
the results of the initial survey.
A conceptual framework to guide study design can be
The mixing of methods might only occur in data
n
developed from a review of relevant empirical and
analysis, perhaps by interpreting different sets of
study results or through converting one type of data theoretical literature. The framework can help to identify
into the other in order to allow statistical analysis of relevant concepts and variables (fixed strategies) or
qualitative data. issues (flexible strategies) for investigation, and to guide
However, whichever approach is used, mixed-method the selection of samples or cases (flexible strategies).
studies involve a focus on a particular phenomenon and In addition, a conceptual framework may be revised
a purposeful combination of methods to achieve justified as the data collected are analysed. Alternatively, it may
goals in the context of the particular inquiry. be generated as a result of the data analysis process.
In either case, the conceptual framework can be put back
into the public domain to be questioned and perhaps
Step 3: Ensure research
used to support future research. Such theory building is quality and rigour
a process of knowledge generation.
The criteria used to make judgements of research quality
Therefore, HPSR is not solely concerned with generating
and rigour differ between paradigms of knowledge.
empirical evidence to inform policy decisions. Rather,
Whereas positivist research emphasizes validity and
HPSR can combine theoretical and empirical work or
reliability – ensured through careful study design, tool
be primarily theoretical and still maintain its policy
development, data collection and appropriate statistical
relevance.
analysis – relativist research considers the trustworth-
Combined theoretical and empirical work has, for iness of the analysis – whether it is widely recognized to
example, aided understanding of the norms and customs have value beyond the particular examples considered.
influencing the decision-making of health system actors The different criteria and questions used in assessing the
in particular contexts (such as Riewpaiboon et al., 2005; quality of research based on fixed and flexible designs
Sheikh and Porter, 2010). It has also traced the patterns are summarized in Table 7. Table 8 indicates how trust-
and influences over time of policy change across sub- worthiness can be established by providing information
national, national and global levels (for example Walt, on study design, data collection, and the processes of
Lush & Ogden, 2004). Theory-driven evaluation, mean- data analysis and interpretation.
while, supports research that seeks to explain how new
Ultimately, good quality HPSR always requires a critical
policies and interventions influence health system
and questioning approach founded on four key processes
operations (Marchal, Dedzo & Kegels, 2010). Combined
(Gilson et al., 2011):
theoretical and empirical work can also generate ideas
about how to influence policy agendas (for example n n active process of questioning and checking during
Shiffman, 2007: advocacy in agenda setting) or manage the inquiry (Thomas, 1998): ask how and why things
happened – not only what happened; check answers
policy change (for example Walker & Gilson, 2004:
to questions to identify additional issues that need to
managing front line providers acting as street-level be followed up in order to deepen understanding of
bureaucrats). Such ideas have relevance beyond the the experience.
original settings in which the research was conducted. n constant process of conceptuali ing and recon-
Purely theoretical research can also lead to new ways to ceptuali ing (Thomas, 1998): Use ideas and theory
to develop an initial understanding of the problem,
describe the nature and organization of health systems,
or situation of focus, in order to guide data collection
or what influences their performance, and to understand but use the data collected to challenge those ideas
what drives particular policy actors in their decision- and assumptions and, when necessary, to revise your
making (for example Bloom, Standing & Lloyd, 2008 ideas in response to the evidence.
(plural health systems); de Savigny & Adam, 2009 Crafting interpretive udgements (Henning, 2004)
n
(systems thinking); Gilson, 2003 (trust and health based on enough evidence, particularly about context,
systems); Kutzin, 2001 (financing); Mackian, Bedri & to justify the conclusions drawn as well as deliberate
consideration of contradictory evidence (negative
Lovel, 2004 (health seeking behaviour)). Through such
case analysis) and review of initial interpretations by
work HPSR informs policy by expanding our under- respondents (member checking).
standing of what strengthening a health system involves, Researcher refle ivity: be explicit about how your
n
and identifies research questions for empirical investi- own assumptions may influence your interpretation
gation. and test the assumptions in analysis (Green &
Thorogood, 2009).
Part 3 of the Reader presents references to some con-
ceptual frameworks that are valuable in HPSR.
Table 7 Criteria and questions for assessing research quality (Source: adapted from Robson, 2002)
i ed designs le i le designs
Reliability: Is your variable measure reliable? Confirmability: Do the data confirm the general findings
and lead to their implications?
Construct validity: Are you measuring what you think
you are measuring? Dependability: Was the research process logical and well
documented?
Internal validity: Does the study plausibly demonstrate
a causal relationship? Credibility: Is there a match between participants’ views
and the researcher’s reconstruction of them?
External validity: Are the findings statistically
generalizable? Transferability: Do the findings generate insights that are
transferable to other settings?
Table 8 Processes for ensuring rigour in case study and qualitative data collection
and analysis (Source: Gilson et al., 2011)
ample
Principle A study of the influence of trust in workplace relationships over
health worker motivation and performance, involving in-depth
inquiry in four case studies (Gilson et al., 200 )
Prolonged engagement with the subject of inquiry Case study:

Although ethnographers may spend years in the field, A period of three to four weeks spent in each case study
HPSR tends to draw on lengthy and perhaps repeated facility
interviews with respondents, and/or days and weeks of Respondents
engagement within a case study site Informal engagement & repeated formal interviews
Use of theory Conceptual framework derived from previous work

To guide sample selection, data collection and analysis, (Gilson et al., 2005)
and to draw into interpretive analysis Case study selection based on assumptions drawn from
framework (see below)
Theory used in triangulation and negative case analysis
(see below)
Case selection Four primary health care facilities: two pairs of facility
Purposive selection to allow prior theory and initial types, & in each pair one well and one poorly performing
assumptions to be tested or to examine average’ or as judged by managers using data on utilization and
unusual experience tacit knowledge (to test assumptions that staff in well
performing’ facilities have higher levels of motivation
and workplace trust)
Sampling In small case study facilities, sampled all available staff;

Of people, places, times etc, initially, to include as many as in larger facilities for interviews: sampled staff of all
possible of the factors that might influence the behaviour groupings and with a range of staff in each group
of those people central to the topic of focus (subsequently (considering e.g. age, sex, length of time in facility);
extend in the light of early findings) random sample of patients visiting each facility; all facility
Gather views from wide range of perspectives and supervisors & area manager
respondents rather than letting one viewpoint dominate
Multiple methods (case studies) For each case study site:

se multiple methods for case studies Two sets of formal interviews with all sampled staff
Researcher observation & informal discussion
Interviews with patients
Interviews with facility supervisors and area managers
Table 8 (Continued) Processes for ensuring rigour in case study and qualitative data collection
and analysis (Source: Gilson et al., 2011)
ample
Principle A study of the influence of trust in workplace relationships over
health worker motivation and performance, involving in-depth
inquiry in four case studies (Gilson et al., 200 )
Triangulation ithin cases:

Looking for patterns of convergence by comparing results Initial case reports based on triangulation across all data
across multiple sources of evidence (e.g. across interviewees, sets for that case (and across analysts in terms of
and between interview and other data), between individual staff members’ experience), generating overall
researchers, across methodological approaches, with theory judgments about facility-wide experience as well as noting
variation in individual health worker experience
Cross-cases:
Initial case reports compared with each other to look for
common and different experiences across cases, and also
compared with theory to look for convergence or divergence
Negative case analysis ithin cases:

Looking for evidence that contradicts your explanations Triangulation across data identified experiences that
and theory, and refining them in response to this evidence contradicted initial assumptions (e.g. about the influence
of community interactions over motivation, and about the
association between low motivation and poor caring
behaviour), and identified unexpected influences (e.g. a
general sense of powerlessness among health workers)
Cross-cases:
Cross-site analysis identified facility-level experience that
contradicted initial assumptions underpinning study (e.g.
about link between high levels of workplace trust, strong
health worker motivation and positive caring behaviour),
and identified unexpected conclusions (e.g. about the
critical importance of facility level management over trust
and motivation)
Report notes weak evidence to support links between
levels of workplace trust and client perceptions, but also
stronger evidence of links between levels of workplace
trust and motivation
Peer debriefing and support Preliminary case study reports initially reviewed by other
Review of findings and reports by other researchers members of the research team
Respondent validation (Member checking) Preliminary cross-case analysis fed back for review and
Review of findings and reports by respondents comment to study respondents; feedback incorporated into
final reports
Clear report of methods of data collection and Report provides clear outline of methods and analysis
analysis (Audit trail) steps as implemented in practice (although more could be
Keeping a full record of activities that can be opened to fuller and reflexive)
others and presenting a full account of how methods
evolved to the research audience
Step 4: Apply ethical Be concerned about safeguarding:
1. the scientific validity and trustworthiness of the data
principles – through careful and deliberate training for all
research staff, including fieldworkers, to equip them
As with all research, it is important to take account of
with the attitudes and communication skills necessary
ethical issues in conducting HPSR. Although the focus of
to conduct good quality interviews and get beyond
the research differs from other health research, there are
their differences in race, class, nationality, gender or
always issues of power at play between those doing the
education with respondents; and treatment of field-
research and those being researched, and so there is real
workers as true partners in the research inquiry,
potential for disrespectful and unfair treatment. Robson
recognizing their essential role in shaping the nature
(2002) suggests that all ‘real world researchers’ need to
and quality of data.
watch out for the following ten questionable ethical
2. social value and a favourable risk–benefit ratio of the
practices:
study – by careful consideration of the individual
involving people without consent
n
and community-level risks and benefits of participa-
coercing them to participate
n
tion in the study, through engagement with a range
withholding information about true nature of the
n
of stakeholders at the start of the study and constant
research
review and reflection during the study.
otherwise deceiving participants
n
3. informed consent and respect for participants and
inducing participants to commit acts diminishing of
n
their self-esteem communities – by ensuring that all team members
are familiar with the study’s key messages and can
violating rights of self-determination
n
call for assistance when unexpected ethical issues
exposing participants to physical or mental stress
n
arise; are able to, and do, demonstrate respect for
invading privacy
n
participants in all their engagements with commu-
withholding benefits from some participants
n
nities; and re-negotiate relationships as and when
not treating participants fairly or with respect.
n
necessary rather than concentrate efforts only on
These are similar to the concerns of all health research. formal consent procedures (which may be infeasible
The challenges may be particularly acute in cross-cultural in an HPSR study or impact negatively on the rela-
research, such as when HPSR is undertaken in lower- tionships with study participants that are essential to
income countries by researchers or others from higher- gathering honest information).
income settings (Molyneux et al., 2009). Thus, one of the 4. independent review – by supporting ethics commit-
eight ethical principles proposed by Emanuel et al. (2004) tees to pay particular attention to the proposed
for clinical research is collaborative partnership between process of research and interactions among different
investigators and research sponsors in higher-income
actors within HPSR work, rather than primarily exami-
countries and researchers, policy-makers and commu-
ning study design and tools.
nities in lower-income countries (see Box 7).
Ultimately, however, “the social relationships established
However, as HPSR differs in nature from medical research,
between researchers and field-teams and community
there are some particular ethical debates in, and peculiar
members, are critical to fulfilling the moral (as opposed
ethical challenges for, this area of work. From reflection
to legal) aspects of ethics guidelines” (Molyneux et al.,
on the experience of conducting household-level HPSR
2009:324). Such relationships will always be important
studies in different countries, for example, Molyneux et
in HPSR, whether the interviewees are community
al. (2009) make the following four sets of proposals on
members or policy elites.
how to implement the principles of Box 7 in this form of
research.
Box : ight ethical principles for clinical research
in low and middle income countries
• Collaborative partnership • Social value

• Scientific validity • Fair selection of study population
• Favourable risk-benefit ratio • Independent review
• Informed consent • Respect for recruited participants and study communities
Source: manuel et al.,
References
Alliance for Health Policy and Systems Research (AHPSR) Gilson L, Khumalo G, Erasmus E, Mbatsha S and McIntyre
(2009). Priority setting for Health Policy and Systems D (2004). Exploring the influence of workplace trust over
Research. AHPSR Briefing Note 3. Geneva, World Health health worker performance: Preliminary national overview
Organization (http://www.who.int/alliance-hpsr/en/ , report South Africa. Johannesburg: Centre for Health
accessed 22 August 2011). Policy. Unpublished report.
Bloom G, Standing H, Lloyd R (2008). Markets, informa- Gilson L et al. (2011). Building the field of health policy
tion asymmetry and health care: Towards new social and systems research: social science matters. PLoS
contracts. Social Science Medicine, 66 (10):2076-2087. Medicine 8(8):e1001079.
de Savigny D, Adam T, eds (2009). Systems thinking for Green A, Bennett S, eds (2007). Sound choices:
health systems strengthening. Geneva, World Health enhancing capacity for evidence-informed health policy.
Organization (http://www.who.int/alliance-hpsr/resources Geneva, World Health Organization.
/9789241563895/en/index.html, accessed 10 July 2010).
Green J, Thorogood N (2009). ualitative methods for
Emanuel E et al. (2004). What makes clinical research in health research. London, Sage Publications.
developing countries ethical? The benchmarks of clinical
Henning E (2004). Finding your way in qualitative
research. ournal of nfectious iseases, 189:930–937.
research. Pretoria, Van Schaik Publishers.
Gilson L (2003). Trust and the development of health
Klopper H (2008). The qualitative research proposal.
care as a social institution. Social Science Medicine,
Curationis. 31(4):62–72.
56(7):1453–1468.
Kutzin J (2001). A descriptive framework for country-level
Gilson L, Palmer N and Schneider H. (2005) Trust and
analysis of health care financing arrangements. Health
health worker performance: exploring a conceptual
Policy, 56:171–204.
framework using South African evidence. Social Science
and Medicine 61(7): 1418-1429 Mackian S, Bedri N, Lovel H (2004). Up the garden
path and over the edge: where might health-seeking
Gilson L, Raphaely N (2008). The terrain of health policy
behaviour take us? Health Policy and Planning,
analysis in low- and middle-income countries: a review
19(3):137–46.
of published literature 1994–2007. Health Policy and
Planning, 23(5):294–307. Marchal B, Dedzo M, Kegels G (2010). A realist evalu-
ation of the management of a well-performing regional
hospital in Ghana. MC Health Services Research, 10:24.
Molyneux C et al. (2009). Conducting health-related Thomas A (1998). Introduction. In: Thomas A, Chataway J,
social science research in low-income settings: ethical Wuyts M, eds. Finding out fast: investigative skills for
dilemmas faced in Kenya and South Africa. ournal of policy and development. London, Sage Publications:1–18.
nternational evelopment, 21(2):309–325.
Varkevisser CM, Pathmanathan I, Brownlee A (2003).
Parsons W (1995). Public policy: an introduction to Designing and conducting health systems research
the theory and practice of policy analysis. Aldershot, projects: olume : proposal development and fieldwork
Edward Elgar. e-book . Amsterdam, KIT Publishers, International
Development Research Centre (http://www.idrc.ca/en/ev-
Potter S, Subrahmanian R (1998). Information needs and
33011-201-1-DO_TOPIC.html, accessed 7 July 2010).
policy change. In: Thomas A, Chataway J, Wuyts M, eds.
Finding out fast: investigative skills for policy and Walker L, Gilson L (2004). We are bitter but we are
development. London, Sage Publications:19–40. satisfied: nurses as street level bureaucrats in South
Africa. Social Science Medicine, 59(6):1251–1261.
Riewpaiboon W et al. (2005). Private obstetric practice
in a public hospital: mythical trust in obstetric care. Social Walt G, Lush L, Ogden J (2004). International organ-
Science Medicine, 61(7):1408–1417. izations in transfer of infectious diseases: iterative loops
of adoption, adaptation and marketing. overnance.
Rist RC (1998). Influencing the policy process with
17(2):189–210.
qualitative research. In: Denzin NK, Lincoln YS, eds.
Collecting and interpreting qualitative materials. Walt G et al. (2008). ‘Doing’ health policy analysis:
Thousand Oaks, California, Sage Publications:400–424. methodological and conceptual reflections and challenges.
Health Policy and Planning, 23(5):308-317.
Robson C (2002). Real world research: a resource for
social scientists and practitioner-researchers, 2nd ed. Yin RK (2009). Case study research: design and methods,
Oxford, Blackwell Publishing. 4th ed. Thousand Oaks, California, Sage Publications.
Sandelowski M (2000) Combining qualitative and

quantitative sampling, data collection, and analysis
techniques in mixed-method studies. Research in ursing
Health, 23:246–255.
Schneider H (2011). Introduction to Health Systems

Research and Evaluation. Masters in Public Health course
power point slides. Cape Town, University of Cape Town.
Sheikh K, Porter J (2010). Discursive gaps in the imple-

mentation of public health policy guidelines in India: the
case of HIV testing. Social Science Medicine, 71(11):
2005–2013.
Sheikh K et al. (2011). Building the field of health

policy and systems research: framing the questions.
PLoS Medicine 8(8):e1001073.
Shiffman J (2007). Generating political priority for

maternal mortality reduction in 5 developing countries.
merican ournal of Public Health, 97(5):796–803.
Part
der ta d g ealth ol y
a d y te
As indicated in Part 1 of this Reader, a defining charac- Additional references
teristic of Health Policy and Systems Research (HPSR) is
that it focuses on issues or problems related to health Atun R, Menabde N (2008). Health systems and
policy and health systems rather than, for example, systems thinking. In: Coker R, Atun R, McKee M, eds.
exploring particular disciplinary questions or perspectives. Health systems and the challenge of communicable
In other words, it is the research question, or issue of diseases: e periences from urope and Latin merica.
focus, that guides the research. Maidenhead, Open University Press (European Observa-
tory on Health Systems and Policies Series):121–140.
This section of the reader presents key references to two
Available at: http://www.euro.who.int/__data/assets/
sets of papers that support HPSR by providing concep-
pdf_file/0005/98393/E91946.pdf
tual frameworks that can inform our understanding of
issues related to health policy and systems. Rationale for selection: Draws on system thinking
n
perspectives.
Health system frameworks Frenk J (2010). The Global Health System: strengthening
national health systems as the next step for global progress.
These references give insight and understanding about PLoS Medicine, 7(1):1–3. Available at: http://www.plosme
the nature of health systems. dicine.org/article/info 3Adoi 2F10.1371 2Fjournal.
pmed.1000089
Two key references
Rationale for selection: A succinct statement of
n
Bloom G, Standing H, Lloyd R (2008). Markets, infor- current thinking by a world leader in the field.
mation asymmetry and health care: Towards new social
World Health Organization (2007). verybody s business:
contracts. Social Science Medicine, 66(10): 2076–2087
Strengthening health systems to improve health outcomes:
which recognizes the plurality of health systems
n
H s framework for action. Geneva, World Health
(i.e. the variety of providers that comprise health
Organization:iv–vii, 1–30. Available at: http://www.who.int/
systems) and the importance of understanding
entity/healthsystems/strategy/everybodys_business.pdf
their institutional dynamics.
Rationale for selection: This is the most recent
n
de Savigny D et al. (2009). Systems thinking: What it is
statement of the influential World Health
and what it means for health systems. In: de Savigny D,
Organization framework.
Adam T, eds. Systems thinking for health systems
strengthening. Geneva, World Health Organization:
37–48. Available at: http://www.who.int/alliance-
hpsr/resources/ 9789241563895/en/index.html
which is the most recent and more nuanced

n
version of the World Health Organization’s
building blocks approach to health systems
(which focuses on the six functions of service
delivery, health workforce, information, medical
products, vaccines and technologies, financing,
leadership/governance) - this work also seeks to
understand health systems from the perspective
of systems thinking.
Part 3 nderstanding Health Policy and Systems 63

Conceptual frameworks Gilson L (2003). Trust and health care as a social
institution. Social Science Medicine, 56(67):1452
for HPSR –1468.
http://dx.doi.org/10.1016/S0277-9536(02)00142-9
These references provide a range of conceptual frame-
works that can be used to guide careful and systematic Rationale for selection: highlights the importance
n
investigation of health policy and health systems’ issues, of relationships within health systems and the
and so lead to a deeper understanding of their complexity. institutional influences over them, and specifically
trust; provides concepts for understanding the
References nature and role of trust in health systems
Atun R et al. (2010). Integration of targeted health Kutzin J (2001). A descriptive framework for country-level
interventions into health systems: a conceptual analysis of health care financing arrangements. Health
framework for analysis. Health Policy and Planning, Policy, 56(3):171–204.
25(2):104–111. http://dx.doi.org/10.1016/S0168-8510(00)00149-4
http://dx.doi.org/10.1093/heapol/czp055 Rationale for selection: conceptual framework for
n
Rationale for selection: integration is an enduring
n understanding and investigating financing issues
theme in HPSR and management as part of wider system
Bossert T (1998). Analyzing the decentralization of health
Vian T (2007). Review of corruption in the health sector:
systems in developing countries: decision space, inno-
theory, methods and interventions. Health Policy and
vation and performance. Social Science Medicine,
Planning, 23(2):83–94.
47(10):1513–1527.
http://dx.doi.org/10.1093/heapol/czm048
http://dx.doi.org/10.1016/S0277-9536(98)00234-2
Rationale for selection: conceptual framework
n
Rationale for selection: conceptual framework for
n
for understanding and investigating corruption,
understanding and investigating health system
central to governance
from decision-making authority perspective
Walt G, Gilson L (1994). Reforming the health sector in
Brinkerhoff D (2004). Accountability and health systems:
developing countries: the central role of policy analysis.
toward conceptual clarity and policy relevance. Health
Health Policy and Planning, 9(4):353–370.
Policy and Planning, 19(6):371–379.
http://dx.doi.org/10.1093/heapol/9.4.353
http://dx.doi.org/10.1093/heapol/czh052
Rationale for selection: simple heuristic for under-
n
n
standing influences over policy decision-making,
understanding & investigating accountability
that is widely used to guide related research
issues, central to governance
Franco LM, Bennett S, Kanfer R (2002). Health sector

reform and public sector health worker motivation:
a conceptual framework. Social Science Medicine,
54(8):1255–1266.
http://dx.doi.org/10.1016/S0277-9536(01)00094-6
n
understanding and investigating HR motivation
and performance
Part
r al a er
he papers presented in this part of the Reader are e amples of good quality and innovative research in the field
of health policy and systems.
Overview: research
strategies and papers
Doing good quality Health Policy and Systems Research The introduction to each group of papers includes:
(HPSR) demands an understanding of what research an overview of the research strategy or approach, its
n
strategy is appropriate to the questions of focus. The relevance to HPSR and brief clarification about how
strategy is neither primarily a study design nor a method, to ensure rigour when conducting such research;
but instead represents an overarching approach to con- a brief description or overview of the selected papers;
n
ducting the research that considers the most appropriate a summary of papers with reference details, focus of
n
methods of data collection and sampling strategy for the the study, the perspective it takes, and the rationale
research purpose and questions. for its selection in the Reader.
The papers provided here are grouped by research A summary of the papers is given in Table 9.
strategy in order to encourage critical and creative
thinking about the nature and approach of HPSR, and
to stimulate new research that goes beyond the often
quite descriptive cross-sectional analyses that form the
bulk of currently published work in the field. The research
strategies were chosen to demonstrate the breadth
of HPSR work, covering both dominant and emerging
approaches in the field. They are:
1. Cross-sectional perspectives
2. The case-study approach
3. The ethnographic lens
4. Advances in impact evaluation
5. Investigating policy and system change over time
6. Cross-national analysis
7. Action research
Part 4 m irical Pa ers 67

68
Table 9 Overview of papers presented
Page no Paper System function(s) Policy/System Disciplinary perspective Key features Country
of focus level addressed (or ey approach)
CR SS-S CT P RSP CT S
5 Blauuw et al., Human resources Micro: Health workers Health economics • se of discrete choice experiments Multi-country
2010 (incentive packages) and economic evaluation
• Example of analysis for policy
Glassman et al., Governance and Macro: National Policy analysis • Application of policy-maker’ in Dominican
1999 financing (policy change, analysis Republic
health systems reform) • Example of analysis for policy
95 Morrow et al., Service delivery Meso and micro: Public health • Mixed-method study Viet Nam
2009 (malaria control) Primary level/community • Considers both demand and supply
issues
105 Ramanadhan et al., Human resources Micro: Health workers (Social network analysis) • se of network analysis and Ethiopia
2010 (capacity development) exploration of social capital issues
11 Ranson, ayaswal Financing (household Micro: Households Health economics • Sequential use of methods in mixed- India
& Mills, 2011 expenditures) method study
Gilson L, ed. (2012). Health Policy and Systems Research: A Methodology Reader
129 Riewpaiboon et al., Service delivery (provider Micro: Hospital and Sociology/Anthropology • Theory building Thailand
2005 – patient interactions, individual • Examination of institutions of health
obstetric care) system
1 9 Rwashana, Williams Vaccines and service Macro: System (Systems thinking) • Rare example of use of systems ganda
& Neema, 2009 delivery (immunization thinking
programme, nested in
health system)
152 Sheikh & Porter, Governance and Micro: Individual Policy analysis • Detailed and theory-driven India
2010 service delivery examination of decision-making
(HIV clinical guideline
implementation)
TH C S ST D PPR CH
1 Atkinson et al., Governance Meso: Districts and Anthropology • Districts as cases Brazil
2000 (decentralization) facilities • Theory building
• Examination of complex causality
Murray & Elston, Financing Cross-level Sociology • Integrated analysis of policy change Chile
2005 (private insurance) across system layers
1 Mutemwa, Health information and Meso: District Management • Exploratory case analysis ambia
2005 governance (decision-
making at district level)
19 Rolfe et al., Human resources Meso: District and Sociology • Strong example of analysis in case nited
200 (private sector) facility study work Republic of
• Analysis for policy Tanzania
210 Russell & Gilson, Financing (household Micro: Households Development economics • se of longitudinal household cases Sri Lanka
200 expenditure) • Examination of complex causality
22 Shiffman, Stanton Governance (policy Macro: National/global Policy analysis • se of theory and generation of Honduras
& Salazar, 200 change, Safe Mother- questions from analysis
hood Initiative)
TH TH R PH C S
Aitken, 199 Human resources Micro: Health workers Anthropology/Sociology • Theory building Nepal
(training programmes)
Part 4
2 9 Behague & Governance and service Macro: Global debates Anthropology/Sociology • Discourse analysis N/A
Storeng, 200 delivery (debates about
approaches to maternal
health care provision,
and evidence-based
policy-making)
m irical Pa ers
69
70
TH TH R PH C S (C T D)
George, Human resources and Micro: Health worker– Anthropology/Sociology • Rich analysis of key health system India
2009 governance (manage- supervisor interactions functions
ment, accountability) and influences
2 5 Lewin & Green, Service delivery Micro: Clinic, provider– Anthropology/Sociology • se of concepts South Africa
2009 (primary care clinic) patient interactions • Programme and facility focus
4 D C S P CT T
25 Björkman & Governance (community Meso: Facility/community Health economics • Quasi experimental evaluation ganda
Svensson, 2009 accountability mechanism) • nusual focus for this evaluation approach
292 Macinko et al., Service delivery Macro: National Epidemiology • Ecological analysis using available Brazil
200 (primary care model) panel data
0 Marchal, Dedzo Human resources Meso: Facility (Policy evaluation Critical • Rare example of critical realist evaluation Ghana
& Kegels, 2010 (management) realism)
Wang et al., Financing (community- Micro: Household Health economics • nusual quasi-experimental evaluation, China
2009 based health insurance) using propensity matching scores
Gilson L, ed. (2012). Health Policy and Systems Research: A Methodology Reader
ST T P C D S ST CH RT
20 Brown, Cueto Governance (global Macro: Global History • Historical analysis N/A
& Fee, 200 organizations and • Global organization focus
discourse)
1 Crichton, Governance and service Macro: National Policy analysis • se of theory Kenya
200 delivery (policy change,
family planning)
Masanja et al., Service delivery (child Macro: System Epidemiology • Rich trend analysis with explanation nited
200 mortality trends and around system development Republic of
explanations) Tanzania
ST T P C D S ST CH RT (C T D)
51 Van Ginneken, Human resources Macro: National History • Historical analysis South Africa
Lewin & Berridge, (community health • nusual use of witness seminars
2010 workers)
CR SS- T SS
Bryce et al., Service delivery (Inte- Meso: Districts and Epidemiology • Seminal paper Multi-country
2005 grated Management facilities • Careful system evaluation
of Childhood Illness
approach)
9 Gilson et al., Governance and financing Cross-level Policy analysis • Conceptual framework used to guide Multi-country
2001 (implementing policy study
change, Bamako Initiative • Opportunistic country cases selected
community financing • Explanatory focus
schemes)
00 Lee et al., Governance and service Macro: National Policy analysis • Deliberate country cases selected Multi-country
199 delivery (sustaining • Careful analysis
family planning policy • Explanatory focus
implementation)
11 O’Donnell et al., Financing (public Macro: National Health economics • Rigorous cross-country analysis, with Multi-country
200 spending incidence) explanation
CT R S RCH
5 Khresheh & Health Information Meso: Hospital Action research • Rare application of research strategy ordan
Barclay, 200 (hospital records
Part 4
system)
1 Khresheh & Health Information Meso: Hospital Action research • Account of action research ordan
Barclay, 200 (hospital records
system)
m irical Pa ers
71
1. Cross-sectional To extend the analysis and interpretation, different
n
studies may be triangulated to provide different
perspectives perspectives on the same question or may answer
different kinds of questions (for example ‘what’
versus ‘why’ questions).
Helen Schneider
niversity of the estern Cape, South frica Depending on the purpose, data collection in mixed-
method studies can be either concurrent or sequential
and
(Creswell & Plano-Clark, 2007).
Sara Bennett
ohns Hopkins loomberg School of Public Health, The findings of such studies often involve what can be
altimore, M , nited States of merica described as a ‘bricolage’, a “pieced together close-
knit set of practices that provide solutions to a problem
Cross-sectional studies may seek to explore, describe or
in a concrete situation” (Denzin & Lincoln, 1998:3).
explain a phenomenon at a particular moment in time
The study components provide different insights into a
(see Part 2: Step 2 of this Reader). This distinguishes
phenomenon and are combined as pieces in a puzzle to
them from longitudinal and other studies which describe
explain the phenomenon of focus.
or analyse change over time, and experimental studies
which involve interventions. As cross-sectional studies
generally require fewer resources than other research Rigour in cross-sectional
strategies, they are the most frequently performed and studies
reported type of research in HPSR.
As with other research strategies, research validity/
Cross-sectional studies encompass a wide universe of
trustworthiness and reliability are important in cross
disciplinary perspectives and methods from both the
sectional studies, whether from the fixed or flexible
fixed and flexible research traditions. They range from
traditions. Such concerns are especially important in
single to mixed (quantitative and qualitative) and multi-
HPSR seeking to shed light on the complex dynamics and
method forms of data collection (when the phasing of
relationships between system actors and dimensions (see
fixed and flexible research designs allows triangulation
Part 2: Step 1).
from one data collection approach to inform the other
and epistemiological triangulation, as well as use of The validity of cross-sectional studies may be undermined
secondary data sources). While mixed-method cross- by (Robson, 2002:171):
sectional studies may share features of the case study inadequate or insufficient description of a phenomenon;
n
method they do not necessarily follow the same analytic problematic interpretation through selective use of, or
n
procedures. inappropriate meanings imposed on, data;
As also noted in Part 2: Step 2, HPSR mixed-method explanations drawn without considering alternatives or
n
‘counterfactuals’;
studies serve a number of purposes (Pope & Mays,
failure to draw on existing concepts and theory in the
n
2009):
literature.
In the process of tool design, qualitative interviews may
n
precede the development of quantitative instruments, The validity of cross-sectional studies can be enhanced
in instances where standardized tools may not exist by (Pope & Mays, 2009):
or the context specificity of the phenomenon requires triangulation of data, observers, methodological
n
tailored approaches. approaches, and with theory;
A quantitative survey may be conducted to provide
n member checking (asking respondents to validate the
n
a sampling frame to select cases for qualitative study. findings and analysis);
clear description of methods of data collection and
n
analysis;
Overview of selected papers
reflexivity by the author (reflecting on how their own
n For this Reader we have specifically selected cross-
personal or intellectual biases may have influenced the sectional studies which demonstrate data collection or
study and analysis);
analytic techniques that go beyond the most commonly
attention to, and discussion of, negative cases (incidents
n
used approaches of key informant interviews or straight-
or experiences that are unusual in terms of the domi-
forward content analysis. The selection includes examples
nant pattern of findings and the possible explanations
of which are then specifically discussed in analysis to of:
clarify their implications for the broader set of findings). discrete choice experiments (DCEs), derived from the
n
economic theory of demand, examining nurses’
References preferences for policy interventions that would attract
them to rural areas in three countries (Blaauw et al.,
Creswell JW, Plano-Clark VL (2007). esigning and 2010) – this innovative study also shows the context
conducting mi ed methods research. Thousand Oaks, specificity of health policy and systems interventions
and offers guidance for policy-makers;
California, Sage Publications.
the use of PolicyMaker, a computer-assisted political
n
Denzin NK, Lincoln YS (1998). Introduction: Entering the analysis tool to study health policy reform in the
field of qualitative research. In: Denzin NK, Lincoln YS, Dominican Republic and draw out guidance for policy-
eds. Collecting and interpreting qualitative materials. makers (Glassman et al., 1999);
Thousand Oaks, California, Sage Publications:1–34. a multi-method study that includes observations, use of
n
routine data and multi-stakeholder interviews to
Pope C, Mays N (2009). Critical reflections on the rise of construct a model of the demand and supply side
qualitative research (research methods and reporting). dimensions of poor malaria control in Viet Nam
(Morrow et al., 2009);
ritish Medical ournal, 339(b3425):737–739.
the application of social network analysis, an unusual
n
Robson C (2002). Real world research: resource for and interesting analytic approach for HPSR, to evaluate
social scientists and practitioner-researchers, 2nd ed. the impact of health management training in Ethiopia
Oxford, Blackwell Publishing. (Ramanadhan et al., 2010);
a mixed-method study in which qualitative and
n
quantitative methods are used sequentially to examine
the coping strategies used by households to manage
the costs of hospital inpatient care in India (Ranson,
Jayaswal & Mills, 2011);
building explanatory frameworks for the choice of
n
public or private obstetric care provider among women
of different socio-economic status in Thailand, informed
by trust theory (Riewpaiboon et al., 2005) – this study
also illustrates the approach and value of theory
building in HPSR;
the use of systems theory to explain uptake of immuni-
n
zation in Uganda, drawing on causal loop diagram
methodology to model the relationships in a complex
system (Rwashana,Williams & Neema, 2009);
the use of detailed interpretive analysis in a study of
n
how policy actors’ understandings influence HIV policy
implementation in India (Sheikh & Porter, 2010).

Some of the different purposes of mixed or multi-method Ranson MK, Jayaswal R, Mills AJ (2011). Strategies for
approaches are highlighted in two of these papers. coping with the costs of inpatient care: a mixed methods
Ranson, Jayaswal & Mills (2011) report a study in which study of urban and rural poor in Vadodara District,
focus group discussions were conducted to develop a Gujarat, India. Health Policy and Planning, 1–13.
closed-ended survey tool. The survey, in turn, identified a http://dx.doi.org/10.1093/heapol/czr044
group of poorer patients for further in-depth interview. Reproduced by permission of Oxford University
n
The study reported by Morrow et al. (2009), meanwhile, Press. Copyright Oxford University Press, 2011.
involved 17 different forms of data collection, sequenced
in a ‘formative’ stage that assisted in the design of a Riewpaiboon et al. (2005). Private obstetric practice in a
subsequent ‘assessment’ phase. The paper draws public hospital: mythical trust in obstetric care. Social
together data, like pieces of a puzzle, to present an Science Medicine, 61:1408–1417.
explanatory model of the systems and social (non- http://dx.doi.org/10.1016/j.socscimed.2004.11.075
biological) factors underlying pockets of poor malaria Reproduced by permission of Elsevier. Copyright
n
control. Elsevier, 2005.
Rwashana AS, Williams DW, Neema S (2009). System dy-

References for selected papers namics approach to immunization healthcare issues in
Blaauw D et al. (2010). Policy interventions that attract developing countries: a case study of Uganda. Health
nurses to rural areas: a multicountry discrete choice nformatics ournal, 15(2):95–107.
experiment. ulletin of the orld Health rgani ation, http://dx.doi.org/10.1177/1460458209102971
88:350–356. Reproduced by permission of SAGE Publications.
n
http://dx.doi.org/10.2471/BLT.09.072918 Copyright SAGE Publications, 2009.
Glassman A et al. (1999). Political analysis of health Sheikh K, Porter J (2010). Discursive gaps in the implemen-
reform in the Dominican Republic. Health Policy and tation of public health policy guidelines in India: The case
Planning 14(2):115–126. of HIV testing. Social Science Medicine, 71(11): 2005–
http://heapol.oxfordjournals.org/content/14/2/115.full.pdf 2013.
Reproduced by permission of Oxford University
n http://dx.doi.org/10.1016/j.socscimed.2010.09.019
Press. Copyright Oxford University Press, 1999. Reproduced by permission of Elsevier. Copyright
n
Elsevier, 2010.
Morrow M et al. (2009). Pathways to malaria persistence
in remote central Vietnam: a mixed-method study of
health care and the community. MC Public Health 9:85.
http://dx.doi.org/10.1186/1471-2458-9-85
Ramanadhan et al. (2010). Network-based social capital

and capacity-building programs: an example from
Ethiopia. Human Resources for Health, 8:17.
http://dx.doi.org/10.1186/1478-4491-8-17
Special theme – Health workforce retention in remote and rural areas
Policy interventions that attract nurses to rural areas:

a multicountry discrete choice experiment
D Blaauw,a E Erasmus,a N Pagaiya,b V Tangcharoensathein,b K Mullei,c S Mudhune,c C Goodman,c M Englishc &
M Lagarded
Objective To evaluate the relative effectiveness of different policies in attracting nurses to rural areas in Kenya, South Africa and
Thailand using data from a discrete choice experiment (DCE).
Methods A labelled DCE was designed to model the relative effectiveness of both financial and non-financial strategies designed to
attract nurses to rural areas. Data were collected from over 300 graduating nursing students in each country. Mixed logit models were
used for analysis and to predict the uptake of rural posts under different incentive combinations.
Findings Nurses’ preferences for different human resource policy interventions varied significantly between the three countries. In
Kenya and South Africa, better educational opportunities or rural allowances would be most effective in increasing the uptake of rural
posts, while in Thailand better health insurance coverage would have the greatest impact.
Conclusion DCEs can be designed to help policy-makers choose more effective interventions to address staff shortages in rural areas.
Intervention packages tailored to local conditions are more likely to be effective than standardized global approaches.
Une traduction en français de ce résumé figure à la fin de l’article. Al final del artículo se facilita una traducción al español. .‫اﻟﱰﺟﻤﺔ اﻟﻌﺮﺑﻴﺔ ﻟﻬﺬﻩ اﻟﺨﻼﺻﺔ ﰲ ﻧﻬﺎﻳﺔ اﻟﻨﺺ اﻟﻜﺎﻣﻞ ﻟﻬﺬﻩ اﳌﻘﺎﻟﺔ‬
Introduction that allow proper evaluation, but previous calls to strengthen the
monitoring and evaluation of health reforms in low- and middle-
The shortage of health workers in the areas where they are most income countries have had little impact.11
needed is an important problem for health systems. Patients who
In addition, statistically significant evidence of impact in
have the greatest need for health care tend to live in remote and
well controlled trials may not be sufficient for informing practical
rural areas, but attracting skilled health workers to such areas and
policy decisions. The results of many human resource strategies
retaining them there has proved difficult.1 Such an uneven distri-
are, in some measure, self-evident. Rural financial incentives are
bution of health workers contributes directly to the global burden
likely to improve rural recruitment and retention, but the critical
of ill health and inequity in health outcomes. Thus, it will not be
questions are how much money is required to achieve a certain
possible to improve health outcomes globally unless more health
impact and how do financial strategies compare to other policy
professionals are attracted to work in rural and remote areas.2
options, either individually or in combination. The answers to
The factors that often motivate health workers to stay in these questions will certainly vary between settings. What policy-
remote areas have been extensively studied.3–5 Several strategies makers actually need is information on the relative impact and
have been proposed to address the problem, including changing cost-effectiveness of different packages of human resource inter-
student selection criteria; improving educational opportunities ventions in a variety of contexts. Rigorous evaluation methods to
for workers; introducing financial incentives; creating more sup- answer such questions are not currently available.
portive working environments; and making it compulsory for
health professionals to work in underserved areas.6,7 However, the In the meantime, more modelling studies could be carried
potential impact of these policy interventions, either singly or in out to determine the probable outcomes of different policy
combination, remains undetermined. Recent systematic reviews scenarios. Stated preference discrete choice experiments (DCEs)
have invariably concluded that few rigorous studies evaluating the are a promising method for conducting human resource research
impact of rural recruitment and retention strategies have been in low- and middle-income countries.12 They are a quantitative
conducted.3,6–9 In the Cochrane review,6 for example, not a single technique for evaluating the relative influence of different prod-
controlled study met the inclusion criteria. uct attributes on consumer choices13 and have come to be used
What is needed is more evidence, not more reviews, yet widely in health services research, primarily to assess patients’
just how such new evidence will be generated remains unclear, preferences and willingness to pay for different models of health
particularly for low- and middle-income countries. Evaluating the service delivery.14–16 However, DCEs have been used in recent
effectiveness of human resource interventions is not the same as studies to assess the relative importance of different factors on
testing a drug for efficacy. Many human resource strategies require health workers’ job choices.17–19
national policy changes and few are amenable to controlled stud- The objective of this study was to use data from a DCE to
ies.10 Governments and donors should be encouraged to introduce model the relative effectiveness of different policy interventions on
human resource interventions under more controlled conditions the recruitment of nurses to rural areas in three different countries.
a
Centre for Health Policy, University of the Witwatersrand, Private Bag X3, Johannesburg, 2050, South Africa.
b
International Health Policy Program, Bangkok, Thailand.
c
Kenya Medical Research Institute/Wellcome Trust Programme, Nairobi, Kenya.
d
Health Policy Unit, London School of Hygiene and Tropical Medicine, London, England.
Correspondence to D Blaauw (e-mail: duane.blaauw@wits.ac.za).
(Submitted: 1 October 2009 – Revised version received: 29 January 2010 – Accepted: 10 February 2010 )
350 Bull World Health Organ 2010;88:350–356 | doi:10.2471/BLT.09.072918
75
D Blaauw et al. Policy interventions to attract nurses to rural areas
Methods Second, labelled choices allowed us to the most D-efficient choice design, given
design a model with different attribute our design parameters.25 The final design
This study was conducted in Kenya, South levels for the two choices (for instance, had 16 choice sets. The DCE tool was ad-
Africa and Thailand, all three of which the financial incentive applied only to ministered in English in Kenya and South
have documented shortages of professional rural jobs). Third, a labelled design allowed Africa and in Thai in Thailand.
health workers in rural areas. Kenya is for more sophisticated modelling of the Baseline data collection was con-
typical of low-income countries with poor impact of policy interventions on nurses’ ducted with final year nursing students
health outcomes, has limited financial and choice of a rural posting. in a classroom setting. We explained the
human resources for health, and is largely In finalizing the DCE tool we fol- DCE questionnaire to the group, whose
dependent on donors for new human lowed the standard recommended steps for members then completed it on their own.
resource policy interventions.20 South ensuring rigour.24 We began by identifying Students also completed a second ques-
Africa and Thailand are both middle- the attributes and levels to be included in tionnaire with basic demographic informa-
income countries with higher per capita the study. Our explicit intention was to fo- tion. In each college we also held a focus
health expenditure, sufficient numbers of cus on job characteristics influencing rural group discussion that included feedback
skilled health workers, and demonstrated choices that were amenable to policy inter- on the DCE questionnaire (Table 2).
capacity to implement policies that make vention and to test their likely impact in Data from the DCE were entered,
it attractive or compulsory for health pro- different country contexts. To inform the cleaned and analysed using STATA v9.0
fessionals to work in rural areas,21,22 but selection of policy options to be included, (Stata Corp., College Station, TX, USA)
they differ in terms of health outcomes. we reviewed the international literature and Nlogit version 4.0 (Econometric
A comparison of key indicators in the and conducted preparatory qualitative Software, Inc., Plainview, NY, USA). The
three countries is shown in Table 1 (avail- work in each country, as summarized in basic analysis was performed with a multi-
able at: http://www.who.int/bulletin/ Table 2 (available at: http://www.who. nomial logit model. For the cross-country
volumes/88/5/09-072918). int/bulletin/volumes/88/5/09-072918). comparison we used both country-specific
This DCE was part of baseline data Next we completed several iterations and pooled models. Analysis of pooled
collection for a larger longitudinal cohort of design development and consultation DCE data using a multinomial logit model
study we are conducting with recent nurs- across the three countries to arrive at a is problematic because the model’s coeffi-
ing graduates in the three countries. In similar design that allowed comparisons cients are confounded with the scale pa-
accordance with the usual practice in DCE but also addressed local specificities. Pi- rameter (λ), which is inversely proportional
studies,13,17 we estimated that a minimum lot studies were then conducted in each to the error variance of the model.13,26
sample of 300 subjects was needed to allow country, and this resulted in further design This complicates comparisons between
for sub-group analysis. We used a multi- refinements (Table 2). Table 3 summarizes data sets, since observed differences in
stage stratified cluster sampling strategy. the final design used in each country. The coefficients may be scale (variance) effects
Provinces were purposely selected from policy options we evaluated were: rather than real differences. The problem
rural and urban strata, and nursing col- • the introduction of a financial rural alis well known for analyses that combine
leges were subsequently selected from each lowance, using relative salary increases revealed and stated preference data27 and
province until the required sample size was to facilitate cross-country compari- requires more complex statistical model-
achieved. All students nearing the end of sons; ling.28 Following Rose et al.29 we used an
their training as professional nurses at the • the provision of better housing facili- error components mixed logit model for
selected colleges were invited to participate ties; the analysis and the Chow test to formally
in the cohort study. Data collection was • preferential opportunities for special- test differences between coefficients.30
completed during 2008. ist training; Odds ratios (ORs) and their confidence
For the DCE we used a labelled • faster rank promotion; intervals (CIs) were used to compare the
choice design with two choices in each • the provision of a benefit package that relative importance of attributes, while
choice set. In a labelled experiment the differed in each country; and the preferences of different subgroups
options presented have specific labels, in • a change in workplace culture from were evaluated by including interaction
this case rural job and urban job, whereas hierarchical to relational management. terms in the regression models. Finally, the
in an unlabelled or generic design the op- results of the mixed logit models were used
tions are simply labelled job A and job B. Facility type was also included in to predict the effect of different attribute
Unlabelled designs are used to determine the design because it was identified as an (policy) changes on the proportion of
the value of attributes that are assumed to important determinant of health work- nurses choosing a rural job.
be generic, while labelled designs produce ers’ choices. The financial incentive had National and international ethical
alternative-specific valuations. Most of four levels to allow for the evaluation standards were maintained throughout
the DCE studies in the health econom- of nonlinear effects, while all the other the research project. The research protocol
ics literature have used generic designs. attributes had two levels (Table 3). This was reviewed by the ethics committees of
We had several reasons for using labelled specification resulted in a design with the academic institutions of the research-
choices. First, we suspected that particular 8192 (i.e. 211 × 41) possible combinations ers in Kenya, South Africa, Thailand and
job characteristics were not valued to the of attributes and levels. We used DCE the United Kingdom of Great Britain and
same degree in rural versus urban jobs macros for SAS (SAS, Cary, NC, United Northern Ireland. Permission to conduct
(better housing, for instance, appears to States of America) to select combinations the research was also obtained from the
be more highly valued when considering for an orthogonal main effects design, and relevant governmental and educational
a rural posting rather than an urban one). then to organize the selected profiles into authorities in each country.
Bull World Health Organ 2010;88:350–356 | doi:10.2471/BLT.09.072918 351
76
Policy interventions to attract nurses to rural areas D Blaauw et al.
Results tion opportunities in Thailand and a change In our models, age, gender, marital
in management culture in South Africa. status and motherhood were not consistent
Of the 1429 eligible nursing graduates in Of the individual characteristics, only rural predictors of the choice of a rural job. Thai
the selected colleges, 1064 (74.5%) agreed origin showed statistical significance in all graduates were too homogenous to allow
to participate in the study: 345 in Kenya, three countries. us to test some of these factors. Whereas in
377 in South Africa and 342 in Thailand. Fig. 1 suggests that preferences for dif- South Africa students who were younger,
The response rates in the three countries ferent human resource policy interventions single or had children were more likely to
were 65.2%, 87.9% and 74.7%, respec- vary between countries. Kenyan nurses were choose an urban posting, in Kenya these
tively. The demographic characteristics of indifferent to the type of facility, whereas same groups preferred rural jobs. Female
the participants are shown in Table 4. The Thai respondents were 4.3 (95% CI: 3.3– graduates were less likely to choose rural
Thai nursing students were much younger 5.6) times more likely to choose a job in a ru- postings, but not significantly. However,
and predominantly female, unmarried and ral hospital than in a rural health centre, and in all three countries having been born in a
childless, whereas the students from Kenya the South Africans actually preferred rural rural area was significantly associated with
and South Africa were older, many were clinics. In both Kenya and South Africa, the the choice of a rural job, and the effect was
married and more than half had children. most effective policy interventions to attract comparable to that of a 10% salary increase.
Kenya had the highest proportion of male nurses to a rural job were the introduction of For example, graduates from rural areas in
students. Students of rural origin were in a financial rural incentive and the provision South Africa were more likely to choose
the majority in Kenya and Thailand but of preferential access to specialist nursing a rural job than those from urban settings
made up slightly less than half of the South training. For example, the availability of a (OR: 2.7; 95% CI: 1.9–3.6).
African participants. 30% rural allowance made South African The formal statistical testing for dif-
The results from the mixed logit model and Kenyan nurses 12.4 (95% CI: 9.6–15.9) ferences in model coefficients between
are represented diagrammatically in Fig. 1, and 7.7 (95% CI: 6.0–10.0) times more countries is shown in Table 5. Most of the
which compares the impact of different likely to choose the rural job, respectively. differences were highly significant. This
policy interventions and individual charac- However, Thai nursing students were only confirms that nurses in the three countries
teristics on the odds of choosing a rural job 2.0 (95% CI: 1.5–2.7) times more likely valued the human resource policy interven-
in each country. For simplicity, the figure to do so. In South Africa, allowing nurses tions differently.
does not show the rural constant or urban in rural posts to specialize earlier increased Table 6 presents the proportion of
attributes, but these were included in the the odds of rural uptake 6.7 times (95% CI: nurses who would choose a rural job when
model. The statistical model shown cor- 5.5–8.1) and was a more effective measure the mixed logit model was used to simulate
rectly predicted 60.0% of the responses from than a 20% salary increase. For Thai respon- the effect of different policy interventions
Kenya, 62.6% of the responses from South dents, improved housing and an expanded alone or in combination. Thailand is clearly
Africa and 75.2% of the responses from health benefit package were more important experiencing less difficulty recruiting nurses
Thailand. All policy interventions shown than a 30% salary increase. Overall, faster to work in rural areas than Kenya and South
in Fig. 1 yielded statistical significance as promotion and changes in management Africa. Even in the absence of any human
factors influencing the choice of a rural job culture were the factors that least persuaded resource policy intervention, 84.2% of re-
(at the 0.05 level), except for better promo- nurses to accept a rural posting. cent Thai nursing graduates would choose a
Table 3. Attributes included in discrete choice experiment for assessing the effectiveness of policies to attract nurses to rural areas
in Kenya, South Africa and Thailand, 2006
Kenya South Africa Thailand

Attributea
Rural Urban Rural Urban Rural Urban
Facility Dispensary Dispensary Clinic Clinic Health centre Health centre
Hospital Hospital Hospital Hospital Hospital Hospital
Salary Local entry level Local entry level
Local entry level Local entry level Local entry level Local entry level
+10% +10% +10%
+20% +20% +20%
+30% +30% +30%
Training No study leaveNo study leave 6 years before study 6 years before study 2 weeks study leave 2 weeks study leave
1 years’ study leave
1 years’ study leave leave leave per year per year
after 4 years after 4 years 2 years before study 2 years before study 4 weeks per year 4 weeks per year
leave leave
Housing Basic None Basic None Basic None
Superior Basic Superior Basic Superior Basic
Promotion 4 years before 4 years before 2 years before 2 years before 2 years before 2 years before
promotion promotion promotion promotion promotion promotion
2 years before 2 years before 1 year before 1 year before 1 year before 1 year before
promotion promotion promotion promotion promotion promotion
Additional benefit Permanent contract Permanent contract None None Basic cover Basic cover
Short-term contract Car allowance Expanded cover
Workplace culture Hierarchical Hierarchical Hierarchical Hierarchical Hierarchical Hierarchical
Relational Relational Relational Relational Relational Relational
a
The actual tools used contained more detailed descriptions of each attribute and level.
77
the United Republic of Tanzania showed

Table 4. Demographic characteristics of respondents in discrete choice experiment
for assessing the effectiveness of policies to attract nurses to rural areas in that better educational opportunities and
Kenya, South Africa and Thailand, 2006 salary increases were the most influential
policy levers to attract clinical officers to
Variable Kenya South Africa Thailand remote areas.31 We showed that changes
in management culture are relatively
n 345 377 342
unimportant in South Africa, contrary to
Sex
what previous studies have shown.18 This is
Males (%) 31.9 14.3 4.7
perhaps because young graduates have not
Females (%) 68.1 85.7 95.3
developed clear preferences for different
Mean age (years) 31.0 31.5 22.6 management styles. While many preferred
Marital status more personal, supportive managers, oth-
Single (%) 54.8 65.9 100.0 ers argued that formal, hierarchical man-
Married (%) 41.7 30.4 0.0 agement was needed to maintain discipline
Divorced/Widowed (%) 3.5 3.7 0.0 and manage resources properly.
Any children (%) 51.3 61.0 0.0 Most non-financial strategies have
Born in rural area (%) 66.1 46.7 83.0 budgetary implications. Thus, both finan-
cial and non-financial policy interventions
will require a considerable amount of
rural job, compared with only 43.4% of the ineffective, as it would only result in a 3.3
nurses in Kenya and 36.0% of those in South additional financial resources that are not
percentage points increase in the number
Africa. Therefore, even the most effective currently available in most low- and middle-
of nurses choosing rural posts.
single policy intervention in Thailand (an income countries.32,33 Preferential training
expanded health benefit package) would opportunities are attractive to health
Discussion workers because they also provide future
only increase rural uptake by 8.4 percentage
points. We have used DCE data to quantify the economic returns. This was confirmed in
However, in South Africa and Kenya degree to which nurses in Kenya, South our focus group discussions with nursing
the proportion of nurses prepared to work Africa and Thailand are receptive to various graduates. Interestingly, however, the im-
in rural areas could increase dramatically incentives and to model the likely impact pact of certain benefit packages, such as car
if various human resource strategies were of different human resource strategies on allowances for rural nurses in South Africa,
introduced. For example, the model predicts rural recruitment in those countries. In which are normally reserved for more se-
that a 30% rural incentive would increase the absence of data from rigorous evalua- nior staff, had double the impact expected
the proportion of nurses choosing a rural tion studies, such analyses provide useful from their equivalent financial value. The
job to 75.0% in South Africa and to 79.8% insights into the potential effectiveness of explanation may lie in the prestige attached
in Kenya. Preferential access to specialist different human resource policy interven- to such allowances.
training would also be particularly effective tions. DCEs provide some of the only cur- Of all the individual characteristics
in Kenya and South Africa but would have rent evidence on the relative importance reported here, only rural origin was associ-
no impact in Thailand. health workers attach to different incentives ated with a significant increase in the likeli-
The DCE model can also be used to and human resource strategies.12 Some have hood of choosing a rural job. This suggests,
predict the impact of any combination argued that packages of interventions are however, that preferential selection of rural
of policies. Three examples are shown in essential for improving the distribution of students by training institutions can be an
Table 6. Combining all the non-financial in- human resources,3 and DCEs are one of the effective strategy, and it also lends support
terventions is an effective policy package and few methods available for comparing such to claims that student selection policies
would persuade a total of 86.3% of nurses in packages. In forthcoming publications we are a key component of human resource
South Africa, 82.5% of those in Kenya and will also show how DCE data can be used intervention packages.7
98.1% of those in Thailand to opt for a rural to model the cost-effectiveness of different The limitations of DCEs have been
position. If all the human resource strategies human resource strategies. clearly acknowledged elsewhere:12 they
we included in our design were introduced, Our findings confirm that financial can only include a restricted set of attri-
more than 95% of nursing students would incentives are very important in persuading butes, which limits their range and real-
choose to work in a rural area in all three health workers to choose a rural posting, ism; and they rely on stated preferences,
countries. However, for low- and middle- especially in poorer countries, but only if not actual decisions, but the analysis of
income countries it may be more practical to they are fairly large. In our study, a 10% revealed preference data is not always
introduce a 10% rural allowance combined salary increase was relatively ineffective in straightforward.24 Finally, the complexity
with preferential training opportunities for all three countries (Fig. 1).Non-financial of DCE design and analysis restricts wide-
nurses in rural areas, a strategy that would strategies are just as important. Improved spread application, and failure to keep
increase the rural uptake by 46.0 percentage housing and accelerated promotion were up with methodological developments
points in South Africa and 34.0 percentage moderately effective, but preferential ac- can compromise study rigour and valid-
points in Kenya. In South Africa such a cess to training and career development ity.34,35 Ours is the largest DCE-based
strategy would be more effective than a 30% opportunities were very powerful non-fi- study of human resources in low- and
rural allowance, while in Kenya the impact nancial strategies. Similar results have been middle-income countries to date,12 but
of the two strategies would be similar. In obtained in other human resource DCE- producing nationally-representative data
Thailand, however, the combination of an based studies in low- and middle-income will require larger sample sizes, complex
allowance and training would be relatively countries.12 For example, a recent study in sampling strategies and more resources.
78
Fig. 1. Relative importance of different human resource policy interventions and individual characteristics in discrete choice
experiment to study nurses’ choice of rural postings in Kenya, South Africa and Thailand, 2006
16
15
14 Kenya South Africa Thailand 95% CI
13
12.4
12
11
Odds ratio (95% CI)
10
9
8 7.7
7 6.7
6
5.3
5
4.3
4 3.9
3 2.7 2.7
2.3 2.3 2.4 2.1 2.4
2 2.0 2.0 2.2 1.8 2.0 1.9
1.6 1.4 1.5 1.5 1.4 1.5
1.1 1.3 1.3 1.3 1.2
1 0.8 0.7 0.9
0.5 0.7
0.1 0.6
0
Hospital 10% rural 20% rural 30% rural Improved Benefit
Preferential More rapid Relational Age Female Single Any Rural born
allowance allowance allowance housing package training promotion management < 30 years children
Human resources policy change culture Individual characteristics
CI, confidence interval.
This study, which is one of the first remote and rural areas does not pay suf- we have demonstrated that different sub-
labelled DCE studies and the first mul- ficient attention to the diversity of indi- groups of nurses have different preferences,
ticountry DCE that we could identify in vidual preferences. It should not be the and in future studies we will compare the
the health literature, has demonstrated the aim of human resource policy research to choices of different types of health work-
more advanced modelling that is possible identify a proven set of standard strategies ers. Packages of interventions are likely to
with labelled DCEs. Labelled designs are to be applied in any context. Our model- be more effective than individual policies
of particular relevance to human resource ling study confirms that both financial in attracting health workers to rural areas3
questions but should become more widely and non-financial incentives are effective not only because individual policies have
used in health research.36 Only very few in motivating nurses to move to rural an additive effect, but because different
multicountry studies exist in the entire and remote areas, and that a package of subgroups of health workers respond dif-
DCE literature,29 probably because they interventions is more effective than a single ferently to different components. DCEs
present significant challenges in design and strategy. However, it has also shown that provide an important tool to investigate
analysis. Nevertheless, they could be used to different countries require completely dif- such individual heterogeneity.
investigate contextual differences in health ferent combinations of human resource
worker preferences and responses, an area policies. Furthermore, it is likely that
of research that is underdeveloped to date. nurses and doctors and other categories of
Conclusion
Indeed, much of the discourse and health workers will respond differently to This study confirms that DCEs can
data on health workforce retention in a particular set of incentives. In this study be designed to assist policy-makers in
Table 5. Pairwise comparison of coefficients between countries in pooled results of discrete choice experiment for assessing the
effectiveness of policies to attract nurses to rural areas in Kenya, South Africa and Thailand, 2006
Rural variable South Africa–Thailand South Africa–Kenya Thailand–Kenya

Difference in P-valuea Difference in P-valuea Difference in P-valuea
coefficients coefficients coefficients
Hospital −1.95 < 0.001 −0.59 < 0.001 1.36 < 0.001
Increase in salary 0.26 0.002 0.08 0.201 −0.18 0.022
Improved housing −0.49 < 0.001 −0.23 < 0.001 −0.23 < 0.001
Benefit package −0.11 0.396 2.60 < 0.001 2.71 < 0.001
Preferential training 1.73 < 0.001 0.74 < 0.001 −1.00 < 0.001
More rapid promotion −0.54 < 0.001 −0.59 < 0.001 −0.59 < 0.001
Relational management −0.40 < 0.001 0.02 0.851 0.42 < 0.001
a
Chow test.
79
Table 6. Predicted impact of different policy interventions on nurses’ uptake of rural postings in Kenya, South Africa and Thailand, 2006
Intervention Kenya South Africa Thailand

Change Total uptake Change Total uptake Change Total uptake
(% points) (%) (% points) (%) (% points) (%)
Single interventions
Base uptake – 43.4 – 36.0 – 84.2
10% rural allowance +15.2 58.6 +16.9 52.9 +3.3 87.5
20% rural allowance +27.8 71.2 +30.5 66.5 +4.8 89.0
30% rural allowance +36.4 79.8 +39.0 75.0 +5.8 90.0
Better rural housing +6.5 49.9 +8.1 44.1 +5.4 89.6
Benefit package −28.0 15.4 +15.8 51.8 +8.4 92.6
Preferential training opportunities +21.9 65.3 +35.5 71.5 +1.2 85.4
More rapid promotion +17.0 60.4 +8.6 44.6 +6.7 90.9
Relational management culture +5.0 48.4 +3.1 39.1 +7.5 91.7
Intervention packages
Housing + benefita + training + +39.1 82.5 +50.3 86.3 +13.9 98.1
promotion + relational management
30% allowance + housing + benefita +51.7 95.1 +59.2 95.2 +14.5 98.7
+ training + promotion + relational
management
10% allowance + training +34.0 77.4 +46.0 82.0 +3.3 87.5
a
The benefit package was excluded in Kenya because it decreased rural uptake.
choosing more effective human resource tially between countries. This suggests Funding: This document is an output
policy interventions to address the that intervention packages tailored to from the Consortium for Research on
shortage of health professionals in rural local conditions are more likely to be Equitable Health Systems, funded by
and remote areas. We have quantified effective than standardized global ap- the UK Department for International
the relative importance of different fac- proaches. These insights should inform Development (DFID) for the benefit of
tors in nurses’ career choices and shown the future human resource research developing countries.
that nurses’ receptiveness to various hu- agenda in low- and middle-income
man resource strategies differs substan- countries. ■ Competing interests: None declared.
‫ﻣﻠﺨﺺ‬
‫ ﺗﺠﺮﺑﺔ ﺣﻮل اﺧﺘﻴﺎرات ﻣﻨﻌﺰﻟﺔ ﻣﺘﻌﺪﱢدة اﻟﺒﻠﺪان‬:‫اﻟﺘﺪﺧﻼت ﰲ اﻟﺴﻴﺎﺳﺎت ﻻﺟﺘﺬاب اﳌﻤﺮﺿﺎت إﱃ اﳌﻨﺎﻃﻖ اﻟﺮﻳﻔﻴﺔ‬
‫اﳌﻮﺟﻮدات اﺧﺘﻠﻔﺖ ﺟﻮاﻧﺐ اﻟﺘﻔﻀﻴﻞ ﻟﺪى اﳌﻤﺮﺿﺎت ﺑني اﳌﻮارد اﳌﺨﺘﻠﻔﺔ‬ ‫اﻟﻬﺪف ﺗﻘﻴﻴﻢ اﻟﻔ ﱠﻌﺎﻟﻴﺔ اﻟﻨﺴﺒﻴﺔ ﻟﻠﺴﻴﺎﺳﺎت اﳌﺨﺘﻠﻔﺔ اﻟﺘﻲ ﺗﺴﺘﻬﺪف اﺟﺘﺬاب‬
‫ ﻓﻔﻲ ﻛﻴﻨﻴﺎ‬.‫ﻟﻠﺘﺪاﺧﻼت ﰲ اﻟﺴﻴﺎﺳﺎت اﺧﺘﻼﻓﺎً ﻳﻌﺘﺪ ﺑﻪ إﺣﺼﺎﺋﻴﺎً ﺑني اﻟﺒﻠﺪان اﻟﺜﻼﺛﺔ‬ ‫ وﺗﺎﻳﻼﻧﺪ وذﻟﻚ ﺑﺎﺳﺘﺨﺪام‬،‫ وﺟﻨﻮب أﻓﺮﻳﻘﻴﺎ‬،‫اﳌﻤﺮﺿﺎت إﱃ اﳌﻨﺎﻃﻖ اﻟﺮﻳﻔﻴﺔ ﰲ ﻛﻴﻨﻴﺎ‬
‫وﰲ ﺟﻨﻮب أﻓﺮﻳﻘﻴﺎ ﻛﺎﻧﺖ اﻟﻔﺮص اﻷﻓﻀﻞ ﰲ اﻟﺘﻌﻠﻴﻢ أو ﻟﻠﺘﻌﻮﻳﻀﺎت ﻋﻦ اﻟﺨﺪﻣﺔ‬ .‫ﻣﻌﻄﻴﺎت ﻣﺴﺘﻤ ﱠﺪة ﻣﻦ منﻮذج ﺗﺠﺮﺑﺔ اﺧﺘﻴﺎرات ﻣﻨﻌﺰﻟﺔ ﻣﺘﻌﺪﱢدة اﻟﺒﻠﺪان‬
‫ أﻣﺎ ﰲ‬،‫ﰲ اﻷرﻳﺎف ﻫﻲ اﻷﻛرث ﻓﻌﺎﻟﻴﺔ ﰲ زﻳﺎدة اﻻﺟﺘﺬاب ﻟﻠﻮﻇﺎﺋﻒ ﰲ اﻷرﻳﺎف‬ ‫ﺻﻤﻢ اﻟﺒﺎﺣﺜﻮن ﺗﺠﺮﺑﺔ ﺣﻮل اﺧﺘﺒﺎرات ﻣﻨﻌﺰﻟﺔ ذات ﻟﺼﺎﻗﺎت ﺗﻮﺳﻴﻢ‬ ‫اﻟﻄﺮﻳﻘﺔ ﱠ‬
.‫ﺗﺎﻳﻠﻨﺪ ﻓﻘﺪ ﻛﺎن اﻟﺘﻐﻄﻴﺔ اﻷﻓﻀﻞ ﺑﺎﻟﻀامن اﻟﺼﺤﻲ ﻫﻲ ذات اﻟﺘﺄﺛري اﻷﻋﻈﻢ‬ ‫ﻟﻜﻞ ﻣﻦ اﻻﺳﱰاﺗﻴﺠﻴﺎت اﳌﺎﻟﻴﺔ وﻏري اﳌﺎﻟﻴﺔ‬‫ﻹﻋﺪاد منﻮذج ﺣﻮل اﻟﻔﻌﺎﻟﻴﺔ اﻟﻨﺴﺒﻴﺔ ﱟ‬
‫اﻟﻨﺘﻴﺠﺔ ميﻜﻦ ﺗﺼﻤﻴﻢ ﺗﺠﺮﺑﺔ اﻻﺧﺘﻴﺎرات اﳌﻨﻌﺰﻟﺔ ﳌﺴﺎﻋﺪة أﺻﺤﺎب اﻟﻘﺮار‬ ‫ وﺟﻤﻊ اﻟﺒﺎﺣﺜﻮن اﳌﻌﻄﻴﺎت‬.‫اﳌﺼﻤﻤﺔ ﻻﺟﺘﺬاب اﳌﻤﺮﺿﺎت إﱃ اﳌﻨﺎﻃﻖ اﻟﺮﻳﻔﻴﺔ‬ ‫ﱠ‬
‫اﻟﺴﻴﺎﳼ ﻋﲆ اﺧﺘﻴﺎر اﻟﺘﺪاﺧﻼت اﻟﻔﻌﺎﻟﺔ ﰲ اﻟﺘﺼﺪﱢي ﻟﻨﻘﺺ اﻟﻌﺎﻣﻠني ﰲ اﳌﻨﺎﻃﻖ‬ ‫ واﺳﺘﺨﺪﻣﻮا منﺎذج‬،‫ ﻣﻦ اﳌﻤﺮﺿﺎت اﳌﺘﺨﺮﺟﺎت ﰲ ﻛﻞ ﺑﻠﺪ‬300 ‫ﻣﻦ أﻛرث ﻣﻦ‬
‫اﳌﺼﻤﻤﺔ ﳌﻮاءﻣﺔ اﻟﻈﺮوف اﳌﺤﻠﻴﺔ ذات‬ ‫ وﻳﺒﺪو أن ﻣﻀﻤﻮﻣﺎت اﻟﺘﺪاﺧﻼت ﱠ‬.‫اﻟﺮﻳﻔﻴﺔ‬ ‫ﻟﻮﻏﺎرﻳﺘﻤﻴﺔ ﻣﺨﺘﻠﻄﺔ ﻟﺘﺤﻠﻴﻞ ﺷﻐﻞ اﻟﻮﻇﺎﺋﻒ ﰲ اﻷرﻳﺎف وﺗﻮ ﱡﻗﻊ ﺷﻐﻠﻬﺎ ﺗﺤﺖ‬
.‫ﻓﻌﺎﻟﻴﺔ أﻛرث اﺣﺘامﻻً ﻣﻦ اﻷﺳﺎﻟﻴﺐ اﻟﻌﺎﳌﻴﺔ اﻟﻘﻴﺎﺳﻴﺔ‬ .‫ﺗﻮﻟﻴﻔﺎت ﻣﺨﺘﻠﻔﺔ ﻣﻦ اﻟﺤﻮاﻓﺰ‬
Résumé
Comment attirer le personnel infirmier dans les zones rurales? Résultats d’une expérience à choix discrets
réalisée dans plusieurs pays.
Objectif Comparer l’efficacité de différentes politiques visant à attirer le avec des modèles logit mixtes afin de prédire l’acceptation de postes en milieu
personnel infirmier dans les zones rurales au Kenya, en Afrique du Sud et en rural en fonction de différentes combinaisons de mesures incitatives.
Thaïlande, en utilisant les données d’une expérience à choix discrets. Résultats Les préférences du personnel infirmier pour diverses
Méthodes Une expérience à choix discrets a été conçue pour modéliser interventions de réaffectation des ressources différaient significativement
l’efficacité d’incitations financières et non financières visant à attirer le entre les trois pays. Au Kenya et en Afrique du Sud, des possibilités plus
personnel infirmier dans les zones rurales. Dans chaque pays, des données ont intéressantes sur le plan éducatif ou des primes de ruralité seraient les
été collectées auprès de 300 élèves infirmiers en fin d’études, puis analysées incitations les plus efficaces pour améliorer le recrutement de personnel
80
en milieu rural, tandis qu’en Thaïlande, c’est une meilleure couverture par pour remédier aux pénuries de personnel dans les zones rurales. Des
l’assurance maladie qui aurait le plus d’impact. interventions adaptées aux conditions locales seront probablement plus
Conclusion Des expériences à choix discrets peuvent être conçues pour efficaces que des approches standardisées, définies au niveau mondial.
aider les décideurs politiques à définir les interventions les plus efficaces
Resumen
Intervenciones de política para atraer a las enfermeras a las zonas rurales: modelo de elección discreta
multinacional
Objetivo Evaluar la eficacia relativa de diferentes políticas para atraer a entre los tres países. En Kenya y Sudáfrica, unas mejores oportunidades
las enfermeras a zonas rurales en Kenya, Sudáfrica y Tailandia utilizando educativas o la instauración de subsidios rurales serían la fórmula más
los datos obtenidos mediante un modelo de elección discreta (MED). eficaz para aumentar la ocupación de los puestos rurales, mientras que
Métodos Se diseñó un MED con etiquetas para modelizar la eficacia en Tailandia se conseguiría el máximo impacto ampliando la cobertura
relativa de la aplicación de estrategias financieras y no financieras para del seguro de enfermedad.
atraer a las enfermeras a las zonas rurales. Se recogieron datos de más Conclusión Es posible diseñar MED que ayuden a las autoridades a
de 300 estudiantes de enfermería al término de la carrera en cada país, y elegir las intervenciones más eficaces para hacer frente a la escasez de
se aplicaron modelos logit mixtos para analizar y predecir la ocupación de personal en las zonas rurales. Los paquetes de intervenciones adaptados
los puestos rurales en respuesta a distintas combinaciones de incentivos. a las condiciones locales tienen más probabilidades de ser eficaces que
Resultados Las preferencias de las enfermeras ante diferentes los enfoques mundiales normalizados.
intervenciones en materia de recursos humanos difirieron significativamente
References
1. Dussault G, Franceschini MC. Not enough there, too many here: understanding 18. Penn-Kekana L, Blaauw D, Tint KS, Monareng D, Chege J. Nursing staff dynamics
geographical imbalances in the distribution of the health workforce. Hum Resour and implications for maternal health provision in public health facilities in the
Health 2006;4:12. doi:10.1186/1478-4491-4-12 PMID:16729892 context of HIV/AIDS. Johannesburg: Centre for Health Policy; 2005.
2. Travis P, Bennett S, Haines A, Pang T, Bhutta Z, Hyder AA et al. Overcoming 19. Mangham LJ, Hanson K. Employment preferences of public sector nurses
health-systems constraints to achieve the Millennium Development Goals. Lancet in Malawi: results from a discrete choice experiment. Trop Med Int Health
2004;364:900–6. doi:10.1016/S0140-6736(04)16987-0 PMID:15351199 2008;13:1433–41. doi:10.1111/j.1365-3156.2008.02167.x PMID:18983274
3. Lehmann U, Dieleman M, Martineau T. Staffing remote rural areas in middle- and 20. Adano U. The health worker recruitment and deployment process in Kenya: an
low-income countries: a literature review of attraction and retention. BMC Health emergency hiring program. Hum Resour Health 2008;6:19. PMID:18796157
Serv Res 2008;8:19. doi:10.1186/1472-6963-8-19 PMID:18215313 21. Reid SJ. Compulsory community service for doctors in South Africa – an evaluation
4. Dolea C, Adams O. Motivation of health care workers – review of theories and of the first year. S Afr Med J 2001;91:329–36. PMID:11402906
empirical evidence. Cah Sociol Demogr Med 2005;45:135–61. PMID:15938446 22. Wibulpolprasert S, Pengpaibon P. Integrated strategies to tackle the inequitable
distribution of doctors in Thailand: four decades of experience. Hum Resour Health
5. Willis-Shattuck M, Bidwell P, Thomas S, Wyness L, Blaauw D, Ditlopo P. Motivation
2003;1:12. doi:10.1186/1478-4491-1-12 PMID:14641940
and retention of health workers in developing countries: a systematic review. BMC
23. World Health Organization. World health statistics 2006. Geneva: WHO; 2006.
Health Serv Res 2008;8:247. PMID:19055827
24. Hensher D, Rose JM, Greene WH. Applied choice analysis: a primer. New York:
6. Grobler L, Marais BJ, Mabunda SA, Marindi PN, Reuter H, Volmink J. Interventions
Cambridge University Press; 2005.
for increasing the proportion of health professionals practising in rural and other
25. Kuhfeld WF. Marketing research methods in SAS. Experimental design, choice,
underserved areas. Cochrane Database Syst Rev 2009;1:CD005314. conjoint and graphical techniques. Cary: SAS Institute; 2009.
7. Wilson NW, Couper ID, De Vries E, Reid S, Fish T, Marais BJ. A critical review of 26. Swait JD, Louviere JJ. The role of the scale parameter in the estimation and
interventions to redress the inequitable distribution of healthcare professionals to comparison of multinomial logit models. J Mark Res 1993;30:305–14.
rural and remote areas. Rural Remote Health 2009;9:1060. PMID:19530891 27. Hensher D, Louviere JJ, Swait JD. Combining sources of preference data. J
8. Chopra M, Munro S, Lavis JN, Vist G, Bennett S. Effects of policy options for human Econom 1999;98:197–221.
resources for health: an analysis of systematic reviews. Lancet 2008;371:668–74. 28. Hensher D. Empirical approaches to combining revealed and stated preference
9. Dolea C, Stormont L, Shaw D, Zurn P, Braichet JM. Increasing access to health data: Some recent development with reference to urban mode choice. Res Transp
workers in remote and rural areas through improved retention. Geneva: World Econ 2008;23:23–9. doi:10.1016/j.retrec.2008.10.003
Health Organization; 2009. 29. Rose JM, Hensher DA, Caussade S, de Dios Ortúzar J, Jou R. Identifying
10. Mills A, Gilson L, Hanson K, Palmer N, Lagarde M. What do we mean by rigorous differences in willingness to pay due to dimensionality in stated choice experiments:
health-systems research? Lancet 2008;372:1527–9. PMID:18984174 a cross country analysis. J Transp Geogr 2009;17:21–9.
11. Gilson L, Mills A. Health sector reforms in sub-Saharan Africa: lessons of the last 30. Chow GC. Tests of equality between sets of coefficients in two linear regressions.
10 years. In: Berman P, ed. Health sector reform in developing countries. Boston: Econometrica 1960;28:591–605. doi:10.2307/1910133
Harvard University Press; 1996. 31. Kolstad JR. How can rural jobs be made more attractive to Tanzanian health
12. Lagarde M, Blaauw D. A review of the application and contribution of discrete workers? Results from a discrete choice experiment. Bergen: Department of
choice experiments to inform human resources policy interventions. Hum Resour Economics, University of Bergen; 2008.
Health 2009;7:62. doi:10.1186/1478-4491-7-62. PMID:19630965 32. World Health Organization. World health report 2006: working together for health.
13. Louviere JJ, Hensher DA, Swait JD. Stated choice methods: analysis and Geneva: WHO; 2006.
applications. Cambridge: Cambridge University Press; 2000. 33. Chen L, Evans T, Anand S, Boufford JI, Brown H, Chowdhury M et al. Human
14. Ryan M, Farrar S. Using conjoint analysis to elicit preferences for health care. BMJ resources for health: overcoming the crisis. Lancet 2004;364:1984–90.
2000;320:1530–3. doi:10.1136/bmj.320.7248.1530 PMID:10834905 34. Street D, Burgess L, Vinay R, Louviere J. Designing discrete choice experiment
15. Ryan M, Bate A, Eastmond CJ, Ludbrook A. Use of discrete choice experiments to for health care. In: Ryan M, Gerard K, Maya-Amaya M, eds. Using discrete choice
experiments to value health and health care. Dordrecht: Springer; 2008.
elicit preferences. Qual Health Care 2001;10(Suppl 1):i55–60. PMID:11533440
35. Louviere J, Lancsar E. Choice experiments in health: the good, the bad, the ugly
16. Ryan M, Gerard K. Using discrete choice experiments to value health care
and toward a brighter future. Health Econ Pol Law 2009;4:527–46.
programmes: current practice and future research reflections. Appl Health Econ
36. Kruijshaar ME, Essink-Bot ML, Donkers B, Looman CW, Siersema PD, Steyerberg
Health Policy 2003;2:55–64. PMID:14619274
EW. A labelled discrete choice experiment adds realism to the choices presented:
17. Scott A. Eliciting GPs’ preferences for pecuniary and non-pecuniary job
preferences for surveillance tests for Barrett esophagus. BMC Med Res Methodol
characteristics. J Health Econ 2001;20:329–47. PMID:11373834 2009;9:31. doi:10.1186/1471-2288-9-31 PMID:19454022
81
Table 1. Key indicators used in discrete choice experiment for assessing the
effectiveness of policies to attract nurses to rural areas in Kenya, South
Africa and Thailand, 2006
Indicator Kenya South Africa Thailand

National population × 106 36.5 48.3 63.4
Population in rural areas (%) 79 40 67
GNP per capita (PPP$) 1 470 8 900 7 440
Total expenditure on health (% of GDP) 4.6 8.6 3.5
Per capita expenditure on health (PPP$) 105 869 346
No. of nurses 37 113 184 459 172 477
No. of nurses per 10 000 population 12 41 28
No. of doctors 4 506 34 829 22 435
No. of doctors per 10 000 population 1 8 4
Life expectancy at birth (years) 53 51 72
IMR (per 1 000 live births) 79 56 7
MMR (per 100 000 live births) 560 400 110
HIV infection prevalence (%) 6.1 16.6 11.4
GDP, gross domestic product; GNP, gross national product; HIV, human immunodeficiency virus; IMR, infant
mortality rate; MMR, maternal mortality ratio; PPP$, purchasing power parity dollar.
Data from the World Health Organization.23
Table 2. Methods for selecting attributes included in discrete choice experiment for assessing the effectiveness of policies to attract
nurses to rural areas in Kenya, South Africa and Thailand, 2006
Method Objective(s) Details

International literature review • Identify strategies that have been used to attract • Systematic search and review of relevant literature
health workers to underserved areas in HICs and
LMICs
• Review evidence on the effectiveness of existing
strategies
Review of the HR DCE literature • Identify attributes that have been used in previous • Systematic search and review of relevant literature12
HR DCEs
• Review relative importance of selected attributes
Review of ministry of health HR • Identify policy interventions that have been • Systematic search and review of HR policy documents
policy documents implemented or proposed in each country in each country
Key informant interviews with • Identify policy interventions implemented or • Semi-structured interviews with 3–5 senior policy-
relevant policy-makers proposed in each country. makers responsible for HR in ministry of health in each
• Identify planned policy strategies for the future country
• Assess feasibility of strategies tried in other
countries
Focus group discussions with • Obtain student suggestions on important job • Focus group discussion in each country with 6–9
nursing students characteristics and required policy interventions final year nursing students from nursing college not
selected for final study
• Semi-structured discussion on factors considered in
job choices and attitudes towards working in rural
areas
Pilot study • Test understanding of DCE task and tool format • 10–20 nursing students in each country from different
• Confirm understanding and relevance of DCE colleges completed draft tool
attributes and levels • Semi-structured discussion on responses to DCE tool
and attributes
• DCE results analysed for consistency
DCE, discrete choice experiment; HIC, high-income country; HR, human resource; LMIC, low- and middle-income country.
A Bull World Health Organ 2010;88:350–356 | doi:10.2471/BLT.09.072918
82
HEALTH POLICY AND PLANNING; 14(2): 115–126 © Oxford University Press 1999
Political analysis of health reform in the Dominican Republic

AMANDA GLASSMAN1 MICHAEL R REICH,2 KAYLA LASERSON,3 AND FERNANDO ROJAS4
1Inter-American Development Bank, Washington DC, USA, 2Harvard School of Public Health, Boston, USA, 3US
Public Health Service, Centers for Disease Control and Prevention, Atlanta, USA, and 4Medical Doctor, Dominican
Republic
This article examines the major political challenges associated with the adoption of health reform proposals,
through the experience of one country, the Dominican Republic. The article briefly presents the problems of
the health sector in the Dominican Republic, and the health reform efforts that were initiated in 1995. The
PolicyMaker method of applied political analysis is described, and the results of its application in the Domini-
can Republic are presented, including analysis of the policy content of the health reform, and assessment of
five key groups of players (public sector, private sector, unions, political parties, and other non-governmental
organizations). The PolicyMaker exercise was conducted in collaboration with the national Office of Techni-
cal Coordination (OCT) for health reform, and produced a set of 11 political strategies to promote the health
reform effort in the Dominican Republic. These strategies were partially implemented by the OCT, but were
insufficient to overcome political obstacles to the reform by late 1997. The conclusion presents six factors
that affect the pace and political feasibility of health reform proposals, with examples from the case of the
Dominican Republic.
Introduction an interaction between international lending agencies and

national government bodies. The proposed policies usually
In the early 1990s, countries throughout Latin America initi- seek to build a self-sustaining national health care financing
ated the process of reforming their health sector policies. system as the primary goal. Secondary goals include greater
These efforts received unprecedented levels of financial coverage for basic health services at a lower cost per person,
support from multilateral institutions, especially the World rationalized decision-making within public sector health
Bank and the Inter-American Development Bank (IDB). agencies, institutional reform, and expanded access for dis-
World Bank lending in health in Latin America quadrupled advantaged populations.
in a five-year period, to over US $900 million a year in 1995
(World Bank 1995). IDB activity added to this total, increas- Health sector reforms are politically problematic. In many
ing total lending in the Latin American health sector to nearly countries considering reform, the most powerful health sector
US $2.2 billion in 1995, with the expectation that IDB lending actors are often satisfied with the status quo – despite serious
in education and health would continue to grow. problems in the distribution of health services, quality of care,
patterns of utilization, efficiency, and equity. Moreover, the
Yet little attention has been paid to the major political chal- proposed policy changes are often perceived as politically and
lenges associated with the adoption and implementation of economically painful decisions in the short term. One of the
health reform proposals. Reform is a profoundly political most important and complex problems in the process of
process that affects the allocation of resources in society, and health reforms is the management of these short-term, con-
often imposes significant costs on well-organized, politically centrated costs, and of the powerful groups affected.
powerful groups. This article presents a method of political
analysis for health sector politics, and identifies key patterns Reform proposals create the perception that a major redistri-
in the politics of health reform proposals. The article uses the bution of the benefits and costs within the health system will
experience of one country, the Dominican Republic, to illus- occur, but how and when that redistribution will occur is
trate the political challenges of health reform. unclear. In contrast to education reform, which usually entails
increasing budgets, building new schools, and hiring teachers,
Health sector reform has been variously defined. In this paper, health reform seeks to radically alter the social contract
we define health sector reform as those activities undertaken between citizens and the government, changing physician
cooperatively between the international development banks payment schemes, introducing patient payments, and limiting
and a national government to alter in fundamental ways the reimbursable services to affluent social groups. Politically,
nation’s health financing and health provision policies. health reform proposals resemble structural adjustment poli-
cies, but without the national mandate for change accorded to
This limited definition focuses on the processes around the adjustment.1 In addition, health reform policies confront
design and adoption of new health policies, occurring through more complex obstacles in implementation, compared to
83
116 Amanda Glassman et al.
structural adjustment policies, because of the nature of the taking into account the interactions among policies, players,
decisions and institutions involved. and positions. Finally, the process strengthens the reform
group’s capacity to advocate for reform policies.
Both multilateral institutions and national health reform
teams have experienced some difficulty in understanding and Put another way, this method helps policy-makers and policy
navigating the political economy of health sector reform. This analysts do what they should do anyway: systematically
paper reports on an effort to try to improve the understand- analyze the support and opposition for a proposed policy;
ing and the navigation. The paper first reviews a method for consult with the major stakeholders on their views; analyze
applied political analysis. We then explore the background of opportunities and obstacles to change; design a set of creative
the Dominican health sector, and apply the method to the and effective strategies for change; and assess and track the
reform proposals in the Dominican Republic. Finally, we processes of implementing those strategies.
draw some general conclusions about the political processes
of health sector reform. In the case of the Dominican Republic, three consultants
(AG, KL, MRR) were financed by the IDB to work with the
government’s health reform group (headed by FR) to define
Applied political analysis
the policy, interview key players, and propose strategies. The
The method of applied political analysis known as Policy- analysis was carried out by a team of ‘insiders’ and ‘outsiders’,
Maker was used in this project to assist decision-makers in in order to minimize analyst bias through group discussion
analyzing and managing the politics of health reform in the and collective judgment. As with any social science method-
Dominican Republic. The method provides a systematic ology, however, the method cannot eliminate unpredictable
analysis of the probable consequences of policy reform elements in the policy-making process.
efforts, the positions of support and opposition taken by key
players, the political, financial, and other interests of key In this case, 35 guided interviews with key figures in the health
players. It then assists decision-makers in initiating the sector were conducted in the Dominican Republic between
process to design strategies for managing the politics of policy July and November 1995. Both published and unpublished
reform (Reich 1996; Reich and Cooper 1996). documents were collected and reviewed, and the national
press was monitored closely for one year following the inter-
In the software format, PolicyMaker uses a series of matrices views. This paper presents some results of the analysis, and
to guide the analyst through five steps of political strategizing. the conclusions reached.
The framework prompts the analyst to: (1) define the content
of the policy under consideration; (2) identify political
The health sector in the Dominican Republic
players, their interests and relationships; (3) analyze oppor-
tunities and obstacles to the policy in the political environ-
The Dominican health sector
ment; (4) design political strategies; and, (5) assess the
potential and actual impacts of proposed strategies. The The Dominican health sector exhibits a number of systemic
analyst can complete each matrix, or can be selective accord- problems, typical of many countries in Latin America. These
ing to the objectives of the analysis. problems include inadequate financing, low coverage,
inequitable distribution of services, an emphasis on curative
The method assists policy analysts with the political dimen- care, fragmented vertical programming, redundant and under-
sions of policy change in five ways. First, the method pro- used facilities, inefficient institutions and personnel, corrupt
vides a systematic assessment of the political environment in bureaucracies, and unregulated private health services.
which health sector reform policies are formulated and
implemented. At a minimum, the method provides a tool to By the early 1990s, many Dominicans felt that the health
describe the political dimensions of a policy decision, and sector was in crisis: preventive and curative services were low-
then to organize and prepare the data for analysis. Second, quality, irregular, concentrated in the capital and in tertiary
the method provides practical assistance in the design of care facilities, and highly inefficient. The sector had experi-
political strategies. The software includes a tool box of 31 enced one of the largest and longest (8 months) strikes in the
‘expert-suggested’ political strategies that can be modified country’s history, with the Dominican Medical Association
by the user. Third, if conducted by a team analysis, the (Asociación Médica Dominicana – AMD) showing its power
method helps to make explicit the team’s assumptions about to control the functioning of government health services. As
how a new policy will be adopted, and forces the team to in many other countries, doctors work in both the private and
explain and justify those assumptions. This reflective public sectors, usually squeezing their public sector obli-
process helps to enhance the coherence and feasibility of the gations, where they are poorly paid, in order to attend to their
policy. Fourth, if conducted with interviews of key stake- private practices. Remuneration is not connected to per-
holders, the method helps validate the reform group’s performance. Physicians working in public hospitals regularly
ceptions about other stakeholders, and helps the reform refer their patients to their private clinics for procedures, and
team view the policy from the perspectives of other stake- some physicians use public sector facilities to conduct for-
holders. Fifth, the process of performing the analysis helps profit procedures.
create a sense of common language and mission for some
reform groups. The analysis encourages reform groups to The Secretariat of Health (SESPAS) and the Social Security
make their strategies explicit, and rethink the strategies, Institute (IDSS), the largest institutional actors in the public
84
Applying political analysis 117
sector, have shown little capacity to respond to the major as part of pre-electoral political positioning by PLD, rather
problems in the health sector. Both institutions have been than a genuine reform effort. Debate around the bill lasted
used extensively for political patronage and have limited nearly a year, and then died.
technical capacity. The average stay of a Health Secretary is
less than eight months. Although almost 60% of the popu- At about the same time, in January 1995, an executive decree
lation falls below the poverty line, subsidized government ser- created the National Health Commission (CNS) with a
vices through the Ministry cover only 35% of the population mandate to promote ‘modernization’ of the health sector.
(Santana and Rathe 1994). SESPAS is organized vertically by The Office of Technical Coordination (OCT) was created to
programme, and focuses mostly on curative, tertiary level design a health reform plan under the auspices of the CNS.
care.2 IDSS, with its own networks of hospitals and clinics, The OCT operated primarily with project funds from the IDB
covers only 6% of the Dominican population. Many busi- and the World Bank, with occasional assistance from the Pan-
nesses now pay double for health care – an obligatory American Health Organization (PAHO), the US Agency for
payment to the IDSS, plus payments to cooperatives of International Development (USAID), and other donors.
private providers for health insurance. Evasion of the IDSS Initially, the OCT operated under the CNS; however, in 1997,
scheme is widespread. As a result, the private sector has the OCT was shifted organizationally to the SESPAS,
grown rapidly but with minimal regulation. The private sector although the OCT maintained separate offices in Santo
now represents the primary source of health financing and Domingo away from the ministry.
service provision in the Dominican Republic. While health
service infrastructure is plentiful in both the public and In this third wave of health reform, the OCT was asked to
private sectors, access is highly inequitable since it depends draft a reform ‘white paper’ with technical assistance from
on an individual’s ability to pay. According to the 1991 Demo- consultants in the first half of 1995. The ‘white paper’ was to
graphic and Health Survey, approximately 60% of persons serve as the basis for assessing the technical feasibility of
who reported a serious illness in the past month did not seek various reform initiatives and as a first attempt to change the
medical care, principally for economic reasons. discourse on health sector transformation in the country.
Reform studies were commissioned by the OCT from
national and international consultants using non-reim-
Recent efforts at health reform
bursable technical cooperation monies from the IDB and
The Dominican Republic has experienced several waves of donated funds from the Government of Japan through the
policy responses to problems in the health sector. In Novem- World Bank and the UNDP.
ber 1992, SESPAS received funding from the United
Nations Development Program (UNDP) to undertake a Reform studies addressed the following topics, in chronologi-
project of ‘modernization’ of the Dominican health system. cal order: (1) hospital autonomy; (2) SESPAS re-organization;
For more than a year, a group of Dominican professionals (3) SESPAS financing systems; (4) IDSS reorganization; (5)
elaborated policy proposals for reforms, in consultation with prepaid health systems (igualas); (6) incorporating NGOs into
health sector players and with technical assistance from health sector reform; (7) survey on use of and satisfaction with
UNDP. Late in 1993, the results were disseminated to poli- health services; (8) financing of public expenditure in health;
cymakers. The proposals included recommendations to (9) health expenditure module as part of the DHS; (10) per-
rationalize human resources policies, including the intro- sonnel administration systems; (11) burden of disease and
duction of new forms of physician payment, and a ‘new basic package definition; (12) pharmaceutical and supply
model of care’. For political and financial reasons, including stocks at SESPAS; (13) accreditation and re-equipping health
the absence of a forum in which to continue reform discus- services; (14) decentralization of SESPAS; (15) design of a
sions, no follow-on activities resulted from this first wave of new social security system; and, (16) a legal and regulatory
reform efforts. framework for social security reform in the Dominican
Republic (OCT 1995a). As the product of intensive collabor-
The second wave occurred between October and May 1995, ation between the OCT and the multilateral development
when the health commission of the national legislature intro- banks, with a great deal of autonomous leadership from the
duced a ‘National Health Law’, written by deputies from the OCT, the studies were intended to lay the groundwork for
Partido de la Liberación Dominicana (PLD) with technical implementing reform activities in these 16 specific areas. The
assistance from SESPAS and PAHO advisors. While recog- OCT has monitored the progress of and payment for the 16
nizing many of the problems of the sector, the proposed law studies.
read like a long list of special-interest programmes. Each
disease and programme priority was included, based largely In addition, the OCT expected to manage the process of
on a ‘traditional’ public health paradigm, while little attention reform. For example, the OCT was expected to secure high-
was paid to the methods for financing health services, the level political support for reforms among government
roles of existing health sector institutions, or the regulation of leaders, especially the Secretary of Health, the Director of
the private sector. The bill was intended to replace the Social Security, and the President of the Republic. More
Dominican Republic’s ‘Sanitary Code’, which contains broadly, the OCT was intended to prepare government agen-
special provisions for regular salary raises for doctors cies, other interest groups, and society at large for accepting
working in the public sector. Although these provisions have and implementing the reforms. The reform studies were
never been implemented (since 1956), the new law was intended to play a major role in this preparation, and usually
opposed by the AMD (OCT 1995b). Some perceived the bill involved staff members from the affected institutions.
85
The OCT ‘white paper’ recommended the following reforms: stable bureaucracy in charge of the health system. The
(1) the separation of financing from provision of services bureaucracy, however, is also very conservative, not well
within SESPAS and IDSS; (2) the massive expansion of IDSS trained, accustomed to certain privileges (to offset the low
coverage; (3) the definition of a cost-effective basic package salary) and fearful for their jobs. In this sense, any change in
of services to be financed by the public sector; (4) hospital the system that could increase the degree of formal control or
autonomy; and, (5) linkage of productivity and incentives in the grade of institutionalized procedures implies a significant
the health work force (e.g. through physician contracts). This reduction in the discretionary power of the bureaucracy. The
set of recommendations, published as Salud: Una Visión del bureaucracy, therefore, has tended to oppose reform in prin-
Futuro, was taken as the ‘policy’ for this applied political ciple and in practice.
analysis (OCT 1995a).
Analysis of the 1995 OCT reform proposal
In 1995, the OCT had seven staff members, primarily techni-
cal, with one public relations person part-time. The CNS This section analyzes the political circumstances around
included 33 health sector ‘actors’ and had no clearly defined health sector reform using the PolicyMaker method. The
decision-making structure, but had taken most decisions analysis uses the OCT ‘white paper’ of July 1995 as the reform
through voting. All votes (through November 1995) were proposal, and considers the OCT its primary client. Two
unanimous, and voting was initiated by the chair of the CNS, major objectives are: (1) to assess the political feasibility of
the Secretary of Health. the reform proposal, as of mid-1995, and (2) to propose strat-
egies that could enhance the political feasibility of the reform
process. Before designing strategies, PolicyMaker analyzes
Political climate
policies along three dimensions: policy content, players, and
In June 1995, the Dominican Republic was one of the poorest environment (opportunities and obstacles). These three
countries in Latin America. In 1988, it had the third lowest dimensions frequently intersect. A player’s position may
Gross Domestic Product (GDP) per capita in the Americas, emerge out of a complex combination of its reactions to the
after Haiti and Bolivia. Despite respectable economic growth policy content, the player’s interests, relative power, and
rates in the 1980s, the economic crisis (followed by structural relationships with other policy actors, and the internal and
adjustment policies) impoverished the country in the 1990s. external organizational environment.
The Dominican Republic was one of the last of the aging dic-
tatorships in Latin America. When health sector reform For this case study, we first review the content of the reform
design began, Joaquín Balaguer had been president of the policy under consideration. Second, we analyze the players,
country for more than 50 years, off and on. The political by exploring the interests, power, and position of the domi-
system can be categorized as ‘clientelistic’. As one study of nant policy players, with reference to relevant aspects of the
Dominican political culture put it, ‘The Dominican political reform proposal. Third, we review the external opportunities
system is theoretically organized along formal democratic and obstacles that the OCT faced in the policy environment.
principles, however, it is essentially informal operationally’ Finally, we present the strategies that were designed in the
(AG translation, Cross-Beras 1985). It is a limited pluralist Dominican Republic, using the PolicyMaker method, for
system without accountability, and without an explicit politi- OCT to consider in managing the reform process.
cal ideology. Most decisions, national or otherwise, were
taken by the President personally.
(1) Policy content
Although SESPAS is the major public provider of health ser- Policy proposals for health sector reforms supported by the
vices, in recent years the Secretariat of the Presidency has multilateral development banks are similar across Latin
become a significant source of health financing, especially for America, responding to similar challenges within public
the purchase of plant, equipment, and supplies for SESPAS health bureaucracies. At the time of the analysis, proposals
facilities. In 1991, for instance, the Secretariat of the Presi- followed the ideas presented in the 1993 World Development
dency was the source of 38% of public expenditures on health Report, and built on the World Bank’s seminal 1987 policy
(IDB 1997). An unpublished study on the health sector found study, Financing Health Services in Developing Countries
that SESPAS decisions on even micro-level budgeting and (Akin et al. 1987). The reforms have usually included three
personnel issues lay with the President of the Republic (Perez levels of policy goals and mechanisms.
Uribe et al. 1974). In June 1996, the Dominican Republic held
democratic elections which resulted in the election of Leonel First, the reforms define broad governing principles. In the
Fernández, a young US-educated lawyer. Dominican Republic, the principles were universal access,
equity, solidarity, quality, freedom of choice, efficiency,
In contrast to the longevity of the Presidency, other political efficacy, and transparency.
leaders have a short duration in office. Few political
appointees are able to acquire effective capacity to manage Second, strategic guidelines are developed that set out more
the technical or organizational challenges of their policy specific parameters for a restructured health system. In the
domain. Between 1930 and 1974, 37 people served as Secre- Dominican Republic, these guidelines included: (1) the
tary of Health. A similar turnover has affected the director- design of a single system, organized functionally (regulation,
ship of the IDSS: 21 vice ministers in the past two years. This financing, policy, provision); (2) a shift towards preventive
lack of continuing leadership has left the poorly paid but services; (3) a strengthened regulatory role of the state; (4)
86
increased financing for the health system; (5) guaranteed ben- meet longer-term institutional objectives. Decision-making is
efits for affiliates; (6) efficient systems; and, finally, (7) the usually concentrated in an individual, and accountability is
facilitation of social participation in the health system (OCT diffuse. An attempt at regionalization of SESPAS failed and
1995a). These strategic guidelines represent policy goals, but local officials lack authority. Services are poor in quality, and
they do not specify how to achieve the goals, which may have coverage is low. Human and material management is
contradictory objectives. deficient. Nearly all appointments are made at the central
level by the Minister (or the President) without the know-
The third level provides more specific policy mechanisms. In ledge of division chiefs or facility managers. Mismatches
the Dominican Republic, policy mechanisms were defined in result between human resource supply and service demand.
four areas, according to the OCT in 1995. Similar proposals For example, several SESPAS facilities have up to 50% more
can be found in other Latin American countries undergoing medical personnel than necessary to meet demand, while
health reform: other facilities are closed due to lack of personnel (IDB
1997).
1. development of a new model of rationally determined,
publicly financed health services that would ensure a basic Considerable confusion exists concerning the role of SESPAS
basket of cost-effective interventions, namely preventive within the sector because the Secretariat of the Presidency
services, available to the entire population; administers nearly one-third of government health spending,
2. decentralization and restructuring of the ministry of health and little coordination of any kind exists among public sector
and the social security institute; health institutions. Linkages between the public and private
3. transformation of the state’s role from direct service sector are absent. Each institution makes policies, sets plans
provider to financer and regulator; and, and implements programmes more or less independently.
4. creation of managed competition through government This, in turn, contributes to stratified access to health care,
contracting with both public and private sector providers. concentration of resources in large cities, duplication of infra-
structure and service provision, and overlapping financial
arrangements. For example, household surveys show that
(2) Players
50% of IDSS enrollees do not use IDSS services, while 50%
Assessment of political feasibility requires an analysis of the of users of IDSS services are not enrolled in IDSS. In some
stakeholders – the political actors affected by or affecting a rural areas, NGOs and SESPAS provide similar services to
given policy. These actors are called the ‘players’ in Policy- the same population groups. A significant percentage of the
Maker. The field of policy analysis has not produced a single poor bypass ‘free’ SESPAS facilities, seeking care at fee-for-
or simple method for assessing the characteristics of players service private clinics.
involved in policy change (Reich 1996). PolicyMaker, there-
fore, combines a number of analytical methods. The basic
Hospital autonomy
analysis requires an assessment of each player’s position on
the policy (support, opposition, or non-mobilized position), After the public release of the ‘white paper’ in 1995, the Sec-
power (resources available to use in the policy debate), and retary of Health and the OCT were accused of ‘privatizing’
intensity of position (high, medium, or low, depending on the the health sector. While it is true that the management of pub-
willingness to use available resources in the policy debate). In licly owned hospitals through contracts is not privatization,
this analysis, a player can be either an organization or an indi- especially since the government would guarantee subsidies
vidual, though the analyst might consider weighting these for preventive services and basic ambulatory care (F. Rojas
groups differently, according to their power resources. 1995), elements of autonomization can have (and can be per-
ceived as having) the same political and social effects as pri-
In our analysis of health reform in the Dominican Republic, vatization has had on other state industries. That is, hospital
the players were divided into five key groups: public sector, autonomization does imply that current government employ-
private sector, unions, political parties, and other non- ees become employees working under contract, without a
governmental organizations. lifetime guarantee, which allows for discretionary firing and a
complete break in the traditional relationship between the
state and physicians. Hospital autonomization also implies
Public sector: SESPAS and IDSS
that public sector hospitals would compete with the private
The reform proposal has profound implications for the public sector to provide the basic package of services; that the hos-
sector, especially the Ministry of Health (SESPAS) and the pital director would have discretion over budgets, and that
Dominican Social Security Institute (IDSS). Political resist- the central SESPAS would not; and that any services pro-
ance in the public sector was anticipated particularly around vided in excess of basic ambulatory care would not be subsi-
the issues of hospital autonomy and institutional restructur- dized by the government.
ing.
For all these reasons, the SESPAS bureaucracy, though not
A 1985 evaluation of SESPAS described it as a government fully cognizant of the potential implications of the reform,
agency suffering from ‘overall inoperativeness’. SESPAS and was extremely wary of the proposal. And the AMD was
IDSS lack the internal structures, formal lines of command, highly opposed to hospital autonomization, due to the loss of
functional definition, administrative machinery and policy- job security that physicians would face under this system.
making capacity to effectively execute current mandates or to Hospital directors, who stand to gain in status and control,
87
were pleased with the idea, but were not organized. Overall, In 1994, a private think tank and the association of employers
there were serious concerns about the technical capacity of published a plan for health sector reform that proposed the
hospital staffs to manage the process of autonomization and elimination of IDSS. The new IDSS director accepted the
re-orient the hospital to a competitive environment. proposal, but was fired shortly thereafter. The position of the
subsequent directors was not known officially. At the time of
analysis (1995–96), IDSS’s technical office questioned the
Institutional restructuring
value of contracting and seemed to reject the idea of elimi-
The processes of institutional restructuring present serious nating its role as a direct provider of health services.
challenges. For the Health Secretary, restructuring could
mean political suicide if the AMD were to mobilize against
Transformation of the state’s role
the plan. Any benefits from the policy reform are likely to be
long-term and difficult to perceive as tangible. For the Bank-financed health sector reform is meant to transform the
bureaucracy, restructuring is feared, because it would disturb state’s role from direct service provider to financer and regu-
the status quo, create a threat to job security, and upset estab- lator, but the details of this transformation are unclear. There
lished ways of doing things. For physicians, institutional is some ambiguity on how the state becomes ‘financer and
restructuring places the AMD’s organizational autonomy and regulator’. At the time of our interviews in the Dominican
negotiating power at stake. For hospitals, it represents a Republic in 1995, ‘separation of financing and provision’ was
change from the status quo, which is so negative at present, interpreted in the press as the ‘privatization’ of health ser-
that any change is perceived positively. vices provision and created reluctance among political
leadership to support health sector reform with enthusiasm:
IDSS faces many of the same issues. Restructuring for IDSS political leaders of SESPAS (SecSal) and IDSS (IDSSDir)
has similar implications as for SESPAS, but with the added were thus classified as high-power actors in low support of the
nuance that IDSS would be forced to stop its direct service white paper. In the pre-presidential election period (Septem-
provision altogether. Under the reform, all financing of ber 1995 to May 1996), this reluctance was expected. In the
health services for formal and informal sector employees post-election period, the issues were still unresolved. A dis-
would be provided through the IDSS. Many observers tinction was also made between political leadership and
outside of IDSS were surprised that IDSS could be con- SESPAS and IDSS bureaucracies in the analysis, as these
sidered a responsible controller for funds, given its history of groups had contrasting interests in the process. The SESPAS
political patronage. Most likely, the reform proposal would bureaucracy (SESPBur) was considered high-power and low-
be revised to remove IDSS (not government) from the col- support at the time of the white paper, while the IDSS
lection and disbursement of funds.3 However, if this were to bureaucracy (IDSSBur) was medium-power and low-opposi-
occur, then IDSS would have few tasks remaining in health tion, with potential to move to high opposition in the near
services. future.
This is not the first attempt to restructure the IDSS. Created

Private sector: private clinics and igualas
in 1948 during a wave of Bismarckian-style social security,
the IDSS was primarily a response to pressures from the The private sector is highly opposed to regulation, having
cane-cutters union. Its political patronage functions have operated profitably during the progressive decay of public
persisted over time. In 1982, the President, three of the major sector services. Approximately 15% of Dominicans, pri-
trade unions, and the main employers’ association, with marily formal sector employees, belong to employer pre-
assistance from the International Labor Organization, payment plans, known as igualas médicas, which cover a basic
endorsed a legislative draft to expand IDSS health care package of ‘equal’ services. The plans compete on price,
coverage to all salaried workers in both the private and service quality, and completeness. In principle, consumers of
public sectors regardless of salary level (removing a cut-off iguala health plans would welcome government financing of
that exempted most white and blue collar workers from these services, but would resist any attempt to be incorpor-
obligatory payment). The reform presented a politically ated into government-provided services. If formal sector
viable solution, given the power of the AMD. A last-minute employees were obliged to contribute to the public sector (in
revision of the bill, prepared by a group of civil servants who order to finance the rest of the health system), then formal
did not want to be required to pay into IDSS instead of their sector employers who are not already evading payment
current private insurers, restricted care to IDSS-owned facil- would be expected to resist further. This practice (the so-
ities, which were notoriously poor quality. This revision elim- called doble cotización or double payment) has been identi-
inated free choice to use private facilities, a central element fied as an agenda item for small and large business
of the bill. The strong opposition of the private sector then organizations.
defeated the bill in Congress. High-salaried employees, who
would have been incorporated into IDSS after the projected Employer discontent (and evasion), along with the private
elimination of the salary ceiling, also opposed the amend- sector’s resistance to regulation and the formal sector con-
ment because they would have been required to pay contri- sumer’s aversion to government-related (financed or pro-
butions but did not want to use IDSS services. The bill’s vided) services, make the decision to move towards a
defeat strengthened the private sector and contributed to managed competition model difficult for the government.
further erosion of the IDSS public image (Mesa-Lago 1978, While the private sector is expected to gain under managed
1989, 1992). competition, the igualas would probably be more profitable if
88
they can continue to restrict plan entry to the relatively to make concessions on most issues, rather than delegating
healthy and wealthy, which would probably occur more easily this task to the Secretary of Health, thereby undermining the
without reform. Private health sector players (private efforts of the Secretary of Health to be firm with the AMD.
clinic/iguala owners – PrivClin – and employers – This position agrees with Dominican political culture, as
EMPLOYER), while expressing basic agreement with the described earlier, where power is concentrated in the Presi-
reform’s principles, were lukewarm towards the white paper, dent.
and based on an analysis of player interests were classified as
moderately opposed, high-power players. These expressions of position and power do not necessarily
indicate that the President is fully opposed to the OCT ‘white
paper’. However, they do indicate that he is not supportive,
Unions: the AMD
and that he will not serve as an advocate. The Secretary of
A key feature of the Dominican health sector is the near- Health, who was replaced in January 1997, could be an impor-
omnipotence of the physicians’ association (AMD).4 In the tant factor in the reform process. Thus far, however, the Sec-
past, every negotiation between the government and the retary has been remarkably uninvolved in planning for
AMD has ended with government concessions. As part of this reform. The passivity of Balaguer’s last Secretary of Health
process, the AMD strikes frequently and for long periods of could be linked to a protracted ‘lame duck’ period prior to the
time. In 1996, for example, the AMD held an eight-month elections. In the case of the new government, the Secretary’s
strike for higher wages and increased job security. This strike tepid support is notable and could have significant conse-
came after an extremely generous settlement, in which the quences for feasibility. The President (PRES) at the time of
government promised to double all doctors’ salaries in the analysis was classified as high-power with a non-mobilized
public sector, waive import taxes on vehicles, and provide position. The PLD, currently in office, was classified as a high-
public housing. The strike was perhaps precipitated by the power, low-support player.
government’s inability to finance its health services, much less
provide housing to doctors. During this time, the government
Non-governmental organizations (NGOs)
agreed to pay doctors their salary for the time missed, and
still, the AMD remained on strike pending resolution of the While NGOs were expected to be supportive of reform plans
‘situation’ of IDSS doctors. This situation is particularly to expand coverage to the entire population and provide
deplorable since physicians are supposed to work eight hours more preventive services, the interviews did not find much
a day legally, but typically work only two hours a day and support for reform among NGOs. NGOs initially focused
spend an average of two minutes per patient (Mesa-Lago attention on the creation of a basic package of services using
1992). In addition, they are frequently absent, delay hospital cost-effectiveness criteria. NGOs focusing on preventive care
dismissals, violate rules, and reject any effort to introduce services felt that many elements of equity were not well
planning, set work schedules, or enforce the budget (Mesa- served by an application of cost-effectiveness criteria, which
Lago 1992). were not connected to a concept of health as a right. NGO
staff published press articles criticizing the OCT for using
The AMD is led by an experienced union organizer, and the ‘economic’ criteria where they ‘don’t belong’, that is, in the
Secretary of Health, usually inexperienced in negotiation health sector (O. Rojas 1995a, 1995b). This criticism had the
given his short tenure, is the AMD’s primary target. If the potential of associating reform with particular ‘victims’, such
Secretary is unable to meet the AMD’s demands, the organiz- as children who would not receive emergency interventions
ation has often been able to pressure the President to remove that fall outside of the basic package. Other groups, which
the Secretary. The AMD is also able to mobilize quickly provided specialized forms of care and received government
against journalists and policy-makers who attack their inter- monies, such as the Asociación Dominicana de Rehabilitación
ests publicly. The AMD was considered a high-power actor, or the Liga Dominicana Contra el Cáncer, feared that reform
highly opposed to the white paper in principle and in practice. would decrease resources available to their work. While
NGOs are generally not very influential on the national politi-
cal scene in the Dominican Republic, they have sufficient
Political parties: Fernández and the PLD
resources to access the media, to shape public perceptions of
Leonel Fernández, who was elected President as a member of health reform, and thereby to influence the reform process in
the Partido de Liberación Dominicana (PLD) in 1996, pro- the CNS. NGOs were classified as low-power, low-opposition
duced an elaborately detailed, Bill Clinton-style governing players.
plan. The plan placed health reform at the bottom of a 24-
item list of priorities and left it undefined (Partido de At the time of our analysis (July–November 1995), most politi-
Liberación 1996). During an interview conducted in August cal players were essentially non-mobilized with regard to the
1995 with the current vice-president, Dr Jaime Fernández health reform proposal (the National Health Commission –
Mirabal (then a PLD senator), the reform group was advised CNS; beneficiaries – BENEFS; the press – PRESS; universi-
to stop using ‘economic terminology’ in their proposals and to ties – UNIV; and the Church – CHURCH), although many
focus on ‘decentralization issues’, consistent with the democ- players’ interests clearly conflicted with the white paper. Even
ratization rhetoric favoured by the PLD. Leonel Fernández’s when players expressed nominal support (such as the Secre-
position on the AMD strike, which occurred before he took tary of Health), the interviews suggested that most players
office, was that the President of the Republic should negoti- preferred to wait for completion of the studies and proposals
ate directly with the head of the AMD, and should continue before taking a position. This lack of involvement forced the
89
High Medium Low Non- Low Medium High

support support support mobilized opposition opposition opposition
OCT PRES UNIV NGO PrivClin AMD
IntlBank PLD Church IDSSBur EMPLOYER

IDSSDir Press
SecSal Benefs
SESPBur CNS
Key to power gradient: white box = low power; grey box = medium power; black box = high power
Figure 1. Position map for major players in health reform in the Dominican Republic in 1995 (abbreviations in text)
OCT and the Banks to become the main advocates for health had recently completed a negotiation that resulted in a
reform, a politically problematic situation. (See Figure 1 for a favourable outcome for the AMD, which might facilitate the
position map for the major players in the Dominican health changes required in health sector reform.
reform.)
(4) Strategies
(3) Opportunities and obstacles
PolicyMaker provides a tool box of 31 basic political strat-
The PolicyMaker analysis also produces a systematic assess- egies for enhancing the feasibility of policy reform and a
ment of opportunities and obstacles to the policy change matrix for defining strategy actions and associated risks, prob-
under consideration. As many of the obstacles were discussed lems, and benefits. Strategies are usually designed with the
above, this section focuses on the opportunities. An impor- client’s full collaboration, to assure that the proposed strat-
tant opportunity lay with the OCT, which had ample financial egies are relevant and realistic under the time and resource
resources from international agencies, relative independence constraints. Table 1 provides a summary of the strategy
from other health sector players, excellent technical staff, and design exercise for health reform in the Dominican Republic.
a vision of how health reform could work. These strengths
created a good negotiating position for the OCT within the Facing competing priorities and upcoming elections, the OCT
health sector and civil society. The small staff, however, executed a selection of the strategies in Table 1. The OCT
limited the OCT’s ability to respond quickly to political chal- created common ground and vision (strategies 1 and 2)
lenges in the media and the health sector. The reform process through the official debate and publication of the white paper
was just beginning, which gave the OCT substantial flexibility by the CNS. An indicator of this success was the subsequent
in planning. publication of institution-specific (SESPAS and IDSS)
reform proposals that differed minimally from the original
In the larger political environment, there was broad consen- white paper (strategy 8). The mobilization and preparation of
sus that the Dominican public health system suffered from key actors was limited by the pre-electoral environment
multiple problems and needed serious improvement. This (strategy 4); however, the passive role of the Secretary of
realization was occurring at the same time as the Dominican Health during this period seemed to activate debate within
Republic approached its historic transition to democracy and the CNS. Contrary to expectations during the analysis about
the country’s first democratic elections. The pre-electoral the ineffectiveness of external commissions, the CNS pro-
environment in late 1995 and early 1996 limited the willing- vided an excellent sector-wide forum for discussion (strategy
ness of politicians and political appointees to take a leader- 10, 11). But the decision-making processes in the CNS were
ship position on health reform, but created the hope that a never fully defined (strategy 3), and its existence depended on
new government might adopt proposals developed by the a presidential decree issued under Balaguer. Consequently,
OCT. Another opportunity was that the AMD and SESPAS the CNS did not survive under the new Administration.
90
Table 1. Summary of PolicyMaker strategies
Strategy name Actions
Strategy #1 1. Seek common ground with other organizations.

OCT: Create Common Ground 2. Identify common interests.
3. Link different interests – invent new options.
4. Make decisions for opponents easier.
Strategy #2 Keeping in mind that the principal obstacles to reform are not only technical:
OCT: Create a Common Vision 1. Create an atmosphere of shared values, unified leadership.
2. Articulate a common vision of equity and the respective roles of the public and private sectors.
Strategy #3 At the time of the analysis, there was no formal procedure for decision-making in the CNS, so:
Define the Decision-Making 1. Formalize process for the approval of the ‘white paper.’
Process 2. Legal efforts to formalize this process may be fruitful.
Strategy #4 1. The Secretary of SESPAS should be positioned to take a strong position of leadership.
CNS: Mobilize and Prepare Key 2. The Director of the IDSS should be prepared to take a clear position on the reform of the IDSS.
Actors 3. Key actors within SESPAS, IDSS, and the CNS should limit their discussion to the specific
components under consideration.
Strategy #5 1. Select pilot study sites according to technical and political exigencies.
OCT: Initiate Pilot Studies
Strategy #6 1. Meet with political candidates and their technical staffs.

OCT: Political Parties/ 2. Attempt to integrate health reform policies and the ideas of the ‘white paper’ into political debate
New Government and discourse.
Strategy #7 1. Initiate strategic contacts with the press, responding to critical attacks (except those of the AMD).
OCT: Initiate Strategic 2. Place key decision-makers in the media.
Communications
Strategy #8 1. Identify possible opposition and involve them in the technical design of the reform.
SESPAS and IDSS: Manage
the Bureaucracy
Strategy #9 1. Request technical-political assistance from the IDB and the WB in order to respond more
OCT: Strengthen Alliances with effectively to common critiques of the WDR-style reforms.
International Organizations 2. Work together with PAHO in concrete areas.
3. Ask for donor support for the vision of reform articulated by the OCT and define their active
participation in influencing key actors in the health sector.
Strategy #10 1. Hold informal consultations with ‘friends’ of the reform on the sequencing of actions and
OCT: Involve ‘Friends’ in political strategy; draw on the experience of the education reforms.
Planning 2. Bring together public hospital directors to articulate an agenda.
Strategy #11 1. Create strategic alliances with key actors not usually involved in health sector policy debate
AMD and IDSS: Create (nurses’ union, igualas, other unions, business associations, NGOs, churches, universities).
Strategic Alliances
Pilot projects (strategy 5) in hospital autonomy were initiated accurate over time, and the OCT was able to respond to edi-
during the pre-electoral period as planned, but the demon- torials and attacks in a timely manner. Alliances with inter-
stration effects of the studies were limited due to the OCT’s national agencies were strengthened during the design phase
weakening relationships with the new Administration, through the creation of working groups on specific themes
especially with the new leadership in SESPAS. Subsequently, such as human resources (strategy 9, 11).
key technical staff in the OCT and in the Secretariat of the
Presidency were replaced, reducing the feasibility of reform Overall, the PolicyMaker exercise produced a set of strategies
proposals as originally conceived. The limited political that achieved some success for the OCT, especially with
support of the new Administration for health reform showed regard to common ground, vision, and work with the SESPAS
the OCT’s mixed success in working with political parties and IDSS bureaucracies. However, relationships with key
(strategy 6). A communications strategy (strategy 7) was political actors were particularly precarious in the post-elec-
launched with success; the debate in the press became more tion period, and presented an insurmountable challenge to
91
reformers. This exercise in systematic applied political analy- important than the high and immediate costs felt by small
sis helped move the health reform process forward in the groups that are highly concentrated (and politically strong).
Dominican Republic, but did not result in full adoption of the Explicit political strategies are needed to manage this distri-
health reform package. In short, applied political analysis bution of the political costs and benefits of reform, especially
may be necessary to promote WDR-style health reform in relation to key interest groups (the medical association and
efforts, but analysis alone is not sufficient for success, for health workers’ union), the government bureaucracy, and
reasons discussed below. In late 1997, the OCT repeated the international agencies. Reformers need short-term concrete
PolicyMaker analysis, updating the position maps and setting gains that can satisfy key constituents, especially if the
out modified strategies. Whether this additional analysis will expected benefits of reform are perceived as long-term,
provide sufficient guidance to produce political and social uncertain, or intangible. In short, reform advocates require
acceptance of health reform in the Dominican Republic in the political strategies to manage the perceived interests of key
near future is an open question. stakeholders. If there is a political leadership vacuum, then
reform groups need substantial human and financial resources
to plan for these non-technical dimensions of the reform
Conclusions
process. Applied political analysis can assist the process of
This analysis of the political dimensions of health sector generating strategies for promoting reform, but analysis must
reform processes in the Dominican Republic suggests some be supported with the skills and resources for on-going consul-
generalizations that may be relevant to other nations. Six tation and negotiation with major stakeholders.
factors seemed to affect the pace and feasibility of the health
sector reform proposal in the country in 1995.
Factor 3: The location of the reform group
A structural dilemma exists in the organizational location of
Factor 1: The leadership of the reform
the reform group, reflecting a general dilemma about the
The leadership vacuum in the Dominican Repubic in 1995 location of advisory or policy analysis groups. A location
made decision-making on health reform difficult and incre- within the agency can restrain the group’s autonomy and
mental at best. The Secretary, facing the progressive decay of ability to question basic assumptions of the leadership,
institutions and the near certain loss of his party in the coming making the advice serve the preconceptions of the leadership.
elections, was unwilling to tackle health system change. On the other hand, a location outside the agency can produce
Comprehensive health sector reform usually requires the full weak links to decision-makers with a tendency to marginal-
commitment of the Secretary of Health. In the Dominican ization and irrelevance, while allowing the reform group
Republic and elsewhere, leadership capacity is deeply more autonomy and capacity for independent analysis. At the
affected by the system of government (new democracy versus time of this analysis, the OCT was located outside SESPAS
aging dictatorship), the credibility of the government, politi- and was seen as an outsider by the Health Secretariat’s
cal timing (the approach of elections), and the political effects bureaucracy. This allowed critics of health reform to link the
of the technical content of reforms. OCT symbolically with the development banks, and helped
weaken the OCT’s political legitimacy. After the election, the
If the political leadership is inactive on health sector reform, OCT was brought into SESPAS, only to be separated again
the technical reform group and the Banks themselves become several months later.
the policy advocates. To play this role effectively, leadership
and resources are required within the reform group. The
Factor 4: The ownership of the reform
reform group must receive technical, strategic, and political
support, above and beyond the standard studies conducted For health reform to be adopted, the reform package needs
under Bank pre-loan processes. In a personalized political to have strong ownership, usually by the Minister and by the
system in which decision-making is highly centralized, the government. But a dilemma also exists with ownership. If a
reform group must create a critical mass of reform support- reform is closely associated with a government, and the
ers, who can promote reform despite a turnover of leaders. government changes, then a common political response of the
Reform groups may need to create incentives for the Minis- new regime is to reject or reverse the reforms. The new
ter to become a fully engaged advocate for reform. Politicians government needs its own reforms, with material and sym-
need to find ways to navigate the political costs and benefits bolic benefits, and also needs to distinguish itself from
of health reform, through a combination of short-term gains previous power-holders.
and a supportive environment. In situations of uncertain
political leadership, as shown by the case of the Dominican The dilemma is this: an effort to raise ownership above the
Republic, the prospects of health reform are greatly handi- current government-in-power (through a multi-partisan com-
capped. mission, for example) may successfully diffuse ownership, but
this process could lower the probability of achieving success-
ful acceptance and implementation. Minister-driven reform
Factor 2: The political strategies adopted by the reform
can tie the change closely to one person and thereby raise the
group
chances of adoption now and reversal later; but if not tied
Health sector reform confronts a collective action dilemma: closely, then the reform may not happen at all. The goal is to
the small and delayed benefits for many people who are create a reform with sufficient ownership by the current
highly dispersed (and politically weak) are perceived as less power-holders that it is likely to be accepted, and without so
92
much ownership that the next government will reject the political challenges that are of significant size. These chal-
reform and seek its own. Achieving this goal requires the cre- lenges require political leadership that is committed to
ation of strong constituencies, within the bureaucracy and reform and prepared to expend political capital, and political
among interest groups, to mobilize supporters who will have strategies that can manage the political costs of powerful
an interest in continuing the reform and who will persist stakeholders associated with the reform. The experience in
beyond changes in government. the Dominican Republic suggests that applied political analy-
sis can help identify strategic options, which may enhance the
In the Dominican Republic in 1995, prior to elections, the prospects for health reform. But the experience also demon-
potential political owner of health reform had little chance of strates that analysis must be accompanied by an adept use of
continuing in office, and therefore no effort was made to mobi- political power; otherwise the reform package is likely to lan-
lize high-level political support for the reform. The Dominican guish as technically desirable but politically infeasible.
Republic’s approach of technical studies plus wait-and-see
was effective in preserving some elements of the OCT after
Endnotes
the election. But this strategy also reduced the probability that
1 Governments usually started adjustment with the tacit
the reform proposals emerging from the study period would
be adopted and owned by the new administration. consent of the population, having been put into office to ‘reverse
economic collapse’ (Lindenberg and Ramirez 1989). Health sector
reform has not enjoyed such a mandate in Latin America.
2 More than 70% of public (SESPAS, IDSS, Secretariat of the
Factor 5: The political language of reform
Presidency) spending on health is directed to hospital care (IDB 1997).
Reform efforts often require new ideas that can change the 3 In the most recent OCT document, money management
political landscape, provide new perspectives on old prob- would be the responsibility of the Central Bank.
4 While politically powerful, it is interesting to note that the
lems, and create alliances among diverging groups. The politi-
1996 eight-month AMD strike, which resulted in the total shut down
cal language of reform can create legitimacy by connecting
of public services, evoked little interest from the public. Private
the reform to international sources and the experiences of sector services seem to have absorbed most clients willing to pay. In
other countries. The promotion of ‘equity and efficiency in fact, health indicators (infant mortality) actually improved during
health systems’ is hard to oppose. Poor choices of political this same time period.
language can undermine efforts at reform. As shown in the
Dominican Republic, an association with the word ‘privatiza-
tion’, regardless of its technical accuracy, can undermine
References
support for a reform effort and can put reformers in a defen- Akin J, Birdsall N, de Ferranti D. 1987. Financing Health Services in
sive mode that is difficult to overcome. Developing Countries. Washington, DC: The World Bank.
Cross-Beras J. 1985. Cultura Política Dominicana. Santo Domingo:
Instituto Tecnológico de Santo Domingo (INTEC).
Factor 6: The political timing of reform Inter-American Development Bank. 1997. Dominican Republic:
Health Sector Modernization and Restructuring Project Files,
The feasibility of health sector reform is often affected by September 15.
political timing; whether a government is recently elected or Lindenberg M, Ramirez N. 1989. Managing Adjustment in Develop-
is approaching the end of its term will affect its political ing Countries: Economic and Political Perspectives. San Fran-
capital and its willingness to take political risks. The approach cisco: ICS Press.
Macro International and the Government of the Dominican Repub-
of elections can complicate strategies to create political cir- lic. 1991. Demographic and Health Survey. Columbia, MD:
cumstances that would support reform. If the current govern- Macro International.
ment is unlikely to stay in power, or if the current Minister is Mesa-Lago C. 1992. Health Care for the Poor in Latin America and the
unlikely to stay in power, then the power-holders may have Caribbean. Washington, DC: Pan American Health Organiz-
limited political resources and limited interest in attempting ation.
a reform that entails high political costs. Mesa-Lago C. 1989. Ascent to Bankruptcy: Financing Social Security
in Latin America. Pittsburgh: University of Pittsburgh Press.
Mesa-Lago C. 1978. Social Security in Latin America: Pressure
The process of health sector reform involves a continual Groups, Stratification, and Inequality. Pittsburgh: University of
tension between the technical and political dimensions. Pittsburgh Press.
Often, the proposed technical solutions are only partially con- Oficína de Coordinación Tecnica (OCT), Comision Nacional de
structed, with large ambiguities remaining in the institutions Salud, Republica Dominicana. 1995a. Salud: Una Vision del
required and the implementation methods. The reform group Futuro.
may be highly qualified in a technical sense and acutely aware Oficína de Coordinación Tecnica (OCT), Comision Nacional de
Salud, Republica Dominicana. 1995b. Acta Segunda Reunión.
of the political implications of different reform options, but Unpublished document.
may be unprepared for analyzing and managing the highly Partido de Liberación Dominicana. 1996. Programa de Gobierno.
political dimensions of the reform process. Applied political Santo Domingo, Dominican Republic.
analysis can be helpful in organizing political data in a sys- Perez Uribe. Study on SESPAS decision-making. Unpublished docu-
tematic way, in analyzing the political risks of health sector ment, 1974.
reform, and in constructing and selecting political strategies Reich MR. 1996. Applied political analysis for health policy reform.
Current Issues in Public Health 2: 196–91.
to manage the multiple players involved. Reich MR, Cooper D. 1996. PolicyMaker: Computer-Aided Political
Analysis Version 2.0. Brookline, MA: PoliMap.
The case of health sector reform in the Dominican Republic Rojas F. 1995. Autonomía no es privatización. Santo Domingo,
shows that the WDR-style reform package creates multiple Dominican Republic: Hoy, November 9.
93
Rojas O. 1995a. El futuro de la salud dominicana. Santo Domingo, Biographies

Dominican Republic: Hoy, November 14.
Rojas O. 1995b. ¿Paciente o cliente? Santo Domingo, Dominican Amanda Glassman works on health policy and financing reform
Republic: Listín Diario, November 9. issues at the Inter-American Development Bank. She holds a Master
Santana I, Rathe M. 1994. Setting a new agenda for the Dominican of Science from the Harvard School of Public Health, Department of
Republic. In Social Service Delivery Systems: An Agenda for Population and International Health.
Reform. Aedo C and Larrañaga O (eds). Washington, DC:
Inter-American Development Bank. Michael Reich is Taro Takemi Professor of International Health
World Bank. 1993. World Development Report: Investing in Health. Policy and Acting Chair of the Department of Population and Inter-
Washington, DC: The World Bank. national Health, at the Harvard School of Public Health. He received
World Bank. 1995. Background Paper prepared for Regional Con- his Ph.D. in Political Science from Yale University.
ference on Health Sector Reform in Latin America. Washing-
ton, DC: Pan American Health Organization. Kayla Laserson is an epidemiological intelligence officer with US
Public Health Service based in Atlanta at the US Centers for Disease
Control. She holds a doctoral degree from the Harvard School of
Acknowledgements Public Health, Department of Tropical Public Health and Depart-
ment of Population and International Health.
The authors would like to thank the following for their thoughtful
comments on the paper: Gerard LaForgia and Ruth Levine of the Fernando Rojas is a medical doctor currently running a private clinic
Inter-American Development Bank, members of the HSPH Health in the Dominican Republic. Dr Rojas was head of the Oficína de
Policy and Politics Working Group, and an anonymous reviewer. Coordinación Técnica during the PolicyMaker analysis.
The views expressed in this paper are those of the authors and do not Correspondence: Michael R Reich, Harvard School of Public Health,
reflect the views or policies of the Inter-American Development Bank. 677 Huntington Avenue, Boston, MA 02115, USA.
94
BMC Public Health BioMed Central
Research article Open Access

Pathways to malaria persistence in remote central Vietnam: a
mixed-method study of health care and the community
Martha Morrow*1, Quy A Nguyen2, Sonia Caruana3, Beverley A Biggs3,
Nhan H Doan2 and Tien T Nong2
Address: 1Nossal Institute for Global Health, University of Melbourne, Melbourne, Vic 3010, Australia, 2National Institute of Malariology,
Parasitology and Entomology, BC 10200 Tu Liem, Hanoi, Vietnam and 3Department of Medicine, University of Melbourne, Melbourne, Vic 3010,
Australia
Email: Martha Morrow* - martham@unimelb.edu.au; Quy A Nguyen - quyanhnguyen@s.vnn.vn; Sonia Caruana - sonia2rose@yahoo.com.au;
Beverley A Biggs - babiggs@unimelb.edu.au; Nhan H Doan - hanhnhan@netnam.vn; Tien T Nong - nongthitien@yahoo.com
* Corresponding author
Published: 23 March 2009 Received: 25 January 2008

Accepted: 23 March 2009
BMC Public Health 2009, 9:85 doi:10.1186/1471-2458-9-85
This article is available from: http://www.biomedcentral.com/1471-2458/9/85
© 2009 Morrow et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
Background: There is increasing interest in underlying socio-cultural, economic, environmental and
health-system influences on the persistence of malaria. Vietnam is a Mekong regional 'success story' after
dramatic declines in malaria incidence following introduction of a national control program providing free
bed-nets, diagnosis and treatment. Malaria has largely retreated to pockets near international borders in
central Vietnam, where it remains a burden particularly among impoverished ethnic minorities. In these
areas commune and village health workers are lynchpins of the program. This study in the central province
of Quang Tri aimed to contribute to more effective malaria control in Vietnam by documenting the non-
biological pathways to malaria persistence in two districts.
Methods: Multiple and mixed (qualitative and quantitative) methods were used. The formative stage
comprised community meetings, observation of bed-net use, and focus group discussions and semi-
structured interviews with health managers, providers and community. Formative results were used to
guide development of tools for the assessment stage, which included a provider quiz, structured surveys
with 160 community members and 16 village health workers, and quality check of microscopy facilities and
health records at district and commune levels. Descriptive statistics and chi-square analysis were used for
quantitative data.
Results: The study's key findings were the inadequacy of bed-nets (only 45% of households were fully
covered) and sub-optimal diagnosis and treatment at local levels. Bed-net insufficiencies were exacerbated
by customary sleeping patterns and population mobility. While care at district level seemed good, about a
third of patients reportedly self-discharged early and many were lost to follow-up. Commune and village
data suggested that approximately half of febrile patients were treated presumptively, and 10 village health
workers did not carry artesunate to treat the potentially deadly and common P. falciparum malaria. Some
staff lacked diagnostic skills, time for duties, and quality microscopy equipment. A few gaps were found in
community knowledge and reported behaviours.
Conclusion: Malaria control cannot be achieved through community education alone in this region.
Whilst appropriate awareness-raising is needed, it is most urgent to address weaknesses at systems level,
including bed-net distribution, health provider staffing and skills, as well as equipment and supplies.
Page 1 of 10
(page number not for citation purposes)
95
BMC Public Health 2009, 9:85 http://www.biomedcentral.com/1471-2458/9/85
Background cases (confirmed and clinical). Over 9% of Quang Tri's

Malaria remains a major global threat and its control is approximately 573 000 inhabitants are ethnic minorities,
one of the Millennium Development Goals. Anti-malarial overwhelmingly Van Kieu and Paco [14], who live astride
drug resistance, linked to both unnecessary and inade- the border with Lao PDR in districts with rugged, lush ter-
quate drug intake, creates risks for malaria resurgence, and rain, frequent rainfall, and large infrastructure projects
is a major challenge for malaria control [1]. Insecticide- within forests (including a bridge and road along the old
treated bed-nets and effective anti-malarial drug combina- Ho Chi Minh Trail). These contextual aspects, together
tions are essential components of control programs [2,3]. with poverty, low education levels, cross-border mobility,
However, effective control requires consistent action from and cultural diversity, made this an appropriate study site
both health systems and community, and an understanding for malaria social science research.
of features that precipitate risk, such as development
projects bringing labourers through forested areas [4]. The This paper reports on a collaborative study aiming to con-
broad social (i.e. non-biological) aspects of control are tribute to malaria control in Vietnam by documenting the
thus critical to success [5]. non-biological pathways to malaria persistence in two dis-
tricts. The objectives were to identify the role and nature
Overextension or poor training of health staff can under- of health system and community factors directly linked to
mine diagnosis and treatment, while bed-net distribution malaria persistence, and underlying influences that help
requires reliable systems that are difficult for impover- explain the direct factors. The study was undertaken by
ished, isolated settings. Patients may not seek care if they Vietnamese and Australian researchers from March 2004
lack knowledge or money for treatment or transportation, to April 2005.
or may seek care from multiple sources [6-8,3], making
implementation of treatment guidelines – and health Methods
information systems – problematic [9]. In order to meet the study objectives we chose a flexible
study design with multiple methods (both qualitative and
Vietnam is considered a Mekong region malaria 'success quantitative). Mixed-method approaches permit explica-
story' after the introduction of a National Malaria Control tion of complex interrelationships between actors and sys-
Program (NMCP) in 1991 to address a spike in cases and tems, and have been used for malaria social research
increasingly widespread drug resistance. The NMCP pro- [15,16,5]. Data were collected in two stages. The forma-
vided free anti-malarials (especially artemisinin to treat P. tive stage used mainly qualitative tools to help define and
falciparum, Vietnam's most prevalent – and most deadly if expand thematic areas of enquiry; these data were rapidly
untreated – strain, impregnated bed-nets, twice-yearly reviewed to inform the (mainly quantitative) tools used
home insecticide spraying, and early diagnosis and treat- for the assessment stage. An overview of methods and
ment. Declines from 1.2 million clinical cases (without samples appears in Table 1. NIMPE investigators were
confirmed blood tests) and 4646 recorded deaths in 1991, trained by Australian colleagues and collected all data
to 185 529 clinical cases and 50 deaths in 2002, are attrib- during 3 field visits.
uted to the program [9-12]. Managed by the National
Institute for Malariology, Parasitology and Entomology Choice of Study Sites
(NIMPE), the NMCP relies on vertical and horizontal col- Among Quang Tri's 8 districts, two (hereafter, A and B)
laborations from central to village level. were selected for their greater malaria caseload and prox-
imity to the Lao border. At the 1999 census, district A's
By 2000 malaria had largely retreated to 84 districts in 15 population was 54 547 and B's was 27 000; the vast
(out of 61) provinces, with an estimated population of 7.2 majority were Van Kieu and Paco. For the formative stage
million (vs 83 million nationwide). Over 90% of severe we chose 3 border communes per district.
cases and deaths occurred in mountainous, forested and
largely ethnic minority areas of central Vietnam [9,12], For the assessment phase we used 2 of these communes
often along international borders, where the common per district (i.e. total 4 communes) in order to ensure suf-
mosquito vectors that transmit malaria, Anopheles dirus and ficient sample recruitment within the timeframe in view
An minimus, are plentiful [13]. These ethnic minorities tend of the low population density and transportation difficul-
to be impoverished, poorly educated, culturally and lin- ties. From each commune's approx 10 villages we selected
guistically distinct, and living in dispersed, less accessible 4 with varying ease of access as well as distance from the
settlements; such areas 'represent a real challenge for the commune health station (i.e. total 16 villages).
[NMCP]' [9] (p.217). Therefore, it is both instructive and
timely to investigate persistent malaria in such settings. Development and Use of Instruments, Sampling and Ethics
In the formative stage we held community meetings with
In 2002–03 the central province of Quang Tri was among district stakeholders to establish rapport and elicit local
Vietnam's highest malaria burden provinces with 4178 information and views. Semi-Structured Interviews (SSIs)
Page 2 of 10
96
Table 1: Summary of methods and data sources (both stages)
Sample/source/focus Location Method (number)
FORMATIVE STAGE
MC officials, local government, mass 2 district capitals Community Meetings

organisations, hospitals (approx 20 participants each)
Provincial MC officials Provincial capital SSIs (2)
District MC secretaries 2 district capitals SSIs (2)
District Hospital managers 2 district capitals Informal group discussion (2)
Anti-malarials sold at market 1 border commune Observation (1)
Village environment 6 villages Observation (6)
District Hospital staff 2 District Hospitals SSIs (3)
Commune health staff 6 Commune Health Stations FGDs (5)

SSIs (4)
Village Health Workers 6 communes FGD (1)

SSIs (13 in 5 communes)
Community members 6 communes FGD (1 with women)

SSIs (14 men, 6 women)
ASSESSMENT STAGE
District Hospital staff 2 District Hospitals Tests (11 open questions) (14)
Malaria patient record cards 2 District Hospitals Case numbers year-to-date (one DH);
Review of previous month's cards DH-A (53) &
DH-B (35)
Microscope points 2 District Hospitals & 4 Commune Health Observation (6)

Stations
Patient treatment logs 4 Commune Health Stations Breakdown of year-to-date case numbers (one
CHS);
Review of previous 3 months (4 CHSs)
Village Health Workers 16 villages KAP survey (16) (1/village)
Community members 16 villages KAP survey (160) (10/village)
Bed-net use and quality 16 villages Use observed in night-time home visits (55);
quality observed in KAP survey home visits
(160)
MC = malaria control; FGD = Focus Group Discussion; SSI = Semi-structured Interview; KAP = Knowledge, Attitudes and Practices
and Focus Group Discussions (FGDs) using flexible members (village heads and adult men and women,
guides were held to explore beliefs, attitudes, awareness, recruited purposively).
care seeking/providing and circumstances relevant to
malaria exposure and control with all provincial and dis- For the assessment stage we developed and administered
trict MC managers and Commune Health Stations (CHS) face-to-face structured knowledge, attitudes and practices
staff, a convenience sample of VHWs, and community (KAP) surveys in the 16 villages, one with every Village
Page 3 of 10
97
Health Worker (VHW) (n = 16) and another with10 com- (4178). District A recorded 2131 cases (vs 2246 in 2003)
munity members per village (n = 160), respectively. The and District B 608 cases (vs 571 in 2003). Below we
community sample size was determined on the basis of present evidence of direct and underlying influences on
time, resources and feasibility, along with power to con- malaria persistence in both districts at health system levels
duct tests of significance on some demographic variables. (district, commune, village) and community level, in
Sampling was undertaken randomly from village house- turn.
hold lists, stratified for equal numbers of men and women
aged 18–48. Van Kieu interpreters (one male and one District hospital level: satisfactory standards of malaria
female) were used for nearly all community surveys after care but early discharge for some patients
training by NIMPE researchers. We also devised observa- Record review from the first 9 months of the year showed
tion check-lists to assess visibility and currency of malaria that DH-A treated 433 malaria cases. Review of a total of
treatment guidelines, quality of CHS microscopy, and 88 patient cards from the two DHs showed close adher-
bed-net quality during KAP survey home visits. Actual ence to the most recent national malaria guidelines [17].
bed-net use was determined by unannounced night visits Just 3 patients were treated for malaria despite having a
to 55 homes in 2 communes. To obtain an impression of parasite-negative slide. Most DH malaria control staff
provider knowledge and guidelines adherence, we were trained in the guidelines and generally knowledgea-
quizzed (11 open questions) district hospital (DH) staff ble. Of the 11 questions, the 8 staff at DH-A correctly
involved in malaria control and available on the day, and answered all but 3, with 1–3 staff incorrect on each. Of the
reviewed one month of patient records from both DHs 6 DH-B respondents, all got 5 questions correct, with one
and 3 months of treatment logs from all 4 CHSs; compre- wrong answer apiece for the remaining 6 questions.
hensive malaria case record numbers for the first 9 Microscopes were in good condition, microscopists had
months of the year were collated from one DH and one specialist training, and results were reportedly usually
CHS. available within 30 minutes. There was one notable prob-
lem noted by DH staff during a community meeting:
Potential participants were assured that participation was about one-third of inpatients discharged themselves prior
voluntary and confidential and refusal would have no to completion of treatment. Staff attributed this to inabil-
negative consequences. As is common in Vietnam, all ity to afford 'extra' charges for in-patient care, e.g. antibi-
agreed to participate; verbal informed consent was taken. otics and vitamins. Many were lost to follow-up, making
Participants were given a t-shirt with a malaria control it impossible to verify their adherence or recovery. How-
message in appreciation. The study was approved by ever, most patients presented first to lower levels (though
NIMPE's Human Research Ethics Committee for Medical- some were referred to DHs). At their last bout of malaria,
Biological Research, and the University of Melbourne's 38% of community members reported they sought care
Human Research Ethics Committee. Instruments were from the VHW and 60% from the CHS; just 10% travelled
developed in English, translated into Vietnamese (and to the DH (>one answer possible).
back-translated) and pre-tested with a convenience sam-
ple in the study area. Commune Health Stations: deficiencies linked to resources
Each commune in Vietnam has a health station in a fixed
Data Management and Analysis facility serving the commune's villages. National policy
Notes were taken during SSIs and FGDs; transcripts were states CHSs should have at least 4 staff, including a fully-
not prepared due to time constraints. Researchers qualified doctor, nurses and/or midwives, and should
reviewed the formative data to finalise the assessment implement all basic preventive and curative care under
stage tools. Check-list data, health record reviews and quiz DH direction. Just 2 of our 4 communes had the full staff
results were collated. KAP survey data were analysed using complement, but also had larger populations than usual.
Stata v8.0 (descriptive statistics and chi square tests), and The others had 3 staff, though some were not qualified to
community level differences calculated for location, sex offer routine services.
and education. Interpretation of findings was iterative and
involved all data sources and researchers; together we dis- Checks found deficiencies at most CHSs in malaria diag-
tilled a subset of triangulated findings that offered a nosis, treatment and microscopy. During FGDs and indi-
coherent picture of the interplay between direct and vidual interviews, staff at all 4 communes acknowledged
underlying influences on persistent malaria. that presumptive treatment frequently occurred. A
detailed record review for the first 9 months of 2004 was
Results undertaken in one CHS (popn 2618) in District A; staff
Provincial records showed a continued high malaria bur- treated 100 parasite-positive and 82 'clinical' cases
den in Quang Tri in 2004, with a total of 3958 cases (both (unconfirmed by microscopy and diagnosed by symp-
clinical and slide-confirmed), a slight decline from 2003 toms). Thus nearly half of all cases (i.e. 82/182) were
Page 4 of 10
98
treated presumptively. Review of the past 3 months of logs ways, steep ravines and a dearth of telephones hinder
in all 4 CHSs showed that in 2 communes, staff gave communications and transportation. Home visits, refer-
appropriate treatment per guidelines. In the other 2, staff rals and patient follow-up were particularly difficult, espe-
sometimes gave CV8 for P. vivax cases (when chloroquine cially considering understaffing and (at the time of the
temporarily ran out) and primaquine + artesunate for study) lack of telephones in some CHSs, leading at times
clinical cases; moreover, workers at these CHSs did not to local management of severe cases who would have
recognise these treatments were contrary to guidelines. been referred to the DH.
Laminated treatment guidelines intended for display to
facilitate their use were locked out of sight in 3 of the 4 Poorly trained Village Health Workers and lack of
CHSs. appropriate drugs
Among the 16 VHWs surveyed, most (14) were men, 12
Although CHS staff discharged patients with instructions were Van Kieu, 3 were Kinh (ethnic Vietnamese), and one
to report to their VHW during treatment, staff (at both lev- Kazo. Median age was 31 years (range: 21–45 years). All
els) said patients often failed to do so, making it impossi- had regular occupations as farmers (14) or traders (2). The
ble to monitor adherence to treatment and course of 2 female VHWs had the highest education (10–12 years),
illness, both of which are important for effective malaria 10 of the men had 6–9 years, and the other 4 had the min-
control at the population level. imum required (5 years) for VHWs. Median length of serv-
ice was 5 years (range: 7 months-15 years).
Several underlying influences apparently contributed to
CHS-level weaknesses, including deficiencies in human The VHW (one per village) is a volunteer working across
resources, training, equipment and supply, all exacerbated all primary health care programs following very basic
by geographic isolation. In SSIs and FGDs most CHS staff training. For MC alone, VHWs are expected to prepare
said they found it difficult to accomplish their duties blood films, make referrals for severe cases, treat with
given current staffing levels. Understaffing placed particu- (free) anti-malarials, educate the community, manage
lar pressures on microscopy services. Blood films would cases discharged from higher levels, and assist with spray-
arrive haphazardly via VHWs or outpatient CHS services. ing and net impregnation. The study found that some
Slides should be prepared and read immediately, which VHWs lacked confidence in their clinical MC duties (see
takes 30–45 minutes, but this rarely happened because of Table 2).
competing tasks, e.g. queues of infants awaiting immuni-
sation, disease outbreaks, meetings with district health KAP analysis revealed that 11/16 VHWs prepared blood
staff, or absence of the microscopist. For each slide the films, but only 6 delivered these the same day to the CHS,
microscopist is paid an 'incentive' of just 300 dong (about with 4 waiting >72 hours. Ten said they 'rarely or never'
USD two cents), which is low even by local standards. stayed to obtain results; only parasite positive results were
This situation may help explain why staff frequently pre- reported back to them from the CHS, often after a few
scribed anti-malarials according to symptoms, rather than more days. Most (11/16) commenced treatment without
after microscopic confirmation, as is preferred. For quality microscopic-confirmed diagnosis, prescribing partly by
assurance, district staff periodically collected slides for symptoms, and partly by the type of drug currently on
review at the provincial capital; the percentage of incorrect hand within their kits. In 10 villages VHWs did not carry
readings was reported back to the district, and thence to artesunate, the recommended drug of choice for P. falci-
each CHS, but without specifics on individual slides. One parum malaria at the time of the study (see Table 3).
commune was told that 20% were incorrect after awaiting
feedback for 4 months. Of the 6 who carried artesunate, all believed it was appro-
priate for 'serious' malaria cases. The main indication for
Although the MC program stipulates a properly trained chloroquine offered by the 14 who carried it was 'light'
microscopist for each CHS, most CHSs relied on one of their fever, not its usefulness for P. vivax malaria. Hence, use of
staff who was designated for this role but undertook the anti-malarials for non-malarial fever may have occurred.
usual CHS workload, and typically had just a week of train- Despite the fact that 12 VHWs reported confidence in case
ing. Few had in-service training. As well, quality was under- management, 8 admitted they never followed up.
mined by ageing microscopes, lack of stain solution in one
commune, improper storage of materials in another, and Triangulation of data sets suggests that VHW weaknesses
inadequate pure water and filtering equipment in several. in malaria management were attributable to a number of
underlying influences, including insufficient time to com-
The geographic features that make malaria viable in this plete duties outside normal working hours, inadequacies
region, coupled with low population density, present in pre- and in-service training and some delays in rolling
great challenges for its control. Poor roads, many water- out the new guidelines for drugs in VHW kits.
Page 5 of 10
99
Table 2: VHW self-reported confidence in aspects of malaria control role
n (%)
Diagnose malaria through symptoms (n = 15*) 10 (66)

Take a blood sample from patients (n = 15*) 9 (60)
Make a blood film for microscopy (n = 15*) 9 (60)
Give correct anti-malarial medication (n = 15*) 9 (60)
Adequately manage malaria cases (n = 15*) 12 (80)
Undertake malaria information-education-communication activities (n = 16) 13 (81)
* missing data
In Vietnam, individuals often become VHWs out of civic refer patients. These circumstances presented ongoing
duty or the appeal of further education and occasional – risks that some seriously ill patients would be treated in
if small – incentives for particular health care tasks. Apart the village, possibly with a less effective anti-malarial.
from their MC duties, VHWs must keep abreast of chang-
ing, relatively complex, treatment guidelines. This is Community level: sub-optimal prevention linked to
daunting for volunteers with low education levels residing insufficient bed-nets and socio-cultural context
in remote locations. When asked to name the role's disad- Demographic information from the community KAP
vantages, our sample mentioned low remuneration, lack appears in Table 4. Most were Van Kieu, and education
of time, and difficulties with transportation and distance, levels were low, with females more likely to be
all of which could undermine case identification and unschooled (!2 = 28.22, p = 0.01). Median household size
management. About one-third felt frustrated by the villag- was 6 persons (range 2–13 persons). Sixty percent had a
ers' 'refusal to take advice'. 'Poor Card', which denotes low-income status and ena-
bles free medical care and basic drugs. Most (66%) survey
Although policy dictates that each VHW is trained pre- respondents reported having had malaria, including
service for at least three months, just 4 (one-quarter) had about one-third at least once in 2004.
such training; 3 had 12–45 days, 6 had 1–5 days, and 3
reported no training. Only 5 reported training during Our findings suggest the direct risks operating at commu-
2004, although provincial policy requires annual nity level were sub-optimal bed-net use and early self-dis-
refresher training. Only 12 VHWs knew about the new charge from care. The national MC program calculates net
guidelines and 10 carried the new treatment table. Most, sufficiency on a ratio of one net per two people, with a tar-
however, knew correct dosage for the drugs they carried. get of consistent use by at least 80% of the population in
At the time of the study NIMPE was disseminating new endemic areas. Quang Tri health staff at all levels believed
diagnosis and treatment guidelines, which include some this target was not met in the study communes, a view
devolution of decision-making on local treatment to pro- based on irregular day-time spot checking by provincial
vincial MC managers. Some confusion appeared to persist and district survey teams. We undertook our estimates dif-
during this transition, because informants at various lev- ferently, i.e. by observation during unannounced night-
els provided inconsistent information about policy for time visits, coupled with survey questions on bed-net use.
anti-malarials in VHW drug kits, and a range of explana- Night visits to 55 homes in two communes found no nets
tions for what was actually in the kits. were used in 20% of households and some nets did not
reach the floor or were used as blankets. The 160 survey
The terrain and isolation that hinder optimal care by respondents, however, reported very high usage: 145
CHSs act as greater barriers for the VHW MC role, because (92%) claimed to have slept under a net on the previous
VHWs typically have even less access to reliable transpor- night, and 136 (86%) said that all family members had
tation. It takes time, effort and – at the least – opportunity done so, whether singly or (more frequently) sharing.
costs for these part-time volunteers to remain in close Respondents cited adolescents and the elderly as less
touch with higher health system levels, to follow up or to likely to use and/or share nets, with just 50% of teenagers
consistently using, among whom 70% shared. Report-
Table 3: Number of VHWs carrying different types of anti- edly, 91% of elders 'always' used nets, but only 57%
malarial drugs
shared. Whilst 16% of respondents claimed to travel occa-
n = 16 sionally or often into Laos, and about half went into for-
ests at varying frequency, just a handful carried bed-nets
Number who carry only chloroquine 8 on overnight trips.
Number who carry only artesunate 0
Number who carry chloroquine & artesunate 6 While 66 (41%) sometimes (n = 58) or always (n = 8)
Number who do not carry any anti-malarial drugs 2
consulted traditional healers for 'health problems', the
Page 6 of 10
10 0
Table 4: Description of the KAP community sample, by sex
Males Females Total

n (%) n (%) n (%)
Sex 80 (50) 80 (50) 160 (100)
Age (mean, range) in years 34, 18–48 30.6, 18–45 32.5, 18–48
Ethnicity:
Van Kieu 69 (86) 71 (89) 140 (87.5)
Kinh (Vietnamese) 8 (10) 9 (11) 17 (10.6)
Other 3 (4) 0 3 (1.9)
Education level reached:

No schooling 31 (39) 64 (80) 95 (59.3)
Some primary (1–5 years) 27 (34) 4 (5) 31 (19.3)
Some secondary (6–9 years) 22 (27) 12 (15) 34 (21.3)
Occupation
Farmer 67 (84) 68 (85) 135 (84.3)
Other 11 (14) 9 (11) 20 (12.5)
missing 2 (2) 3 (4) 5 (3.1)
Poor Card
Yes 44 (55) 52 (65) 96 (60)
No 35 (44) 27 (34) 62 (38.8)
missing 1 (1) 1 (1) 2 (1.3)
survey showed high awareness of recommended help- Family configuration and cultural sleeping patterns also
seeking for suspected malaria. Respondents claimed to act affected net adequacy. In FGDs we heard that some teen-
accordingly (Table 5), although this could not be verified. agers refuse to use nets, and that elders (with reportedly
Some malaria patients with Poor Cards said they were high net usage) strongly prefer to sleep alone, thus poten-
charged for extras like vitamins at DHs (6/20) and CHSs tially leaving other family members short. As well, over-
(12/113), leading some to borrow money or discharge night socialising among male neighbours is so normal
themselves early. that Van Kieu houses contain a nominated 'guest' space in
the living room, but just 19% of respondents had a spare
There was considerable evidence that insufficient bed- net for guests.
nets, cultural sleeping norms, low education and poverty
acted as underlying influences on sub-optimal commu- Table 5: Community responses about care-seeking for suspected
nity behaviours. Provincial staff told us that Quang Tri malaria
had comprehensive bed-net coverage through the NMCP,
and MC staff at all levels attributed persistent malaria in What to do first for fever or suspected malaria (n = 149*) n (%)
Quang Tri mainly to community 'refusal' to use bed-nets,
arguing the need for more 'information, education and Do nothing 1 (0.7)
Pray 3 (2.0)
communication'. While enough nets may have been dis- Buy drug in market 4 (2.7)
tributed, our survey respondents reportedly received Go to Village Health Worker 77 (52)
theirs prior to 2003, and many were no longer intact. Go to Commune Health Station 63 (42)
Some purchased additional nets, usually cheaper single Go to District Hospital 1 (0.7)
bed size. Using MC guidelines on bed-net ratios (one net/
2 people) and data on household size, we calculated that How long do you wait before seeking care? (n = 129*)
among the 160 households represented by survey
respondents, just 72 (45%) had sufficient nets to cover Immediately 81 (63)
their needs and 88 (55%) did not. In addition, checks of One day 41 (32)
Two days 6 (4.7)
net quality when conducting the survey found 62% of
More than two days 1 (0.8)
households had at least one ripped or damaged net. Thus,
even if all available nets were used, less than half of all * Missing data
households were fully protected.
Page 7 of 10
101
A lack of spare nets also contributes to exposure risk dur- map the non-biological 'causal pathways' that led to the
ing periods of mobility – usually by foot – into Laos, for- problem of persistent malaria in a remote ethnic minority
est or fields for overnight stays. This mobility is culturally population. As Hawe et al argue, exploring the underlying
and economically driven, as families seek reunions with influences that precipitate, amplify or mitigate direct
relatives across the Lao border, and individuals collect for- health risks provides evidence that can assist program-
est products for consumption or sale due to lack of mers to design and target comprehensive interventions to
employment options. bring about and sustain necessary changes; the same
approach used in program evaluation can pinpoint spe-
As Table 6 illustrates, most respondents had basic under- cific opportunities to address quality concerns [18].
standing of malaria symptoms and causation, and knew
malaria is curable. However, about one-quarter were Strengths and weaknesses of the study
unsure about causation and prevention. Among those Particular strengths of the study were the involvement of
who said malaria is not preventable, 28 (55%) had no stakeholders from various health levels, including the
schooling, versus just 5 (18%) with one or more years of community itself, and the triangulation of data through
schooling (!2 14.33, p = 0.001); this misperception was use of multiple methods (quantitative and qualitative),
held by 17.5% of men and 50% of women(!2 6.60, p = including self-report and the more objective tools of
0.01). The lower education levels of women in particular observation and record review. Malaria social scientists
may explain gaps in preventive behaviours. have noted the need for community-level malaria investi-
gations to commence with qualitative methods that help
Ethnic minorities in western Quang Tri have little involve- explain behaviours, thus permitting grounded develop-
ment with mainstream society. Whilst VHWs tend to be ment of structured surveys [19]. This formative approach
the same ethnicity as villagers, this is less true for other was one of our study's strengths. However, due to lack of
providers. A third of respondents 'sometimes' had lan- resources and expertise, systematic preparation and anal-
guage problems with district or commune providers, and ysis of complete transcripts were not conducted, prevent-
one ethnic Vietnamese commune health worker who ing full utilisation of qualitative data to illuminate the
spoke Van Kieu felt neither fully accepted nor fully study's quantitative findings.
trusted.
Another limitation was a lack of definitive data from
In theory, cost should not deter care-seeking because CHWs on case management and microscopy quality,
malaria diagnosis and treatment are free. However, these which reflects the more rudimentary health reporting
involve transportation, opportunity and (sometimes) often found in remote settings. However, our objective
medical 'extras' costs that this community could ill-afford, was to map pathways in one study site and not to produce
which may help explain why some discharged themselves generalisable findings, which in any case would be inap-
from care and were lost to follow-up. Such charges are propriate given the small number of communes explored
imposed increasingly as Vietnam's health system is decen- and relatively small sample of providers and community
tralised. members. This study also did not attempt to identify the
role of biological factors such as vector prevalence or drug
Discussion sensitivity; thus preventing us from arguing conclusively
This mixed-method study in Quang Tri province in central the relative importance of all potential factors.
Vietnam was designed by a multi-disciplinary team that
included malaria experts and social scientists. It set out to Systems and the community: a dual focus for malaria
control in remote settings
Table 6: Community knowledge about malaria transmission, Figure 1 summarises relationships and pathways to
prevention and cure malaria persistence drawn from this study and lays out the
underlying influences that apparently explained weak-
n (%)
nesses found at both health systems and community lev-
Have heard of malaria (n = 158*) 143 (91) els. This model excludes vectors, weather events and drug
Mosquitoes main 'cause' of malaria (n = 158*) 113 (72) sensitivity. We present this as a conceptual framework for
'Don't know' what causes malaria (n = 158*) 40 (25) mapping our findings, and for possible adaption by
Fever is a symptom of malaria (n = 160) 124 (77) researchers wishing to investigate such pathways in other
Malaria can be cured (n = 156*) 134 (84) complex settings.
Malaria can be prevented (n = 140*) 107 (76)
Bed-net is best way to prevent malaria (n = 160) 98 (61)
Previous studies in Vietnam have found widespread mis-
'Don't know' best way to prevent malaria (n = 160) 53 (33)
understanding about malaria treatment and prevention
* missing data among populations in similar isolated endemic areas
Page 8 of 10
102
[20,21]. Our study found around a quarter of the commu- nam's commune-level services found poorer quality CHSs
nity shared these misunderstandings, and our model sug- in remote areas, especially where ethnic minorities live.
gests this may have contributed to poor health We found that local providers often lacked diagnostic
behaviours. Health systems managers often assume (as skills, time, equipment and/or appropriate drugs for pop-
here) that minority group customs, culture or knowledge ulations in this remote region. Even temporary shortfalls
'barriers' account for poor behaviours (and outcomes), in the supply of anti-malarial drugs, especially during out-
assumptions that typically lead solely to community edu- breaks, could have serious impacts. Additionally, District
cation interventions. The national program's ratio for bed- Hospital staff estimated that one-third of malaria patients
net sufficiency also rests upon assumptions about net- discharged themselves early for cost reasons (medical
sharing, and about where people actually sleep. Our 'extras'), and were usually lost to follow-up. Thus, pre-
major finding – that over half of households surveyed sumptive, under-treatment and unnecessary treatment
lacked sufficient bed-nets – illustrates the risks of untested probably occurred, which are known to endanger individ-
assumptions, particularly in view of population sleeping ual patients and may contribute to the emergence of drug
patterns and mobility through forests and borders, which resistance [1].
increases net requirements while enhancing exposure risk.
A recent study in Vietnam found that regular forest work Conclusion
accounted for 53% of P. falciparum infections, with A recent multi-country analysis found increasing use of
increased risk if people used nets at home but not in for- income-generating malaria services and reductions in free
ests [22]. Another found that movement of infrastructure services, with low provider salaries associated with inap-
project workers within forests (which was occurring in our propriate care-giving [26]. Regional disparities in revenue-
site) was a source of ongoing malaria [4]. While respond- raising and human resources can result in uneven imple-
ents – particularly women and the unschooled – require mentation of control programs [27]. In a poor province
an appropriate educational program, it is clear that with limited revenues like Quang Tri, care must be taken
responsibility for non-use of bed-nets, and/or ongoing to ensure that pressures to charge additional service fees
malaria, cannot fully be placed at the feet of this commu- do not discourage people from seeking and completing
nity. malaria treatment. Malaria control in this site cannot be
achieved through community education alone. Focused
A review by Williams and Jones [23] found that malaria training, strategies to attract staff to remote areas, appro-
studies typically focused on the role of mothers or care priate transportation and communication systems, greater
givers in malaria management, while few looked at health efforts to keep (often impoverished) patients under care,
care quality. This is surprising given the pivotal role and robust supply chains for drugs and impregnated bed-
played by both providers and rational drug use. The nets – with regular monitoring of use, quality and suffi-
World Health Organization [24] has noted that health ciency – are among the responses that can further
worker shortages – an increasing global problem and one strengthen Vietnam's efforts to address malaria persist-
found in our site – are linked to higher mortality rates. A ence in this isolated region.
recent review [25] of the impact of health reforms on Viet-
Abbreviations
CHS: Commune Health Station; DH: District Hospital;
FGD: Focus Group Discussion; KAP: Knowledge, Attitudes
and Practices; MC: Malaria Control; NMCP: National
Malaria Control Program; NIMPE: National Institute for
Malariology, Parasitology and Entomology; SSI: Semi-
structuredInterview; VHW: Village Health Worker.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MM conceptualised and designed the study, trained co-
investigators, led the analysis process and was primarily
responsible for drafting the manuscript. QAN coordinated
Figureof
Model
Quang Tri
1 non-biological
province pathways to malaria persistence in the field work and conducted the majority of field
Model of non-biological pathways to malaria persist- research, entered and analysed quantitative data and con-
ence in Quang Tri province. tributed to the analysis process. SC made substantial con-
tributions to training of co-investigators, data analysis and
Page 9 of 10
103
revision of manuscript drafts. BAB contributed to the 18. Hawe P, Degeling D, Hall J: Evaluating Health Promotion: a Health Work-
ers Guide Sydney: Maclennan & Petty; 1990.
analysis process and revision of the manuscript. NHD and 19. Agyepong IA, Manderson L: Mosquito avoidance and bed net use
TTN contributed to analysis of data and revision of the in the Greater Accra region, Ghana. J Biosoc Sci 1999, 31:79-92.
manuscript. All authors read and approved the final man- 20. Khai PN, Van NT, Lua TT, Huu VT, Dang DT, Huong PT, Salazar N,
Sukthana Y, Singhasivanon P: The situation of malaria along the
uscript. Vietnam-Lao PDR border and some related factors. Southeast
Asian J Trop Med Public Health 2000, 31(Suppl 1):99-105.
21. Nguyen QA, Le XH, Hoang NT, Tran QT, Caruana S, Biggs BA, Mor-
Acknowledgements row M: KAP surveys and malaria control in Viet Nam: Find-
The authors are grateful to Atlantic Philanthropies for providing financial ings and cautions about community research. Southeast Asian
support for the project, and to all research participants for their time and J Trop Med Public Health 2005, 36(3):572-577.
inputs. The authors would also like to acknowledge and thank Ms Pham Thi 22. Erhart A, Thang ND, Hung NQ, Toi LV, Hung LX, Tuy TQ, Cong LD,
Xuan and other NIMPE colleagues for their important contribution to field Speybroeck N, Coosemans M, D'Alessandro U: Forest malaria in
Vietnam: a challenge for control. Am J Trop Med Hyg 2004,
data collection, and the Quang Tri provincial and district health staff for 70(2):110-118.
their logistical support. We also thank the reviewers for their useful com- 23. Williams HA, Jones COH: A critical review of behavioural issues
ments on a previous draft of the manuscript. related to malaria control in sub-Saharan Africa: what con-
tributions have social scientists made? Soc Sci Med 2004,
59:501-523.
References 24. World Health Organization: Towards a Scientific Basis for
1. Bloland PB: Drug resistance in malaria. WHO/CDS/CSR/DRS/2001.4, Health Systems. In World Report on Knowledge for Better Health
2001 Geneva: World Health Organization; 2001. Geneva: World Health Organization; 2004.
2. Editorial. BMJ 2004, 328:1086-1087. 25. Fritzen S: Legacies of primary health care in an age of health
3. White NJ, Nosten F, Looareesuwan S, Marsh K, Snow RW, Kokwaro sector reform: Vietnam's commune clinics in transition. Soc
G, Ouma J, Hien TT, Molyneux ME, Taylor TE, Newbold CI, Ruebush Sci Med 2007, 64:1611-1623.
TK 2nd, Danis M, Greenwood BM, Anderson RM, Olliaro P: Avert- 26. Blas E: The proof of the reform is in the implementation. Inter-
ing a malaria disaster. Lancet 1999, 353:1965-1967. national Journal of Health Planning and Management 2004, 19:S3-23.
4. Nam NV, de Vries PJ, Toi LV, Nagelkerke N: Malaria control in 27. Espino F, Beltran M, Carisma B: Malaria control through munici-
Vietnam: the Binh Thuan experience. Trop Med Int Health 2005, palities in the Philippines: struggling with the mandate of
10(4):357-365. decentralized health programme management. Int J Health
5. Heggenhougen HK, Hackethal V, Vivek P: The Behavioural and Social Plann Manage 2004, 19:S155-166.
Aspects of Malaria and its Control UNDP/World Bank/WHO Special
Programme for Research and Training in Tropical Diseases (TDR):
TDR/STR/SEB/VOL/03.1. Geneva: World Health Organization; 2003. Pre-publication history
6. Breman JG: The ears of the hippopotamus: manifestations, The pre-publication history for this paper can be accessed
determinants, and estimates of the malaria burden. Am J Trop here:
Med Hyg 2001, 64(1–2 Suppl):1-11.
7. Foster S: Treatment of malaria outside the formal health
services. J Trop Med Hyg 1995, 98:29-34. http://www.biomedcentral.com/1471-2458/9/85/prepub
8. Tanner M, Vlassoff C: Treatment-seeking behaviour for
malaria: a typology based on endemicity and gender. Soc Sci
Med 1998, 46(4–5):523-532.
9. Erhart A, Thang ND, Xa NX, Thieu NQ, Hung LX, Hung NQ, Nam
NV, Toi LV, Tung NM, Bien TH, Tuy TQ, Cong LD, Thuan LK, Coose-
mans M, D'Alessandro U: Accuracy of the health information
system on malaria surveillance in Vietnam. Trans R Soc Trop
Med Hyg 2007, 101:216-225.
10. Ettling MB: The control of malaria in Vietnam from 1980 to 2000: What
went right? Report of Consultancy for World Health Organization
Regional Office for the Western Pacific, Manila.
11. Hung LQ, de Vries PJ, Giao PT, Nam NV, Tran QB, Chong MT, Quoc
NTTA, Thanh TN, Hung LN, Kager PA: Control of malaria: a suc-
cessful experience from Viet Nam. Bull World Health Organ
2002, 80(8):660-666.
12. Schuftan C: A story to be shared: The successful fight against malaria in
Vietnam WHO WPRO, 11/6/2000. Geneva: World Health Organiza-
tion; 2000.
13. Trung HD, Van Bortel W, Sochantha T, Keokenchanh K, Quang NT,
Cong LD, Coosemans M: Malaria transmission and major
malaria vectors in different geographical areas of Southeast Publish with Bio Med Central and every
Asia. Trop Med Int Health 2004, 9(2):230-237.
14. World Health Organization: Health and Ethnic Minorities in scientist can read your work free of charge
Viet Nam. In Technical Series No 1 Hanoi: World Health Organiza- "BioMed Central will be the most significant development for
tion Viet Nam; 2003.
disseminating the results of biomedical researc h in our lifetime."
15. Agyepong IA, Aryee B, Dzikunu H, Manderson L: The malaria manual:
guidelines for the rapid assessment of social, economic and cultural aspects Sir Paul Nurse, Cancer Research UK
of malaria UNDP; World Bank/Special Programme for Research and
Your research papers will be:
Training in Tropical Diseases (TDR). Geneva: World Health Organi-
zation; 1995. available free of charge to the entire biomedical community
16. World Health Organization WHO/HTM/RBM: Protocols and methods peer reviewed and published immediately upon acceptance
for malaria situation analysis 2003 [http://whqlibdoc.who.int/hq/2003/
WHO_HTM_RBM_2003.47.pdf]. Trial edition. Geneva: World cited in PubMed and archived on PubMed Central
Health Organization yours — you keep the copyright
17. Ministry of Health: Guidelines for Malaria Diagnosis and Treatment in
Vietnam Hanoi: Ministry of Health, The National Malaria Control Submit your manuscript here: BioMedcentral
Project; 2003. http://www.biomedcentral.com/info/publishing_adv.asp
Page 10 of 10
10 4
Ramanadhan et al. Human Resources for Health 2010, 8:17
http://www.human-resources-health.com/content/8/1/17
CASE STUDY Open Access
Network-based social capital and capacity-building

Case study
programs: an example from Ethiopia

Shoba Ramanadhan*1, Sosena Kebede2, Jeannie Mantopoulos2 and Elizabeth H Bradley2
Abstract
Introduction: Capacity-building programs are vital for healthcare workforce development in low- and middle-income
countries. In addition to increasing human capital, participation in such programs may lead to new professional
networks and access to social capital. Although network development and social capital generation were not explicit
program goals, we took advantage of a natural experiment and studied the social networks that developed in the first
year of an executive-education Master of Hospital and Healthcare Administration (MHA) program in Jimma, Ethiopia.
Case description: We conducted a sociometric network analysis, which included all program participants and
supporters (formally affiliated educators and mentors). We studied two networks: the Trainee Network (all 25 trainees)
and the Trainee-Supporter Network (25 trainees and 38 supporters). The independent variable of interest was out-
degree, the number of program-related connections reported by each respondent. We assessed social capital
exchange in terms of resource exchange, both informational and functional. Contingency table analysis for relational
data was used to evaluate the relationship between out-degree and informational and functional exchange.
Discussion and evaluation: Both networks demonstrated growth and inclusion of most or all network members. In
the Trainee Network, those with the highest level of out-degree had the highest reports of informational exchange, χ2
(1, N = 23) = 123.61, p < 0.01. We did not find a statistically significant relationship between out-degree and functional
exchange in this network, χ2(1, N = 23) = 26.11, p > 0.05. In the Trainee-Supporter Network, trainees with the highest
level of out-degree had the highest reports of informational exchange, χ2 (1, N = 23) = 74.93, p < 0.05. The same pattern
held for functional exchange, χ2 (1, N = 23) = 81.31, p < 0.01.
Conclusions: We found substantial and productive development of social networks in the first year of a healthcare
management capacity-building program. Environmental constraints, such as limited access to information and
communication technologies, or challenges with transportation and logistics, may limit the ability of some participants
to engage in the networks fully. This work suggests that intentional social network development may be an important
opportunity for capacity-building programs as healthcare systems improve their ability to manage resources and
tackle emerging problems.
Introduction greatly hinders the improvement of facilities and systems

The global health agenda is increasingly focused on and the ability to provide needed services [2,4-6].
strengthening health systems to improve population-level Successful management and leadership training pro-
health outcomes in low- and middle-income countries grams have improved process-related outcomes (such as
[1]. One component of this strategy focuses on the devel- planning and coordination, delivery of services, and
opment of sufficient workforce capacity, a target area that resource management) in a range of countries, including
has been somewhat resistant to intervention thus far The Gambia, Ethiopia, and Nicaragua [7-9]. Such capac-
[2,3]. The chronic shortage of skilled leadership in the ity-building programs typically target human capital, or
healthcare sectors of low- and middle-income countries increased value of a professional from acquiring knowl-
edge, skills, and other assets that may benefit an employer
* Correspondence: shoba_ramanadhan@dfci.harvard.edu
1 Center for Community-Based Research, Dana-Farber Cancer Institute, 44 or system. Another benefit of these programs, which is
Binney St., LW 703, Boston, MA 02115 USA seldom evaluated, may be the development of social capi-
Full list of author information is available at the end of the article
© 2010 Ramanadhan et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Com-
mons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduc-
tion in any medium, provided the original work is properly cited.
10 5
Ramanadhan et al. Human Resources for Health 2010, 8:17 Page 2 of 11
tal, or resources that exist in a social structure and can be one's own) and effectiveness (development of a strong set
retrieved and utilized to meet specific goals [10]. of key contacts) [18].
Taking a broad view of potential benefits is consistent Social network analysis provides the necessary tools for
with current perspectives on capacity-building, which our analysis as the methodology allows for the assess-
focus on processes that assist individuals, organizations, ment of structures in social relationships, as well as the
and societies in efforts to manage, develop, and utilize the resources exchanged through those relationships [26].
resources at their disposal to solve problems [3,11], here Additionally, given that successful capacity-building
those related to healthcare. This view represents an relies on changes at the individual, organizational, and
intentional shift away from programs focused on techni- system levels [27], the ability to assess relationships and
cal assistance and knowledge transfer towards an endoge- resource flow at multiple levels allows for a holistic
nous process, owned and driven by those who will assessment. For example, a network in which all members
ultimately benefit from and sustain changes in their sys- are connected prompts members to develop trust and a
tems [3]. Capacity-building program participants (and sense of obligation towards each other and encourages
the organizations for which they work) can benefit from the generation of social capital [28]. At the same time, at
increased social capital as participants are able to utilize the individual level, connections to other network mem-
relationships to increase their effectiveness and perfor- bers are expected to provide new access to resources for
mance [10,12,13]. In this way, participants can leverage program participants. If a capacity-building program
relationships to improve communication and collabora- results in network structures that support resource
tion across and within organizations to reach a common exchange, network-based social capital can have an
goal [14,15]. Such benefits are particularly important in impact on the ultimate goal of management training pro-
low-resource settings as organizations are expected to grams: the improvement of trainee performance.
turn to external sources to find needed resources [16]. Despite the number of programs focused on building
healthcare worker capacity [2,7-9] and the understanding
A network perspective on social capital that increased collaboration and partnerships are impor-
Although there are a wide range of conceptualizations of tant outcomes of capacity-building efforts [29], we are
social capital [17], we take a network perspective, which not aware of previous studies examining how such pro-
holds that the extent to which an individual can realize grams may affect the structure and functioning of result-
the benefits of social capital is a function of that individ- ing social networks. Examining this potential impact is
ual's position in a given social network [10,18]. This important to our understanding of the full impact of
drives our focus on: a) the resources that can be accessed capacity-building programs in health. Using survey data
by network members (either directly or through con- from hospital executives participating in an executive-
tacts), and b) the structure of relationships or linkages in education program in Ethiopia [30], we conducted a
a network of interest [10]. In a professional network, key social network analysis to examine the growth of the net-
benefits of increased social capital among colleagues work and the social capital generated by the network (in
include increased exchange of information and resources the form of resource exchange) during the first year of the
[17,19]. For example, sharing of appropriate and timely program. Social network development and social capital
information allows individuals to make strategic adjust- generation were not explicit goals of the training pro-
ments to reach their goals [10,20]. Additionally, partici- gram, but we were able to take advantage of this natural
pants can access novel information by developing experiment to test exploratory hypotheses. We expected
relationships with individuals who are dissimilar in terms to find growth and resource exchange within networks as
of experience and professional contacts [21]. By learning well as a positive association between network connec-
in the context of social relationships, network members tions and resource exchange. We tested these assump-
can come together to identify pressing problems, make tions among a network of program participants and
sense of complex changes in the environment, and among a network of participants plus educators and men-
develop innovative solutions [22,23]. Provision of tangi- tors participating in the program.
ble support or material resources from one network
member to another also improves network members' per- Case description
formance [24]. By tapping into relationships, network Study setting
members can gain access to contacts' resources, and per- The capacity-building program under study was a two-
haps more importantly, to the resources held by the orga- year executive-education Master of Hospital and Health-
nization(s) represented by those contacts [25]. The care Administration (MHA) program in Ethiopia devel-
challenge is to balance efficiency (knowing others who oped by the Federal Ministry of Health (FMOH), the
have contacts and resources that are very different than Clinton HIV/AIDS Initiative (CHAI), Jimma University,
and the Yale School of Public Health [9,31]. The program
10 6
was implemented at the request of the FMOH, with the trainees and supporters (educators and mentors).
goal of developing skilled executives to improve hospital Respondents were presented with a roster that listed all
management in Ethiopia, a low-resource, high-demand trainees and supporters. The survey asked all respon-
setting. This program was part of a larger quality dents to identify trainees and supporters with whom they
improvement effort targeting the Ethiopian healthcare interacted for professional purposes. Respondents also
system, which began decentralization in 1994. The course noted whether or not they were acquainted with each
was offered by Jimma University in Jimma, Ethiopia and network member before the MHA program started. From
was the first graduate-level program for hospital manage- these responses, we derived our measures of interest for
ment in the country. The course was administered and each network.
taught jointly by faculty from Jimma and Yale Universi- We measured a series of network characteristics which
ties, with local coordination provided by a Program have been shown in other settings to promote exchange
Director and Program Assistant. As an executive-educa- of information and flow through networks [26]. These
tion program, the course was offered over two years, with measures were based on data about connections (or
three-week long sessions in residence three times per reported relationships) between network members. Some
year, as well as regular progress reports and evaluations measures focus on presence or absence of a connection,
when trainees were working at their hospitals. whereas others include information about the 'direction'
Executives of public hospitals were eligible to apply. of the connection. For the latter, the measure can capture
The course focused on improving trainees' skills in a whether Member X reported a connection to Member Y,
range of management-related areas, such as human Y reported a connection to X, or both reported a connec-
resources, hospital operations, financial management, tion to each other.
strategic planning, and leadership. Trainees also had the To describe the network as a whole, the first measure of
opportunity to develop professional connections with interest was network density, or the proportion of possi-
each other as well as with leaders and mentors in Ethiopia ble relationships between members that were realized,
and the United States. which described the extent to which network members
are connected, regardless of the direction of connections
Study design and respondents [26]. A more dense, or more highly connected, network
We conducted a cross-sectional study at the end of the may be useful for sharing information and resources and
first year of the MHA program to describe the social net- cooperation, whereas a more sparsely connected network
works that developed during the year. Data were col- may provide greater access to diverse contacts and novel
lected with a self-administered survey of two groups of resources [10,18]. A density level of around 15-20% is
respondents: trainees and supporters. Trainees were the expected to support knowledge-sharing in a network of
first Chief Executive Officers (CEOs) of public hospitals about 100 members [32]. We also identified isolates, indi-
in Ethiopia. Supporters comprised educators and men- viduals who reported no connections to other network
tors formally linked with the MHA program through members. Isolates are of interest as their lack of connec-
either Yale or Jimma University or through CHAI. We tions prevents them from contributing to or benefiting
contacted all 25 trainees enrolled in the MHA program from network membership. Last, we identified compo-
and 38 supporters affiliated with the program to com- nents, or subgroups of members that are not connected
plete the survey. All research procedures were approved to each other and therefore cannot share information and
by the Human Investigation Committee at the Yale resources between subgroups [26].
School of Public Health and the Institutional Review Shifting our focus to individual network members, we
Board at Jimma University. calculated degree, the number of connections between a
given network member and all other network members,
Data collection and measures
regardless of the direction of ties [33]. The bulk of our
The self-administered survey was distributed in Decem-
analyses focused on out-degree, or connections from a
ber 2008 and January 2009 and required approximately
given network member to other network members. Thus,
20 minutes to complete. Paper copies of the survey were
if Member X reported three connections with other net-
distributed to all trainees in residence during the Decem-
work members, that member's out-degree value would be
ber course session and electronic copies were distributed
three, regardless of how many network members
to all other respondents. Surveys were administered in
reported connections to Member X. Compared with
English, which was the language of instruction and a
degree, this measure narrows the focus to connections
requirement for participants in the MHA program.
that may be perceived as functionally useful to respon-
For this study, we focused on two networks: 1) the
dents [34]; here, these connections involve the set of indi-
Trainee Network, which was comprised solely of trainees,
viduals from whom respondents may seek and gain skills.
and 2) the Trainee-Supporter Network, which included
In the Trainee Network, 'trainee out-degree' was the
107
number of connections a trainee reported regarding cators and mentors (n = 38) in addition to trainees (n =
other trainees, grouped into tertiles. In the Trainee-Sup- 25). Sociometric analyses assess the connections between
porter Network, 'trainee-supporter out-degree' was the all members of each network of interest, supporting eval-
number of connections to supporters reported by each uation of network growth and resource exchange [36,37].
trainee, grouped into tertiles. Last, geographic homophily Thus, an individual who was invited to participate, but
referred to whether or not pairs of network members did not fill out a survey, could have been noted as a con-
worked in the same region. tact by another respondent and would still appear in the
To assess potential by-products of social network dataset. Although the Trainee Network is wholly con-
development, we measured informational and functional tained within the Trainee-Supporter Network, we ana-
exchanges, which are complementary manifestations of lyzed them separately to be able to isolate resource
social capital that can help trainees achieve work-related exchange among complementary sets of ties that are
goals [10,24]. Informational exchange refers to access to important for trainees.
necessary knowledge, the ability to transmit it to the cor- Network analysis requires dedicated software to assess
rect person, and acquisition of information with suffi- relational data, and we used UCINET-6 [38]. As network
cient time to react [18]. Trainees were asked whether or data are not independent and do not meet the assump-
not they received guidance in non-classroom settings tions of classical statistical techniques, we utilized proce-
from: a) other trainees, and b) supporters on a series of dures developed for network data available in the
subjects. These topics included: problem-solving, human UCINET software package [38,39]. Thus, the significance
resources, finance management and budgeting, basic tests were based on random permutations of matrices as
public health, biostatistics/research methods, hospital is appropriate for relational data. Here, the significance
operations, strategic management, health policy develop- levels were determined based on distributions created
ment and analysis, health ethics and public health law, from 10 000 random permutations. The analytic proce-
leadership, and management information or tools. The dures also supported comparison of matrices of data.
list of topics was defined in the curriculum as critical to Descriptive measures were calculated using standard
the program and most topics, but not all, were covered in UCINET procedures developed for network data. We uti-
the MHA course at the time of the survey. We created a lized UCINET Contingency Table Analysis to assess the
summary score of the total number of exchanges association of out-degree with two types of resource
reported and dichotomized responses at the 50th percen- exchange. We tested the relationship between geographic
tile for each network, resulting in categories of 'low homophily and connection patterns using UCINET QAP
exchange' and 'high exchange' for each network. Based on Relational Cross-Tabulation.
the distribution of data, 'low exchange represents zero
reported informational exchange in the Trainee Network. Results
Functional exchange described the provision of tangi- Trainee network
ble support from one network member to another [24]. Among trainees, 23 of 25 individuals completed the sur-
Such exchange often involves collaboration between vey (92% response rate). Table 1 describes the character-
institutions or individuals that benefit one party to a istics of trainees' hospitals. The trainee hospitals had on
greater degree, e.g., one individual training another on average 204 beds with a range of 40-800 beds, and the
the use of a new tool. Examples of tangible support can average number of employees per hospital was 399
include sharing of useful tools, policies, and materials or employees, with a range of 82-2500 employees. The
serving as a reference for colleagues [25,35]. Trainees majority of hospitals (72%) were classified as regional;
were asked whether or not they received a series of tangi- one-third were rural.
ble resources from: a) other trainees, and b) supporters. The network graphs comparing connections before the
These resources included: materials and goods (such as program started at year 1 (Figure 1) and key network
surplus supplies), connections/introductions, and hands- measures (Table 2) demonstrate network-level growth.
on instruction, such as through site visits. We created a The network transitioned from having seven isolates
summary score of the total number of exchanges (individuals who were not connected to anyone) and two
reported and dichotomized responses at the 50th percen- components (distinct and isolated subgroups) to having
tile for each network, resulting in categories of 'low zero isolates and only one component. At year 1, the net-
exchange' and 'high exchange' for each network. work demonstrated closure, or the ability of all members
to connect with each other, either directly or through
Analysis contacts. The density of connections increased from 4%
We conducted a sociometric network analysis for both to 13% of all potential connections over the year. In terms
the 25-member trainee network and for the larger 63- of resource exchange, 55% of trainees reported that they
member trainee-supporter network, which included edu- had informational exchanges with other trainees during
10 8
Table 1: Descriptive characteristics for hospitals led by degree and trainee in-degree values at year 1 compared
trainees (n = 25). with the beginning of the program, suggesting that the
network became more centralized, or more centred on a
n Range
subset of individuals.
At year 1, trainees in the lowest out-degree tertile aver-
Hospital location aged 0.5 outgoing connections compared with an average
of 2.0 outgoing connections for the middle tertile, and 6.1
Rural 8 (32%) outgoing connections for the highest tertile. Individuals
with the highest level of connections were more likely to
Urban 17 (68%) be working in the capital city of Addis Ababa compared
with other regions (Fisher's exact test, p = 0.03). We
Number of beds: mean 204 40-800 found a significant (p < 0.001) association between
regional homophily and connections reported at year 1.
Number of employees: mean 399 82-2500
Of potential connections among individuals from the
same region, 45% (45 of 100) were reported compared
with 6% (30 of 500) of potential connections among indi-
Hospital classification
viduals from different regions.
As presented in Table 3, we found that at year 1, trainee
Federal 4 (16%) out-degree was positively associated with informational
exchange, χ2(1, N = 23) = 123.61, p < 0.01. Those with the
Regional 18 (72%) highest tertile of trainee out-degree had the highest
reports of informational exchange. We did not find a sta-
Sub-regional/Zonal 3 (12%) tistically significant relationship between trainee out-
degree and functional exchange, χ2(1, N = 23) = 26.11, p >
0.05.
the first year of the program. The same percentage
reported functional exchange with other trainees. We Trainee-Supporter Network
found that trainee out-degree (the number of connec- For the larger network, 41 of 63 individuals completed
tions reported by the trainee regarding other trainees) the survey (65% response rate), with a 47% response rate
increased from 1.0 to 3.0 connections in the first year of among supporters. Network-level growth was assessed
the program, which was not a statistically significant using a pair of network graphs (Figure 2) and a series of
increase. We found increased variation in trainee out- complementary measures (Table 4). The density
Figure 1 MHA trainee network before the program started (left) and at year 1 (right), n = 25. Key: Circular nodes represent trainees. Node size
represents degree (number of connections); nodes in upper left corner of diagram on left represent isolates (individuals who did not report any con-
nections).
10 9
Table 2: Descriptive measures for the trainee-only sociometric network (25-member network)
Measure Pre-MHA Year 1
Network-level measures
Density (proportion of potential ties that were actually realized) 0.04 0.13
Isolates (members of the network not connected to anyone else) 7 0
Components (distinct and isolated subgroups in the network) 2 0
Individual-level measures
Degree (all connections reported to/from the respondent) Mean: 1.92 Mean: 4.88
SD: 1.79 SD: 4.42
Trainee out-degree (number of connections reported by respondent Mean: 1.04 Mean:3.00

regarding others)
SD: 1.43 SD: 4.62
Trainee in-degree (number of connections reported regarding Mean: 1.04 Mean: 3.00
respondent by others)
SD: 1.25 SD: 1.67
increased from 3% to 13% of all potential ties realized at year 1, ignoring isolates. Again, increased variation in
over the first year of the program. We analyzed density out-degree and in-degree values for the full network from
increases among subgroups and found increased ties the beginning of the program to year 1 suggests that the
from trainees to supporters (3% to 20%), from supporters network became more centralized. Assessing the overall
to trainees (0% to 12%) and from supporters to support- network, the individuals with the most connections in
ers (5% to 9%). The number of isolates decreased from 8 this network were mainly faculty and staff that played a
to 2 in this network, and there was only one component central role in program administration and teaching.
Table 3: Relationship between trainee out-degree and resource exchange at year 1, contingency table analysis (n = 23).
Informational: no Informational: some Functional: no Functional: some

exchange (%) exchange (%) exchange (%) exchange (%)
Trainee out-degree
Low 70.00 8.33 40.00 33.33
Medium 10.00 33.33 30.00 16.67
High 20.00 58.33 30.00 50.00
Observed X2 123.61** 26.11

Key: ~ < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
110
Figure 2 Trainee-supporter network before the program started (left) and at year 1 (right), n = 63. Key: Square nodes represent supporters,
circular nodes represent trainees. Nodes in upper left corner of diagrams represent isolates (individuals who did not report any connections).
When we narrowed our focus to relationships between lihood of resource exchange, as hypothesized based on
trainees and supporters, we found that at year 1, 94% of extant social network literature [10,40]. This level of
trainees reported informational exchange with support- growth and exchange may be expected in high-resource
ers and 55% reported functional exchange with support- professional settings, such as corporations, academic
ers. The average trainee-supporter out-degree at year 1 institutions, or hospital systems in high-income countries
was 8.1 connections. In this network, the average number [32,41] but is impressive in a low-resource setting given
of outgoing connections with supporters was 2.3 for the the level of investment required to support network
lowest trainee-supporter out-degree tertile, 5.3 for the development [40]. The growth is also notable given that
middle tertile, and 14.9 for the highest tertile. Trainee- network development was not an explicit goal of the
supporter out-degree did not vary significantly between training program.
regions. Although the network growth and resource exchange
As seen in Table 5, trainee-supporter out-degree was are promising, limited resources for communication may
positively associated with informational exchange, χ2(1, N have inhibited network development of some network
= 23) = 74.93, p < 0.05. Those in the highest tertile of members. We found that the network of program partici-
trainee-supporter out-degree also had the highest reports pants centered on a subset of individuals from the capitol
of informational exchange. We found a similar pattern for city of Addis Ababa. The centralization of the network is
trainee-supporter out-degree and functional exchange, important because the literature suggests that central
χ2(1, N = 23) = 81.31, p < 0.01. members of a network have higher potential to access and
utilize resources than their colleagues [10,42]. The pat-
Discussion and evaluation tern may reflect the relative ease with which individuals
We found substantial development of social networks from Addis Ababa can interact, without communication
within the context of a capacity-building program in impediments such as transportation and logistics that
healthcare management. Through involvement with the individuals from other regions may face. Information and
MHA program, participants developed professional con- communication technologies, such as mobile phones or
nections with each other and with supporters, including internet, can mitigate challenges of physical distance and
faculty in Ethiopia and hospital executives in the United logistics in low-resource settings [25]. At the time of the
States of America. These connections supported valuable study, reliable access to such technologies was limited for
exchanges including information relating to hospital individuals working outside the Addis Ababa region [43],
management and resources such as hands-on assistance. though these technologies may play an important role in
The networks that developed through the first year of network development in the future. Here, reduced oppor-
this program demonstrated several characteristics that tunities to communicate and interact may have had a
have been shown to support resource exchange such as large impact on resource exchange in this network, as
sufficient network density and connections between all or strong connections are required to support exchange of
almost all members [26,32]. We found that the number of complex information [40].
connections within the network was associated with like-
111
Table 4: Descriptive measures for the trainee-supporter sociometric network (63-member network).
Measure Pre-MHA Year 1
Network-level measures
Density (proportion of potential ties that were actually 0.03 0.13

realized)
Density between and within groups of trainees and Ties among trainees: 0.04 Ties among trainees: 0.13
supporters
Ties from trainees to supporters: 0.03 Ties from trainees to supporters: 0.20
Ties from supporters to trainees: 0.00 Ties from supporters to trainees: 0.12
Ties among supporters: 0.05 Ties among supporters: 0.09
Isolates (members of the network not connected to 8 isolates 2 isolates

anyone else)
Components (distinct and isolated subgroups in the 1 component + isolates 1 component + isolates
network)
Individual-level measures
Degree (all connections reported to/from the Mean: 3.52 Mean: 14.22
respondent)
SD: 3.11 SD: 10.87
Out-degree (number of connections reported by Mean: 1.87 Mean: 8.14

respondent re: others)
SD: 2.88 SD: 10.81
In-degree (number of connections reported re: Mean: 1.87 Mean: 8.14

respondent by others)
SD: 1.77 SD: 5.68
Trainee-supporter out-degree (number of Mean: 1.04 Mean: 8.26

connections reported by trainees regarding
supporters)
SD: 1.46 SD: 5.82
We also saw evidence of the benefits of diverse connec- lower. The severe system-level constraints experienced by
tions for program participants and found that program trainees were evident in a recent assessment of public
participants were able to gain different categories of hospitals engaged in a quality improvement initiative,
resources from different types of network members. This including those represented by trainees in this program
is likely a function of differential access to resources by [44].
individuals in different organizations and levels of power Experience with the MHA program suggests that pro-
[10]. In a low-resource setting, other constraints may also grams to build human resource capacity in low-income
be an important driver of resource exchange. For exam- countries can also increase network-based resources.
ple, the material costs and logistical barriers associated However, given the common challenges of geography and
with providing tangible support to colleagues may be too limited communication technologies in such settings,
great for program participants. For mentors and educa- social network development and resource exchange will
tors, the costs of sharing both types of resources may be likely be more effective if they are integrated as explicit
112
Table 5: Reports of resource exchange by trainee-supporter connection level at year 1, contingency table analysis (n = 23)
Informational: low Informational: high Functional: no Functional: some

exchange (%) exchange (%) exchange (%) exchange (%)
Trainee-supporter out-degree
Low 20.00 6.67 19.51 9.76
Medium 13.33 13.33 14.63 14.63
High 20.00 26.67 19.51 21.95
Observed X2 74.93* 81.31**

Key: ~ < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
goals of training programs to develop human resources novel attempt to study network-based social capital in
for health. For instance, curricula can be developed to capacity-building programs targeting healthcare work-
facilitate opportunities for developing new contacts. The force development. Additionally, our assessment of
focus on development of relationships should extend resource exchange uses a broad view of social capital in
both to fellow trainees as well as supporters of the train- public health settings, rather than the typical focus on
ees, given the breadth of resources that can be accessed communication patterns [46].
through diverse contacts. Another important lesson from Developing human resources for health is an interna-
the MHA experience is the importance of an enabling tional priority in global health [47], and our paper high-
environment. This program was developed at the request lights the importance of taking a broad view of outcomes
of the Ethiopian government and was part of a broader of capacity-building programs. Capacity-building pro-
effort to reform the healthcare system, such as adopting grams provide a unique opportunity to direct interactions
new hospital standards. This climate of organizational between participants and potentially useful contacts
and system change was supportive of changing through coursework, mentoring relationships, and other
approaches to hospital management, and thus presented course-related activities. Active promotion of relation-
an environment in which social capital exchange was ship-building by organizations and/or program develop-
warranted and could have impact. Network development ers can support diversity of contacts and development of
and social capital exchange may be particularly critical in strong channels for knowledge transfer [48-50]. In this
low-resource settings as such networks can foster infor- way, the workforce and system will be better equipped to
mation and function exchanges in inexpensive ways. solve problems in healthcare by more effectively manag-
There are several limitations that help place the results ing, accessing, and utilizing resources, thus truly building
in context. First, although we had a high response rate, capacity [10,11].
some trainees and supporters did not complete the sur-
vey potentially influencing our findings. However, we Conclusions
used out-degree as our independent variable, which is This analysis suggests that network-based social capital
robust to missing data [45]. Second, the data are cross- may be a useful addition to the goals and evaluation of
sectional; thus causation cannot be assessed. However, a capacity-building programs. As discussed by Hawe and
connection must exist between individuals before colleagues [11], social capital deserves further attention
resources can be exchanged across that connection, so in capacity-building efforts as it leaves the system under
the directionality assumed seems plausible. Third, social intervention with greater ability to tackle current issues
desirability bias may have resulted in respondents over- as well as those outside the scope of the program and
reporting connections and/or resource exchanges, future issues. Through active development of diverse pro-
although we encouraged frank responses during survey fessional networks and investment in relationship-build-
administration. Despite these limitations, the study is a ing within the context of system resource constraints,
113
capacity-building programs can build stronger healthcare 17. Portes A: Social capital: its origins and applications in modern
sociology. Annual Review of Sociology 1998, 24(1):1-24.
workforces in low- and middle-income countries. 18. Burt RS: Structural Holes: The Social Structure of Competition.
Cambridge, MA: Harvard University Press; 1992.
Competing interests 19. Greenhalgh T, et al.: Diffusion of innovations in service organizations:
The authors declare that they have no competing interests. systematic review and recommendations. Milbank Quarterly 2004,
82(4):581-629.
Authors' contributions 20. Hawe P, Webster C, Shiell A: A glossary of terms for navigating the field
All authors were involved in study and survey instrument design. SR con- of social network analysis. J Epidemiol Community Health 2004,
ducted the data analysis and drafted the manuscript. EHB, SK, and JM provided 58(12):971-975.
intellectual content and manuscript revisions. All authors read and approved 21. Granovetter M: The strength of weak ties. American Journal of Sociology
the final manuscript. 1973, 78:1360-1380.
22. Lave J, Wenger E: Situated Learning: Legitimate Peripheral
Acknowledgements Participation. Cambridge UK: Cambridge University Press; 1991.
The authors acknowledge the invaluable assistance of Ms. Mahlet Gebeyehu of 23. Weick KE: The Collapse of Sensemaking in Organizations: The Mann
the Jimma-Yale MHA program. SR was funded in part by a grant from the Whit- Gulch Disaster. Administrative Science Quarterly 1993, 38(4):628-652.
ney and Betty MacMillan Center for International and Area Studies at Yale. The 24. Israel BA: Social Networks and Social Support: Implications for Natural
work for this research was also supported by the Clinton HIV/AIDS Initiative. Helper and Community Level Interventions. Health Education and
EHB was supported by the Patrick and Catherine Weldon Donaghue Medical Behavior 1985, 12(65):65-80.
Research Foundation Investigator Award. 25. McPherson JM, Smith-Lovin L, Cook JM: Birds of a Feather: Homophily in
Social Networks. Annual Review of Sociology 2001, 27:415-444.
Author Details 26. Wasserman S, Faust K: Social Network Analysis: Methods and Analysis.
1Center for Community-Based Research, Dana-Farber Cancer Institute, 44 New York: Cambridge University Press; 1994.
Binney St., LW 703, Boston, MA 02115 USA and 2Division of Health Policy and 27. LaFond AK, Brown L, Macintyre K: Mapping capacity in the health sector:
Administration, Yale School of Public Health, 60 College St. P.O. Box 208034, a conceptual framework. International Journal of Health Planning and
New Haven, CT 06520 USA Management 2002, 17:3-22.
28. Burt RS: Brokerage and Closure: An Introduction to Social Capital. New
Received: 8 February 2010 Accepted: 1 July 2010 York: Oxford University Press; 2005.
Published: 1 July 2010 29. Crisp BR, Swerissen H, Duckett SJ: Four approaches to capacity building
Human
©
This
2010
is an
Ramanadhan
Resources
Open
article Access
for Health
is availableetfrom:
article
al; licensee
2010,
distributed
8:17
BioMed
under
Central
the Ltd.
terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
in health: consequences for measurement and accountability. Health

References Promotion International 2000, 15(2):99-107.
1. Reich MR, Takemi K: G8 and strengthening of health systems: follow-up 30. Kebede S, et al.: Educating Leaders in Hospital Management: A New
to the Toyako summit. Lancet 2009, 373(9662):508-515. Model in Sub-Saharan Africa. International Journal for Quality in Health
2. World Health Organization: Everybody's business: strengthening health Care 2010, 22(1):39-43.
systems to improve health outcomes: WHO's framework for action. 31. Kebede S, et al.: Educating Leaders in Hospital Management: A New
World Health Organization.: Geneva; 2007. Model in Sub-Saharan Africa. International Journal for Quality in Health
3. Organization for Economic Co-operation and Development: The Care 2009 in press.
Challenge of Capacity Development: Working Towards Good Practice. 32. Parise S: Knowledge Management and Human Resource Development:
Development Assistance Committee, Organization for Economic Co- An Application in Social Network Analysis Methods. Advances in
operation and Development,: Paris; 2006. Developing Human Resources 2007, 9(3):359-383.
4. Filerman GL: Closing the management competence gap. Human 33. Huang K, Provan KG: Resource Tangibility and Patterns of Interaction in
Resources for Health 2003, 1(7):. a Publicly Funded Health and Human Services Network. Journal of
5. Egger D, et al.: Strengthening Management in Low-Income Countries. Public Administration Research and Theory 2007, 17(3):435-454.
Evidence and Information for Policy: World Health Organization: Geneva, 34. Hansen MT: The search-transfer problem: The role of weak ties in
Switzerland; 2005. sharing knowledge across organization subunits. Administrative Science
6. Koplan JP, et al.: Towards a common definition of global health. Lancet Quarterly 1999, 44(1):82-111.
2009, 373(9679):1993-1995. 35. Ozcan P, Eisenhardt KM: Origin of Alliance Portfolios: Entrepreneurs
7. Conn CP, Jenkins P, Touray SO: Strengthening health management: Network Strategies and Firm Performance. Academy of Management
experience of district teams in The Gambia. Health Policy and Planning Journal 2009, 52(2):246-279.
1996, 11(1):64-71. 36. Valente TW: Network Models and Methods for Studying the Diffusion
8. McEwan E, et al.: Developing public health management training of Innovations in Models and Methods in Social Network Analysis.
capacity in Nicaragua. American Journal of Public Health 2001, Edited by: Carrington PJ, Scott J, Wasserman S. Cambridge University
91(10):1586-8. Press: Cambridge England; 2005:98-116.
9. Bradley E, et al.: Hospital quality improvement in Ethiopia: a 37. Kawachi I, Subramanian SV, Kim D: Social Capital and Health: A Decade
partnership-mentoring model. International Journal of Quality in Health of Progress and Beyond in Social Capital and Health. Edited by:
Care 2008, 20(6):392-399. Kawachi I, Subramanian SV, Kim D. Springer: New York; 2008:1-26.
10. Lin N: Social capital: a theory of social structure and action. New York: 38. Borgatti SP, Everett MG, Freeman LC: UCINET for windows: Software for
Cambridge University Press; 2001. social network analysis. Analytic Technologies: Harvard MA; 2002.
11. Hawe P, et al.: Multiplying health gains: the critical role of capacity- 39. Krackhardt D: Predicting with social networks: Nonparametric multiple
building within health promotion programs. Health Policy 1997, regression analysis of dyadic data. Social Networks 1988, 10(4):359-382.
39(1):29-42. 40. Reagans R, McEvily B: Network Structure and Knowledge Transfer: The
12. Bourdieu P: The forms of capital. In Handbook of Theory and Research for Effects of Cohesion and Range. Administrative Science Quarterly 2003,
the Sociology of Education Edited by: Richardson JG. Greenwood: New 48(2):240-267.
York; 1985:241-258. 41. Sparrowe RT, et al.: Social networks and the performance of individuals
13. Scott C, Hofmeyer A: Networks and social capital: a relational approach and groups. Academy of Management Journal 2001, 44(2):316-325.
to primary healthcare reform. Health Research Policy and Systems 2007, 42. Valente TW, Chou CP, Pentz MA: Community coalitions as a system:
5(9):. effects of network change on adoption of evidence-based substance
14. March JG, Simon HA: Organizations. New York: John Wiley & Sons; 1958. abuse prevention. American Journal of Public Health 2007, 97(5):880-886.
15. Perrow C: The short and glorious history of organizational theory. 43. Demeke M, Biru T: ICT Penetration and Usage in Ethiopia: Baseline
Organizational Dynamics 1973, 2(1):3-15. Study. In Department of Economics Addis Ababa University: Addis Ababa;
16. Pfeffer J, Salancik GR: The External Control of Organizations. New York: 2002.
Harper and Rowe; 1978.
114
44. Hartwig K, et al.: Hospital management in the context of health sector

reform: a planning model in Ethiopia. International Journal of Health
Planning and Management 2008, 23(3):203-218.
45. Costenbader E, Valente TW: The stability of centrality measures under
conditions of imperfect data. Social Networks 2003, 25:283-307.
46. Lakon CM, Godette GC, Hipp JR: Network-Based Approaches for
Measuring Social Capital in Social Capital and Health. 2008:63-81.
47. World Health Organization: Report on the WHO/PEPFAR consultation on
maximizing positive synergies between health systems and Global
Health Initiatives through work on building and sustaining health
workforce development. World Health Organization,: Washington DC;
2009.
48. Garvin DA, Edmondson AC, Gino F: Is Yours a Learning Organization?
Harvard Business Review 2008, 86(3):109-116.
49. Massoud MR, et al.: A Framework for Spread: From Local Improvements
to System-Wide Change. IHI Innovation Series white paper. Institute for
Healthcare Improvement: Cambridge MA; 2006:1-20.
50. Li LC, et al.: Use of communities of practice in business and health care
sectors: A systematic review. Implementation Science 2009, 4(27):.
doi: 10.1186/1478-4491-8-17
Cite this article as: Ramanadhan et al., Network-based social capital and
capacity-building programs: an example from Ethiopia Human Resources for
Health 2010, 8:17
115
Health Policy and Planning Advance Access published June 7, 2011
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/2.5/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine Health Policy and Planning 2011;1–13
! The Author 2011; all rights reserved. doi:10.1093/heapol/czr044
Strategies for coping with the costs of

inpatient care: a mixed methods study of
urban and rural poor in Vadodara District,
Gujarat, India
Michael Kent Ranson,1* Rupal Jayaswal2 and Anne J Mills3
1
Technical Officer, Alliance for Health Policy and Systems Research, World Health Organization, Geneva, Switzerland, 2Project Officer,
Indian Institute of Technology (Madras), Chennai, India, 3Professor of Health Economics and Policy, London School of Hygiene and
Tropical Medicine, London, UK
*Corresponding author. Technical Officer, Alliance for Health Policy and Systems Research, World Health Organization, 20 avenue Appia,
1211 Geneva 27, Switzerland. E-mail: ransonm@who.int. Tel: þ41 22 791 54 25. Fax: þ41 22 791 48 17.
Accepted 14 March 2011

Background In India, coping mechanisms for inpatient care costs have been explored in rural
areas, but seldom among urbanites. This study aims to explore and compare
mechanisms employed by the urban and rural poor for coping with inpatient
expenditures, in order to help identify formal mechanisms and policies to
provide improved social protection for health care.
Methods A three-step methodology was used: (1) six focus-group discussions; (2) 800 exit
survey interviews with users of public and private facilities in both urban and
rural areas; and (3) 18 in-depth interviews with poor (below 30th percentile of
socio-economic status) hospital users, to explore coping mechanisms in greater
depth.
Results Users of public hospitals, in both urban and rural areas, were poor relative to
users of private hospitals. Median expenditures per day were much higher at
private than at public facilities. Most respondents using public facilities (in both
urban and rural areas) were able to pay out of their savings or income; or by
borrowing from friends, family or employer. Those using private facilities were
more likely to report selling land or other assets as the primary source of coping
(particularly in rural areas) and they were more likely to have to borrow money
at interest (particularly in urban areas). Poor individuals who used private
facilities cited as reasons their closer proximity and higher perceived quality of
care.
Conclusions In India, national and state governments should invest in improving the quality
and access of public first-referral hospitals. This should be done selectively—
with a focus, for example, on rural areas and urban slum areas—in order to
promote a more equitable distribution of resources. Policy makers should
continue to explore and support efforts to provide financial protection through
insurance mechanisms. Past experience suggests that these efforts must be
carefully monitored to ensure that the poorer among the insured are able to
access scheme benefits, and the quality and quantity of health care provided
must be monitored and regulated.
Keywords Hospitalization, expenditures, coping strategies, insurance, social capital, India
116
2 HEALTH POLICY AND PLANNING
KEY MESSAGES
! In both urban and rural areas, respondents using public hospitals were poorer than those using private hospitals.
! While most of those hospitalized at public facilities were able to pay out of savings or income, or by borrowing from
friends, family or employer, individuals using private facilities were more likely to sell land or other assets (rural areas)
and they were more likely to have to borrow money at interest (urban areas).
! In India, national and state governments should invest in improving the quality and access of public first-referral
hospitals. This should be done selectively—focusing on rural areas and urban slum areas—to promote a more equitable
distribution of resources.
Introduction financial shocks. The strategies have been categorized as

informal (or traditional) and formal, i.e. market-based or
The aim of this study is to explore mechanisms employed by
publicly provided (World Bank 2001). In India, formal mech-
the urban and rural poor for coping with inpatient expend-
anisms are largely inaccessible to the poor. In theory, govern-
itures, in order to help identify formal mechanisms and policies
ment provision of universal and free health care should
to provide improved social protection for health care for these
cover the poor, but in practice it often does not.
populations.
Private-for-profit health insurance and government social
insurance are geared primarily towards India’s formal sector,
Background comprising less than 10% of the total population (Gupta and
In India, health care costs, and those for inpatient care in Trivedi 2005).
particular, pose a barrier to seeking health care, and can be a There have been many studies of the informal mechanisms
major cause of indebtedness and impoverishment. Three- used by poor rural households for coping with financial shocks,
quarters (1999 to 2003 estimates) of total health care spending and more specifically those related to medical expenditure
flows from individuals and households directly to health care (Rosenzweig 1988; Townsend 1994; Kochar 1995; Krishna et al.
providers (predominantly private) in the form of out-of-pocket 2003). In rural areas, when a shock hits, people cope by selling
payments (World Health Organization 2006a). On average, the livestock or other assets, or calling on support networks
poorest quintile of Indians is 2.6 times more likely than the for transfers or loans. If these mechanisms fail or fall short,
richest to not seek medical treatment when ill, and only households may increase their labour supply, working
one-sixth as likely to be hospitalized (Peters et al. 2002). more hours or involving more household members (women or
In India, 31.1% of individuals fall below the World Bank’s children), or borrow from a private lender at high rates
absolute poverty threshold of US$1.08 per day. Subtraction of interest. In Gujarat, the rates of interest on these
of out-of-pocket payments for health care increases the loans start as high as 5–7% per month (Krishna et al. 2003).
poverty headcount to 34.8%, representing over 37 million If all else fails, households reduce consumption—including
people (van Doorslaer et al. 2006). A study of more than 3000 the consumption of medical services and goods—and go
households in 20 Gujarati villages found that, over 25 years, hungry.
among all households that fell into poverty, ill health and Far less is known about coping strategies among the poor
related expenses were critical in 85% of cases (Krishna et al. living in urban areas, including those who have migrated from
2003). rural areas for work. It has, however, been argued that the
In Gujarat (as in all India), the poor are more likely than the strong social networks that provide mutual insurance in India
rich to choose public rather than private inpatient facilities are actually a deterrent to mobility, and explain rates of
(Mahal et al. 2000). Nonetheless, 54.9% of all hospitalizations urbanization that are low relative to other low-income countries
among the rural population below the poverty line (BPL) of (Munshi and Rosenzweig 2005). These authors show, based on
Gujarat, and 51.1% of all hospitalizations among the urban BPL 1982 and 1999 survey data, that migration (not necessarily to
population, are in private hospitals.1 Hospital charges faced by urban areas, but away from one’s native village) is associated
the urban BPL population are higher than those faced by their with a significantly lower probability of receiving loans from
rural counterparts, at both public and private facilities. For friends or family. International literature suggests that those
example, the average charge per day of hospitalization in a who have migrated for work are particularly vulnerable as they
private facility, for BPL patients, is Rs. 125.7 in urban Gujarat are more likely to work longer hours, live and work in poorer
compared with only Rs. 57.8 in rural Gujarat (Mahal et al. conditions, be socially isolated and lack access to basic
2000). amenities (International Organisation for Migration 2005).
The term ‘coping strategies’ was coined during the 1980s in In 2002, a World Bank consultation highlighted the fact that,
the literature on household responses to lack of food and despite a growing population of urban poor (27% of 285 million
nutrition. The term refers to the mechanisms or activities people), for whom many health indicators are as bad or
undertaken by members of a household that help them worse than for poor rural populations, there has been little
survive through a crisis situation (McIntyre and Thiede 2007). analytical work on health issues of the urban poor (World Bank
Figure 1 lists some of the measures available for coping with 2002).
117
STRATEGIES FOR COPING WITH INPATIENT CARE COSTS 3
Figure 1 Mechanisms for coping with financial shocks. Source: World Bank (2001)
Methods migrants were expected to be the most vulnerable to the costs

of health care. In rural areas, three sub-districts (out of 12 in
Study setting
Vadodara) were randomly selected, and within each sub-district
The study was conducted in Vadodara District, Gujarat State, a poor residential area (usually a ‘para’ or neighbourhood) on
between October 2007 and December 2008. Vadodara District the periphery of a village was purposefully selected. Group size
has a population of 3.6 million people, 52.1% of whom are male and inclusion/exclusion criteria were the same as in urban
and 47.9% female (Office of the Registrar General 2001). The areas, with the exclusion of the migration criterion.
district’s only city, Vadodara, is the third largest city in the state FGDs were conducted in the Gujarati language by the
of Gujarat and the sixteenth largest in India (UNDP and World Principal Investigator (RJ) and with the permission of respond-
Bank 1999). In 2001, the urban population of Vadodara ents, recorded using a digital video recorder. They were
accounted for 45% (1 647 317) of the total district population. transcribed in English, and analysed and coded in MS Word.
Like many other Indian cities, Vadodara is growing rapidly,
from only 950 000 in 1981, to 1.3 million in 1991 and 3.6 million
in 2001 (Office of the Registrar General 2001). The dominant Step 2: Exit surveys
industries in Vadodara city are petrochemicals, fertilizers, The goals of the exit survey were: (i) to document costs of
pharmaceuticals, cotton textiles and machine tools. hospitalization (both medical and non-medical) at private and
Agriculture predominates in rural areas, with the major crops public facilities, and in urban and rural areas; (ii) to explore the
being: rice, wheat, yellow peas, grams, oilseeds, groundnut, utilization of different coping strategies; and (iii) to identify
tobacco, cotton and sugarcane. poor households who could be interviewed, in-depth, in the
final step of field-work.
Eight-hundred respondents were sought, with equal numbers
Step 1: Focus group discussions in urban and rural areas, and equal numbers using the public
Focus group discussions (FGDs) were conducted as the first and private hospitals that were most commonly mentioned in
step in a three-step methodology. The FGDs aimed to: the focus-group discussions. In rural areas, respondents had to
(i) identify commonly used sources of inpatient care among be resident in the three sub-districts included in Step 1. Urban
poorer populations; and (ii) explore strategies for coping with respondents had to be resident in Vadodara city. In both urban
the costs of inpatient care, to help develop a list of options for and rural areas, hospitals were purposively selected based on
questions in a hospital exit survey. Three FGDs were conducted frequency of use reported by respondents in the FGDs. The
in urban areas and three in rural areas. In Vadodara city, three rural hospitals tended to be much smaller than the urban
different urban slum areas were purposefully selected. facilities; hospitals had to have a minimum of 15 inpatient beds
Vadodara slum areas are quite segregated according to the in order to be included in the study.
State of origin of the residents; our FGD groups consisted of One hundred exit surveys were conducted at each of four
migrants from Uttar Pradesh, Rajasthan and Maharashtra. urban hospitals (two public and two private). Given the smaller
Eight to 12 adult respondents (>18 years of age) were included size of rural hospitals, interviews had to be conducted at six
in each FGD, including both men and women, and only those facilities (three public and three private), with 65–70 respond-
who had experienced hospitalization within the previous year ents per hospital. Potential respondents were identified by
and had migrated to the city within the last 5 years, as recent having hospital administrators provide a list of patients to be
118
discharged on the day of exit interviews. Exit interviews were

restricted to those hospitalized in general wards (thus excluding
those who paid extra—both at public and private facilities—to
stay in private rooms). Patients (and their families) were
approached for interviews immediately after they made their
payments and had received their discharge cards. In order to be
included in the exit surveys, respondents had to be: (i) older
than 18 years of age; (ii) hospitalized for more than 24 hours;
and (iii) resident in the corresponding area (either urban or
rural) at the time of the interview (for example, residents of a
rural village hospitalized in urban facilities were excluded from
the study).
Interviews were conducted inside hospital premises by RJ and
three trained investigators. Data were collected using an
interview schedule which was filled out by the interviewer.
Figure 2 Distribution of in-depth interview respondents
The following data were collected:
! Place of residence and place of origin;
! Details as to when they moved to their current place of areas, migration status (6 non-migrants and 6 recent migrants)
residence; (Figure 2).2
! Cause of hospitalization; All interviews were conducted in Gujarati by RJ with the
! Expenditures on hospitalization, with breakdown by type of assistance of one trained investigator. For all interviews, the
costs, e.g. medicines, doctors’ fees, etc; spouse or other family members were present as well as
! Indicators of socio-economic status. the hospitalized person. A semi-structured interview guide was
In most cases the patient was interviewed (generally with used. Interviews were recorded, with the permission of
their accompanying family present). In those cases where respondents, using a digital video recorder, transcribed in
patients were unable to respond (for example, if the patient English, and analysed and coded in MS Word.
remained ill or unresponsive at the time of discharge) we
interviewed an accompanying household member instead. As
Ethical approval
anticipated, 800 exit interviews were conducted. In no case did
potential respondents refuse to participate in the interviews. Ethical approval was obtained from the Ethics Committee of
Data were double-entered into an Excel database, and The London School of Hygiene and Tropical Medicine and from
cross-checked for any inconsistencies. Analyses were conducted the Health Commissioner of Gujarat State. Free and informed
using the statistical software STATA. As a proxy for wealth, we consent of all respondents was obtained; this consent was
constructed a socio-economic status (SES) index based on taken in verbal rather than written form, given high rates of
household assets and utilities, allowing the weights of these illiteracy in Gujarat State.
assets to be determined by principal components analysis (PCA)
(Filmer and Pritchett 2001). All 26 assets and utilities variables
from the survey were retained in the index (see Appendix 1)
and weighted based on PCA. Twenty-one categorical variables Results
were converted to dichotomous variables as this provided for Exit surveys
greater discrimination amongst poorer households. Ultimately, Table 1 describes the surveyed population, and highlights the
the index comprised 25 dichotomous variables and one main differences between urban and rural respondents, and
continuous variable (number of rooms). The index was users of public and private facilities. A majority of respondents
validated by examining the likelihood of ownership of specific were male. Urban respondents and users of private facilities
assets (or utilities) by decile. For example, it can be seen that were more likely to be male than their counterparts in rural
no respondent below the 50th percentile reported owning a areas and at public facilities. Distribution of the surveyed
refrigerator, compared with 65% of respondents in the wealth- population by quintiles of SES suggests that rural respondents
iest decile (Appendix 1). Respondents were grouped by quintile were poorer than urban, and users of public facilities poorer
or decile; in both cases the 1st was the poorest. than those using private facilities. A slightly higher percentage
of urban residents reported having moved to their current place
of residence within the last year (8% vs 2%). The broad
Step 3: In-depth interviews categories of illnesses reported by respondents differed little
After the exit surveys, 18 in-depth interviews were carried out between urban and rural areas. However, respondents at public
in order to explore further household coping strategies. We facilities were more likely to report infectious ailments as the
aimed to interview people from the poorest three deciles by primary cause of hospitalization (in rural areas), were more
SES. The respondents were stratified according to type of likely to report non-infectious ailments (both in urban and in
hospital used (half had used public facilities and half private), rural areas) and were less likely to report accidents and injuries.
place of residence (6 rural and 12 urban), and within urban The median duration of hospital stay differed little between
119
Table 1 Description of study population, by place of residence and type of hospital used (percentage distributions unless otherwise indicated),
Vadodara, India
Urban Rural
Public Private Total Public Private Total
Gender (n ¼ 200) (n ¼ 200) (n ¼ 400) (n ¼ 200) (n ¼ 200) (n ¼ 400)
Male 55.5 75.5 65.5 52.5 68.5 60.5
Female 44.5 24.5 34.5 47.5 31.5 39.5
SES (n ¼ 199) (n ¼ 200) (n ¼ 399) (n ¼ 199) (n ¼ 199) (n ¼ 398)
1 22.61 2.50 12.53 38.19 18.09 28.14
2 23.62 9.50 16.54 23.62 23.12 23.37
3 26.13 14.50 20.30 13.07 26.13 19.60
4 17.59 30.50 24.06 17.09 14.07 15.58
5 10.05 43.50 26.57 8.04 18.59 13.32
Migrant status (n ¼ 200) (n ¼ 200) (n ¼ 400) (n ¼ 200) (n ¼ 200) (n ¼ 400)
Recent migrant (<1 year) 5.5 10.5 8 1 3 2
Non-migrant (resident >1 year) 94.5 89.5 92 99 97 98
Primary cause of hospitalization (n ¼ 200) (n ¼ 200) (n ¼ 400) (n ¼ 200) (n ¼ 200) (n ¼ 400)
Infectious 42 41.5 41.75 49.5 38.5 44.00
Non-infectious 48 30.0 39.00 43.0 34.5 38.75
Injuries 10 28.5 19.25 7.5 27.0 17.25
Days of hospitalization (median) (n ¼ 200) (n ¼ 200) (n ¼ 400) (n ¼ 200) (n ¼ 200) (n ¼ 400)
5 7 6 5 7 5
urban and rural areas, but was shorter at public facilities slightly greater reliance on savings and income amongst the
(5 days) than at private facilities (7 days). less poor 70%. Among rural, private hospital users, the poorest
Table 2 describes the hospital expenditures reported by exit 30% were much more likely than the less poor 70% to have
survey respondents. The median expenditure per day was borrowed from friends, family and relatives, or to have
almost three times as high among urban residents (Rs. 395) borrowed money on interest, and less likely to have relied on
relative to rural residents (Rs. 138). Expenditures per day were savings and income. It is difficult to comment on ‘poor–less
higher at private vs public facilities, and this difference was poor’ differences among those using urban, private hospitals, as
especially marked in urban areas. Medicine fees were an only 11 respondents falling below the 30th decile used these
important component of total costs in both urban and rural facilities.
areas. Doctors’ fees were a more important component of the
costs at private vs public facilities. Among non-medical fees, the In-depth interviews
costs of food and travel were particularly important compo- Table 3 provides a description of the 18 in-depth interview
nents of total costs at rural public facilities (at 12.4% and 17.6% respondents (references to respondents provided below corres-
of total costs, respectively). pond to the respondent identification numbers in this table).
Figure 3 provides an overview of the different (primary)
coping strategies reported by exit survey respondents. It is clear Poor people choose public facilities due to lower cost
that for hospitalizations at public facilities (both in urban and In-depth interview respondents who used public facilities
in rural areas) most respondents were able to pay out of their generally reported that they did so because these were
savings or income, or by borrowing from friends, family or perceived to be less costly than private hospitals.
employer. For hospitalization at private facilities, these were
also common means of paying. But relative to users of public ‘‘Because we did not have money, we had to go to the public
facilities, those using private facilities were more likely to report hospital. We heard that they do not charge services at the public
selling land or other assets as the primary source of coping hospital.’’ (Respondent 1)
(particularly in rural areas), and they were more likely to have
to borrow money at interest (particularly in urban areas). While this was the most common reason for choosing a public
Differences in coping strategies between the poorest 30% of facility, some reported that even the fees charged at the public
respondents and the less poor 70% are illustrated in Figures 4a facility were unaffordable.
and 4b.3 In general, there seems to be remarkably little
difference in primary coping strategies between the poor and ‘‘We chose the public hospital because they would provide treatment
the less poor who used public facilities, although there was a for free. But even they charged money . . . five hundred
120
Table 2 Expenditures on hospitalization, by place of residence and type of hospital used, Vadodara, India
Urban Rural
Public Private Total Public Private Total
Observations 198 198 396 200 200 400
Median total expenditures (Rs.) 677.5 4330.5 2525 390 1530 910
Median daily total expenditures (Rs./d) 119.5 629.72 394.44 93.75 277.64 138.33
% breakdown of total expenditures
Medical fees
Doctors’ fees 0.2 31.8 24.9 2.6 19.2 16.3
Medicine fees 55.5 25.3 31.9 51.5 40.4 42.3
Bed fees 0.5 14.4 11.4 4.3 14.2 12.5
Laboratory fees 21.6 9.5 12.1 4.8 8.3 7.7
Ambulance charges 1.1 0.5 0.6 0.4 0.1 0.1
Other 5.5 8.8 8.1 6.0 6.9 6.7
Sub-total 84.5 90.3 89.1 69.7 89.1 85.7
Non-medical fees 0.0 0.0 0.0 0.0 0.0 0.0
Food 6.3 5.4 5.6 12.4 5.4 6.6
Travel 9.0 4.3 5.3 17.6 5.5 7.6
Lodging 0.0 0.0 0.0 0.0 0.0 0.0
Other 0.2 0.0 0.0 0.3 0.0 0.1
Sub-total 15.5 9.7 10.9 30.3 10.9 14.3
Total fees 100.0 100.0 100.0 100.0 100.0 100.0
Figure 3 Strategies used for coping with hospitalization costs, by place of residence and type of hospital used, Vadodara, India. Legend (for x axis):
1 ¼ Savings or income; 2 ¼ Borrowed from friends, family or employer; 3 ¼ Borrowed on interest, from moneylender or bank; 4 ¼ Sold land or other
assets; 5 ¼ Other (including ‘did extra labour’, ‘don’t know’) (n ¼ 200; 200; 200; 200)
121
(a)
(b)
Figure 4 (a) Strategies used for coping with hospitalization costs: poorest 30%, by place of residence and type of hospital used (n ¼ 108; 53; 68; 11),
(b) Strategies used for coping with hospitalization costs: wealthiest 70%, by place of residence and type of hospital used (n ¼ 91; 146; 131; 189).
Legend (for x axis): 1 ¼ Savings or income; 2 ¼ Borrowed from friends, family or employer; 3 ¼ Borrowed on interest, from moneylender or bank;
4 ¼ Sold land or other assets; 5 ¼ Other (including ‘did extra labour’, ‘don’t know’).
122
Decile
rupees . . . How can we afford to pay this amount?’’

(Respondent 14)
3
1
2
3
3
4
4
1
2
1
3
2
4
3
3
1
Casual wage labourer
Casual wage labourer
Reasons for using private facilities
Domestic worker
Domestic worker
Self-employed
Self-employed
Self-employed
Respondents cited a number of reasons for using private
Farm worker
Occupation
Unemployed
Unemployed
Rag picker
Housewife
Housewife
Housewife
Housewife
hospitals. The most commonly cited reason was that the private
Labourer
Salaried
Retired
hospital was nearby to the respondent’s home:
‘‘(We chose the private hospital) only because it was nearby and
she (Respondent 11) was in a great deal of pain. It was an
Age
55
30
40
18
60
19
28
21
33
30
18
32
28
47
20
40
40
32
emergency and the first hospital that I thought of was this nearby
private hospital.’’ (Relative of respondent 11)
Borrowed from friends/relatives/employer



Borrowed from moneylender on interest

Several respondents also expressed the belief that private
Primary coping strategy used
hospitals are of higher quality than public:
‘‘We do not use the public hospital, as the treatment given there is
Sold land or other assets

Sold gold/silver jewellery
not good. It is very dirty and unhygienic. The private hospital is

very clean and well maintained.’’ (Respondent 18)
Savings/income
Savings/income
Savings/income
Savings/income
Poor people rely on multiple coping strategies.

Respondents generally reported using multiple (two or more)
different strategies for coping with the costs of hospitalization.
Respondent 8, for example, was a young woman from
Tot. Exp. (Rs.)
Vadodara city, hospitalized in a public facility for gastroenter-

itis. In order to cover the cost of the hospital stay—Rs. 600—the
family drew on their savings and borrowed money from
985
535
330
660
1550
2250
5300
900
600
890
275
1500
4640
1560
2500
23300
12700
26500
relatives:
‘‘I bought medicines using the money I borrowed from him (my
Cause of hospitalization
Fever of unknown origin
brother) . . . I took 400 rupees from my brother, and I gave

Accident/injury/fracture
Accident/injury/fracture
approximately 100 rupees from my own (savings). As well, I later

Diarrhea/dysentery
General weakness
borrowed 100 rupees from another relative in order to pay the

Dental problem
Blood pressure
Gastroenteritis
medical shop.’’ (Mother of respondent 4)

Hysterectomy
Hysterectomy
Heart disease
Dengue fever
Anaemia
Abortion
Diabetes
Delivery
Asthma
Malaria
Respondent 1, a recent migrant to Vadodara, was also

hospitalized in a public hospital, for fever of unknown origin.
His family, who earn a living by selling rags and scraps
Hospital
salvaged from garbage, drew on savings and sold a small

Private
Private
Private
Private
Private
Private
Private
Private
Private
Public
Public
Public
Public
Public
Public
Public
Public
Public
amount of jewellery:
Table 3 Description of 18 in-depth interview respondents
‘‘. . . we paid using our own money—money that we had saved for
Place of origin
Diwali . . . We also had to pawn our belongings. We pawned a silver

Within Gujarat
Within Gujarat
(non-migrant)
(non-migrant)
(non-migrant)
(non-migrant)
(non-migrant)
(non-migrant)
Uttar Pradesh
ornament that was on my son’s leg. We removed (that ornament)

Rajasthan
Rajasthan
Rajasthan
and pawned it . . . Now it is gone. I could not repay the money (to
the pawn-broker) so we lost the ornament.’’ (Sister-in-law of
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
respondent 1)
Gender
Poor who use public facilities can pay from savings, income or
M
M
M
M
M
M
M
F
F
F
F
F
relying on social networks

As was suggested by the exit survey data, respondents who
City/sub-dist.
used public hospitals could generally cope with the costs either
Sankheda
Sankheda
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
Vadodara
through their savings and income or through borrowing from

Dabhoi
Dabhoi
Padra
Padra
friends, family or employers:
‘‘We paid using money that we had saved for Diwali. We do not
No.
10
11
12
13
14
15
16
17
18
have any relatives from whom we can borrow.’’ (Respondent 1)

1
2
3
4
5
6
7
8
9
123
Poor who use private facilities use different coping strategies Discussion and conclusions
(viz. a viz. those who use public)
Poor respondents who used private facilities were more likely Summary of findings
to report having borrowed money on interest or selling assets. Users of public hospitals, in both urban and rural areas, were
This was the case, for example, for the following respondent poor relative to users of private hospitals. Median expenditures
who reported total expenditures of Rs. 26 500: per day were higher at private vs public facilities, and this
difference was especially marked in urban areas. For hospital-
‘‘He (my brother) owns land, which he mortgaged. We paid the izations at public facilities (both in urban and in rural areas),
money that we got after mortgaging the land. Even today I don’t most respondents were able to pay out of their savings or
have any money saved. I roll cigarettes every day and earn income; or by borrowing from friends, family or employer.
money from this. But I spend this money on medicines.’’ Those using private facilities were more likely to report selling
(Respondent 16) land or other assets as the primary source of coping (particu-
larly in rural areas) and they were more likely to have to
Key differences in coping strategies between urban and borrow money at interest (particularly in urban areas).
rural poor In-depth interview respondents (those below the 30th percent-
In the in-depth interviews, respondents from rural areas were ile of SES) who used public hospitals often did so because of
more likely to report having borrowed from family members, their lower perceived cost. Poor respondents reported using
while those in urban areas were more likely to have borrowed multiple different coping strategies. Respondents from urban
from friends or employers. The following urban respondent, for areas were more likely to have borrowed from friends or
example, borrowed from neighbours: employers, while those in rural areas were more likely to report
having borrowed from family members. Recent migrants to
‘‘I borrowed some money from my neighbours . . . And we don’t urban areas were less likely than permanent residents to report
even have much income so that we can save (to repay this loan). borrowing from relatives, but in some cases were able to borrow
Our neighbours are our best friends . . . We have not been here for from neighbours or employers.
long, however they have been very helpful.’’ (Respondent 4)
In several cases, the rural poor reported selling or mortgaging

Discussion
land in order to cope with the costs of hospitalization (e.g. Studies that have focused on the costs of inpatient care in
respondents 16 and 18) while some urbanites reported selling India, and the related coping strategies, are relatively few.
other assets: Consistent with previous studies, the current study suggests
that the poor depend predominantly on the public sector for
‘‘Sister, what can people like us do? She (Respondent 3) had two inpatient care. Peters et al. (based on National Sample Survey
gold earrings, and I had to sell off one of these. I got 1400 rupees Organisation data) concluded that the poor rely on public
for the earring.’’ (Spouse of respondent 3) hospitals more than the rich (Peters et al. 2002). For the poorest
25% of the population, 61% of hospitalizations are in public
Key differences in coping strategies among recent migrants hospitals. Despite this, wealthier populations capture a dispro-
Recent migrants to urban areas were less likely than permanent portionate share of public health spending. For example, the
residents to report borrowing from relatives. For example, the richest quintile received more than three times the public
following two respondents, both long-term residents of subsidy received by the poorest quintile. In part, this reflects
Vadodara city, borrowed from family: the much higher rates of hospitalization among the non-poor—
the richest quintile of the population is six times more likely to
‘‘My brother helped me a lot. I bought medicines with the money I have been hospitalized (in either a public or private facility)
borrowed from him.’’ (Respondent 8) (Peters et al. 2002, p. 219).
‘‘We told my cousin about her (Respondent 11’s) hospitalization On the one hand, use of (nominally) free public health care
and he immediately came over. I told him that I would need some services is a common strategy for coping, protecting households
money, so he went back and arranged for money.’’ (Spouse of from potentially burdensome health care costs. This is consist-
respondent 11) ent with findings in other countries where consumers can
choose between public and private sectors. Russell, for example,
In contrast, the following respondents, who had recently found that ‘public health care services, free at the point of
migrated to Vadodara city, used other coping strategies: delivery in Colombo (Sri Lanka) protected the majority of poor
households against high direct cost burdens, particularly the
‘‘We don’t know anyone here, so who would give us money? . . . All potentially high costs of hospital inpatient care and regular
our relatives are back at our village . . . So we had to sell our treatment of chronic illness’ (Russell 2008, p. 112–3).
belongings.’’ (Respondent 3) Nonetheless, as is the case in many other low- and
middle-income countries, there are significant out-of-pocket
‘‘I cook at someone’s home. I asked this employer for money to costs even at public facilities (McIntyre et al. 2006).
pay for my hospital expenses. They were very helpful . . . Our Given the much higher total costs incurred by those using
employers are (like) our relatives. Our real relatives are far away private hospitals, it is interesting that the poorest (particularly
so they (the employers) are the ones who help us first.’’ in rural areas) do not rely on public hospitals to an even greater
(Respondent 5) extent. The findings suggest that access and quality are among
124
the main reasons that poorer people turn to the private sector schemes is increasing. The private and community-based
for inpatient care. Many previous studies have found that schemes primarily cover hospital care, and are usually subject
India’s public health care sector is rife with problems (Peters to caps (i.e. limited indemnity) or deductibles.
et al. 2002). Among these problems are poor management, low The findings of this study suggest that there are several policy
service quality, staffing limitations (particularly in remote, rural options that should be pursued in order to protect the poor
areas), and limited drugs and supplies. from the costs of inpatient care. Governments should invest in
This study finds the median daily hospital expenditures to be improving the quality and access of public first-referral
almost three times higher among urban than rural respondents, hospitals. This should be done in a selective manner—with a
and that this difference is due largely to higher daily expend- focus, for example, on rural areas and urban slum areas—in
itures at private hospitals. This is consistent with the findings order to promote a more equitable distribution of resources. In
of some previous studies. For example, the World Health fact, the Government of India is already making efforts to
Survey, conducted in 2003, found that the mean annual improve access to quality health care in urban slum areas under
household expenditure on health care (goods and services) the National Urban Health Mission (2008–2012) (MOHFW
was Rs. 3304 in rural areas and Rs. 6384 in urban areas (World 2008). Similarly, the National Rural Health Mission (2005–
Health Organization 2006b), and the 60th round of the 2012) aims to improve health care in rural areas of 18 states
National Sample Survey (conducted in 2004) found that the deemed to have the weakest infrastructure, in part by
average cost of a hospitalization was Rs. 6225 among rural strengthening rural hospitals (MOHFW 2005; Mudur 2005).
respondents and Rs. 9367 among urban residents (National Given that these two schemes are still being implemented, it is
Sample Survey Organisation 2006). The difference found in our unclear how successful they will be.
study may be explained in part by higher urban incomes. While Efforts must also be made to reduce the cost of inpatient care
data are not available for Vadodara district, according to the at public facilities; this can be done in part by ensuring the
63rd round of the National Sample Survey (2006–07) the availability of basic drugs and supplies at first-referral hospitals.
monthly per capita consumption expenditure (MCPE) was Policy makers should continue to explore and support efforts to
Rs. 797 in rural Gujarat and Rs. 1422 in urban Gujarat provide financial protection through insurance mechanisms.
(National Sample Survey Organisation 2008). The difference The Indian government and individual state governments are
may also be due in part to higher quality (or more intensive) indeed pursuing expansions in publicly funded (or subsidized)
services provided at private hospitals in urban vs rural areas. coverage for rural populations as well as other vulnerable
This study confirms that people often borrow from friends, populations. For example, in April 2008, the Government of
family or employers to cope with the direct costs of inpatient India launched the Rashtriya Swasthya Bima Yojana (RSBY).
care. Borrowing from one’s social networks to cope with This voluntary scheme targets the 300 million people who are
medical costs is common in other settings (McIntyre et al. below the poverty line. In return for a premium of Rs. 30 per
2006). The current study suggests that the poor in urban areas person per year, coverage is provided for hospitalization (either
(including recent migrants) may be more likely to rely on at public or private facilities), capped at Rs. 30 000 per family
employers or neighbours rather than relatives, but this finding per year (Bhattacharjya and Sapra 2008). The balance of the
is based upon very few observations. This study did not add premium, Rs. 750 per person per year, is to be borne by central
evidence to Russell’s finding that lower-income households had and state governments. By the end of 2009, almost 9 million
weaker social networks and could access fewer financial households were enrolled in the scheme (Ministry of Labour
resources (Russell 2008). The findings are, however, consistent and Employment, undated) though this still represents a tiny
with those of Flores et al. (2008), who found that even the proportion of the target population.
poorest households in the poorest districts of India coped with Past experience suggests that these efforts must be carefully
medical expenditures through borrowing and drawing on monitored to ensure that the poorer among the insured are able
savings. to access scheme benefits, and that the quality and quantity of
Aside from use of public health services, very few survey health care provided has to be monitored and regulated. In
respondents reported use of formal social protection mechan- particular, there needs to be further discussion and debate as to
isms. India’s National Health Policy (2002) encouraged the whether or not it is a good idea to use public funding to
improve financial access to private hospitals (as is occurring
setting up of private insurance companies and the introduction
under the RSBY). In the absence of systems for monitoring and
of government-funded district-based insurance schemes on a
regulating private hospitals, such schemes risk exposing poor
pilot basis (MOHFW 2002). In India, 75–85 million people are
people to care that is unnecessary, of poor quality, or unneces-
at present covered by health insurance—approximately 8% of
sarily expensive. A study by Ranson and John (2001) docu-
the total population (Matthies and Cahill 2004; Gupta and
mented the problem of unnecessary hysterectomies, often of
Trivedi 2005). Social (mandatory) health insurance in India
poor quality, performed on members of a community-based
consists of coverage by the Employee State Insurance Scheme
health insurance scheme.
(ESIS) and Central Government Health Scheme (CGHS). The
This study also suggests several areas where additional
private insurance sector has grown tremendously in recent
research is required. These include:
years; the number of people covered under voluntary, private
health insurance schemes increased rapidly from 1995–96 to ! Further study of health care costs in urban areas, to extend
2002–03, by 29% per annum (Gupta 2004). Community-based understanding beyond the one city studied here;
health insurance schemes (CBHI) cover only 3 million people ! Longitudinal assessments of the costs of medical care—both
(Gupta and Trivedi 2005), although the number of such direct and indirect—and coping strategies;
125
! Documenting interventions that have been successful in Acknowledgements

improving the quality (or reducing the cost) of care at public
We wish to thank Dr V R Muraleedharan (IITM, Chennai) and
facilities, and the factors that have contributed to success;
his team for their support throughout this project. Thank you to
! Further research—possibly including larger, representative
Nicola Lord (LSHTM) for logistical support. We thank
samples—exploring the challenges faced, and coping strate-
Dr Amarjeet Singh, Commissioner of Health, Gujarat, for
gies used, by migrant workers and their families;
permitting us to carry out the study at the public hospitals.
! Studies of social networks—for example, how these net-
We are particularly grateful to the hospital representatives and
works are affected by new, formal risk-sharing mechanisms
staff who gave us support for the exit survey interviews. Also,
(like the RSBY) and factors that help or hinder social
thanks to the staff of Baroda Citizens Council, who helped us
networks in rapidly expanding urban areas;
identify recent migrants and conduct focus group discussions
! Further studies on strategic purchasing, to ensure that
with them. We also take this opportunity to extend our
health care purchases by insurers are of acceptable quality
gratitude to the Faculty of Social Work, Vadodara who provided
and reasonable cost.
work space during the project. Most importantly, we wish to
thank the field investigators who assisted in data collection;
and all the respondents for sharing their experiences and
Strengths and weaknesses of the methodology
responding to our questions.
The main strength of the methodology is that it employed both
qualitative and quantitative methods. This helped, for example,
to ensure that the questions on coping strategies included in
the exit survey were culturally and contextually appropriate (as Funding
they had been reported by participants in the preceding focus This work was funded by the Wellcome Trust, United Kingdom
group discussions), and it enabled verification of exit survey (Grant number GR080101MA).
findings through in-depth interviews with select respondents.
Data on hospital expenditures and coping strategies are likely to
have been recalled with good accuracy given that they were
collected right at the time of discharge from hospital. This is Conflict of interest
one of very few Indian studies where investigators have been None declared.
granted permission to conduct such an exit survey on the
premises of both public and private hospitals. Finally, because
the exit survey and in-depth interviews purposefully included
strata of urban respondents, this is one of very few Indian Endnotes
studies that provide data on coping strategies among urbanites. 1
The poverty line is set for each state, and represents the level of
The study has a number of methodological shortcomings. consumer expenditure per capita required to ensure a calorie intake
First, because respondents were identified by exit survey (as of 2100 per day in urban areas and 2400 per day in rural areas. In
opposed to a house-to-house survey, for example) the study 1995–96, it was estimated that 20.5% of Gujarat’s rural population,
and 30.7% of the urban population, were BPL.
excludes those who may have required hospitalization, but who 2
For the purpose of this study, we considered recent migrants as those
chose not to seek care or who sought outpatient care only. who had moved from their place of origin/birth (or usual
Thus, the study may have excluded the very poor who did not residence) within the previous year (Census 2001). For the purpose
seek care. Second, the exit survey was non-random, and the of the in-depth interviews, non-migrants were those who had lived
at their current place of residence since birth.
respondents therefore may not represent the wider population 3
The latest poverty statistics for India suggest that in 1999–2000, 28.6%
of hospitalized persons in Vadodara district. There is likely to be of the total population were living below the ‘national poverty line’
considerable variation in costs between different private hos- (World Bank 2010). We chose to compare coping strategies among
pitals, so selection of different facilities may have led to the poorest 30% of respondents (vs the less poor 70%) as they
might roughly be thought of as representing those who live below
different results. Respondents at any one facility were
the poverty line (vs those who live above the poverty line). This is
non-randomly selected; respondents were interviewed during a rather arbitrary and imperfect cut-off, given that Gujarat does
the same season and over a period of only a few days, and tend to perform slightly better than all-India on measures of
patients were excluded from the sample if they had used a poverty (and so is likely to have a poverty line lower than 30%)
private inpatient room rather than a general hospital ward. and, more importantly, given that respondents to the exit survey
are not representative of the general population.
Third, there may be limits to the extent to which study findings
can be generalized to other districts or states, given, for
example, that Vadodara district is fast-growing and urban
relative to many others. Fourth, the exit survey did not explore
indirect costs (i.e. loss of household productive labour time and References
income). And the in-depth interviews, because they were
Bhattacharjya AS, Sapra PK. 2008. Health insurance in China and India:
cross-sectional, were more likely to capture fairly discrete and
segmented roles for public and private financing. Health Affairs
memorable coping strategies (e.g. borrowing from a money (Millwood) 27: 1005–15.
lender), and less likely to capture, for example, small reductions Filmer D, Pritchett LH. 2001. Estimating wealth effects without
in household consumption or increases in time spent in expenditure data—or tears: an application to educational enroll-
productive labour. ments in states of India. Demography 38: 115–32.
126
Flores G, Krishnakumar J, O’Donnell O, Van Doorslaer E. 2008. Coping Munshi K, Rosenzweig MR. 2005. Why is mobility in India so low? Social
with health-care costs: implications for the measurement of insurance, inequality, and growth. BREAD Working Paper No. 097.
catastrophic expenditures and poverty. Health Economics 17: Durham, NC: Bureau for Research in Economic Analysis of
1393–412. Development (BREAD).
Gupta A. 2004. Wealth through health insurance: the way to sustainable National Sample Survey Organisation. 2006. Morbidity, health care and the
development. IRDA Journal 2: 24–5. condition of the aged: NSS 60th round. New Delhi: National Sample
Gupta I, Trivedi M. 2005. Social health insurance redefined: Health for Survey Organisation, Government of IndiaOnline at: http://www
all through coverage for all. Economic and Political Weekly 40: .mospi.gov.in/mospi_nsso_rept_pubn.htm, accessed 5 January
4132–40. 2010.
International Organisation for Migration. 2005. Migration, Development National Sample Survey Organisation. 2008. Household Consumer
and Poverty Reduction in Asia. Geneva: International Organisation for Expenditure in India, NSS 63rd round, 2006-07. Report No. 527(63/
Migration. 1.0/1). New Delhi: National Sample Survey Organisation,
Government of India.
Kochar A. 1995. Explaining household vulnerability to idiosyncratic
income shocks. American Economic Association Papers and Proceedings Office of the Registrar General. 2001. Census of India. New Delhi:
85: 159–64. Government of India.
Krishna A, Kapila M, Porwal M, Singh V. 2003. Falling into poverty in a Peters DH, Yazbeck AS, Sharma RR et al. 2002. Better Health Systems for
high-growth state: escaping poverty and becoming poor in Gujarat India’s Poor: Findings, Analysis, and Options. Washington, DC: The
villages. Economic and Political Weekly 38: 5171–9. World Bank.
Mahal A, Singh J, Afridi F et al. 2000. Who Benefits from Public Health Ranson MK, John KR. 2001. Quality of hysterectomy care in rural
Spending in India? New Delhi: National Council of Applied Gujarat: the role of community-based health insurance. Health
Economic Research (NCAER). Policy and Planning 16: 395–403.
Matthies S, Cahill KR. 2004. Lessons from across the World: how India Rosenzweig MR. 1988. Risk, implicit contracts and the family in rural
can break barriers to develop health insurance. IRDA Journal 2: areas of low-income countries. The Economic Journal 98: 1148–70.
7–12. Russell S. 2008. Coping with the costs of illness: vulnerability and
McIntyre D, Thiede M. 2007. Illness, health service costs and their resilience among poor households in urban Sri Lanka. In:
consequences for households. In: Bennett S, Gilson L, Mills A Bennett S, Gilson L, Mills A (eds). Health, Economic Development
(eds). Health, Economic Development and Household Poverty: From and Household Poverty: From Understanding to Action. London:
Understanding to Action. London: Routledge. Routledge.
McIntyre D, Thiede M, Dahlgren G, Whitehead M. 2006. What are the Townsend RM. 1994. Risk and insurance in village India. Econometrica
economic consequences for households of illness and of paying for 62: 539–91.
health care in low- and middle-income country contexts? Social UNDP, World Bank. 1999.
Science & Medicine 62: 858–65. Van Doorslaer E, O’Donnell O, Rannan-Eliya RP et al. 2006. Effect of
Ministry of Labour and Employment. Undated. Rashtriya Swasthya payments for health care on poverty estimates in 11 countries in
Bima Yojna. Online at: http://www.rsby.gov.in/, accessed 4 January Asia: an analysis of household survey data. The Lancet 368:
2010. 1357–64.
MOHFW. 2002. National Health Policy 2002. New Delhi: Ministry of World Bank. 2001. World Development Report 2000/2001: Attacking Poverty.
Health and Family Welfare, Government of India. Oxford: Oxford University Press.
MOHFW. 2005. National Rural Health Mission (2005–2012). New Delhi: World Bank. 2002. Health of the poor in urban India: a consultation, July,
Ministry of Health and Family Welfare, Government of India. 2002. New Delhi: World Bank.
Online at: http://www.mohfw.nic.in/NRHM/Documents/Mission_ World Bank. 2010. World Development Report 2010: Development and Climate
Document.pdf, accessed 4 January 2010. Change. Washington, DC: World Bank.
MOHFW. 2008. National Urban Health Mission (2008–2012). New World Health Organization. 2006a. The World Health Report 2006: Working
Delhi: Ministry of Health and Family Welfare, Government of Together for Health. Geneva: World Health Organization.
India. Online at: http://www.uhrc.in/downloads/Reports/NUHM-
World Health Organization. 2006b. World Health Survey: Report of
Draft.pdf, accessed 4 January 2010.
India. Online at: http://www.who.int/healthinfo/survey/whsind-
Mudur G. 2005. India launches national rural health mission. British india.pdf, accessed 4 January 2010.
Medical Journal 330: 920.
127
Appendix 1 List of 26 assets and utilities variables included in the socio-economic status (SES) index, showing responses for each decile (1st being
the poorest and 10th being the least poor)
No. Variable Deciles of SES

1 2 3 4 5 6 7 8 9 10
Number of observations (n ¼ 797) 80 82 78 81 78 81 81 77 80 79
1 Ownership of house 68.75 65.85 73.08 70.37 73.08 71.60 58.02 63.64 66.25 75.95
2 Housing with brick/stone with cement plaster 12.50 15.85 20.51 41.98 62.82 58.02 58.02 87.01 81.25 96.20
3 Owns electricity connection 52.50 86.59 85.90 92.59 89.74 83.95 75.31 80.52 77.50 84.81
4 Gas/kerosene as cooking source 1.25 18.29 19.48 39.51 48.72 54.32 72.84 68.83 90.00 94.94
5 Own mattresses 8.75 20.73 44.87 53.09 60.26 79.01 74.07 89.61 96.25 98.73
6 Own cot/bed 27.50 54.88 89.74 87.65 93.59 97.53 95.06 97.40 98.75 100
7 Own wristwatch 37.50 64.63 88.46 95.06 93.59 90.12 95.06 98.70 98.75 100.00
8 Own fans 26.25 87.80 87.18 96.30 100.00 98.77 100.00 98.70 100.00 100.00
9 Own radios 8.75 37.80 28.21 41.98 47.44 61.73 51.85 48.05 57.50 79.75
10 Own television 0.00 6.10 8.97 25.93 46.15 69.14 85.19 93.51 100.0 100.0
11 Own refrigerator 0.00 0.00 0.00 0.00 0.00 1.23 6.17 15.58 45.00 64.56
12 Own sewing machine 0.00 1.22 0.00 1.23 5.13 6.17 8.64 15.58 5.00 8.86
13 Own telephones 0.00 0.00 0.00 1.23 0.00 1.23 3.70 15.58 31.25 94.94
14 Own mobile-phones 0.00 3.66 11.54 29.63 34.62 58.02 83.95 89.61 97.50 100.00
15 Own two-wheelers 0.00 0.00 0.00 0.00 1.28 1.23 9.88 36.36 83.75 100.00
16 Own tractors 0.00 0.00 0.00 0.00 0.00 0.00 1.23 2.60 3.75 12.66
17 Own buffaloes 17.50 24.39 24.36 30.86 24.36 23.46 12.35 14.29 11.25 15.19
18 Own cows 1.25 7.32 8.97 9.88 6.41 11.11 4.94 5.19 6.25 6.33
19 Own bulls 1.25 8.54 12.82 19.75 16.67 11.11 6.17 11.69 7.50 16.46
20 Own pada 0.00 2.44 1.28 1.23 0.00 0.00 0.00 0.00 0.00 0.00
21 Own goats 33.75 21.95 14.10 17.28 14.10 6.17 2.47 6.49 1.25 1.27
22 Own hens 2.50 2.44 1.28 0.00 0.00 0.00 0.00 0.00 0.00 0.00
23 Own bullock carts 0.00 0.00 3.85 6.17 2.53 3.70 6.17 6.49 3.75 11.39
24 Own shops 0.00 1.22 3.85 2.47 6.41 9.88 18.52 23.38 27.50 26.58
25 Own lauri 1.25 7.32 5.13 7.41 6.41 6.17 12.35 11.69 10.00 6.33
26 Total no. of rooms 1.1 1.3625 1.6875 1.5949 1.7375 1.9024 2.1299 2.125 2.325 3.3797
128
ARTICLE IN PRESS
Social Science & Medicine 61 (2005) 1408–1417

www.elsevier.com/locate/socscimed
Private obstetric practice in a public hospital:

mythical trust in obstetric care
Wachara Riewpaiboona,b,!, Komatra Chuengsatiansupc,d,
Lucy Gilsone,f, Viroj Tangcharoensathiend,g
a
Health System Research Institute, Ministry of Public Health, Tivanond Road, Nonthaburi 11000, Thailand
b
Sirindhorn National Medical Rehabilitation Center, Ministry of Public Health, Thailand
c
Society and Health Research Institute, Health System Research Institute, Ministry of Public health, Thailand
d
Bureau of Health Policy and Strategy, Ministry of Public Health, Thailand
e
Centre for Health Policy, School of Public Health, University of Witwatersrand, South Africa
f
Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine, UK
g
International Health Policy Program, Health System Research Institute, Ministry of Public health, Thailand
Available online 25 January 2005
Abstract
There is evidence to suggest the decline of trusting relationships in modern healthcare systems. The primary aim of
this study was to investigate the role of trust in medical transactions in Thailand, using obstetric care as a tracer service.
The paper proposes an explanatory framework of trust for further investigation in other healthcare settings.
The study site was a 1300-bed tertiary public hospital in Bangkok which it provides two forms of obstetric care:
regular obstetric practice (RP) and private obstetric practice (PP). Forty pregnant women were selected and interviewed
using a set of guiding questions. A thematic analysis of the interviews was undertaken to generate understanding and
develop an explanatory framework.
It was found that patients’ trust in obstetric services was influenced by their perceptions of risk and uncertainty in
pregnancy and childbirth, and that these perceptions were linked to their social class. Social class also influenced the
accessibility and affordability of care to patients. Middle class pregnant women with relatively high-level concerns
about risk and uncertainty preferred using PP service as a means to achieve interpersonal trust. These women thought
that an informal payment would provide the basis for interpersonal trust between themselves and the chosen
obstetricians.
In practice, however, obstetricians involved in PP rarely acknowledged this reciprocal relationship and hardly
expressed the additional courtesy expected by patients. As a result, PP service only created an expensive impersonal
trust that was mistaken as interpersonal trust by patients. Negative outcomes from PP often caused disappointment
that could eventually lead to medical litigation.
The study suggests that there are some negative impacts of PP within the health system. Negative experiences among
PP users may undermine trust not only in the specific doctor but also trust in health professionals and hospitals more
generally. Steps need to be undertaken to protect and strengthen existing impersonal trust, which combine institutional
trust based on good governance and service quality with trust in the professional standard of practice. The explanatory
!Corresponding author. Health System Research Institute, Ministry of Public Health, Tivanond Road, Nonthaburi 11000,
Thailand. Tel./fax: +66 2 9510735, +66 2 9510830.
E-mail address: wachara@health.moph.go.th (W. Riewpaiboon).
0277-9536/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2004.11.075
129
ARTICLE IN PRESS
W. Riewpaiboon et al. / Social Science & Medicine 61 (2005) 1408–1417 1409
framework developed through this study provides a foundation for further studies of trust in different specialties and
care settings.
r 2005 Elsevier Ltd. All rights reserved.
Keywords: Trust; Relationships; Obstetric care; Private practice; Public hospital; Thailand
Introduction Gilson (2003) categorized trust into interpersonal

trust, where two individuals known to each other rely
In modern healthcare systems, the traditional, pater- faithfully on each other, while impersonal trust refers to
nalistic doctor–patient relationship is gradually being trust in strangers or in a social system. These two forms
transformed to a provider–customer or consumerist of trust are dynamic. A stranger may become a known
doctor–patient relationship (Mechanic, 1998) through individual, as a result of personal interactions and
the application of new public management and market accessing information by which one can judge how the
principles (Hunter, 1996). Medicine has been increas- other will behave in relation to one’s interests. The
ingly commoditized and medical litigation, involving behavioral characteristics that underpin interpersonal
confrontation of providers by patients, is increasing in trust include technical competence, openness, concern
many countries including Thailand (Mechanic, 1998; and reliability (Coulson, 1998; Mechanic, 1998). Alter-
Wibulpolprasert, Hempisut & Pittayarungsarit, 2002). natively trust in strangers can be rooted in institutions
The changing nature of healthcare systems may under- that allow delegated or fiduciary trust to develop.
mine the role of trust within them. Yet without a However, trust always involves an element of risk
trusting doctor–patient relationship, the healthcare derived from uncertainty regarding the motives, incen-
system cannot achieve its goal of improving health in a tives and future actions of another on whom one
holistic sense, considering its physical, mental, social, depends (Coulson, 1998; Gambetta, 2000; Kramer,
and spiritual dimensions (Mechanic, 1998). 1999; Lewicki & Bunker, 1996).
With regard to healthcare seeking behavior, trust has This study investigated trust in obstetric care. We
an important influence over the choice of healthcare selected obstetric care as the focus of inquiry for three
provider (Balkrishnan, Dugan, Camacho, & Hall, 2003; reasons. First, it involves a continual contact between
Hall, Camacho, Dugan, & Balkrishnan, 2002) and is patient and doctor over at least six to ten months. It is
based on beliefs or expectations about how others will possible that the same provider will be used for several
behave in relation to oneself in the future (Gambetta, pregnancies. Past experiences may also lead to selection
2000; Gilson, 2003). In addition to monetary incentive, of a new provider for each pregnancy. Obstetric care,
then, trust can be viewed as an important non-financial thus, allows opportunities to build up, or break down,
incentive affecting care providing behaviour (Franco, patient-provider trust. Secondly, pregnant women and
Bennett, & Kanfer, 2002). It is therefore crucial for policy their social networks have a certain degree of health
makers to understand the roles of trust (and distrust) in information and know how to negotiate with providers
shaping patients’ experience and their satisfaction. over decision-making around antenatal care and the
This study aimed to understand the roles of trust in childbirth process. Thirdly, there are two different ways
medical transactions in the Thai healthcare system. It in which patients pay their providers for obstetric care in
was expected that an analysis of trust (or distrust) in Thailand, namely regular and private practice. Differ-
transactions between doctors and patients would pro- ence in payment mechanism provides good grounds for
vide the basis for developing a conceptual framework to investigating the role of trust in mothers’ decision-
allow better understanding of the roles of trust and making and for developing a related explanatory
other non-financial incentives in healthcare transactions. framework. In addition, it will allow some initial
The conceptual framework developed through this assessment of how the existence of private practice in
study may be useful for investigating such issues in public hospital impacts on trust in doctors and the
future studies. public obstetric care system.
To trust someone is to believe that they are honest, Private obstetric practice (PP) can be described as an
sincere and will not deliberately harm you. In addition informal relationship between a pregnant woman and an
to trust, risk and uncertainty also play crucial roles obstetrician in which the pregnant woman voluntarily
in health care decision-making and medical choices pays money in exchange for personalised obstetric
(Kapferer, 1976; Mechanic, 1998). While risk and services. These services include the provision of antena-
uncertainty are inherent in sickness and illness, trust tal care, support for delivery and postpartum care by the
can be built and managed by patients and their relatives doctor him/herself. In contrast, regular obstetric prac-
as well as by providers. tice (RP) does not involve any such special and
130
ARTICLE IN PRESS
1410 W. Riewpaiboon et al. / Social Science & Medicine 61 (2005) 1408–1417
personalised agreement. It relies on the general and 50th anniversary just before the study. Originally
routine hospital obstetric services, payment for which providing only maternity services, the hospital has now
depend on the patient’s health insurance status and the expanded its services to include all other specialties.
related maternity benefit scheme. However, obstetric services remain particularly impor-
Both PP and RP are commonly available in public tant. Thirty-six percent of all in-patients attend for
hospitals. PP services are considered a source of extra obstetric care and the hospital carries out approximately
income for obstetricians. An average public sector salary 45 deliveries a day.
for an obstetrician was around 20,000 Baht per month in The hospital’s catchment area is extensive, 55% of all
20031 while the financial income derived from private patients being Bangkok residents and the remainder are
practice was generally 3000–5000 Baht per birth for an migrant workers and cases referred from other hospitals.
average of 20–80 births per month. The financial income It has a well-established residency training program in
derived from PP may, therefore, significantly boost all specialties. It also provides training for medical and
doctors’ income given the relatively low level of public nursing students in collaboration with other colleges and
sector salaries. universities. Private obstetric practice is informal and
In Thailand, there are approximately 800,000 births has been commonly practiced for decades (indeed, two
each year. More than 95% of births take place in of the PP users mentioned that their mothers had PP
hospital (Health Information Center, 2001). Public services some thirty years ago at this hospital), with the
hospitals serve as the main provider of antenatal, PP fee changing over the time.
intrapartum and post-partum services. In 2001, 41% During December 2002, 40 pregnant women, with at
of total deliveries took place in Ministry of Public least one childbirth, attending the antenatal clinic
Health (MOPH) district hospitals, 34% in MOPH (ANC) were randomly chosen. Informed consent was
provincial hospitals, 18% in other public hospitals and obtained before recruitment. Choosing informants with
only 7% in private hospitals (Teerawattananon, Tang- previous pregnancy experience allowed us to study
charoensathein, Srirattana, & Tipayasoti, 2003). A 1998 experiences on previous pregnancy and childbirth. An
survey of private practice in 29 MOPH provincial in-depth interview using a guiding list of questions was
hospitals showed that 37% of all public hospital conducted in private in a comfortable room close to the
deliveries involved PP (Hanvorawongchai, Lertiendum- ANC clinic by one of the researchers. Each interview
rong, Teerawattananon, & Tangcharoensathein, 2000). took about 60 min and was tape-recorded, with permis-
However, the financial gains to providers resulting from sion. This was subsequently transcribed. Home tele-
PP may encourage its growth both for obstetric care and phone numbers were obtained in case further interview
for other specialties. It is, therefore, crucial to under- was needed.
stand if and how PP affects patient trust in providers Interview sessions were conducted in an informal
and the health system more generally, in order to manner. The informants were invited to describe their
formulate further policy development. experiences on pregnancies and childbirths in their own
narratives. Interviewer would ask questions to elicit
further information to cover the following key areas:
demographic and socio-economic data; general percep-
Methods tions of pregnancy and childbirth; fear, worry or
anxiety; expectations; past experiences of obstetric
As trust is a relational state, its investigation requires services; clinical outcomes and satisfaction. Informants’
an in-depth analysis of people’s relationships and use of PP or RP was identified for the current and
interactions considering, for example, how patients feel previous pregnancy. Those mothers with PP experience
and what they expect from their doctors and how both were specifically asked to describe: (1) the process
of them behave in relation to each other. Such by which they engaged in PP, (2) the reasons and
investigation is most effectively undertaken through an factors influencing their decision to use PP, and (3)
interpretive analysis of patients’ narratives of their their understanding of trust (distrust) in the chosen
obstetric encounters. Through the examination of obstetrician.
patients’ stories, trust-related words or themes can be Those mothers with RP-only experience were also
identified and used in the construction of an explanatory asked: whether they knew about PP; if so, why they did
framework. This study, therefore, applied a qualitative not choose PP service; and their opinions on private
approach, involving thematic analysis of in-depth inter- practice. By the end of the interview when informants
view data. became more comfortable, all informants were invited to
The study site was a well-known 1300-bed MOPH disclose their feelings and opinions on their doctor–
tertiary hospital in Bangkok, which had celebrated its patient relationship; whether they made gratitude
payments, either in cash or in kind; or provided any
1
Exchange rate 42 Baht per dollar in 2003. other forms of non-material reciprocity. In two final
131
ARTICLE IN PRESS
questions, we asked our informants to describe the level service for at least one childbirth, and 26 had experience
of trust in their choices of PP or RP. For those having of RP. Among the 40 informants there were 43 events of
PP experience, they were asked if they would engage in completed pregnancies and childbirth.
PP again and/or would recommend their friends to do Compared to RP users, those patients choosing PP
so. In addition to interviewing pregnant women, a few were generally older, more highly educated, earned
referring nurses, obstetricians and other staff members higher incomes, owned their own homes, were employed
working in ANC and the labor room were also with permanent jobs and entitled to medical benefits
informally interviewed about their perceptions and such as the Civil Servant Medical Benefit Scheme and
practices related to RP and PP services. Social Security Scheme. There were more unemployed
The meanings of trust were examined by identifying housewives among RP users. A higher proportion of RP
groups of associated words and phrases in Thai that
have a closely related meaning to trust: for example,
confidence (chueymun), being assured (munjai), sense of
security (rusuek ploudpai), certainty (naenon), belief Table 1
(chuey), reliance or dependence on (the obstetrician) Sample Characteristics
(laewtae mow). The themes that were commonly
PP (n RP (n Total
found and emerged from a majority of cases were 14) 26) (n 40)
further analyzed in order to develop the explanatory
framework of trust. Mean age (year) 32.8 29.9 29.4
Parity including this pregnancy
(%)
Second 64 73 70
Results Third 21 23 23
Other 14 4 8
Service arrangement in the hospital Education (%)
Primary 14 77 55
In the hospital studied, all ANC visits were seen by an Secondary 29 23 25
Undergraduate 29 0 10
obstetrician. While PP cases were always seen by their
Graduate 29 0 10
chosen doctors, the RP cases were seen by the
Occupation/insurance coverage
obstetrician in-charge on the day. The premises and (%)
basic amenities available for labor and delivery were Government employeea 36 0 13
similar for both groups. In labor, PP cases were Private employeeb 14 42 33
normally attended by in-charge nurses and were Own account work or self 50 15 28
frequently visited by their personal obstetricians. The employedc
use of PP service implied that the whole delivery process House wifed 0 42 28
would be conducted by the chosen obstetrician. The fee Estimate monthly income in Baht
paid for PP services ranged from 3000–5000 Baht, based (%)
430,000 14 0 5
on informal agreement between patient and provider.
20,000–29,999 43 12 23
This fee was not covered by medical insurance.
10,000–19,999 43 35 38
RP cases were normally attended by the obstetric o10,000 0 54 35
nurse in-charge of the day and deliveries were assisted by Home owner (%)
any one of the available nurses. An episiotomy could be Yes 64 4 25
done by a nurse-teacher or obstetric resident or Home town (%)
obstetrician on duty whilst suturing of the perineum Bangkok 50 15 28
would only be done by the resident or obstetrician on a
duty. A consultant obstetrician was available for Government employees and their dependents such as
parents, spouse and children are covered by the Civil Servant
complicated cases, if called in by the attending nurses.
Medical Benefit Scheme on a fee for service basis. Their medical
The principles and guidelines of clinical services as well expenses are fully reimbursed.
as the availability of basic and advanced obstetric b
Private employees are covered by the Social Security
equipment provided to patients using the RP system Scheme, which provides a lump sum reimbursement of 4000
was similar to those using the PP system. Baht per confinement.
c
Own-account workers and self employed are not covered by
Social Security and pay out of pocket for care.
Socio-demographic profile d
This group is covered by the recent universal health care
coverage scheme, which includes a flat rate. capitation fee for
The socio-economic profile of our sample is described ambulatory care and a case-based payment fee within a global
in Table 1. Fourteen out of forty cases had used the PP budget for hospital services.
132
ARTICLE IN PRESS
users were migrant workers whose hometown was and had strong communal ties with their hometowns.
outside Bangkok. They usually attended the ANC clinic in Bangkok,
but at around the 8th month of pregnancy, they
would return to their hometowns, carrying their ANC
Concerns and aspirations
records with them, to deliver their babies. Important
reasons for returning home for childbirth were the
Although all respondents viewed pregnancy and availability of social support and their preference
childbirth as natural processes they also expressed for the traditional postpartum care provided by their
serious concerns about the risk and uncertainty of the parents. Two RP users expressed the following typical
events. Tracing the words ‘‘fear’’ or ‘‘afraid of’’ and accounts:
‘‘worry’’ or ‘‘anxiety’’ throughout the narratives indi-
cated considerable concerns over the risk of newborn Last pregnancy, I continued my work until the 8th
abnormalities, labor pain and difficulties, prolonged month of pregnancy. My father came to take me
delivery processes, injuries and bleeding, and life back to give birth in my hometown. My parents,
threatening experiences for both mother and child. The especially my mother, were quite worried about me.
most common concern was the risk to newborn They were afraid that the baby and me would be at
abnormality, as many respondents said, risk. They also went to a shrine and prayed for my
safe childbirth.
I am very much afraid that my baby will not be krob-
I would like to give birth in my hometown, especially
sam-sib-song2. I think only obstetricians and tech-
for the first pregnancy, where my mother lives, since I
nologies can help to prevent or ensure early detection
will be able to ‘‘yu fai3‘‘(stay in a fire place) for fifteen
of the problems for proper management.
days and during that time my mother would look
Anxiety and serious concerns about having a healthy after my baby.
baby were more strongly identified among PP users.
Most of the RP users that did not return to their
These mothers often requested additional services and
hometowns were in their second pregnancies and had
technological interventions. For example, some PP users
prepared to have their mothers come to stay with them
said that because of anxiety about their newborn’s
in Bangkok.
health they sought reassurance by attending the ANC
Overall, most informants accepted that in modern
clinic early and regularly, as well as seeking frequent
medical encounters patients and doctors were almost
ultrasonographies and an amniocentesis test. A few
strangers to each other. In line with Benoliel’s findings
mothers also expressed apprehension about labor pain
(Benoliel, 1993), our study suggests that patients felt
and wanted a quick response to their requests for
they were powerless and lacked confidence to voice their
analgesia by having a particular provider to count on. In
problems or to request information, reassurance or any
general, these strong concerns about safety and risks had
specific care from their physicians. They were uncertain
driven PP users to seek the best care by employing
if the system would respond satisfactorily to their needs,
obstetric specialists particularly in PP service.
and so felt they could not fully trust in such a system. PP
The interviews also reveal that while PP users
users who expressed strong concerns over risk and
saw cesarean section as important in assuring a safe
uncertainty sought an adequate level of reassurance by
delivery, RP users were generally more hesitant
attempting to establish an interpersonal relationship
about this medical intervention. They often mentioned
with a chosen obstetrician. RP users, in contrast,
that many pregnant women delivered babies every
relied more on the support of their traditional and
year without any problem. They felt that if they were
kinship networks.
healthy, attended antenatal visits regularly and strictly
followed the doctor’s recommendations, any abnorm-
ality and problem would be detected at an early stage. Expectations of care
RP users strongly believed that the hospital and
personnel in-charge had the capacity to manage any All pregnant women we interviewed expressed their
problem with regard to their pregnancies. For these desire to maintain their autonomy in their pregnancies
women, additional services were, therefore, considered and childbirths. They not only sought clinically compe-
unnecessary. tent providers but also specifically chose providers who
Moreover, it was noticed that although many had they considered reassuring and responsive to their felt
migrated to work in Bangkok, 19 out of 26 RP users needs. These findings confirm that interpersonal trust
were still bound to their traditional kinship networks
3
yu fai is a traditional practice whereby the post-partum
2
‘‘Krob-sam-sib-song’’ means a person who has no physical women stays on a bench with a wood fire nearby in order to
abnormalities nor is missing any body parts. ‘‘dry up the womb’’ and expedite involution.
133
ARTICLE IN PRESS
was perceived as crucial for assuring good care. Trust in Discussion

providers was constituted by the patient’s perception of
clinical competency, as well as the attitudes and Trust in health care is a comparatively new area of
reliability of the provider (Coulson, 1998; Mechanic, inquiry. The aim of our research was to build a better
1998). understanding of various aspects of trust. The proposed
framework is shown in Fig. 1 below. We found that
perceptions, values and concerns about risk and
Determinants of the PP service uncertainty among healthcare users were important
attributes differentiating trust in obstetric encounters.
The process of seeking PP services began with strong Trust was differently conceived by different groups of
concerns about the risk and uncertainty of pregnancy users and the roles various forms of trust played in the
and childbirth. The concerns of risk and uncertainty doctor–patient relationship were different between
drove pregnant women, on the one hand, to demand private patients and regular patients.
higher quality of care and, on the other hand, to seek Since PP involved a relatively large amount of
greater control over their situations. Feeling that regular additional expense, ability to pay was an important
services could not provide a satisfactory level of determining factor for mothers who sought private care.
response to their needs, they sought PP services by first Most PP users who found the private care affordable
choosing an obstetrician on whom they could count. In can be said to belong to the middle class stratum which
choosing their obstetrician, PP users sought information has steadily expanded since Thailand’s rapid industria-
and advice from friends or relatives who previously had lization in the latter half of the 1980s. This newly
PP experience. RP users, in addition to their limited constituted middle class (MC) has been defined as an
ability to pay an extra expense for PP service, were short intermediate hub between the capitalist and the working
of information regarding PP options. class (WC) in Thai society (Piriyarangsan & Phongpa-
However, poor experience of either type of care nichit, 1993). The Thai MC stratum can be divided into
had a strong influence on whether they chose the two major groups: the businessmen who mostly rely on
same care in a later pregnancy. Three respondents who the market for their wealth accumulation, and the
had previously used PP services had changed to RP professionals and white-collar workers characterized by
services for the current pregnancy due to problems high levels of education, or non-manual workers who
experienced in the past. These problems included the rely on their professional skills and expertise for social
doctor: arriving late for the delivery without a satisfac- mobility.
tory explanation; suggesting a caesarean section against As described by Piriyarangsan and Phongpanichit
the mother’s wish (who subsequently had a normal (1993), the Thai MC subculture has an inclination
vaginal delivery); setting the time and date for induction toward an individualistic worldview. They usually
of labor or an elective cesarean section at his own place a high value on education and base their decisions
convenience without asking the mother’s opinion. A on systematic calculation and economic rationality.
mother’s decision to change from PP to RP seemed to As their cultural worldview is also strongly influenced
indicate her inability to maintain her autonomy or fulfill by the ruling class’s concepts of merit, patronage
her expectations. and power (or boon, baramee, and amnat in Thai),
Four past RP users had changed to PP care for their they are not only more assertive and self-determined but
current pregnancy. Bad experiences of the regular also less committed to the idea of social equality
obstetric systems in other hospitals, such as the death (Aewsriwong, 1993).
of the baby and failure to diagnose an ectopic As shown in Table 1, the characteristics of PP users,
pregnancy, were reasons for the change. These mothers such as high education, white-collar work and high
felt more confident under the PP system. However, income, suggest that they belonged to the middle class,
affordability was also an important enabling factor in while RP users fell within the working class stratum. In
two cases where the mothers now had more money and addition to ability to pay, which enabled middle class
so could better afford PP than in the past. mothers to attain private care, middle class cultural
Answers to the interview questions confirmed that PP perceptions and values also played a determinant role in
users who were satisfied with the outcomes of their the decision to choose PP. Not only were PP users, with
experiences of care trusted their choices and would not their higher education, more trustingly confident in
change service type in the future. They said they would specialist care and advanced medical technology, but
recommend this option to friends and relatives. In also they were more assertive, self-determined, and
contrast, although most RP users expressed satisfaction wanted to be in control of their perceived risk and
with the services they had received, it seemed possible uncertainty.
that there may be a trend of shifting to PP when and if As the accounts in our findings also indicate, their
increased income level allows. middle class view and value was also evident in the belief
134
ARTICLE IN PRESS
Fig. 1. Framework of healthcare trust.
of most PP patients that their engagement in PP was noted that since RP users belonged to a different social
economically rational. As one PP case told us, ‘‘ymy class to the obstetricians. They sometimes felt alienated
doctor told me a fee of 5000 Baht for his service, I and unable to bridge the class divide gap. With strong
thought it was affordable for me. It was not expensive support from their family, relatives and friends who
compare to expenses in private hospital.’’ While another belonged to the same social class, the working class
said, ‘‘yspending 5000 Baht for the safety of my baby mothers were able to limit their interaction with the
and myself was worth spending. I was not pregnant that hospital and professional system to as little as clinically
frequently.’’ necessary. For RP users, perhaps the basic trust in the
In contrast, RP users were bounded by a different hospital system in conjunction with their trusting social
cultural background. Instead of middle class individu- support system was adequate to reinforce their con-
alism, the working class maintained traditional commu- fidence in times of uncertainty and risk.
nal connections even after they migrated to urban areas
for job opportunities (Piriyarangsan & Phongpanichit,
1993). Most RP users interviewed in this study main- The dynamic interplay of trust and the myth of inter-
tained strong ties with their traditional social network, personal trust
friends and relatives, even whilst they worked in
Bangkok. As various accounts in our findings indicate, As Fig. 1 outlines, this analysis of transactions in
maternal and child-care was traditionally viewed as a obstetric care suggests that pregnant women viewed
collective activity to be managed and assisted by parents trust in obstetric services at two distinctive levels: (1)
and members of extended families in their community trust in the health care system (including trust in the
settings. hospital and professional competency) and (2) trust in a
It is clear from the findings that RP users did not rely particular person in-charge of caring.
entirely on professional and institutional care. The In this study, all respondents knew about the
traditional knowledge and warm communal interaction reputation of the hospital and their basic trust in the
formed a supportive system that gave them confidence. hospital led them to choose its services for their current
Such communal support means that it was more likely pregnancies. This level of trust in the hospital can be
for them to return home for their obstetric care than to seen as a basic condition for seeking care. In addition,
engage in PP care. Their use of RP services was, thus, there was also trust in clinical competency, as expressed
partly based on their comparatively lower expectation by the common preference for an obstetrician (with
from healthcare care system. However, it should also be higher levels of training and licensing) rather than a
135
ARTICLE IN PRESS
general practitioner or a nurse. Together these two and the short encounter couldn’t make us familiar
categories of ‘basic trust’, namely trust in the hospital with each other.
system and trust in professionals, are derived from a
sense of assurance external to personal familiarity. This Third, the doctors viewed their relationship with PP
basic trust can be called institutional or impersonal trust users as an ordinary professional matter and rarely
(Gilson, 2003). acknowledged any agreement, or expressed particular
Trust in a particular person in charge of caring courtesy in these relationships. This asymmetrical
involved some combination of impersonal and inter- relation could be viewed as a ‘‘negative exchange’’
personal trust. Among RP cases, most patients trusted (Kapferer, 1976) in which PP users explicitly committed
the persons in charge without knowing them personally. to provide financial incentive to the obstetricians but the
PP cases, however, assumed that by offering additional obstetricians neither acknowledged nor acted in reci-
payments to an obstetrician they would be recognized procity. This meant that while the obstetricians could be
and would develop a personal relationship with their certain that they would be paid an additional fee, the PP
doctors. This interpersonal trust represents the second users could not shore up their confidence and thus still
level of trust identified by respondents. experienced some degree of uncertainty regarding the
However, PP users’ expectations of an interpersonal reliability of the doctors.
relationship were often not met. By offering a financial Overall, although PP respondents may have felt that
incentive PP users hoped to open the door to an they had trusting, interpersonal relationships with their
interpersonal relationship with their obstetricians, en- chosen doctors, the limits on these relationships suggest
suring that the obstetricians would be honest, sincere, that all pregnant women, PP and RP users, eventually
and do their best to serve the patients’ best interests. ended up relying more on impersonal trust. The financial
However, in practice, three sets of experiences suggested incentives involved in PP simply did not provide an
that the interpersonal trust perceived by PP users might adequate basis for building interpersonal trust. Given
only have been an expensive impersonal trust mistaken also that there is wider evidence to suggest that financial
as interpersonal trust. incentives may even undermine trust (Hunter, 1996), this
Firstly, most PP users said they had never known their study suggests that it is important to develop the
chosen doctor before. In other words, there had not mechanisms that can strengthen impersonal trust. These
been initial interpersonal relationship. Most obstetri- include professional control, an accreditation system,
cians chosen by PP users were suggested by the patients’ and the enforcement of ethical codes (Mechanic, 1998).
friends or relatives. As the quote below indicates, most
PP users were kept at arm’s length during the course of The impact of private practice on the healthcare system
their encounters, and never got a chance to be
acquainted with their obstetricians.
Both the negative and positive patient experiences of
PP may reflect deeper structural problems within the
It was a suggestion by one of my relatives. When I
health system as a whole. The negative experiences of PP
asked to be his private case, the obstetrician did not
led to distrust in the obstetrician. The following instance
say any words of acceptance. He just nodded and
from a PP user is illustrative:
wrote his name and phone number on a corner of my
ANC record. ‘‘When my pregnancy was near term, the doctor told
me to deliver before the New Year day, as he would
Secondly, even by the end of their pregnancies most not be around during holidays. I did not come as
obstetricians remained strangers to PP users, and vice appointed because I did not have a labor pain. My
versa. Although the interactions between doctors and labor pains started on the second of January during
pregnant women lasted over a period of more than 6 that long holiday, but my doctor did not come to
months, each interaction involved only a short visit to attend immediately as I expected. Two days later in
the hospital, too short to build up interpersonal trust. the labor room, a nurse delivered my baby and my
Not surprisingly, two PP cases indicated that when they doctor just came at the end to suture the perineum.
met the doctors who had attended them as PP users the There was not a single word of apology from him.
previous year, the doctors could not remember them. However, I paid for his PP fee since I felt that I had
promised to pay him even if not in words. No more
I was given the doctor’s telephone number and was PP for me.’’
told that I could call him for a consultation when I
felt I needed. But I dare not call him because I was Given the high expectations of PP patients, the
afraid I would be disturbing him too much. When I impacts of any negative experiences with PP will be
saw him at the ANC, most of our conversation was exacerbated when there are serious complications, such
about obstetric care. There was very little courtesy as injuries, disabilities or fatalities. Such outcomes have
136
ARTICLE IN PRESS
negative repercussions on trust in the hospital where the strong basis for a trusting, personal relationship.
obstetrician works, and the wider healthcare system. Although the unofficial financial payment associated
The positive outcomes of PP may, however, also have with PP is intended to provide a foundation for
negative impacts on the healthcare system. As the use of interpersonal trust, the doctors involved rarely acknowl-
PP services mostly begins with advice or suggestions edge or commit themselves to the reciprocal relation-
from friends or relatives, positive experiences of PP can ship. The interpersonal trust associated with PP services,
encourage more patients to engage in this practice. But if any, is, therefore, quite fragile.
as a doctor’s time is a finite resource, it is likely to be Our findings also suggest that the presence of private
impossible for a doctor to provide better care for his/her practice in public hospitals could have serious negative
private patients while maintaining adequate care for repercussions for healthcare generally. Positive private
complicated RP cases. There is a real danger that two- experiences might lead to an expansion of private
tier care will result. practice not only in obstetric care but also in other
It is noteworthy that among patients there were specialties. As private practice expands, doctors are
conflicting opinions about private obstetric practice. unlikely to be able to manage their time to ensure fair
Some endorsed it as a personal right as long as it was treatment of both PP and RP users. Negative PP
affordable. Others opposed PP on the grounds that it experiences may not only erode trust in a particular
would lead to unequal access to public resources, with doctor but also in professionals and the hospital system.
the obstetricians acting as gatekeepers. As doctors in Finally, financial payments may undermine genuine
state hospitals are fully paid by the public sector, reciprocities, sincere expressions of gratitude, and the
additional payments for privileged access to PP care maintenance of societal non-monetary value within the
were seen as similar to a bribe and, thus, as an unethical healthcare system.
or corrupt practice. The practice was left unperturbed Given the fragility of interpersonal trust in the current
partly because most PP users preferred to keep it context and the potential negative impacts of private
informal, while the doctors gained their extra income practice for impersonal trust, we suggest that steps must
without any formal obligation or acknowledgment. be taken to protect and build the current levels of
It is possible that the continued provision of unregu- impersonal trust in Thailand’s obstetric services. We
lated PP services may undermine institutional and specifically recommend that the two components of
professional trust. When the outcomes of PP service are impersonal trust must be strengthened: namely organi-
less than satisfactory, the patients may become disillu- zational or hospital trust through hospital quality
sioned not only with a particular obstetrician but also with accreditation process and trust in healthcare profes-
the hospital which allows PP in the first place, under- sional institutes to function as a regulatory control body
mining the basic trust in the whole healthcare system. of medical ethical conduct. Although it is difficult to
stop private practice completely in public hospitals,
policy makers and hospital administrators must develop
guidelines to limit the extent of private practice allowed,
Conclusions and so ensure social accountability to users of the
regular services. In promoting trust in the healthcare
This study illuminates the important role of trust in system, account must be taken not only of the socio-
the health care seeking behavior of pregnant mothers in cultural characteristics of patients and providers, but
Thailand. Two levels of trust are important: impersonal also of the organizational cultures of hospitals, which
trust in the hospital or healthcare system and strongly shape provider-patient interactions.
interpersonal trust rooted in a personal relationship. Further investigation of trust in health care settings is
Among the middle class mothers, attaining inter-personal required to provide a stronger foundation for policy
trust is the main motive underlying their obstetric care guidance. We hope that the explanatory framework
seeking behaviour. Their ability to pay and the fears and developed here can be applied more widely in other
anxieties about childbirth lead them to offer obstetricians specialties and settings. The future research agenda
an informal fee in return for personalised care (PP around private obstetric practice includes generating
services). Working class mothers who are more content greater understanding about obstetricians’ trust in
with regular obstetric care (RP services) rely on an patients and hospitals, the magnitude of reciprocity in
impersonal trust in the hospital standard care. They also private practice, and consideration of how to contain or
receive additional support during childbirth from their prevent the negative impacts of private practice. At a
traditional kinship networks, and can neither pay for more general level, further studies are needed to
private care nor feel able to bridge cultural gap between understand the role of trust at different healthcare levels
them and the obstetricians. (primary, secondary, and tertiary care) and in different
Even among PP users, the interactions they have with healthcare settings (rural, urban, local cultural orienta-
their obstetricians are usually inadequate to provide a tion), as well as perceptions of trust among healthcare
137
ARTICLE IN PRESS
providers and healthcare administrators. Understanding Franco, L. M., Bennett, S., & Kanfer, R. (2002). Health sector
the process of trust formation and factors that could reform and public sector health worker motivation: a
undermine trust will also be important in strengthening conceptual framework. Social Science & Medicine, 54,
non-financial incentives in the healthcare system. 1255–1266.
Gambetta, D. (2000). Can we trust trust? In: D. Gambetta
(Ed.), Trust: making and breaking cooperative relations,
electronic edition. (pp. 213–237). http://www.sociology.ox.
Acknowledgements ac.uk/papers/gambetta213-237.pdf: Department of Sociol-
ogy, University of Oxford.
This study was supported by the Thailand Research Gilson, L. (2003). Trust and the development of health care as a
Fund to Senior Research Scholar Program in Health social institution. Social Science & Medicine, 56, 1453–1468.
Systems and Policy Research. We appreciate its con- Hall, M., Camacho, F., Dugan, E., & Balkrishnan, R. (2002).
tinuous supports to this Program. Institutional grant Trust in the medical profession: conceptual and measure-
support to the IHPP from the Health Systems Research ment issues. Health Services Research, 37(5), 1419–1439.
Hanvorawongchai, P., Lertiendumrong, J., Teerawattananon,
Institute is also highly appreciated. Lucy Gilson is a
Y., & Tangcharoensathein, V. (2000). Implication of private
part-time member of the Health Economics and Finan- practice in public hospitals on the cesarean section rate
cing Programme of the London School of Hygiene and in Thailand. Human Resources for Health Development
Tropical Medicine, which is supported by the UK Journal, 4(1), 2–12.
Department for International Development. The UK Health Information Center. (2001). Health Statistics. Bangkok,
Department for International Development (DFID) Thailand: Bureau of Health Policy and Planning, Ministry
supports policies, programmes and projects to promote of Public Health.
international development. A workshop on Trust and Hunter, D. J. (1996). The changing roles of health care
Health Systems organized by the University of Witwa- personnel in health and health care management. Social
tersrand and London School of Hygiene and Tropical Science & Medicine, 43(5), 799–808.
Kapferer, B. (1976). Transaction and meaning: directions in the
Medicine, in September 2002, inspired us to investigate
antropology exchange and symbolic behavior. Philadelphia:
trust in healthcare. Institution for the Study of Human Issue.
Kramer, R. M. (1999). Trust and distrust in organizations:
emerging perspectives, enduring questions. American Review
References of Psychology, 50, 569–598.
Lewicki, R. J., & Bunker, B. B. (1996). Developing and
Aewsriwong, N. (1993). Culture of Thai middle class. (in Thai). maintaining trust in work relationships. In R. M. Kramer,
In S. Piriyarangsan, & P. Phongpanichit (Eds.), The middle & T. R. Taler (Eds.), Trust in organizations: frontiers of
class and Thai democracy (pp. 49–65). Bangkok: Political theory and research. Thousand Oaks, CA: Sage.
Economy Center, Faculty of Economics, Chulalongkorn Mechanic, D. (1998). Public trust and initiatives for new health
University and Feiedrich Ebert Stiftung. care partnerships. The Milbank Quarterly, 76(2), 281–302.
Balkrishnan, R., Dugan, E., Camacho, F., & Hall, M. (2003). Piriyarangsan, S., & Phongpanichit, P. (1993). The middle class
Trust and satisfaction with physician, insurers, and the and Thai democracy. Bangkok: Political Economy Center,
medical profession. Medical Care, 41(9), 1058–1064. Faculty of Economics, Chulalongkorn University and
Benoliel, J. Q. (1993). Personal care in an impersonal world. In Friederich Ebert Stiftung.
J. D. Morgan (Ed.), Personal care in an impersonal world: a Teerawattananon, Y., Tangcharoensathein, V., Srirattana, S.,
multidimensional look at bereavement (pp. 3–13). Amityville, & Tipayasoti, P. (2003). Twelve years pattern of hospital
NY: Baywood Pub. Co. delivery in Thailand, 1990–2001: a national survey. Journal
Coulson, A. (1998). Trust and contract in public sector of Health Science, 12(1), 1–18.
management. In A. Coulson (Ed.), Trust and contract: Wibulpolprasert, S., Hempisut, P., Pittayarungsarit, S. (2002).
relationships in local government, health and public services. Impact of free trade on health manpower development.
Bristol: The Polity Press. (In Thai). Bangkok.
138
Article
Health Informatics Journal

•••••
© The Author(s), 2009. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav
Vol 15(2): 95–107 [1460-4582(200701)15:2; 95–107; DOI: 10.1177/1460458209102971] www.sagepublications.com
System dynamics approach to immunization

healthcare issues in developing countries:
a case study of Uganda
Agnes S. Rwashana, Ddembe W. Williams and Stella Neema
This article critically examines the challenges associated with demand for immun-
ization, including the interplay of political, social, economic and technological forces
that influence the level of immunization coverage. The article suggests a frame-
work to capture the complex and dynamic nature of the immunization process
and tests its effectiveness using a case study of Ugandan healthcare provision.
Field study research methods and qualitative system dynamics, a feedback and
control theory based modelling approach, are used to capture the complexity and
dynamic nature of the immunization process, to enhance a deeper understanding
of the immunization organizational environment. A model showing the dynamic
influences associated with demand and provision of immunization services, with
the aim of facilitating the decision making process as well as healthcare policy
interventions, is presented.
Keywords
causal loop diagrams, healthcare services, immunization demand, immunization healthcare,
system dynamics, Uganda
Introduction
Stagnant and falling immunization rates in most sub-Saharan African countries have re-
sulted in renewed international attention, and the effectiveness and sustainability of immun-
ization programmes have become key issues of policy debate [1]. Increasing immunization
coverage to prevent childhood diseases is an important developmental issue [2–5] and an
area of critical research [6–11].
95
139
••••• Health Informatics Journal 15 (2)
In a study to evaluate new trends and strategies in international immunization, Martin
and Marshall [12] suggest that ‘failure to immunize the world’s children with life saving
vaccines results in more than 3 million premature deaths annually’.
There is an urgent need to improve immunization coverage around the world. The World
Health Organization has targeted measles for eradication in several regions of the world
by the year 2010 but, despite an effective vaccine, there are still estimated to be 30–40
million measles cases and 800,000 deaths per year [13]. In Uganda, despite numerous
immunization campaigns through the media, health visits and improved health services,
coverage rates are generally still low (less than 60%) [13].
Various approaches have been applied to understand immunization coverage problems.
However, there are still acknowledged deficiencies in these approaches, and this has given
rise to research into alternative solutions, including the need to adopt new technologies
to address the imbalance between immunization demand and provision of health services.
Understanding of the immunization coverage system and its problems may be helped
through system dynamics methods [14]. System dynamics provides us with tools which help
to better understand difficult management problems such as faced by the immunization
programme in Uganda.
Background to immunization coverage in developing countries

Preventable childhood diseases such as measles and premature deaths still occur, particularly
in the developing countries, due to low immunization coverage [5]. According to the World
Health Organization (WHO) [15], global coverage for measles immunization stands at 77 per
cent. In 2005, 28 million infants worldwide had not been vaccinated with DPT3 (diptheria
pertussis tetanus, third dose), with 75 per cent of these being in developing countries.
Immunization coverage is lowest in poor countries and among poor populations such as
Africa and Asia [16]. Globally, it was reported that the goal of fully immunizing 80 per cent
of the world’s children was reached in 1990; however, coverage in Africa for that year was
55 per cent. The United Nations Children’s Fund (UNICEF) and the World Health Organ-
ization [17] further state that immunization coverage rose significantly since the launch
of the expanded programme for immunization in 1974, from less than 5 per cent of the
world’s children in the first year to around 76 per cent by the end of 2003. Governments,
donor agencies and projects have made contributions towards the improvement of immun-
ization rates through the improvement of health infrastructure, financing, supplies, staffing
and management of national immunization programmes.
Figure 1 shows the immunization coverage rates in Uganda over time. In Uganda, BCG
immunization rates are higher than those of polio 3 (polio, third dose), measles and DTP3
due to the fact that BCG is administered at birth, while the rest are administered after
some weeks as scheduled. There is a general upward trend in immunization coverage
rates which is sustained by continuous campaigns to challenge negative responses towards
immunization.
Research design
The study employed the dynamic synthesis methodology (DSM) developed by Williams
Ddembe which combines two powerful research strategies: system dynamics (SD) and
96
14 0
Rwashana et al. Immunization in Uganda
Figure 1 Ugandan immunization rates 1981–2003
case study research methods [18]. Combining simulation and case study methods is
beneficial in that the case study enables the collection of on-site information about the
current immunization system, owners and user requirements and specifications used to
develop the generic model.
The system dynamics methodology illuminates key principal effects such as exogenous
shocks, systemic feedback loops, systemic delays and unintended consequences typical of
the immunization system as follows:
1 The immunization system presents exogenous shocks (factors external to the
system), such as changes in demand for immunization (which may occur as a result
of immigration) and the emergence of epidemics such as measles.
2 The immunization system contains feedback loops, communication paths and
methods that impact behaviour. A feedback loop is a control system where the
output of the system is fed back into the system [19]. For example, immunization
knowledge enhances utilization of immunization services which in turn results in
more knowledge.
3 The immunization system has systemic delays (time lags) which develop over time
in response to internal and external influences. Examples of such delays are those
arising from delivery of health services and cold chain maintenance (ensuring that
vaccines maintain the right temperature during distribution), especially to rural
communities, as well as delays in uptake of immunization.
4 Policy changes, feedback loops and behavioural changes in the immunization
system result in both intended and unintended consequences which can be
investigated using the SD methodology.
The problems faced by the nation’s immunization system policy can be interpreted in terms
of the information, actions and consequences which the system dynamics viewpoint pre-
sents [20–23]. The research design is shown in Figure 2.
97
141
Figure 2 The research design (source: [28])
In order to understand factors that influence immunization coverage and their rela-
tionships, survey research supported by semi-structured interviews was conducted to
understand the intricate information flows, delays and other competitive challenges. In
stage 1 (Figure 2), information related to immunization issues and associated problems
was initially collected from related literature and documents. Management and staff of
the national immunization programme (UNEPI) and Mukono District Health Services were
interviewed in order to establish the current problems faced by delivery and uptake of
immunization services. Field studies were used to determine the full range of activities
and challenges associated with immunization coverage (stage 2). Data obtained from
the study were analysed with the SPSS statistical package (stage 3). The factors affecting
immunization coverage, as well as national immunization policies used for immunization
coverage, were critically analysed and used to develop causal loop diagrams (Figures 4
and 5) using Vensim modelling software.1 The causal loop diagrams were presented to
stakeholders for comments and improvements. Feedback from consultations was then
used to develop the quantitative model. Stage 4 involved empirical investigation into the
current Ugandan immunization healthcare services. Data obtained from the empirical
investigation was used to populate the model. Stage 5 involved scenario modelling and
testing of various policies as well model validation. Stage 6 involved the proposition of
98
142
intervention strategies towards improvement of immunization coverage. Key information

and processes required for immunization coverage improvement were derived.
Field studies
Field studies were carried out to determine the full range of activities and events that
are associated with immunization coverage, and to examine the various acknowledged
factors associated with the provision and utilization of immunization services [24]. The study
was both qualitative and quantitative. The study was carried in Mukono district which lies
in the central region of Uganda. Mukono was selected as the area of study since it has a
good mix of both rural and urban populations. The people of Mukono district reside both
on the islands (one county) and the mainland (three counties) and the population consists
of more than 18 tribes. Data were collected through interviews using semi-structured
questionnaires from various stakeholders interested in the current immunization system:
mothers, health workers, district health officials, implementers of policy (UNEPI), policy
makers (government) and community leaders. Field observation of some activities was also
carried out, and other sources of data, especially those that would be able to highlight
historical, social, political and economic contexts, were collected.
Mothers. In each county of the selected district, 200 mothers were interviewed. A multi-
stage sampling method was used to define a target sample size of 800 mothers. The
sample size was determined as follows:2
where:
n is the required sample size.
z is the standard normal deviation corresponding to the level or degree of confidence
selected. Two confidence intervals normally used for the population mean are
95 and 99 per cent. This study selected the 95 per cent confidence interval as
suggested by Hutchins et al. [25]. For 95 per cent confidence interval, z = 1.96.
p is the fraction of population normally covered by immunization, i.e. 0.7.
q is the fraction of population not covered by immunization, i.e.
(1 – p) = (1 – 0.7) = 0.3.
e is the error caused by observing a sample instead of the whole population or the
permissible error which is less than or equal to 10 per cent.
Hence n = 80.7. Taking into account a non-response rate of 20 per cent, this gives a figure
of 100 respondents. A design effect consideration resulted in 200 (100 × 2) respondents
for each county, thus making the number of respondents in the four counties equal to 800.
In each county, the planned number of interviews was at least 200 mothers. A structured
questionnaire was used.
Health workers. Three private and five government health facilities were selected by simple
random sampling from the district. Those that were selected included one government
hospital and one private hospital, and the rest were health centres and dispensaries. At
each sampled health unit, two people were interviewed, one vaccinator and one officer-
in-charge of vaccines; this brought the total interviewed to 16.
99
143
Officials. At the district level, several meetings with various officials from the health and
administrative services were held. Local community leaders and national officials, as well
as consultants with UNICEF, were interviewed.
Data analysis
A thematic approach was used to compile and analyse the qualitative data. Socio-economic
and demographic variables were treated as independent variables, while attitudes and
knowledge were treated as both dependent and independent variables. The data were
analysed using SPSS 10.0 for Windows. The analysis employed descriptive statistics, including
frequencies, percentage distributions, cross-tabulations and correlations. Cross-tabulations
were used to further analyse the data by considering a combination of information on two
or more variables.
Analysis of the Ugandan immunization system

Healthcare services in Uganda, including immunization services, are provided through
a decentralized system consisting of geographically spread health centres and regional
hospitals which are categorized into health districts and health subdistricts. A system
diagram, a high level map showing the overall architecture of the immunization system,
was developed as shown in Figure 3.
The immunization system diagram shows four key subsystems which are explained
as follows:
1 The immunization management subsystem is responsible for the management,
monitoring and supervision of immunization services at the national level.
Collaboration with the government/donor agencies effects proper budgeting and
flow of funds. Management ensures that vaccines are ordered and delivered to the
vaccines management subsystem for distribution.
2 The vaccines management subsystem is responsible for the management and
delivery of quality vaccines to the healthcare service subsystem for distribution
to the districts and health facilities. This involves maintenance of vaccine efficacy
which is done by ensuring that the vaccines are kept at the right temperature.
3 The healthcare service subsystem is responsible for the provision of immunization
services to the population. The health service is concerned with the management
of resources (health workers, vaccines and equipment) as well as providing health
education to the community.
4 The community subsystem represents the population that utilizes the immunization
services. The community is obliged to take the children for immunization, follow
the immunization schedule and report any incidences concerning immunization.
The key external agents (outside the boundary) include the government and donor agencies
that provide the funds necessary to run the immunization programmes, the vaccine
manufacturing organizations, and the community leaders who carry out mobilizations in
the communities.
100
14 4
Figure 3 Immunization system diagram for Uganda
The system diagram conveys information on the boundary and levels of aggregation in
the model by showing the number and type of different organizations or agents repre-
sented. Key processes and flows of information are shown. However, the diagram does
not show the influences and causality which provide a deeper understanding of the
immunization system.
Causal loop diagrams

Causal loop diagrams (influence diagrams) are circular chain diagrams of cause and effect
which are used to represent relationships between variables which are often difficult to
101
145
describe. A relationship between two variables is represented by an arrow showing the
direction of influence. A positive sign on a link implies that a change in one variable results
in a change in the same direction, whereas a negative sign denotes a change in the opposite
direction. A feedback loop occurs when arrows connect a variable to itself through a series
of other variables. A feedback loop may be reinforcing (R) or balancing (B). A reinforcing
loop is defined as a positive feedback system that represents a growing or declining
action, while a balancing loop is a negative feedback system that is self-regulating [26].
Findings from the field study, as well as immunization studies of other researchers [3, 5,
7, 27], are represented in the causal loop diagrams of Figures 4 and 5. The figures show
the factors associated with demand for immunization and the provision of healthcare ser-
vices, as well as the key issues that need to be taken into consideration.
Figure 4 illustrates the intricate and complex relationships among factors affecting
immunization coverage from a parental participation perspective and a number of
feedback loops which may help to explain different immunization coverage levels [28].
It is this feedback structure that gives rise to complexity, non-linearity and time delays in
immunization coverage.
Figure 4 presents two balancing loops B1 and B2 and one reinforcing loop R1. Loop B1
is a balancing loop which shows that, with increased level of awareness, the demand for
immunization increases, which increases the number of children immunized, thus creating
a herd immunity which, in turn, results in fewer occurrences of epidemics [27]. Increased
epidemic occurrences, on the other hand, result in an increased disease burden; this neces-
sitates more awareness campaigns which, in time, lead to increased awareness levels.
Figure 4 Causal loop diagram for demand for immunization dynamics
102
146
Figure 5 Causal loop diagram for healthcare service provision
Loop B2, a balancing loop, represents the dynamics involved in the effectiveness of
healthcare systems. With a time delay, increased effectiveness results in increased level of
trust, thus increasing the demand for immunization services. However, as the demand for
immunization services increases, the resources are depleted and the workload increases,
thus causing a reduction in the effectiveness of the health systems. Loop R1 is a reinforcing
loop which shows a growing decline in the number of immunizations performed due to
inadequate provision of immunization services. As the number of children to be immunized
increases, there is need to increase the capacity of the healthcare services.
Figure 5 presents two balancing loops B3 and B4. Loop B3 seeks to achieve the set
immunization targets by focusing on health service delivery. The difference between the
targeted number of children and the actual number of children immunized creates a gap
which triggers an increase in government funding for immunization programmes. An
increase in funding results in increased resources and health centres which, when coupled
with a high level of service, will increase effectiveness; this, in turn, improves the demand
for immunization.
Loop B4 is a balancing loop, which represents the limiting factor resulting from increased
demand as far as the effectiveness of the healthcare system is concerned. An increase in
demand results in increased workload which reduces health worker motivation, resulting
in reduced level of service, which affects the provision of healthcare services.
103
147
The effectiveness of the health system can be achieved through a combination of factors
such as availability of resources (health centres, vaccines, transport), level of service of health
workers (skills, workload) and effectiveness of monitoring systems.
Insights from the causal loop diagram

From the causal loop diagram, a broad integrated view of the system is provided for stake-
holders to prioritize and set policies. The different policies and interventions that need to
be developed for the improvement of immunization coverage can then be generated from
a clear understanding of the complexity of the system.
Parental participation subsystem

The parental participation subsystem is based on the case study that was undertaken
and the immunization studies of other researchers [7, 27, 29, 30]. From the study, the key
issues that affect parental participation are grouped under the following, as shown in the
causal loop diagram in Figure 4:
• Effectiveness of health centres results in increased availability of immunization
services to mothers, which in turn increases the number of performed
immunizations.
• Mothers’ availability is associated with family problems (single parenting, number
of children in the household), high poverty levels and social status (mothers’ work,
work schedule).
• Level of trust in the health system is increased as the effectiveness of the
health centres increases. Increase in level of trust results in a change in attitude
towards increased demand in immunization. Issues that are associated with the
effectiveness of health centres include hygiene, levels of injection safety, number of
health workers at the health centre, and health workers’ response to the mothers.
• Level of immunization awareness is associated with mothers’ level of literacy,
belief in myths, effect of media, level of education and effectiveness of community
mobilization.
• Immunization dropouts (infants who take initial doses but do not complete the
immunization schedule) are associated with the level of civil unrest (presence of
wars), children’s health, level of complexity of immunization schedule, provision of
health cards and reminders.
Healthcare subsystem
The healthcare subsystem is based on field studies carried out in a number of health centres in
the Mukono district and various other studies. The key issues associated with the healthcare
system are grouped under the following, as shown in the causal loop diagram of Figure 5:
• Level of service is associated with health worker motivation resulting from the
provision of allowances, level of safety and workload. Increased skill level resulting
from the provision of quality training increases the level of service which in turn
increases the effectiveness of the health system.
104
148
• Effectiveness of monitoring of immunization activities involves the following:

monitoring systems for adverse events, documentation of immunization activities,
display of immunization activities, reporting of immunization activities and reviews
of immunization plans.
• Effectiveness of immunization campaigns is affected by the number of campaigns
in a year, availability of allowances, sufficient time for planning and effectiveness
of communication.
• Efficiency of health facilities is affected by the availability of resources (finance,
equipment, logistics) as well as monitoring of immunization activities.
Validation
The initial causal loop diagram generated from the field studies was presented to various
stakeholders for their comments and feedback on understanding of immunization coverage
problems. Thereafter, specific interviews were conducted with health workers, district health
workers and mothers to improve the resulting causal loop model, and further qualitative
analysis led to refinement of Figures 4 and 5.
Conclusions and future work

Systems dynamics has been used to capture and analyse complex interactions between
behavioural, technical, policy and cultural issues. This provides a broad integrated view of
the immunization system which facilitates communication and caters for the different
stakeholder viewpoints. The synthesis of the various theoretical concepts through the use
of causal loop diagrams facilitates the understanding of the immunization system which
enables agreement on different policies and priorities. Examining causal loop diagrams
enables decision makers to focus on the root causes of shortcomings and not the symptoms
alone. It is through such understanding that effective decisions and policy interventions
that are suitable for communities such as those in developing countries can be designed.
Ongoing work involves development of the quantitative model and applying simulation
modelling to test different healthcare policies using ‘what if’ analysis, with the aim of im-
proving policy analysis in immunization coverage. The causal loop diagrams are converted
into stock and flow diagrams which are defined by mathematical equations where vari-
ables are given initial numerical values emanating from historical data. The aim of
developing the model will be to show the relationships, trends and effects of key variables
by testing various scenarios.
Notes
1 http://www.vensim.com/software.html.
2 How to determine a sample size: http://www.extension.psu.edu/evaluation/pdf/TS60.pdf.
References
1 IDS. The social dynamics of infant immunisation in Africa. Working Paper 262. 2006.
2 Bryce J, Arifeen S, Pariyo G, Lanata F C, Gwatkin D, Habicht J, Multi-Country Evaluation of IMCI Study
Group. Reducing child mortality: can public health deliver? The Lancet 2003; 362; 159–64.
105
149
3 DISH II Project. Childhood Immunisation in Uganda: A Report of Qualitative Research. K2-Research
Uganda, 2002.
4 World Health Organization. Vaccines: summary country profile – Uganda. Immunisation and
Biologicals Global, 2002.
5 World Health Organization. Measles: progress towards global control and regional elimination.
Weekly Epidemiology 1999; 74 (50); 429–34.
6 Bozzette S A, Boer R, Bhatnagar M D, Brower J L, Keeler E B, Morton S C, Slora M A. A model for
smallpox vaccination policy. New England Journal of Medicine 2003; 348 (5).
7 Drain P K, Ralaivao J S, Rakotonandrasana A, Carnell M A. Introducing auto-disable syringes to the
national immunisation programme in Madagascar. Bulletin of WHO 2003; 81 (8).
8 Edmunds W J, Brisson M, Melegaro A, Gay N J. The potential cost effectiveness of a cellular pertussis
booster vaccination in England and Wales. Vaccine 2002; 20; 1316–30.
9 Fairbrother G, Hanson H K, Friedman S, Butts G C. The impact of physician bonuses, fees and
feedback on childhood immunisation coverage rates. American Journal of Public Health 1999;
89; 171–5.
10 Stafford E F Jr, Aggarwal S C. Managerial analysis and decision-making in outpatient health clinics.
Journal of Operations Research Sociology 1979; 30; 905–15.
11 Subramanyam K, Sekhar S C. Improving vaccine supplies to rural India. Socio-Economic Planning
Science 1987; 21 (2); 131–8.
12 Martin J F, Marshall J. New tendencies and strategies in international immunisation: GAVI and
The Vaccine Fund. Vaccine 2002; 21; 587–92.
13 World Health Organization. Measles Mortality Reduction and Regional Elimination Strategic Plan
2001–2005. 2001.
14 Forrester J W. Industrial Dynamics. Portland, OR: Productivity, 1961.
15 WHO/UNICEF (2006). Global immunisation coverage. http://www.who.int/immunisation_monitoring/
data/en, 17 March 2007.
16 Gwatkin D. The need for equity-oriented health sector reforms. Internal Journal of Epidemiology
2001; 30; 720–3.
17 UNICEF/WHO. Global Immunisation Vision and Strategy. 2005.
18 Williams D. Dynamic Synthesis: A Theoretical Framework for Research in Requirements Engineering
Process Management. Operational Research Society, 2000.
19 Sterman J D. Business Dynamics: Systems Thinking and Modelling for a Complex World.
McGraw Hill, 2000.
20 Williams D, Kennedy M. Towards a model of decision-making for systems requirements engineering
process management. In International Systems Dynamics Conference, Bergen, Norway, 6–10
August 2000.
21 Ritchie-Dunham J L, Rabbino H T. Managing from Clarity: Identifying, Aligning and Leveraging
Strategic Resources. Wiley: 2001.
22 Wolstenholme E F, Monk D, Smith G, McKelvie, D. Using System Dynamics to Influence and Interpret
Health and Social Care Policy in the UK. OLM Consulting, 2004.
23 Rwashana A S, Williams D W. Enhancing immunisation coverage through health information systems:
a system dynamics approach. Studies in Health Technology and Informatics 2007; 247–56.
24 Borooah V K. Gender bias among children in India in their diet and immunisation against disease.
Social Science and Medicine Journal 2003; 58; 1719–31.
25 Hutchins S S, Dezayas A, Blond K L, Heath J, Bellini W, Audet S, Beeler J, Wattigney W, Markowitz
L. Evaluation of an early two dose measles vaccination schedule. American Journal of Epidemiology
2001; 154 (11); 1064–71.
26 Maani E K, Cavana R Y. Systems Thinking Modelling: Understanding Change and Complexity.
Prentice Hall, 2000.
27 Hanratty B, Holt T, Duffell E, Patterson W, Ramsay M, White J M, Jin L, Litton P. UK measles outbreak
in non-immune anthroposophic communities: the implications for the elimination of measles from
Europe. Epidemiology and Infection 2000; 125; 377–83.
28 Rwashana A S, Williams D W. An evaluation of healthcare policy in immunisation coverage in Uganda.
In Proceedings of the 24th International System Dynamics Conference, System Dynamics Society.
October 2006.
106
15 0
29 Leask J, McIntyre P. Public opponents of vaccination: a case study. Vaccine JVAC 2002; 4056; 1–4.
30 Ymba A, Perrey C. Acceptibility of tetanus toxoid vaccine by pregnant women in two health centres in
Abidjan (Ivory Coast). Vaccine 2003; 21; 3497–3500.
Correspondence to: Agnes S. Rwashana
Agnes S. Rwashana BSc MSc, Assistant Stella Neema BA MA PhD, Medical

Lecturer Anthropologist/Researcher
Faculty of Computing and Information Makerere Institute of Social Research
Technology Makerere University
Makerere University PO Box 16022 Kampala, Uganda
PO Box 7062 Kampala, Uganda E-mail: sheisim@yahoo.com
E-mail: asemwanga@cit.mak.ac.ug
Ddembe W. Williams Senior Lecturer

Faculty of Computing and Information
Technology
Makerere University
PO Box 7062 Kampala, Uganda
E-mail: d.williams@cit.mak.ac.ug
107
151
Social Science & Medicine 71 (2010) 2005e2013
Contents lists available at ScienceDirect
Social Science & Medicine

journal homepage: www.elsevier.com/locate/socscimed
Discursive gaps in the implementation of public health policy guidelines in India:

The case of HIV testing
Kabir Sheikh a, *, John Porter b
a
Public Health Foundation of India, Gurgaon, India
b
London School of Hygiene & Tropical Medicine, United Kingdom
a r t i c l e i n f o a b s t r a c t
Article history: The implementation of standardized policy guidelines for care of diseases of public health importance
Available online 29 September 2010 has emerged as a subject of concern in low and middle-income countries (LMIC) globally. We conducted
an empirical research study using the interpretive policy analysis approach to diagnose reasons for gaps
Keywords: in the implementation of national guidelines for HIV testing in Indian hospitals. Forty-six in-depth
India interviews were conducted with actors involved in policy implementation processes in five states of
Public health policy making
India, including practitioners, health administrators, policy-planners and donors. We found that actors’
Implementation
divergences from their putative roles in implementation were underpinned by their inhabitation of
Public health guidelines
HIV testing
discrete ‘systems of meaning’ e frameworks for perceiving policy problems, acting and making deci-
Policy analysis sions. Key gaps in policy implementation included conflicts between different actors’ ideals of perfor-
Interpretive analysis mance of core tasks and conformance with policy, and problems in communicating policy ideas across
systems of meaning. These ‘discursive’ gaps were compounded by the lack of avenues for intellectual
intercourse and by unaccounted interrelationships of power between implementing actors. Our findings
demonstrate the importance of thinking beyond short-sighted ideals of aligning frontline practices with
global policymakers’ intentions. Recognising the deliberative nature of implementation, and strength-
ening discourse and communications between involved actors may be critical to the success of public
health policies in Indian and comparable LMIC settings. Effective policy implementation in the long term
also necessitates enhancing practitioners’ contributions to the policy process, and equipping country
public health functionaries to actualize their policy leadership roles.
! 2010 Elsevier Ltd. All rights reserved.
Introduction public health initiatives and programmes (Brugha, 2003; Das &
Hammer, 2004).
The policy-practice gap in public health There is however significant evidence that the practices of
frontline health care providers do not always correspond with
This paper investigates the problem of gaps in the imple- standard policies for care of important diseases. For instance,
mentation of standardized public health practice guidelines in insisting on HIV tests before hospital admission or surgery, breach
India, using the issue of HIV testing as an illustrative case study. The of confidentiality of HIV status and testing patients without specific
formulation of evidence-based guidelines has emerged as a key consent count among the common infractions of national policy
approach in health care policy globally, and particularly for diseases guidelines by doctors in India. In a study in Pune city, India, Sheikh
of public health concern. Guidelines and the ideas they contain et al. (2005) documented that private practitioners prescribed HIV
often assume emblematical status in the global public health tests in large numbers and often indiscriminately, and forty percent
community and become a great part of the policy lexicon in their reported mandatory HIV testing before conducting invasive
respective fields (Ogden, Walt, & Lush, 2003). Guidelines are widely procedures. In a multi-centre study of 2200 health care profes-
seen as benchmarks of quality of care, and adherence to them by sionals in private and public hospitals and health centres (Kurien
frontline providers is regarded to be requisite for the success of et al., 2007) 67% of the doctors reported that they screened
patients for HIV before elective surgery, and only 30% reported
obtaining written consent for the test regularly.
Das and Hammer (2004) assessed private and government
* Corresponding author. Tel.: þ91 9911987670 doctors’ practices in treating infant diarrhoea, pharyngitis, tuber-
E-mail address: kabirsheikh@yahoo.co.uk (K. Sheikh). culosis, depression and pre-eclampsia, observing significant
0277-9536/$ e see front matter ! 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2010.09.019
152
2006 K. Sheikh, J. Porter / Social Science & Medicine 71 (2010) 2005e2013
deviations from recommended standards of practice in Delhi. values and beliefs which sometimes come into conflict with policy
Kamat (2001) reported widespread instances of presumptive norms e respectively in the cases of HIV testing (Sheikh & Porter,
treatment of malaria by private practitioners in a Mumbai suburb. 2009) and neonatal care (Miljeteig & Norheim, 2006).
Chakraborty and Frick (2002) have reported shortcomings in Inefficiencies of government health departments and lack of
private practitioners’ treatment of acute respiratory infections in capacity to execute their regulatory functions have also been
children, evaluated against a standard of WHO-recommended implicated as reasons for poor implementation of public health
guidelines. Gross variations in tuberculosis treatment among policies in India (Das Gupta, Khaleghian, & Sarwal, 2003;
private practitioners have been reported frequently in India, Muraleedharan & Nandraj, 2003). Bhat and Maheshwari (2005)
notable studies including Uplekar, Juvekar, Morankar, Rangan, and have highlighted vulnerabilities and lack of capacity of govern-
Nunn (1998) and Prasad et al. (2002). ment departments to engage meaningfully with private sector
The phenomenon is also well documented in other low and actors. Muraleedharan and Nandraj (ibid) also implicate the
middle-income countries (LMIC). Health professionals’ treatment absence of, or lack of detail in, legal frameworks for medical care
of malaria in Sudan, childhood diarrhoea in Thailand and sexually standards (nationally and in different states and municipalities), as
transmitted infections (STI) in South Africa respectively has been a context for perverse practices in the private medical sector.
reported to diverge from standardized norms (Mannan, Malik, & From the existing literature, it can be concluded that while
Ali, 2009; Howteerakul, Higginbotham, Freeman, & Dibley, 2003; documentation of policy violations by frontline practitioners in
Schneider, Chabikuli, Blaauw, Funani, & Brugha, 2005), while in India is common, there is little by way of systematic understanding
Somalia practitioners have been found to disregard global recom- of reasons for the problem. Explanations of policy-practice gaps
mendations for tuberculosis management (Suleiman, Houssein, have tended to be partial and/or conjectural, and in-depth empir-
Mehta, & Hinderaker 2003). There is also evidence of widespread ical investigations of the phenomenon are lacking.
divergence from policy recommendations in the case of dengue
diagnosis (Ng, Lum, Ismail, Tan, & Tan, 2007) in Malaysia, The putative architecture of implementation
management of chronic obstructive pulmonary disease in Morocco
(Benouhoud, Trombati, Afif, Aichane, & Bouayad, 2007), and anti- How are national guidelines for health care supposed to be
microbial prescription for paediatric respiratory tract infections in implemented? Table 1 charts the key groups of actors involved in
Argentina (Aznar, Mejía, Wigton, & Fayanas, 2005) respectively. the implementation of public health guidelines in India, with their
What underlies this ubiquitous phenomenon of policy-practice putative functions indicated in italics. Firstly, medical practitioners
gaps? Frontline practitioners’ divergences from standardized poli- are obliged to conform to national guidelines in managing their
cies have also been reported in high-income country (HIC) contexts, patients. For the purposes of this study, we have focused only on
with developed regulatory mechanisms. These are typically recognised practitioners in urban government and private hospi-
ascribed to contests around policy content e the evidence base for, tals, trained in the allopathic (Western) system of medicine. Within
or appropriateness of a particular set of guidelines (Chiao et al., hospitals, administrators including heads of departments and
2010; Warwick, 2010). However, viewed in LMIC contexts, given superintendents are responsible for staff behaviour. The National
the greater prevalence of diseases of global public health impor- AIDS Prevention and Control Programme (NAPCP) publishes and
tance, and the general recognition of deficient accountability promulgates policies and guidelines for various aspects of HIV care
systems and health market failures, the problem has naturally and control, including for HIV testing. Finally, international donors
assumed a greater significance. Prevailing diagnoses of policy- and technical organizations have a professed role in the develop-
practice gaps from LMIC include the bearing of manifold influences ment and propagation of public health guidelines.
on the behaviour of frontline practitioners (Howteerakul et al., It cannot be assumed that actors’ participation in policy imple-
2003; Paredes, de la Peña, Flores-Guerra, Diaz, & Trostle, 1996), mentation is shaped solely by formal rules and policies. Policy theo-
and the failings of health systems, particularly regulatory systems rists Hjern and Hull (1982) contend that in real-life contexts
(Haines, Kuruvilla, & Borchert, 2004). implementation processes frequently do not coincide with the
In the Indian context e Das and Hammer established that doctors “written constitutions” which define actors’ putative roles. In order to
in Delhi often did not comply with regulations in spite of being investigate problems of implementation, it becomes necessary to
aware of them (2004), and suggest that incentive structures for undertake empirical research to elaborate “how policy problems are
doctors in both private and government sectors do not promote their defined and addressed” by the actors involved in the implementation
uptake of standard guidelines. Kamat (2001) observed that private process (ibid). In this study we adopt precisely such an approach, of
doctors are unduly subject to patients’ expectations in unregulated investigating gaps in the implementation of public health policy
and highly competitive markets, and Kielmann et al. (2005) and guidelines from the perspectives of the participant actors.
Datye et al. (2006), reporting from the same study, have explained
private doctors’ divergences from recommended practices in terms
Interpretive policy analysis
of reactions to varying market, policy and social pressures, under-
lined by the challenge of keeping abreast with knowledge in
In exploring the reasons for policyepractice gaps in this study, the
a rapidly changing field. Other studies have associated Indian
interpretive approach of policy analysis is adopted, in which events
hospital practitioners’ divergent practices with their reference to
and phenomena are viewed through the lens of the interpretations of
Table 1
Policy actors interviewed, and their putative roles in implementing public health guidelines.
Development Propagation and Education Supervision and Monitoring Conformance

Medical practitioners (32 participants) O
Public health authorities National health programme officials O O O
(3 participants)
Hospital administrators O
(7 participants)
International actors (4 participants) O O
153
K. Sheikh, J. Porter / Social Science & Medicine 71 (2010) 2005e2013 2007
participant actors (Yanow, 2000). This approach relies on in-depth e The requirement of specific written informed consent from
qualitative research methods and is derived from constructivist a patient before conducting a HIV test
epistemologies in social research which aim to “include multiple e Prohibition of HIV testing as a pre-condition for performing
voices and views in their rendering of lived experience” (Charmaz, a procedure, such as surgery (also referred to as mandatory or
2000, p. 525). The approach requires the analyst to be immersed in pre-surgical testing)
the beliefs of participants to understand their purposes and motiva- e Strict confidentiality of HIV test results, including from health
tions for actions. In accessing these interpretations of actors, inter- workers not directly involved in the care process.
pretive analysis can account for the role of various factors including
beliefs, assessments of realities, values, self-interest and dominatory Respondents were encouraged to discuss the topics at length,
power in shaping their actions and interactions. and interviews were guided by probes. Data collection was
The concept of ‘systems of meaning’ is integral to the interpre- concluded when representatives of all the groups identified as
tive approach, signifying how different actors construct their real- being involved in implementation had been interviewed e no
ities, define problems and identify solutions for the problems “new” names of organizations were being identified by
(Yanow, 2000). Policy theorist Vickers (1965) had previously respondents.
proposed that policy actors form ‘appreciations’ of given problems, The “framework” approach for applied policy analysis,
a concept analogous to ‘systems of meaning’. Appreciations consist combining inductive and deductive approaches was used to orga-
of actors’ judgements around the facts of the problem (reality nize and analyze the data from interview transcripts (Ritchie &
judgements) combined with value judgements answering the Spencer, 1994). A thematic framework consisting of three levels
question “what ought to be?”, which together inform their deci- of thematic codes was developed: a priori themes drawing on the
sions around action e action judgements. Reality, value and action topic guide, emergent issues arising from interviewees’ responses
judgements represent categories on the basis of which actors’ and analytical themes based on patterning of emergent themes (see
explanations of their actions can be thematically organized (see Table 2). The analytical theme categories were organized around
Fig. 1) - this framework finds application in organizing the results in the meanings that actors attached to implementation processes, in
this paper. Collectives of actors belonging to the same groups and keeping with the interpretive approach (see above). Vickers’
organizations may share cognitive mechanisms and decision- formulation (1965) of the appreciative dimensions of policy
making processes, and use similar language to discuss policy actors’ responses was useful in classifying respondents’ explana-
problem, hence forming ‘communities of meaning’. tions of their actions (see Fig. 1). The framework was applied
systematically to the data, using the qualitative data programme
Atlas Ti 4.2. Coded chunks of data were retrieved, organized into
Methods charts and written up.
The ‘framework’ demonstrates the systematic nature of the
The study was conducted in nine hospitals in five Indian states, analysis e a criterion of reliability. Care was also taken to ensure
using in-depth social science research methods. Principles of that a multiplicity of perspectives is represented, and that the views
maximum variation were applied in respect to identifying hospitals of a particular group are not presented as the sole truth about
for the study, based on two criteria: type of hospital and a situation. Preliminary findings were presented to study partici-
geographical zone. Nine urban hospitals were selected with pants, to ensure their credibility or trustworthiness (Yanow, 2000),
representation from the government, private, and charitable a key standard of quality in interpretive research.
sectors; and located in five cities, one each from the North, West, Institutional ethics clearance was obtained from committees of
South, East and Central Zones of the country. Four government the academic institution where the study was originated, and by
hospitals, three private hospitals and two charitable hospitals were a local ethics review committee. All interviews were conducted and
identified purposefully. In these five hospitals 32 practitioners recorded following verbal consent and presentation of a standard-
working in specializations associated with HIV testing were ized information sheet. All transcripts and recordings were acces-
selected, also purposefully, ensuring distribution across speciali- sible only to the authors, and care was taken while writing up to
zations, gender and experience. Additionally 14 senior officials delete particulars of individuals and institutions which may have
representing hospital authorities, national health programmes, and led to their identification.
international organizations were interviewed.
In-depth interviews were conducted with all study participants
by the first author of this paper over 18 months in the years Results
2005e2007. Appointments were sought telephonically, by email or
by personal visits. All participants were interviewed in hospitals or The perspectives of different groups of actors on their own
offices which were their usual places of work. Topic guides con- participation in the implementation process are presented
sisted of queries around respondents’ participation in the imple-
Table 2
mentation of HIV testing policies. We focused on the following
Thematic framework of code categories.
aspects of the national HIV testing policy (NACO 2003), which have
been reported to be infringed widely by medical practitioners: A priori themes Expected role in implementing policies
Actual experience of participation in
policy implementation (respondents
Value Judgements other than practitioners)
Actual experience of implementing each
aspect of HIV testing policy (practitioners only)
Emergent themes Accounts of own actions in implementing policies, or not
“Meaning” Action Judgements (action Explanations for actions
judgements)
Analytical themes Reality judgements: pragmatic considerations
Reality Judgements influencing actions
Value judgements: value orientations influencing
actions
Fig. 1. Vickers’ framework of judgements to characterize actors’ ‘systems of meaning’.
154
sequentially, thus allowing and leading up to the diagnosis of policy sometimes seen as unwelcome obstacles. A related observation is
implementation gaps. that the doctors generally approached problems from the position
of belief in the innate beneficence of medical intervention, and
hence asking for consent from patients presented a paradox (see
A. Medical practitioners: primacy of performance Fig. 2). Doctors saw diagnosis as a duty towards patients, and part of
their embedded functions within healing institutions.
Actions
While all the doctors were aware of the guidelines, a majority This business of taking informed consent from a patient before
reported that they contravened guidelines for informed consent doing a HIV test. I don’t know where this has come from. A
(30/32) and confidentiality (27/32) on occasion. Pre-surgical HIV patient has come to you. He is sick, he needs your help. Will you
testing although practiced by fewer doctors (20/32) was probably be thinking about this or about treating him? (Physician, private
as frequent a practice. Most surgeons reported routinely ordering hospital)
pre-surgical HIV tests for their patients. If these were proscribed by The impulse to diagnose a patient was also indicative of the high
hospital authorities they resorted to subversive means such as valuation of the scientific challenge of the clinical procedure.
sending their patients to nearby private laboratories to be tested, or Different diagnostic tests were ordered to maximise knowledge
conspiring with hospital pathologists to perform tests unofficially. about a patients’ condition e a gynaecologist reported the impor-
In other instances, practitioners reported that they complied with tance of ‘knowing where we stand’ in order to be able to ‘take all
the national HIV testing guidelines. However, surface conformance the measures’ for further management. This inclination to investi-
did not always imply engagement with underlying principles of the gate may have frequently overridden considerations of patient
guidelines, especially in the case of informed consent. autonomy in choosing to be tested.
We have always taken informed consent. How much informa- In managing their patients, doctors typically followed unwritten
tion the clients have understood is a separate issue. How do we rules of prioritization of time and resources, based on the seri-
validate or verify that? Humne to bata diya [We did what was ousness of patients’ conditions. In general hospitals with patients
required]. Now how much they have ingested, understood, we with a wide range of serious illnesses, the needs of patients with
can’t say that, we can’t guarantee that. (Senior microbiologist, HIV/AIDS were often not the most imminent. For some doctors, the
government hospital) ‘exceptional’ set of rules (e.g. specific written consent, confidenti-
ality) and allocation of resources (e.g. counsellors) around HIV
Paradoxically, practitioners sometimes treated the consent testing militated against prevailing logics of patient equality and
procedure as a means to persuade, even coerce the patient into need-based prioritization.
taking the test.
In their explanations of these different actions, practitioners in For counselling, we need a man, a patient, a sofa and a cup of tea,
both private and public hospitals emerged as a distinct ‘community and a room. And there in the emergency, you have three patients
of meaning’ (Yanow, 2000), sharing particular cognitive mecha- on one bed, one is alive, one is dying and one is dead. I am not
nisms, engaging in similar acts and using similar language to against consent and counselling. What I am saying is the ground
discuss policy problems. Their divergences from policy norms are realities are entirely different. (Senior physician, government
explained by a mix of values which were often at odds with the hospital)
emphasis on autonomy and patient choice which underpin HIV Relationships between patients and medical practitioners were
testing policies, and by pragmatic considerations. An overarching often fundamentally asymmetric, and patients frequently asked
theme is that practitioners tended to be preoccupied with the doctors to make their decisions for them. While this may have been
performance of core clinical tasks, often at the expense of confor- contrary to the reciprocal logic of informed consent procedures
mance with policy guidelines. Table 3 summarizes practitioners’ which required patients and doctors both to be autonomous and
explanations of their divergent actions. mutually aware of their rights, it was seen by a number of doctors as
a sign of essential trust, and valued by them.
Practice values and goals
Informed consent. Doctors’ emphasis on clinical outcomes and cure They all say ‘well doctor if it is your child what will you do?
represented the value placed on expediting clinical tasks with Whatever you would do for your child, do the same. We leave it
efficiency. The first of these tasks is diagnosis and in this context, to you’. So there is a different relationship. That’s one of the
procedures such as informed consent for a HIV test were great things of working [in India]. (Surgeon, private hospital)
Table 3
Medical practitioners’ systems of meaning.
DIAGNOSIS
Action judgements
Value Clinical efficiency Superficial compliance provider
orientations Innate beneficence of medical care Contravention seeker
Duty to diagnose Discretion/flexibility
Scientific challenge of diagnosis
Trust between patient and provider Patient Doctor
Patient equality
Professional thoroughness
Economy
Right to safety provider seeker
Reality Perception of personal risk
judgements Perception of deprivation
(government/charitable)
Patients’ expectations CONSENT
Co-worker expectations
Fig. 2. Paradox of seeking consent to provide a service.
155
Mandatory testing. In the case of surgeons, role perceptions were frequently comparing their conditions against an imagined ideal of
even more narrowly focused on the specific task of completing the standards of facilities in Western countries.
surgical act successfully. Mandatory HIV testing for some was one
of many necessary steps in preparing for the surgery, and regula- Confidentiality. In some instances, staff who worked in operation
tions preventing mandatory testing were widely regarded by theatres with surgeons expressed their objections to participating
surgeons as obstacles in the way of performing their primary in surgery without knowledge of patients’ HIV status. Supporting
defining role. staff are vitally important in the successful undertaking of surgical
In some instances pre-surgical testing was regarded as an procedures, and their perspectives were important in motivating
essential part of a thorough clinical work-up for patients. It was mandatory pre-surgical HIV testing. Concerns around the risk of
advocated in private hospitals as part of a package of infection HIV infection through needle-stick injuries for nurses, paramedical
control interventions, and was seen as a signifier of quality in the staff and hygiene workers were also voiced by doctors in all sectors
workplace, and linked to professional values around hygiene and of hospitals.
safety. In under-resourced public hospitals, costs for protective
Maintaining confidentiality is one issue. but at a lot of times,
equipment for surgeons are typically borne out-of-pocket by
simple waste disposal becomes a problem. We are supposed to
patients, and in government and charitable hospitals, practitioners
have segregated waste, but at times, we do not get the bags. If
promoted mandatory testing as a cost-saving alternative to
we look at a sweeper, taking away the waste from the hospital,
adopting universal precautions during surgery.
you will realise how dangerous it is for him. He is carrying all
that waste which has got a lot of fluids, lot of sharps, he is
Confidentiality. Lastly, a key value consideration upheld by doctors
dripping the waste on the floor. How dangerous it is! Just
was that of teamwork and solidarity between co-workers. The
because he is not aware of these things, and how dangerous it is
rights of all health workers to optimal protection from infection
for him. (Junior surgeon, government hospital) (32)
were invoked in defence of practices of mandatory testing. Fairness
in allowing all health workers access to patients’ HIV status was Reportedly health workers also widely felt that they needed to
a consideration which contradicted confidentiality policies. know which patients were HIV positive, for which they resorted to
labelling of case files or beds. Such practices, while breaching
Reality judgements confidentiality norms, were widely tolerated by doctors with
Informed consent. Patients’ actions and attitudes were key reality a sympathetic perspective of the needs of their co-workers.
considerations for doctors. Reportedly patients often approached Apart from a shortage of resources, doctors described their
the clinical encounter trusting the doctor to make the best deci- constraints in terms of lack of time and manpower and of excess of
sions for them, and hence asking for written consent represented patients. For instance the idea of strict confidentiality in a crowded
a rejection of that expectation. Further, patients’ expectations were consulting room with large volumes of patients and limited hours
said to be usually oriented around alleviation of their physical was described as ‘completely impractical’, by one gynaecologist in
ailments, and formal procedures for consent and counselling were a government hospital. Given a low staff to patient ratio, relatives of
often regarded by them as time consuming diversions. Some patients were usually co-opted to perform various basic tasks of
patients were reportedly offended by the presumed implication of care provision, in the context of which confidentiality of patients’
promiscuity in asking them to consent to a HIV test. HIV status was highly improbable.
Mandatory testing. An important “reality” for doctors was the risk B. Public health functionaries: negating regulatory roles
of infection by a HIV positive person through needle-stick injuries
or during interventional procedures. Although the likelihood of Actions
their being infected in scientific terms was very low, the fear of Hospital administrators, and health and HIV/AIDS programme
infection was considerable among most doctors, and particularly officials too did not perform many of their putative regulatory
surgeons, and motivated indiscriminate HIV testing by doctors, functions in ensuring the implementation of public health policy
especially pre-operatively by surgeons. guidelines. Superintendents and heads of department in govern-
Save ourselves! The patient comes later. There is a saying in ment and private hospitals alike displayed considerable leniency
Hindi “bhookhe pet na hoye bhajan gopala” [I can’t express my towards infringements and took few measures to streamline HIV
devotion to God, with an empty stomach]. If we are hungry, if we testing practices among their staff. HIV/AIDS programme officials
are sick, if we are down, then how we will serve? (Senior too reported widespread problems in implementing guidelines,
surgeon, government hospital) particularly in the private medical sector.
These public health functionaries also cited a broadly homoge-
These notions of high risk were compounded by the variably nous set of meanings and motivations for these divergent actions -
accurate perception that protective equipment available to prevent the specific explanations of two groups of functionaries are
injury and HIV infection was inadequate. This was characteristic of summarized in Table 4. A central observation here is that these
a generalized sense of deprivation that prevailed among doctors, actors found more meaning in supportive and developmental
Table 4
Public health functionaries’ systems of meaning.
Action judgements
Hospital administrators HIV/AIDS programme officials
Value orientations Supporting practitioners’ rights Supporting care provision Focus on developmental tasks
‘Development’ ethic Neglect of regulatory tasks
Disengagement with policy principles
Reality judgements Delicate relationships with staff Lack of regulatory capacity
Lack of support from senior authorities Non-integration with general health services
Intellectual subordination to international actors
156
activities, than in their regulatory functions. The relative neglect of Remarkably, lack of conviction about the appropriateness of the
their roles in enforcing and regulating standards was also linked to policies was a common theme among HIV/AIDS programme offi-
overriding pragmatic considerations such as their inability to cials, who were ostensibly owners and promulgators of the policies.
dictate the behaviour of practitioners and the apparent lack of Many programme officials shared the reservations expressed by
resources and capacity to enact regulations. practitioners around the emphases on patient autonomy and on
procedural formalities such as written consent and counselling.
Value orientations ‘[Practitioners’ ] doubts [around the guidelines] are absolutely
Among hospital authorities, role identities were typically valid, and whatever practices they are doing, they have a reason, it
focused around ensuring the efficient delivery of clinical care, and is not unnecessary’, averred a national programme official.
tasks such as ensuring guideline implementation were seen as
secondary, and sometimes as impediments. Being medical profes- Reality judgements
sionals themselves, it was evident that in practice administrators Frequently, the functionaries’ departure from regulatory tasks
often encouraged or supported practitioner discretion rather than was explained by their incapacity to perform these tasks in the face
uniform compliance to policy. Heads of department also cited of situational constraints. Administrators in both government and
solidarity with their subordinates as explanations for their lack of private-run hospitals emphasised the importance of maintaining
enforcement of policies for mandatory HIV screening and harmonious relationships with the practitioners in their employ, in
confidentiality. order to ensure efficient hospital functioning. Relationships with
subordinates were delicate and some department heads perceived
There are senior doctors and junior doctors and everybody
a lack of acceptance of their regulatory functions, by staff as well as
would like to know about the patients’ (HIV) status. Probably it
administrators. One hospital officer in charge of HIV and infection
would not be fair if I knew and I did not tell my junior doctor,
control reported a lack of support from superior authorities in
because that means I am taking the precautions and she is not.
implementing policy guidelines which prevented him from per-
(Head of department, government hospital)
forming his duties
The HIV/AIDS programme officials also expressed their prefer-
All these things are not on the priority list of administrators.
ence for supportive rather than regulatory-type engagement with
There is no culture of this - public health work is not recognised
medical providers. One official specifically opposed the institution of
[in hospitals] (Head infection control, government hospital)
laws around HIV testing, contradicting the official stance at the time.
Lack of true authority over the behaviour of medical providers,
Frankly speaking if you put some kind of regulatory mechanism,
was also a resonant theme among HIV/AIDS programme officials.
at least in India I think, that may not serve the purpose.
HIV/AIDS programmes represent administrative structures parallel
because it is democratic you know. If we keep insisting on a HIV
to the general health services, and doctors in government hospitals
law, every hospital should have this, should have that, then
are not in direct relationships of accountability to HIV/AIDS pro-
people will come out with their own ways. (Senior official,
gramme officials (see Fig. 3). Due to the prevailing balance of power,
national HIV/AIDS programme)
programme workers (such as technicians and counsellors)
Programme officials generally emphasised their role in stationed in hospitals were unable to exercise control over doctors’
promoting voluntary change in practitioner behaviour, by practices, and programme officials did not have the necessary
providing enhanced educational and workplace resources. They authority over hospital administrators to be able to enforce norms.
idealized the growth and expansion of HIV care services and project In the case of private providers, programme officials renounced
activities, and were preoccupied with financing and instituting new a regulatory role altogether. A state programme official emphasised
services and facilities (for testing, treatment, training), in the that the physical task of regulating private providers’ practices was
context of which the ‘command and control’ philosophy of regu- beyond the resources and means of the programme.
lation appeared inimical. A state level official remarked that it was Another key relational dynamic, which may have contributed to
not desirable to combine the programme’s preferred role of insti- programme officials’ lack of engagement with the principles of HIV
tution builders and resource providers with a policing function. testing policy, is their intellectual subordination to international
State health HIV/AIDS

services officials programme officials
Employment, Employment
Limited
answerability (contractual)
authority
GOVT. HOSPITAL
Hospital administrators Power

asymmetries Programme staff
(technicians,
counsellors)
Hospital practitioners
Fig. 3. Health services and HIV/AIDS programme e parallel authority structures.
157
actors. A senior official commented that national policy guidelines was not always a priority. Said a government hospital gynaecolo-
had not developed through means within his control, and cited the gist: ‘because we know that we are not able to maintain confi-
World Health Organization and other international agencies as dentiality, so we take it loosely.’ ‘These (policy guidelines) are ideals
influential sources of guidance in this respect. which have to be strived for, not necessarily to be achieved’, said
one government pathologist.
C. International actors: streamline and standardize The problem of communicating ideas contained in the policy
guidelines across disparate ‘communities of meaning’ is best
Representatives of technical agencies were influential in deter- exemplified by the case of informed consent. While international
mining the contents of national HIV testing policies. They valued actors e with their value orientations fixed on patient autonomy e
ideals of patient autonomy in medical care as contained in the HIV regarded informed consent norms to be of paramount importance,
testing policies and favoured the standardization of medical prac- practitioners preoccupied with performative goals were confused
tices on these lines, but experienced problems in communicating by the paradoxical logic of consent and regarded it chiefly as an
policy messages to implementers. impediment to care. Their adoption of guideline norms often
remained specious and superficial, without intrinsic engagement or
Value orientations comprehension of their value.
The international actors interviewed widely perceived their role
to be that of setters of norms and standards. They felt that it was Unshared platforms
important to enforce standards for consent and confidentiality, ‘Communities of meaning’ e the discursive worlds of medical
which they considered to be universal principles. They tended to practitioners, of public health functionaries, or of international
place a high value on conformance and on streamlining the prac- actors, are formed in contexts of their shared working environ-
tices of doctors to meet global norms. The officials interviewed ments, common goals and similar logics of action. The insularity of
typically expressed strong belief in the principles underlying the these communities and rigidity of their thoughts and ideas were
HIV testing guidelines, particularly on the importance of patient reinforced by the lack of opportunity to engage in mutual mean-
autonomy in medical care interactions. ingful dialogue. By several accounts, there was little communica-
tion between the different government departments involved in
(T)here should be confidentiality, there should be systems of
implementing the guidelines. Private sector practitioners particu-
counselling or informing the patient. Even that is not happening
larly expressed a sense of intellectual seclusion, with little
here. Why? In the West, people make sure that the patient is the
academic engagement of any description with other institutions,
one who decides. The doctor tells you the options and you
public or private. Said one private hospital practitioner: ‘for private
decide what you want.. Eventually in India, people will
[doctors], there are not many options [to train in HIV care]. There is
demand that you treat me with decency, with respect (Officer in
no effort to involve us’. Government practitioners too reported that
a United Nations (UN) technical agency)
there were few opportunities for formal dialogue and deliberation
They were largely dismissive of doctors’ explanations for within their hospitals and with representatives of the HIV/AIDS
infringements of HIV testing policies, which they believed to be programme, even on contentious aspects of the HIV testing policies.
spurious.
Problems of power and authority
Reality judgements Gaps in implementation of policy guidelines were compounded
Technical agencies have a clear mandate to influence the by complexities of power relationships between groups of actors,
contents of national policy guidelines. However the officials’ efforts which did not reflect the ‘expected’ hierarchy of policy imple-
to exercise this mandate were occasionally impeded by a lack of mentation. The most apparent of power imbalances observed was
communication and comprehension. One UN official recounted medical professionals’ ability to resist authority in all its forms. In
a discussion with HIV/AIDS programme officials in which “routine the poorly regulated private sector, there was little recognition of
offer of testing” was conflated with “routine testing” (with contrary the authority of government agencies, and administrators
connotations) e indicative of generalized problems of conveying perceived no reason to enforce the guidelines among their staff.
underlying principles of guidelines to implementing actors who did ‘[The HIV/AIDS programme] has no jurisdiction over us ’, emphas-
not share the same beliefs and ethical frames of reference. ised a private hospital administrator. In government hospitals,
administrators appeared to express little more than notional
D. Diagnosing the gaps authority over practitioners’ HIV testing practices.
The asymmetric power dynamics in transferring HIV testing
Disparate meanings policies from international agencies to national programmes may
Implementing actors’ inhabitation of discrete world-systems of be important in determining the manner of their implementation.
meaning and purpose e symbolized by amalgams of philosophical The intellectual dominance of international technical agencies in
and pragmatic considerations e represents a fundamental obstacle setting standards for HIV testing was largely unquestioned, and
for the universal implementation of standardized policy guidelines. HIV/AIDS programme officials reported a lack of engagement with
A central observation here is that doctors’ and public functionaries’ the contents of the guidelines. Another critical, related obstacle to
perceptions of the purpose of their work tended to focus on guideline implementation lies in the rift between lines of hierarchy
performance of core tasks rather than on conformance with policy in the HIV/AIDS programme and the general health services (see
guidelines. For both these groups of actors, their decisions were Fig. 3). Hospital personnel are employees of general health services
highly contingent on such factors as their relationships with and not of the HIV/AIDS programme, and programme officials were
contiguous actors, the variable adequacy of resources, and incon- largely unable to exercise real authority in hospitals.
sistent support from administrative structures.
In these contexts of uncertainty, they similarly reported finding Discussion
most meaning in the performative and entrepreneurial aspects of
their roles, and conformity with restrictive and regulative tasks and In the global context of LMIC health policy, frontline practi-
with the precisely articulated rationales of HIV testing guidelines tioners and other health systems actors have been viewed primarily
158
in terms of their instrumental roles in the implementation of public intentions of policymakers, towards deeper, more particular
health policies (Peters, 2003; Rowe, de Savigny, Lanata, & Victora, systemic solutions.
2005). The main contribution of this study is in understanding
health policy implementation in India from the “emic” perspectives A deliberative mindset
of the various participant actors. We have elaborated the perspec- The diagnosis of discursive gaps suggests that policy-planners
tives of discrete ‘communities of meaning’ constituted, respec- would be advised to acknowledge the deliberative possibilities of
tively, of medical practitioners, public health functionaries and implementation, and take steps to enhance the quality of and
international actors, and diagnosed the implementation gaps as opportunities for dialogue between different groups of actors
resulting through a combination of disparities in different imple- implementing public health policies, including between different
menting actors’ systems of meaning, deficient avenues for dialogue government departments engaged in implementation. Deliberation
between actors and unaccounted power balances in the imple- also has the virtue of making actors’ concerns, needs and interests
mentation process. The study is based on individual accounts and transparent and can help curtail the role of prevailing power
hence biased towards individualized explanations, to the relative imbalances (Healey, 1993). An emphasis on communication and
neglect of broader structural factors and contexts. We cannot also cross-learning is critical if public health policies are to be reconciled
claim that the setting of hospitals in five cities in different with the disparate worldviews and motivations of actors engaged
geographical zones encapsulates the entire Indian situation. in administering and delivering health services, and with the
Nevertheless the empirical findings gleaned from a systematic socially entrenched functions and operations of health care insti-
research process represent credible insights into public health tutions through which they are implemented.
policy processes in India.
While there are no other studies in LMIC which attempt to Including voices from the field
understand implementation gaps holistically from the perspec- In high-income countries, it is often the norm that practitioners’
tives of different involved actors, there are a few which have experiences inform the development and revision of practice
focused on particular groups of actors. Howteerakul et al. (2003) guidelines (Chiao et al., 2010; Warwick, 2010). In this analysis
and Paredes et al. (1996) respectively elaborated how Thai and however, practitioners’ engagement with existing guidelines was
Peruvian doctors’ interpretations of policies to be implemented tenuous, and the ‘alternative’ perspectives they espoused remain
are informed by their values and experiences. Kapiriri and Bondy ‘underground’, and untested for ethical and scientific validity.
in a study on Ugandan health planners and practitioners (2006) Support for in-service training and participatory educational pro-
have observed that their decisions were guided as much by grammes for practitioners, and the institution of participatory
personal experience and discussions with colleagues as by formal policy fora to deliberate practice guidelines can help to develop
guidelines. cultures of debate, and also practitioners’ capacities to represent
Viewed in the Indian context, our findings support the legitimate local concerns. Instituting systems for the representation
hypotheses of Kielmann et al. (2005), Datye et al. (2006) and Kamat of users of health care in policy development and refinement
(2001) of the highly contingent nature of medical practice, in which potentially represents a longer-term goal.
following policies is often not a preeminent consideration for
practitioners. Our observations also resonate with Sheikh and Empowering implementers
Porter’s study on mandatory HIV testing (2009) and Miljeteig and Finally, the vulnerable position of public health functionaries
Norheim’s study of neonatal care (2006), each of which docu- vis-a-vis both practitioners and international actors reflects the
ment how Indian doctors’ ethical orientations differ from accepted acute need to strengthen country institutional structures for
Western bio-ethical norms. Our findings also demonstrate that effective stewardship, and to set agenda for national health.
practitioners’ divergence from putative roles cannot automatically Country-level public health functionaries in particular are in an
be conflated with failures of ethics or commitment to service (Das & advantaged position to bridge the practical knowledge of practi-
Hammer, 2004) e disparate values and ‘meanings’ may be as tioners and the universal knowhow of international agencies, and
significant an explanation of divergent practices. must be supported financially, materially and morally to provide
The ‘meanings’ of Indian public health functionaries, also balanced leadership to policy initiatives.
heavily focused around ethics of performance and hinging on
uncertain upstream and downstream relationships with other Acknowledgements
actors, are largely unexplored in the literature, and hence our
findings represent unique contributions in these areas. The The authors are grateful for the financial support received from
vulnerable position of these functionaries, entrusted with the the Aga Khan Foundation’s International Scholarship Programme,
key role of implementing national policies but doubly subjected the DfID TARGETS Consortium at the LSHTM, and the University of
to the obduracy of practitioners and the ascendant voice of London’s Central Research Fund, towards conducting this research.
international actors, is a troubling commentary. The perspectives
of international actors and their relationships with Indian References
national health functionaries also represent a poorly explored
domain. Aznar, M., Mejía, R., Wigton, R., & Fayanas, R. (2005). Predictors of antibiotic
Conventional prescriptions for implementation gaps in LMIC prescription in respiratory tract infections by ambulatory care practitioners.
Medicina, 65(6), 501e506.
are often concentrated on enhancing central control over imple- Benouhoud, N., Trombati, N., Afif, H., Aichane, A., & Bouayad, Z. (2007). [Audit of
mentation of guidelines, by means such as the introduction of management of chronic obstructive pulmonary disease by general practitioner
stronger regulations and laws, and strengthening supervision and in Morrocco. Revue De Pneumologie Clinique, 63(1), 40e44.
Bhardwaj, K., & Divan, V. (2005, August). The HIV/AIDS bill 2005. Positive Dialogue,
audit (Bhardwaj & Divan, 2005; Rowe et al., 2005). These strate-
(20).
gies typically do not account for implementing actors’ varied and Bhat, R., & Maheshwari, S. K. (2005). Human resource issues: implications for health
sometimes divergent ‘meanings’, notions of purpose and accom- sector reforms. Journal of Health Management, 7(1), 1.
plishment, and prevailing dynamics of interrelationships. Our Brugha, R. (2003). Antiretroviral treatment in developing countries: the peril of
neglecting private providers. British Medical Journal, 326(7403), 1382e1384.
analysis highlights that it is pertinent to look beyond short- Chakraborty, S., & Frick, K. (2002). Factors influencing private health providers’
sighted ideals of ensuring that frontline practices mirror the technical quality of care for acute respiratory infections among under-five
159
children in rural West Bengal, India. Social Science & Medicine, 55(9), policy research in South Asia: Building capacity for reform. Washington, D.C.:
1579e1587. World Bank.
Charmaz, K. (2000). Grounded theory: objectivist and constructivist methods. In NACO (National AIDS Control Organization). (2003). National AIDS prevention and
N. K. Denzin, & Y. S. Lincoln (Eds.), Handbook of qualitative research (2nd ed). control policy.
(pp. 509e535). Thousand Oaks. Ng, C. F. S., Lum, L. C. S., Ismail, N. A., Tan, L. H., & Tan, C. P. L. (2007). Clinicians’
Chiao, E. Y., Dezube, B. J., Krown, S. E., Wachsman, W., Brock, M. V., Giordano, T. P., diagnostic practice of dengue infections. Journal of Clinical Virology, 40(3),
et al. (2010). Time for oncologists to opt in for routine opt-out HIV testing? The 202e206.
Journal of the American Medical Association, 304(3), 334e339. Ogden, J., Walt, G., & Lush, L. (2003). The politics of ’branding’ in policy transfer: the
Das Gupta, M., Khaleghian, P., & Sarwal, R. (2003). Governance of communicable case of DOTS for tuberculosis control. Social Science & Medicine, 57(1), 179e188.
disease control services: A case study and lessons from India. World Bank. Paredes, P., de la Peña, M., Flores-Guerra, E., Diaz, J., & Trostle, J. (1996). Factors
Das, J., & Hammer, J. (2004). Strained mercy: the quality of medical care in New influencing physicians’ prescribing behaviour in the treatment of childhood
Delhi. In World bank policy research working paper series. Delhi: World Bank. diarrhoea: knowledge may not be the clue. Social Science & Medicine, 42(8),
Datye, V., Kielmann, K., Sheikh, K., Deshmukh, D., Deshpande, S., Porter, J., et al. 1141e1153.
(2006). Private practitioners’ communications with patients around HIV testing Peters, D. H. (2003). A framework for health policy research in South Asia. In
in Pune, India. Health Policy & Planning, 21(5), 343e352. A. Yazbeck, & D. Peters (Eds.), Health policy research in South Asia: Building
Haines, A., Kuruvilla, S., & Borchert, M. (2004). Bridging the implementation gap capacity for reform. Washington, D.C.: World Bank.
between knowledge and action for health. Bulletin of the World Health Orga- Prasad, R., Nautiyal, R. G., Mukherji, P. K., Jain, A., Singh, K., & Ahuja, R. C. (2002).
nization, 82(10), 724e731, discussion 732. Treatment of new pulmonary tuberculosis patients: what do allopathic doctors
Healey, P. (1993). Planning through debate: the communicative turn in planning do in India? The International Journal of Tuberculosis and Lung Disease, 6(10),
theory. In F. Fischer, & J. Forester (Eds.), The argumentative turn in policy analysis 895e902.
and planning (pp. 233e253). Duke University Press. Ritchie, J., & Spencer, L. (1994). Qualitative data analysis for applied policy research.
Hjern, B., & Hull, C. (1982). Implementation research as empirical constitutionalism. In R. Burgess, & A. Bryman (Eds.), Analyzing qualitative data (pp. 173e194).
European Journal of Political Research, 10(2), 105e115. London/New York: Routledge.
Howteerakul, N., Higginbotham, N., Freeman, S., & Dibley, M. J. (2003). ORS is never Rowe, A. K., de Savigny, D., Lanata, C. F., & Victora, C. G. (2005). How can we achieve
enough: physician rationales for altering standard treatment guidelines when and maintain high-quality performance of health workers in low-resource
managing childhood diarrhoea in Thailand. Social Science & Medicine, 57(6), settings? Lancet, 366(9490), 1026e1035.
1031e1044, (1982). Schneider, H., Chabikuli, N., Blaauw, D., Funani, I., & Brugha, R. (2005). Sexually
Kamat, V. (2001). Private practitioners and their role in the resurgence of malaria in transmitted infections e factors associated with quality of care among private
Mumbai (Bombay) and Navi Mumbai (New Bombay), India: serving the affected general practitioners. South African Medical Journal, 95(10), 782e785.
or aiding an epidemic? Social Science & Medicine, 52(6), 885e909. Sheikh, K., & Porter, J. D. H. (2009). “It’s 100% for me”: hospital practitioners’
Kapiriri, L., & Bondy, S. J. (2006). Health practitioners’ and health planners’ infor- perspectives on mandatory HIV testing. Indian Journal of Medical Ethics, 6(3),
mation needs and seeking behavior for decision making in Uganda. Interna- 132e137.
tional Journal of Medical Informatics, 75(10e11), 714e721. Sheikh, K., Rangan, S., Kielmann, K., Deshpande, S., Datye, V., & Porter, J. (2005).
Kielmann, K., Deshmukh, D., Deshpande, S., Datye, V., Porter, J., & Rangan, S. (2005). Private providers and HIV testing in Pune, India: challenges and opportunities.
Managing uncertainty around HIV/AIDS in an urban setting: private medical AIDS Care, 17(6), 757e766.
providers and their patients in Pune, India. Social Science & Medicine, 61(7), Suleiman, B. A., Houssein, A. I., Mehta, F., & Hinderaker, S. G. (2003). Do doctors in
1540e1550. northwestern Somalia follow the national guidelines for tuberculosis
Kurien, M., Thomas, K., Ahuja, R. C., Patel, A., Shyla, P. R., Wig, N., et al. (2007). management? Eastern Mediterranean Health Journal, 9(4), 789e795.
Screening for HIV infection by health professionals in India. National Medical Uplekar, M., Juvekar, S., Morankar, S., Rangan, S., & Nunn, P. (1998). Tuberculosis
Journal of India, 20(2), 59e66. patients and practitioners in private clinics in India. International Journal of
Mannan, A. A., Malik, E. M., & Ali, K. M. (2009). Antimalarial prescribing and Tuberculosis & Lung Disease, 2(4), 324e329.
dispensing practices in health centres of Khartoum state, 2003e04. Eastern Vickers, G. C. (1965). The art of judgment. A study of policy making. London: Chapman
Mediterranean Health Journal, 15(1), 122e128. & Hall.
Miljeteig, I., & Norheim, O. F. (2006). My job is to keep him alive, but what about his Warwick, Z. (2010). Barriers to the implementation of the UK HIV testing guidelines
brother and sister? How Indian doctors experience ethical dilemmas in in secondary care: how many are medical? International Journal of STD & AIDS,
neonatal medicine. Developing World Bioethics, 6(1), 23e32. 21(3), 205e206.
Muraleedharan, V., & Nandraj, S. (2003). Private health care sector in India - policy Yanow, D. (2000). Conducting interpretive policy analysis. Newbury Park CA:
challenges and options for partnership. In A. Yazbeck, & D. Peters (Eds.), Health Sage.
16 0
2. The case-study For example, individual health workers may respond
differently to the same set of incentives; and patients vary in
approach their response to treatment advice.
Second, as the examples of motivation and health seeking

Lucy ilson behaviours show, HPSR questions often require study of
niversity of Cape own, South frica and London School
the complex behaviours of, and relationships among,
of Hygiene and ropical Medicine, nited ingdom of
actors and agencies; and how those relationships influence
reat ritain and orthern reland
change, including change over time. The case-study
The case-study approach is a research strategy entailing approach is particularly relevant to such experiences
an empirical investigation of a contemporary phenome- (Thomas, 1998).
non within its real life context using multiple sources of
Third, as discussed in Part 2 of the Reader, the case-study
evidence, and is especially valuable when the boundaries
approach can be used both to support and analyse policy
between the phenomenon and context are blurred
development: it can generate information for policy (for
(Yin, 2009). It is widely used in research fields and
example see Rolfe et al., 2008 in this section) or be used to
disciplines of relevance to HPSR, such as political science,
analyse past policy experiences in detail (see, for example,
public administration, planning studies, organizational
Shiffman, Stanton & Salazar, 2004 in this section).
and management studies, community psychology and
sociology. Case-study work is also very flexible. In terms of overarching
research purposes (see Part 2: Step 2), it can:
There are three main reasons why this research approach
support e ploratory inquiry to gain a better under-
n
is particularly relevant to HPSR. First, health policy and
standing of certain situations or to generate ideas and
systems experience is strongly influenced by, and is often concepts for use in follow-up work;
embedded in, contextual factors that must themselves
allow detailed description of particular experiences;
n
become part of the focus of inquiry (Gilson et al. 2011).
enable the investigation of ‘how’ and ‘why’ e planatory
n
For example, health worker motivation is influenced by a questions, supporting analytic generalization through
range of personal, organizational and societal factors, as cross-case analysis (see Part 1: Section 7);
well as relationships with others; and, in turn, many be used as a study approach in emancipatory work, such
n
aspects of the provision of health care are influenced by as action research and participatory inquiry.
the motivation of health workers (Franco, Bennett &
Finally, case-study work can involve either single cases
Kanfer, 2002). Similarly, patients’ decisions to use
(of health policies, for example) or a number of individual
services or adhere to treatment advice represent
cases of the same type (a case-study of different health
responses to many influences, such as:
facilities, for example), or an embedded case approach,
their own understandings of illness, and how best to
n where one type of case is nested within a broader case or
treat it
encompasses other cases. An example of the latter would
advice received from friends and family
n be the case of a single health policy process that is inves-
past experience of health providers
n tigated by examining the overall process and experience at
the availability of cash to cover costs
n a number of case-study sites within the health system (such
the gender dynamics influencing household decision-
n as regions, districts, and/or facilities); or the case of a
making. primary health care facility that is recognized as nested in a
On any health policy and systems issue there are also district health system, requiring investigation of the case at
multiple interpretations of the same experience as different both levels.
people bring their own contexts to bear on its interpretation.

The range of ‘cases’, the unit of focus, relevant to and an overview of procedures within the different phases
considered in HPSR, therefore, is quite varied. It includes of case-study work that help to ensure trustworthiness
(Robson, 2002; Thomas, 1998; Gilson & Raphaely, 2008): (see also Gilson et al., 2011).
individuals, communities, social groups, organizations;
n Given the areas of weakness in the current body of HPSR
events, relationships, roles, processes, decisions, parti-
n work (Gilson & Raphaely, 2008), key areas that require
cular policies, specific policy development processes, attention in future case-study research in the field
research studies;
include:
health system decision-making units, particular health-
n
care facilities, particular countries. the use of theory to support and enable analysis
n
case selection to support analysis
n
case contextualization, especially in single cases
n
Rigour in case-study work in studies with multiple cases, comparative analytical
n
strategies that support analytic generalization.
In general terms, the rigour of case-study work is secured
These issues are discussed further below, in relation to
by full reporting on the methods of data collection and
the papers selected for this section.
analysis, so that readers can assess whether the analysis
and interpretation is credible. As discussed in Part 2: Readers are also encouraged to review available texts
Step 3, the judgement of credibility is, in essence, one of (for example Yin, 2009; Thomas, 1998) on good case-
whether the research procedures suggest that the study practice to strengthen HPSR case-study work.
conclusions derived are trustworthy. Table 10 provides
Table 10 Procedures to ensure trustworthiness in case-study research (Source: Yin, 2009)
Criterion of
trustworthiness Case-study tactic Phase of research
Confirmability • Conduct literature review, identify key concepts Research design

• se multiple sources of evidence Data collection
• Establish chain of evidence Write up of analysis
• Ask key informants to review draft research report (member checking)
Dependability • Develop case-study protocol (so that others can see the decisions made in Data collection
developing the study, and why you made them)
• Develop case-study database (complete set of data, that others could review)
Credibility • Look for patterns in data and across cases (pattern matching) Data analysis
• Consider explanations for experiences analysed (explanation building)
• Consider rival explanations (alternative explanations for the patterns
identified)
• se logic models to think through causal mechanisms
• Triangulation – compare and contrast data across respondents, data sources,
data types and cases
• Consider negative cases (explicitly seek out experiences that contradict
your main line of argument, to test that argument and refine it)
Transferability • se theory in single case studies Research design

• se replication logic in multiple case studies (test ideas from one case
against subsequent cases)
References Overview of selected papers
Gilson L, Raphaely N (2008). The terrain of health policy The papers included in this section were chosen to
analysis in low- and middle-income countries: a review of address a range of issues related to health policy and
published literature 1994–2007. Health Policy and systems and to show the different cases that can be
Planning, 23(5):294–307. used in HPSR case-study work, as shown below.
Gilson L et al. (2011). Building the field of Health Policy Atkinson et al. (2000) examine experiences of Brazi-
n
lian decentralization in three local settings, seeking to
and Systems Research: Social Science Matters. PLoS
understand the ways in which the contextual features
Medicine 8(8):e1001079. of social organization and political culture influence
these experiences.
Franco LM, Bennett S, Kanfer R (2002). Health sector
Murray & Elston (2005) examine the single case of
n
reform and public sector health worker motivation: a
obstetric care in Chile, to understand the influence of
conceptual framework. Social Science Medicine, a macro level intervention (privatization in both
54(8):1255–1266. financing and provision of care) over health system
organization (meso level) and clinical practice (micro
Robson C (2002). Real world research: a resource for level).
Mutemwa (2005) examines multiple cases of district
n
Oxford, Blackwell Publishing:3–15. level decision-making in the context of Zambian
decentralization and in relation to information
Thomas A (1998). Challenging cases. In: Chataway J,
systems.
Wuyts M, eds. Finding out fast: investigative skills for
Rolfe et al. (2008) document and categorise the
n
policy and development. London, Sage Publications: existing experience of private midwifery care across
307–332. multiple districts in the United Republic of Tanzania,
to generate information to guide future regulatory
Yin RK (2009). Case study research: design and methods, policy development.
4th ed. Thousand Oaks, California, Sage Publications.
Russell & Gilson (2006) examine, across multiple
n
households, the consequences of health care seeking
behaviour for the economic situation, or livelihoods,
of households in a low-income Sri Lankan community
and the factors influencing this behaviour.
Shiffman, Stanton & Salazar (2004) examine the single
n
case of the safe motherhood policy in Honduras to
understand how and why this policy became a political
priority.
Although most papers primarily draw on qualitative data,

Russell & Gilson (2006) report a mixed-method study
(see also cross-sectional papers) in which an initial
structured cross-sectional household survey, repre-
sentative of the local community, generated findings that
provided an overview of household experiences related
to the key concerns of the study and the basis for more
detailed qualitative work. The survey was specifically
used to inform the selection of a small number of
household cases for inclusion in a second phase of work,
in which detailed understanding of the households’

experiences was generated through application of Contextualization. All descriptive and explanatory case
multiple data collection methods (a combination of study work requires ‘thick description’, that is, interpre-
qualitative and quantitative data). The analysis also tation of the phenomenon of focus by reference to con-
combines data from both phases of the study. textual features (see the section on the ethnographic
lens; also see Atkinson et al. 2000; Murray & Elston,
These papers also offer insights into rigorous practice for
2005; Russell & Gilson, 2006; Shiffman, Stanton &
case-study work, in relation to the four key current areas
Salazar, 2004);
of weakness in HPSR case-study work, as outlined below.
Analysis and generalization. Rich analysis of context,
The use of theory. Exploratory and descriptive case-
as well as clarification of conflicting perspectives and
study work may build theory as the basis for more
interpretations of different actors, is particularly impor-
detailed, future inquiry into the issue of focus (see
tant in single-case studies as the value of such work lies
Mutemwa, 2005). However, explanatory work should seek
in unpicking the complexity of the phenomenon of focus
to use theory to design the investigation as well as seeing
in a detailed narrative of how and why things happen so
it as a product of research (Atkinson et al., 2000). When
they can be seen more clearly (Murray & Elston, 2005).
designing the investigation and conducting the analysis
Single-case studies can also generate persuasive and rich
(Shiffman, Stanton & Salazar, 2004) theory can help to
insights when combined with theory testing (Shiffman,
gain a deeper understanding of the issue, as well as to
Stanton & Salazar, 2004). Meanwhile, analysis of mul-
contribute to the longer term process of theory testing
tiple case studies is based on the principle of replication.
and building (see also Part 1: Section 7).
Data are not pooled across cases and then analysed by
Selecting cases. Unlike survey work, case selection is issue; instead each case is treated as a unitary whole
never based on the logic of representivity. Instead, the and comparison and contrast across these cases supports
choice depends on the main aim of the study and some the development of general insights and conclusions that
examples are given below. are considered to have sufficient universality to apply to
In exploratory work, the aim may be to find as many
n other settings (see Atkinson et al., 2000; Rolfe et al.
different types of case as possible to allow limited 2008; Russell & Gilson, 2006). The principle of replication
description of many cases and the generation of is central to this process of analytic generalization in that
categories (see Mutemwa, 2005 and Rolfe et al. 2008).
the process of analysis is undertaken iteratively, to see if
In a single case, the aim is to explain how and why
n the analysis of the first case is replicated as expected in
something happens by looking in detail at the inner
the second, third, fourth case, etc. (see Rolfe et al. 2008).
workings of the case. Therefore, the case may be
chosen because it is broadly interesting; or is thought
to be typical of that type of case (Shiffman, Stanton &
Salazar, 2004); or because it is not typical and, indeed,
may represent an extreme case that challenges existing
ideas or the theory guiding the study (Murray & Elston
2005).
In multiple cases the aim may be to test theoretical
n
ideas through comparing and contrasting different
cases (see Atkinson et al., 2000) or to select different
cases to allow analytic generalization on an issue
(Russell & Gilson, 2006).
References for selected papers
Atkinson S et al. (2000). Going down to local: incorpo- Shiffman J et al. (2004). The emergence of political priority
rating social organisation and political culture into for safe motherhood in Honduras. Health Policy and
assessments of decentralised health care. Social Science Planning, 19(6): 380–390.
Medicine, 51(4): 619–636. http://dx.doi.org/10.1093/heapol/czh053
http://dx.doi.org/10.1016/S0277-9536(00)00005-8 Reproduced by permission of Oxford University
n
Reproduced by permission of Elsevier. Copyright
n Press. Copyright Oxford University Press, 2004.
Elsevier, 2000.
Murray SF, Elston MA (2005). The promotion of private

health insurance and its implications for the social
organisation of health care: a case study of private sector
obstetric practice in Chile. Sociology of Health and llness
27(6):701–721.
http://dx.doi.org/10.1111/j.1467-9566.2005.00470.x
Reproduced in the print version only, by permission
n
of Wiley. Copyright Wiley, 2005.
Mutemwa RI (2005). HMIS and decision-making in

Zambia: re-thinking information solutions for district
health management in decentralised health systems.
http://dx.doi.org/10.1093/heapol/czj003
n
Press. Copyright Oxford University Press, 2005.
Rolfe B et al. (2008). The crisis in human resources for

health care and the potential of a ‘retired’ workforce:
case study of the independent midwifery sector in
Tanzania. Health Policy and Planning, 23(2):137–149.
n
Russell S, Gilson L (2006). Are health services protecting

the livelihoods of the urban poor in Sri Lanka? Findings
from two low-income areas of Colombo. Social Science
Medicine, 63(7): 1732–1744.
http://dx.doi.org/10.1016/j.socscimed.2006.04.017
n
Elsevier, 2006.

5",#-+ 5,#*$,* W <*&#,#$* \X URSSS[ ]XY^]_]
'''H*+1*B#*/H,"3b+",-(*b1",1,#3*&
!"#$% &"'$ (" ()* +",-+. #$,"/0"/-(#$% 1",#-+ "/%-$#1-(#"$

-$& 0"+#(#,-+ ,2+(2/* #$(" -11*113*$(1 "4 &*,*$(/-+#1*&
)*-+() ,-/*
5-/-) 6(7#$1"$ -898 :*%#-$$* ;*#+- :"+#3 <*&*#/"1 -8=8 >-2+" ?*$/#@2*
;#3- A+#B*#/- -8 :#,-/&" C#-1 &* 6+3*#&- -
-
!"#$%&'"(& )* +"),%$#-./ 0$(12"34 5))#"% 678348(,/ 9:*)%4 ;)$4/ <(8="%18&. )* 0$(>-"1&"%/ 0$(>-"1&"% 0?@ ABC/ <D
=
E1>)3$ 4" F$7G4" ;HGI38>$/ J)%&$3"K$/ 6%$K83
!"#$%&'$
D)* 1",#-+ "/%-$#1-(#"$ -$& 0"+#(#,-+ ,2+(2/* "4 ()* 1",#*(E #$ ')#,) -$ "/%-$#1-(#"$ #1 *3=*&&*& ,-$ )-B* 3-F"/
*G*,(1 "$ ()* '-E #$ ')#,) "/%-$#1-(#"$-+ 0"+#,E #1 #30+*3*$(*& -$& "$ )"' ()-( "/%-$#1-(#"$ 42$,(#"$1H :*1*-/,)
"$ )*-+() 1*,("/ /*4"/31 )-1 0-#& 1,-$( -((*$(#"$ (" ()#1 -10*,(H I4 ()* ,+-#31 3-&* 4"/ &*,*$(/-+#1*& 3-$-%*3*$( #$
()* )*-+() 1*,("/ -/* (" =* *B-+2-(*& 1*/#"21+E8 #( #1 ,/#(#,-+ (" &*B*+"0 ,"$,*0(1 -$& 3*()"&1 (" *B-+2-(* $"( "$+E
()* 4"/3-+ "/%-$#1-(#"$ -$& ()* "2(02(1 "4 ()* )*-+() 1E1(*38 =2( -+1" ()* -10*,(1 "4 +",-+ 1",#-+ "/%-$#1-(#"$ -$&
0"+#(#,-+ ,2+(2/* '#()#$ ')#,) ()-( +",-+ )*-+() 1E1(*3 #1 *3=*&&*& ()-( 3-E 3*&#-(* ()*#/ /*+-(#"$1)#0H
D)* 0-0*/ *J0+"/*1 ()/** ,-1*1 "4 &#1(/#,( )*-+() 1E1(*31 #$ K"/()*-1( L/-M#+ #$ "/&*/ (" #&*$(#4E -10*,(1 "4 +",-+
1",#-+ "/%-$#1-(#"$ -$& 0"+#(#,-+ ,2+(2/* ()-( -00*-/ (" #$N2*$,* ()* #30+*3*$(-(#"$ "4 ()* /*4"/31 -$& ()*/*=E
0"(*$(#-++E #30-,( 20"$ ()* @2-+#(E "4 ()* ,-/* 0/"B#&*&H D)* /*12+(1 "4 ()* 1(2&E #$&#,-(* ()* #30"/(-$,* ()-(
-10*,(1 "4 +",-+ 1",#-+ "/%-$#1-(#"$ -$& 0"+#(#,-+ ,2+(2/* 3-E *J*/( "$ ()* "0*/-(#"$1 "4 - &*,*$(/-+#1*& )*-+()
1E1(*3H O*E -10*,(1 #&*$(#P*& -/*. ()* 10-,* 4"/ -2("$"3EQ ()* 10-,* 4"/ +",-+ B"#,* #$ 0"+#(#,-+ +#4*Q 0*/1"$-+#1*&
-$& #$1(#(2(#"$-+#1*& #$N2*$,*1 "$ -2("$"3E -$& +",-+ B"#,*Q &#G*/*$,*1 "4 #$B"+B*3*$( "4 )*-+() 1(-G '#() ()*
&#1(/#,(Q &#G*/*$( 10-,*1 "4 -,,*0(-=+* 0/-,(#,* -$& -,,"2$(-=#+#(EH
D)*1* 4-,("/1 -/* 1**$ (" 3"&*/-(* ()* #$(*$( "4 ()* )*-+() /*4"/31 -( -++ 1(-%*1 #$ ()*#/ #30+*3*$(-(#"$H D)/**
0"11#=#+#(#*1 -/* &#1,211*& 4"/ ()* $-(2/* "4 ()* #$(*/-,(#"$ #$ (*/31 "4 ,-21* -$& *G*,( =*('**$ ()* 4"/3-+
"/%-$#1-(#"$ "4 ()* )*-+() 1E1(*3 -$& #(1 +",-+ ,"$(*J(H 5**#$% ()#1 /*+-(#"$1)#0 -1 "$* "4 - &#-+"%2* "G*/1 1"3*
,-2(#"21 "0(#3#13 4"/ ()* 0"(*$(#-+ "4 ()* /*4"/3 -%*$&-H D)* 0-0*/ ,+"1*1 '#() 12%%*1(#"$1 "$ )"' (" (-7* ()#1
+#$* "4 /*1*-/,) 4"/'-/&H ! RSSS T+1*B#*/ 5,#*$,* ;(&H 6++ /#%)(1 /*1*/B*&H
D".L)%41M ?*-+() /*4"/31Q 5",#-+ "/%-$#1-(#"$Q >"+#(#,-+ ,2+(2/*Q C*,*$(/-+#1-(#"$Q L/-M#+
C*,*$(/-+#1*& 3-$-%*3*$( "4 )*-+() ,-/* #1 - 7*E

1(/-(*%E 4"/ /*1(/2,(2/#$% )*-+() 1E1(*31 ()* '"/+&
9 V"//*10"$&#$% -2()"/H D*+. eZZaX]XaR`\aR]Z`Q 4-J. eZZa "B*/H 6+()"2%) 1"3* /*1*/B-(#"$1 )-B* =**$ $"(*& U1**
X]XaR`\a`d`dH 4"/ *J-30+* V"++#$1 W !/**$8 XYYZ[8 &*,*$(/-+#1-(#"$ #1
EN'$83 $44%"11M 1-/-)H-(7#$1"$f3-$H-,H27 U5H 6(7#$1"$[H 3"1(+E 0/*1*$(*& -$& -,,*0(*& -1 ()* '-E (" %" =E -++
SR`àY\_]bSSbc a 1** 4/"$( 3-((*/ ! RSSS T+1*B#*/ 5,#*$,* ;(&H 6++ /#%)(1 /*1*/B*&H
>II. 5 S R ` ` a Y \ _ ] U S S [ S S S S \ a d
166
gIM !" #$%&'()' *$ +," - !).&+, !.&*'.* / 0*1&.&'* 23 456667 839:8;8
!"#"$%& "' ()* +"#,(,!-# &+*!(%$./ 0"1*2*%3 -& 4*(3 '*1 547-.,!& "' #"!-# +"1*%/ e*( *2*7 ;"&&*%( ),.&*#'
&($5,*& )-2* (-6*7 $+ ()* !)-##*78* "' -&&*&&,78 (*75& ("1-%5& +$((,78 #"!-# &"!,-# 547-.,!& (" "7*
1)*()*% ()* +%".,&* 5*!*7(%-#,&-(,"7 &**.& (" "9*% &,5*/ ^"% *P-.+#*3 ,7 %*#-(,"7 (" Q"J*%( U$(7-.L&
(%-7&#-(*& ,7(" %*-#,(4 :;"&&*%(3 <==>? @"##,7&3 <==>A/ 1"%6 "7 &"!,-# !-+,(-# ,7 V(-#43 ;"&&*%( :<==TA 1%,(*&3
;"&&*%( &-4& ()-( ()*%* -%* (1" 6*4 B$*&(,"7& ,7 -&&*&C
&,78 ()* ,.+-!( "' 5*!*7(%-#,&-(,"7D EY)* 1*-67*&& "' (),& -++%"-!) ,& ()-( ,( 5"*& 7"(
+%"2,5* *-&4 +"#,!4 %*#*2-7( !"7!#$&,"7& ! ! ! \* -%*
EE:<A 5"*& 5*!*7(%-#,F-(,"7 ,.+%"2* *B$,(43 *G!,*7!43 #*'( ()*7 1,() ()* +"&&,J#* !"7!#$&,"7 ()-( 5*!*7(%-C
B$-#,(4 "' &*%2,!*&3 )*-#() "$(!".*& -75 5*."!%-(,! #,F-(,"7 1,## 1"%6 "7#4 ,7 -%*-& 1,() &(%"78 ),&("%,*&
+%"!*&&*&H -75 ,' ,( 5"*&3 :IA 1),!) '"%.&3 .*!)-7C "' &"!,-# !-+,(-# -75 ()-( ()* %*&( "' ()* !"$7(%4
,&.& -75 +%"!*&&*& "' 5*!*7(%-#,F-(,"7 -%* ."&( &)"$#5 J* !*7(%-#,F*5 f - !"7!#$&,"7 ()-( ,& 7"(
*9*!(,2* ,7 -!),*2,78 ()*&* "$(!".* -75 "$(+$( #,6*#4 (" J* +"#,(,!-##4 2,-J#*L :+/ <><gA
"JK*!(,2*&LL :;"&&*%(3 <==>3 +/ <=MA/
;$( ,' '-!("%& &$!) -& 1)*()*% - 5,&(%,!( )-& - #"78
N#()"$8) ()* 5*O7,(,"7 "' !"7!*+(& -75 .*()"5& (" (%-5,(,"7 "' #"!-# !,2,! "%8-7,&-(,"7& -75 (%$&( -%* ()*
.*-&$%* ()* 5*+*75*7( "$(+$( 2-%,-J#*& 8,2*7 J4 ;"&C .-K"% ,7c$*7!*& "7 1),!) 5,&(%,!(& 1,## '-%* J*&( $75*%
&*%( .-4 J* 5*J-(*53 ()*%* ,& 7*2*%()*#*&& - 8%"1,78 5*!*7(%-#,&*5 8"2*%7.*7(3 &$%*#4 1* )-2* (" "+*7 -
J"54 "' 1"%6 "7 5*&,87,78 "% -5-+(,78 *P,&(,78 ,7&(%$C 5,-#"8$* %*8-%5,78 ()* +"#,!4 ,.+#,!-(,"7& "' (),& -75
.*7(& '"% -++#,!-(,"7 (" )*-#() &*%2,!*& %*&*-%!) ,7 +"&&,J#* #,7*& "' -!(,"7/ V( ,& "$% !"7(*7(,"7 ()-(
5*2*#"+,78 !"$7(%,*& :Q"*.*% R S"7("4-CN8$,#-%3 -#()"$8) 1* )-2* 7" -7&1*%& -& (" 1)-( (" 5" -J"$(
<=TT? UQV@WQ3 <=TT? X-%7*%3 Y)".-&"7 R Z"7-#5C ()* ,7c$*7!* "' #"!-# &"!,-# "%8-7,&-(,"7 -75 +"#,(,!-#
&"73 <==M? [78*#6*&3 <==M? ;%4!*3 Y""#*3 \-#5.-7 R !$#($%*3 ,( ,& ,7-5*B$-(* (" #*-2* ()*. $75*%C5,&!$&&*5
]",8(3 <==I? ^"%&J*%83 ;-%%"& R ],!("%-3 <==I? U-,7* -75 $75*%C%*&*-%!)*5 J4 ()"&* ,7(*%*&(*5 ,7 ()* +%-!C
R \%,8)(3 <=TT? ;%$!*3 <==M? ;%$!* R _-,73 <==<? _-,73 (,!* "' )*-#() !-%*/
<==I? `-7K,3 `,#,.- R S$7,&),3 <==I? N(6,7&"73 <==a? Y),& +-+*% ()*7 -,.& (" !".+#*.*7( ;"&&*%(L&
\*-6#,-.3 <==b? 0-%-73 Z"2#" R W9*,3 <==b? X-((,C '%-.*1"%6 J4 *P+#"%,78 ,7 ."%* 5*+() ()* &"!,-# -75
7-%-3 VJ-!-!)*3 U$*7(*3 X,-!"7, R @-+%-%-3 <==>A/ +"#,(,!-# *72,%"7.*7( ,7 1),!) )*-#() %*'"%.& -%* (-6C
\)-( )-& J**7 -55%*&&*5 2*%4 #,((#* ,& ()* 5*O7,(,"7 "' ,78 +#-!* -( ()* #"!-#3 1,(),7 !"$7(%4 &!-#* ,7 "%5*% ("
!"7!*+(& -75 .*()"5& '"% ()* ,75*+*75*7( 2-%,-J#*& ,7 ,5*7(,'4 -&+*!(& 1),!) ,7c$*7!* ()* ,.+#*.*7(-(,"7 "'
&$!) -7 *2-#$-(,"73 ()-( ,& '"% ()* -&+*!(& "' "%8-7,&C ()* %*'"%.& -75 ()*%*J4 +"(*7(,-##4 ,.+-!( $+"7 ()*
-(,"7-# -%%-78*.*7(& ()-( .-4 2-%4 J*(1**7 -75 B$-#,(4 "' ()* !-%* +%"2,5*5/
1,(),7 )*-#() &4&(*.&/ \)-( )-& J**7 -55%*&&*5 *2*7
#*&& ,& ()* +"(*7(,-# ,7c$*7!* ()-( #"!-# &"!,-# "%8-7,&C
-(,"7 -75 +"#,(,!-# !$#($%* "' ()* *72,%"7.*7( ,7 1),!) !"#"$%&'()*'%)+$, -"'(%- *.*%"/ 0"&1+&/'$#" '$2 %-"
- )*-#() &4&(*. ,& *.J*55*5 .-4 )-2* "7 )"1 5*!*7C (+#'( #+$%"3%
(%-#,&-(,"7 (%-7&'"%.& ,7(" +%-!(,!* ,7 5,9*%*7( #"!-#
!"7(*P(& :N(6,7&"73 <==>A/ Y)* -2"1*5 J*7*O(& "' 5*!*7(%-#,&*5 .-7-8*.*7(
d,(*%-($%* "7 )*-#() &*!("% %*'"%. .-7,'*&(& - !*%C -%* +%*&*7(*5 1,() J"() '%*B$*7!4 -75 !"7&,&(*7!4 -75
(-,7 $7*-&* 1,() %*8-%5 (" ()* &".*1)-( 7*J$#"$& !-7 J* 5*O7*5 -& J"() .-7-8*%,-# -75 +"#,(,!-# :@"7C
-&+*!(& "' #"!-# &"!,-# "%8-7,&-(,"7 -75 +"#,(,!-# !$#C
($%*/ V7!%*-&,78#43 ()* ,7c$*7!* "' ()*&* '-!("%& ,&
-!67"1#*58*5 J$( 1* 5" 7"( %*-##4 67"1 1)-( (" 5"
1,() (),& "J&*%2-(,"7 ,7 +%-!(,!-# (*%.&/ Y)* %*&$#( ,& -
&!),&. ,7 ()* #,(*%-($%*/ W7 ()* "7* )-753 ()*%* ,& ()*
&"!,-# &!,*7(,&( ,##$.,7-(,78 #"!-# +"#,(,!& ,7 -## ,(& '-&!,C
7-(,78 5*(-,# J$( 1,() #,((#* ,75,!-(,"7 '"% +%-!(,!*/ W7
()* "()*% )-753 ()* )*-#() &4&(*.& %*&*-%!)*% '"!$&*&
"7 ()* .$(-J#* -75 .-7-8*-J#* -&+*!(& "' )*-#() !-%*
-75 1),#&( -!67"1#*58,78 +"#,(,!-# -&+*!(&3 &".*(,.*&
1,() *2,5*7( '%$&(%-(,"73 $#(,.-(*#4 #*-2*& ()*&* (" "7*
&,5* -& &".*(),78 ()-( ,& 7"( -.*7-J#* (" !)-78* '%".
()* )*-#() &*!("%/ ;"&&*%( :<==TA )-& #-,5 "$( - &!)*.*
'"% !".+-%,78 ()* *9*!(& "' 5*!*7(%-#,&*5 .-7-8*.*7(
"' )*-#() !-%* "7 )*-#() &4&(*. 8"-#& "' *B$,(43 *GC
!,*7!43 B$-#,(4 -75 O7-7!,-# &"$757*&& J*(1**7 5,9*%C ^,8/ </ Y)* -2"1*5 J*7*O(& "' 5*!*7(%-#,&*5 .-7-8*.*7( "'
*7( !"$7(%,*&3 -75 1),!) 5"*& ,7!#$5* %*'*%*7!* (" ()* )*-#() !-%*/
167
!" #$%&'()' *$ +," - !).&+, !.&*'.* / 0*1&.&'* 23 456667 839:8;8 )T&
!"#$% &'()* +,--$% ./012/3% 45,62 7 8/9,9:/;"2% E@5I-"5"36 3/6,@3/- E@5I/#,$@3$B 82"#" ,$ / 62,#;
&''<* =-!33% &''>* ?@$$"#6% &''(AB 82" C"! D"/60#"$ @D 1#@0I @D $60;,"$ H2"#" 62" #"/-,6,"$ @D ,5I-"5"36,31 /
62" /#105"36$ D@# ;"E"36#/-,$/6,@3 $011"$6 62/6 ;"E"36#/-,$/6,@3 I@-,E! 2/F" 9""3 ;@E05"36"; /3; /3/G
,3E#"/$"; -@E/- /06@3@5! @F"# ;"E,$,@3G5/C,31 E@5G -!:"; 62#@012 ;"6/,-"; E/$" $60;,"$B U3 62"$" E/$"$% 62"
9,3"; H,62 ,3I06$ @D F@,E" D#@5 62" I@I0-/6,@3 6@ 9" #"$"/#E2"#$ @D6"3 E@5" D#@5 62" $@E,/- /3; I@-,6,E/-
$"#F"; H,-- ,3E#"/$" 62" #"$I@3$,F"3"$$ @D 2"/-62 E/#" 6@ $E,"3E"$ H,62 / $60;! D@E0$ @3 5,E#@GI@-,6,E$% /$I"E6$
-@E/- 3"";$% /EE@036/9,-,6! @D 62" /E6,@3$ @D 62" 2"/-62 @D 9/#1/,3,31 /3; 3"1@6,/6,@3 62/6 @EE0# ,3 @#1/3,$G
$!$6"5 6@ ,6$ E-,"36 I@I0-/6,@3 ,3 6"#5$ 9@62 @D 62" /6,@3$ /$ I/#6 @D 62" ,5I-"5"36/6,@3 I#@E"$$ MD@#
J0/-,6! @D E/#" @K"#"; /3; 62" 0$" @D 2"/-62 $!$6"5 "N/5I-"% =-!33% &''>* Q6C,3$@3% &''ZAB 820$ 62" -,3C
#"$@0#E"$ /3; /-$@ 6@ $@E,/- ;"F"-@I5"36 1@/-$ @D I@IG 6@ 62" J0/-,6! @D 62" 2"/-62 E/#" I#@F,;"; 2/$ 3@6 9""3
0-/# "5I@H"#5"36B 82,$ /#105"36 ,$ $055/#,$"; ,3 / E@3E"#3B
=,1B &B L" ,;"36,]"; @3-! 6H@ $60;,"$ H2"#" @#1/3,$/6,@3/-
L2"#" ;"E"36#/-,$"; 5/3/1"5"36 @D 2"/-62 E/#" 2/$ E@5I@3"36$ @D ;"E"36#/-,$"; 03,6$ 2/F" 9""3 -,3C"; 6@
9""3 "F/-0/6";% $60;,"$ 5@$6 @D6"3 $,5I-! E@5I/#" J0/-,6! @D E/#" ,3;,E/6@#$B U3 U$#/"-% 62" #"$"/#E2"#$
/$I"E6$ @D 2"/-62 E/#" I#@F,$,@3 H,62,3 / E@036#! ;"F"-@I"; / 5@;"- 6@ -,3C 62" D@#5/- $6#0E60#"$ @D
9"6H""3 ;,$6#,E6$ ;"E"36#/-,$"; /3; 62@$" 3@6 MD@# ;"E"36#/-,$"; 5/3/1"5"36 @D ,3;,F,;0/- I#,5/#! E/#"
"N/5I-"$ 82@5/$@3% O"H9#/3;"# 7 P@-"25/,3"3GQ,6G E-,3,E$ 6@ 5"/$0#"$ @D J0/-,6! @D E/#" ,3 62@$" E-,3,E$
C"3% &''&* 4"E#"6/R#,/ ;" 4/0R;"% S"/#/R% &''TAB 82"#" Mc#@$$ "6 /-B% &''TAB X62"# $60;,"$ @D J0/-,6! @D E/#"
/#" 6H@ 5/,3 I#@9-"5$ H,62 62,$ /II#@/E2B =,#$6% 62@$" 5/! ,5I-,E,6-! 2/F" #"F"/-"; #"-/6,@3$2,I$ 9"6H""3
;,$6#,E6$ E2@$"3 D@# "/#-! ;"E"36#/-,$/6,@3 H,62,3 / ;"E"36#/-,$"; 5/3/1"5"36 /3; J0/-,6! H,62@06 2/F,31
#"1,@3 5/! 2/F" ;,$6,3E6,F" J0/-,6,"$ D#@5 62@$" 3@6 3"E"$$/#,-! -/9"--"; ,6 /$ $0E2B 460;,"$ @D ,3;,F,;0/-
;"E"36#/-,$";B U3 I/#6,E0-/#% 62"! /#" -,C"-! 6@ 9" 62@$" 2"/-62 D/E,-,6,"$% H2"62"# I#,5/#! E/#" E-,3,E$ @# 2@$I,G
E@3$,;"#"; 6@ 9" I#@F,;,31 2"/-62 E/#" H"-- /3; 6@ 6/-$% /#" -,C"-! 6@ I#@F" "/$,"# 62/3 $60;,"$ @D -@E/- ;,$G
2/F" / $6#@31 -@E/- 5/3/1"5"36 6"/5B 4"E@3;-!% /3; 6#,E6 $!$6"5$% /3; 5/! D0#3,$2 $"66,31$ ,3 H2,E2 6@
5@#" ,5I@#6/36-! ,3 62" -@31G6"#5% 62" 6"#5 ;"E"36#/-,G ;"F"-@I E@3E"I6$ #"1/#;,31 ,3D@#5/- $6#0E60#"$B W@HG
$/6,@3 /E6$ /$ $2@#62/3; D@# -,66-" 5@#" 62/3 62" D@#5/- "F"#% 62" 5/,3 D@E0$ @D ;"E"36#/-,$/6,@3 ,3 5@$6
-"1/- $6/60$B ?"!@3; 62,$ 6"#5% / H,;" #/31" @D @#1/3,$G E@036#,"$ #"5/,3$ 62" ;,$6#,E6 /3; 620$ ,6 ,$ 62" $6#0EG
/6,@3/- /##/31"5"36$ 5/! 9" I06 ,36@ "K"E6 -@E/--! /$ 60#"$ /6 62" ;,$6#,E6 -"F"- 62/6 50$6 9" ;"$E#,9"; ,3
I/#6 @D ;"E"36#/-,$/6,@3B Q $,5I-" ;,E2@6@5! @D ;"E"3G @#;"# 6@ /$$"$$ 62",# #"-/6,@3 6@ 5"/$0#"$ @D 2"/-62
6#/-,$";V3@6 ;"E"36#/-,$"; D/,-$ 6@ "NI-@#" 62" F/#,/6,@3 E/#" J0/-,6!B U3 d62,@I,/% ;"E"36#/-,$"; 2"/-62 E/#" D/--$
,3 @#1/3,$/6,@3/- /##/31"5"36$ 03;"# H2,E2 2"/-62 $"#G 03;"# 62" 5/3/1"5"36 @D 62" "-"E6"; -@E/- 1@F"#35"36B
F,E"$ 5/! 9" ,5I#@F,31 /3; 620$ 2/$ -,5,6"; I#/E6,E/- ?/#3/9/$ M&''ZA "NI-@#"; 62" #"-/6,@3$2,I 9"6H""3 D@#G
F/-0" ,3 ,;"36,D!,31 H2/6 ;@"$ /3; ;@"$ 3@6 H@#CB 5/- @#1/3,$/6,@3/- $6#0E60#"$% $0E2 /$ E@5I@$,6,@3 @D
82"#" /#" $60;,"$ H2,E2 2/F" /,5"; 6@ ;@E05"36 62" -@E/- 1@F"#35"36 /$$"59-!% ,3 6"#5$ @D 1"3;"#% @EG
$@5" @D 62" F/#,/6,@3 /$$@E,/6"; H,62 ;"E"36#/-,$/6,@3% E0I/6,@3% ";0E/6,@3/- -"F"- /3; $@ D@#62% /3; I"#D@#5G
0$0/--! 9! E@5I/#,31 62" D@#5/- -"1/- /3; @#1/3,$G /3E" ,3 6"#5$ @D 9/$,E E@F"#/1" ,3;,E/6@#$ M/36"G3/6/-
/6,@3/- $6#0E60#"$ /3; I#@E";0#"$ I06 ,36@ I-/E" /E#@$$ E/#" /3; F/EE,3/6,@3$AB
;,K"#"36 E@036#,"$ MLWX% &''Y* S@--,3$ 7 W036"#% O@ $60;! 2/$ 9""3 D@03; 62/6 6#,"$ 6@ ,;"36,D!
&''ZAB 82"$" $60;,"$ 5@F" 03;"#$6/3;,31 / -@31 H/! /$I"E6$ @D $@E,/- @#1/3,$/6,@3 /3; I@-,6,E/- E0-60#"$
D@#H/#; 9! #"E@13,$,31 62/6 62" D@#5/- @#1/3,$/6,@3/- /3; 62",# I@6"36,/- ,3_0"3E" @3 2"/-62 E/#" J0/-,6!B
/##/31"5"36$ /$$@E,/6"; H,62 62" 9-/3C"6 6"#5 @D
;"E"36#/-,$/6,@3 5/! $2@H 1#"/6 F/#,/6,@3B W@H"F"#%
H@#C,31 /6 62" $E/-" @D E#@$$GE@036#! E@5I/#,$@3$ !"#$% $&'()*
5"/3$ 62/6 62" 5/,3 D@E0$ H,-- 9" @3 62" @[E,/- D@#G
5/- $6#0E60#"$ /3; I#@E";0#"$ 6@1"62"# H,62 /11#"G X#1/3,$/6,@3/- 62"@#! 2/$ -@31 $6#"$$"; 62" ,5I@#6G
1/6"; ,3;,E/6@#$ @D @06I06$B 82" 1/I 62/6 ,$ H"-- /3E" @D 62" ,3_0"3E" @D 62" H,;"# "3F,#@35"36 ,3
C3@H3 6@ "N,$6 9"6H""3 H2/6 ,$ $/,; 6@ 9" 9",31 ;@3" H2,E2 /3 @#1/3,$/6,@3 ,$ -@E/6"; Ma/H#"3E" 7 a@#$E2%
9! @[E,/- I@-,E! /3; 2@H @#1/3,$/6,@3$ /E60/--! D03EG &')'* W@D$6";"% &''&AB L" 2/F" D@03; ,6 0$"D0- 6@
6,@3 ,3 I#/E6,E" E/3 9" I,EC"; 0I% 906 @3-! 9#@/; ;,F,;" 62,$ @#1/3,$/6,@3/- "3F,#@35"36 ,36@ 6H@ E/6G
9#0$2 ,3;,E/6,@3$ @D 5/\@# I#@9-"5$ E/3 0$0/--! 9" "1@#,"$ H,62 #"1/#; 6@ "NI-@#,31 ,3_0"3E"$ @3 62" ,5G
,;"36,]";B ^"$"/#E2 D#@5 @#1/3,$/6,@3 $60;,"$ /3; I@-G I-"5"36/6,@3 @D -@E/- 2"/-62 $!$6"5$B =,#$6% 62"#" /#"
,E! /3/-!$,$ 2/$ ,3;,E/6"; 6,5" /3; 6,5" /1/,3 62" 62" 1"@1#/I2,E /3; ;"5@1#/I2,E E2/#/E6"#,$6,E$ @D ;,$G
I@H"# @D -@E/- $E/-" $@E,/- /3; I@-,6,E/- I#@E"$$"$ 6@ ,3G 6#,E6$ ,3E-0;,31 62" D@--@H,31e 0#9/3V#0#/-% ;,$6#,E6 $,:"%
_0"3E" ,5I-"5"36/6,@3 M`#"$$5/3 7 L,-;/F$C!% &'Z>* "E@-@1,E/- 6!I"% I@I0-/6,@3 $,:" /3; E@5I@$,6,@3% 5/,3
a,I$C!% &'(<* ?@-5/3 7 b"/-% &''&* `D"K"#% &''TA% ,3G "E@3@5,E /E6,F,6,"$B dNE"I6 ,3 6,5"$ @D $0;;"3 I@I0G
;,E/6,31 62" 3""; D@# ;"6/,-"; 5,E#@G$E/-" $60;,"$ 6@ -/6,@3 5@F"5"36$% 62"$" /#" -/#1"-! $6/9-" @F"# 62"
168
TBB !" #$%&'()' *$ +," - !).&+, !.&*'.* / 0*1&.&'* 23 456667 839:8;8
!"#$%!&'"(! )*# $' $+ (",)'$-",. ")+. '/ 0/,,"0' $*1/(& )0'$-$'$"+ )*# 6"(+/**", )*# /* +/0$), $*'"((",)'$/*&
!)'$/* /* '2"!3 42" +"0/*# 0)'"5/(. 0/!6($+"+ '2" +2$6+3
#.*)!$0 )+6"0'+ /1 ,/0), +/0$), /(5)*$+)'$/* )*# 6/,$'$& A/( '2" 6%(6/+"+ /1 '2$+ 6)6"(9 '2" #)') )*),.+$+ 2)+
0), 0%,'%(" 7$'2$* 72$02 ) *"7 6/,$0. 7$,, 8" $!& 1/0%++"# /* /*" +6"0$D0 )+6"0'9 '2)' $+ $#"*'$1.$*5
6,"!"*'"#9 $*0,%#$*5 '2" 1/,,/7$*5: +/0$), *"'7/(;+9 2"),'2 +"(-$0"&(",)'"# )0'$-$'$"+ /1 #$G"("*' )0'/(+ /(
+/0$), !/("+ )*# -),%"+9 *)'%(" /1 ,")#"(+2$69 *)'%(" 5(/%6+ /1 )0'/(+ 72$02 !$52' $!6)0' %6/* '2" H%),$'.
/1 <$*=%"*0">9 (",)'$/*+2$6 /1 ,/0), 6%8,$0 7/(;"(+ '/ /1 '2" 2"),'2 +"(-$0"+ )*# 72$02 +""! '/ -)(. 8"'7""*
'2" ,/0), #$+'($0'3 ?%02 1)0'/(+ !). ),+/ 6(/-" +')8," #$G"("*' #$+'($0'+3 42" #)') 2)-" 8""* 0/,,)'"# 1/( '2$+
)*# ("+$+')*' '/ 02)*5" $* '2" !"#$%!&'"(! 8%' $' 0)* 6%(6/+" $* '2("" 7).+3 A$(+'9 '2" D(+' )%'2/( 2",# 1/(&
8" #$@0%,' '/ 5"' )00%()'" $*1/(!)'$/* $* '2" +2/(' !), $*'"(-$"7+ 7$'2 '2" '2("" D",# ("+")(02"(+ /* '2"
'$!"&1()!" /1 !/+' +%(-".+3 A$53 B #"6$0'+ '2" 1/(!), "I6"($"*0"+ '2". 2)-" 8""* 7$'*"++$*53 ?"0/*#,.9 7"
+'(%0'%(" /1 '2" ,/0), 2"),'2 +.+'"! "!8"##"# 7$'2$* 2)-" (")# 8)0; '2(/%52 '2" D",# ("+")(02"(+> /7* #$)&
'2"+" '7/ #/!)$*+ /1 '2" 5"/&#"!/5()62$0 )*# '2" ($"+ '/ $#"*'$1. 6(/0"++"+ )*# 6(/0"#%("+ '2)' )(" ');&
+/0$/&6/,$'$0),3 $*5 6,)0" )(/%*# '2" 2"),'2 +"0'/( /1 ")02 #$+'($0'3
42" ("+%,'+ 1/( '2$+ 6)6"( )(" /*" 6)(' /1 ) !%02 42$(#,.9 7" 2)-" )*),.J"# '2" '()*+0($6'+ /1 ')6"#
,)(5"( +'%#. /1 '2" $*'"()0'$/*+ /1 5/-"(*!"*' 6/,$0. $*'"(-$"7+ '/ $#"*'$1. '2" 0/*0"(*+ /1 '2/+" $*'"(-$"7"#
("1/(! '2(/%52 '2" 2"),'2 +.+'"! 7$'2 ,/0), +/0$), /(& 1/,,/7$*5 )* )66(/)02 6%' 1/(7)(# 8. ?6()#,". KLMNMO
5)*$+)'$/* )*# 6/,$'$0), 0%,'%("3 42("" ,/0), +/0$), )*# P';$*+/* )*# P8% Q, R)S KLMMTO3
+0$"*'$+'+ +6"*' 1/%('""* !/*'2+ ")02 ,$-$*5 $* /*" /1 42" *"I' +"0'$/* +"'+ '2" 0/*'"I' /1 2"),'2 ("1/(! $*
'2" #$+'($0'+ $* /(#"( '/ )00/!6)*. '2" #). '/ #). 2)6& U()J$, 1/,,/7"# 8. ) #"+0($6'$/* /1 '2" '2("" +'%#. +$'"+
6"*$*5+ $* '2" #$+'($0' )*# $* '2" #$+'($0' 2"),'2 +.+'"!3 $* '"(!+ /1 '2" 5"/&#"!/5()62$0 "*-$(/*!"*' )*#
C)') 2)-" 8""* 0/,,"0'"# '2(/%52 /8+"(-)'$/*+9 $*1/(& +/!" 8)+$0 $*#$0)'/(+ /1 2"),'2 0)(" H%),$'.3 42" !)$*
!), 0/*-"(+)'$/*+ )*# !/(" 1/(!), /6"* $*'"(-$"7+9 #$!"*+$/*+ /1 +/0$), /(5)*$+)'$/* )*# 6/,$'$0), 0%,'%("
)*# 2)-" 8""* ("0/(#"# $* '2" 1/(! /1 '()*+0($6'$/*+ 7" 0/*+$#"( !/+' $!6/(')*' )(" #"D*"# )00/(#$*5 '/
/1 ')6"# $*'"(-$"7+9 #)$,. D",# */'"+ )*# #$)($"+ )*# '2("" 0($'"($): '2"$( ,$;",. -)($)'$/* 8"'7""* #$G"("*'
'2" /@0$), #/0%!"*')'$/* /1 '2" ,/0), 5/-"(*!"*'3 #$+'($0'+V 6/'"*'$), 1/( $!6)0' /* 2"),'2 0)(" H%),$'.V
E*'"(-$"7+ 2)-" 8""* 2",# 7$'2 2"),'2 6(/1"++$/*),+9 '2" ")+" /1 $#"*'$D0)'$/* )*# 0/,,"0'$/* /1 $*1/(!)'$/*
,/0), 0/%*0$,,/(+9 ,")#"(+ /1 ,/0), /(5)*$+)'$/*+9 0/!& ",+"72"(" 7$'2/%' "I'"*#"# 6"($/#+ /1 D",#7/(;3
!%*$'. 2"),'2 7/(;"(+ )*# 7/!"* ,$-$*5 $* !/("
#"6($-"# )(")+ /1 '2" '2("" #$+'($0'+3 F8+"(-)'$/*+ 2)-"
8""* !)#" )' 2"),'2 0"*'("+9 )' !""'$*5+ 7$'2$* '2"
!"#$%$&'# &(#%()* '+, -*'#%- )*.")/ $+ 0)'1$#
2"),'2 +"0'/( +%02 )+ +')G !""'$*5+9 ,/0), 2"),'2 0/%*&
0$, !""'$*5+9 )' '2" 6%8,$0 !""'$*5+ /1 '2" ,/0), 5/-"(*&
P *"7 0/*+'$'%'$/* 7)+ 6)++"# $* U()J$, $* LMWW )+
!"*' 0/%*0$, 02)!8"( )*# )' 6/,$'$0), !""'$*5+ #%($*5
6)(' /1 '2" ("'%(* '/ 0$-$,$)* (%," )1'"( +/!" '7"*'.
'2" ,/0), 5/-"(*!"*' ","0'$/* 0)!6)$5*3 ?$*0" '2" D",#
.")(+ /1 !$,$')(. #$0')'/(+2$63 42" "I$+'$*5 2"),'2 +.+&
("+")(02"(+ ,$-"# $* '2" #$+'($0'9 '2". ),+/ 2)-" 2)#
'"! 7)+ 2$52,. +'()'$D"# 8. +/0$/&"0/*/!$0 5(/%6+
!)*. $*1/(!), 0/*-"(+)'$/*+ 7$'2 ,/0), 6"/6," 6(/-$#&
$*-/,-$*5: 6($-)'" 0)(" 1/( '2/+" )8," '/ )G/(# $'V )
$*5 $*1/(!)'$/* /* ,/0), /6$*$/*+ )8/%' 2"),'2 +"0'/(
6%8,$0 *"'7/(;9 ,)(5",. /1 0%()'$-" 0)("9 6(/-$#"# 8.
'2" X$*$+'(. /1 ?/0$), ?"0%($'. 1/( '2/+" 1/(!),,.
"!6,/."# )*# '2%+ 6).$*5 $*'/ ) +/0$), $*+%()*0"
+02"!"V ) !$I'%(" /1 8)+$0 2"),'2 0,$*$0+ )*# 2"),'2
6(/5()!!"+ 6(/-$#"# 8. '2" X$*$+'(. /1 R"),'29 +')'"
)*# #$+'($0' 2"),'2 +"0'/(+ )*# 02)($')8," $*+'$'%'$/*+
1/( '2/+" /'2"(7$+" */' 0/-"("#3 C%($*5 '2" #$0')'/(&
+2$6 .")(+9 0/!!%*$'.&8)+"# /(5)*$+)'$/*+9 2"),'2 6(/&
1"++$/*),+9 +'%#"*'+ )*# +/!" +"5!"*'+ 7$'2$* '2"
Y)'2/,$0 02%(02 1/(!"# -)($/%+ ,/0), ),,$)*0"+ '/
,/88. 1/( $!6(/-"# +"(-$0"+ 1/( '2" 6//( KX)02)#/9
LMMZO3 42$+ 8%$,' $*'/ 7$#"( +/0$), !/-"!"*'+ )!/*5+'
72$02 '2" 2"),'2 ("1/(! !/-"!"*' 7)+ 6)('$0%,)(,.
+'(/*5 )*# )02$"-"# )* "I6,$0$' 0/!!$'!"*' $* '2"
*"7 0/*+'$'%'$/* '/7)(#+ 5(")'"( "H%$'. $* 2"),'2 0)("
A$53 B3 [/0)'$*5 /1 '2" 1/(!), +'(%0'%(" /1 ) #$+'($0' 2"),'2 6(/-$+$/*3 42" /-"(),, 6/,$0. -$+$/* "I6("++"# $* '2"
+.+'"! 7$'2$* $'+ 5"/&#"!/5()62$0 )*# +/0$/&6/,$'$0), "*-$(/*& Y/*+'$'%'$/* $+ '2)' 2"),'2 $+ '2" ($52' /1 "-"(. 0$'$J"*
!"*'), #/!)$*+3 )*# '2" #%'. /1 '2" ?')'"3 42" ;". )('$0,"+ 7$'2$* '2"
169
/0 1&2!"+." %& (*0 3 /.#!(* /#!%"#% 4 5%6!#!"% 78 9:;;;< =8>?=@= DFZ
!"#$%&%'%&"# (") *+,-%* )+("). ,)+ %),#$-,%+/ &# *+,-%* 8,)+ 8".1&#+$ , .&9%')+ "( $+8%"),- ,'%"#".7
0,1-+ 23 ()". ,#/ $'1")/&#,%&"# %" %*+ -"8,- /&$%)&8% 5"=+)#?
4"') .,&# $%),%+5&+$ 6+)+ -,&/ /"6# 17 6*&8* .+#%3 0*+ /&$%)&8% *+,-%* $+8%") 8"#%),8%$ &%$ (").,-
)+("). "( %*+ +9&$%&#5 *+,-%* $+8%") 6"'-/ 8".+ ,1"'%: /+8+#%),-&$+/ $%,%'$ 6&%* %*+ (+/+),- '#&"# ,#/ &$ ,--"?
&#%+5),%&#5 %*+ /&;+)+#% <'1-&8 <)"=&/+)$ &#%" "#+ 8,%+/ S#,#8&,- )+$"')8+$ &# ,88")/,#8+ 6&%* %*,% 8"#?
$&#5-+ $7$%+.> 5)+,%+) +.<*,$&$ "# <)+=+#%,%&=+ ),%*+) %),8%3 0*+ %*)++ /&;+)+#% (").$ "( 8"#%),8% @!"#!$!%"&%'
%*,# 8'),%&=+ 8,)+> /+8+#%),-&$+/ .,#,5+.+#% "( *+,-%* $()#!(*' +%,!-$*%"( I "<+),%&#5 ,% %*+ %&.+ "( %*+ S+-/
8,)+ <)"=&$&"# %" %*+ /&$%)&8% -+=+- ,#/ <"<'-,) <,)%&8&? $%'/7 ,)+ $*"6# &# 0,1-+ F @%*+$+ 6+)+ )+=&$+/ &#
<,%&"# &# %*+ .,#,5+.+#% "( %*+ *+,-%* 8,)+ $7$%+. ,% 2CCTI3 4'#/$ 5" %*)"'5* , /&$%)&8% *+,-%* ('#/ 6*&8*
,-- -+=+-$ @A),B&-&,# !"#$%&%'%&"#> ,)%&8-+$ 2CDEFGG> (,--$ '#/+) %*+ ,/.&#&$%),%&=+ <)"8+/')+$ "( %*+ /&$%)&8%
2CHHI3 0*+ -+5,- (),.+6")J (") *+,-%* $+8%") )+("). -"8,- 5"=+)#.+#%3 0*'$ '-%&.,%+-7> %*+ =&$&"# "( /+8+#?
#,%&"#,--7 6,$ /+=+-"<+/ &# %*+ ("--"6&#5 7+,)$ %),-&$,%&"# &$ %" , 5+"5),<*&8,--7 /+S#+/ -"8,- 5"=+)#?
%*)"'5* %6" *+,-%* -,6$ @K+& L)5,M#&8, /+ N,'O/+> #"$3 .+#%> 8"")/&#,%&#5 ,#/ &#%+5),%&#5 %*+ =,)&"'$ $+8%")$>
H3GHGPCG ,#/ H32QFPCGR /+ !,)=,-*" ,#/ N,#%"$> 2CCFI3 ),%*+) %*,# %" , $+8%"),--7 /+S#+/ /&$%)&8% *+,-%* $7$?
0*+ ."/+- "( /+8+#%),-&$,%&"# (") %*+ .,#,5+.+#% "( %+. )+-,%&=+-7 &#/+<+#/+#% "( %*+ "%*+) -"8,- ,8%&=&%&+$3
A),B&-&,# $"8&+%7 *,$ 1++# 1'&-% "#> ,#/ *,$ 1'&-% '<>
0,1-+ 2 , %),/&%&"# "( $"8&,- ,#/ <"-&%&8,- )+-,%&"#$ 8*,),8%+)?
c+7 ,)%&8-+$ &# %*+ A),B&-&,# !"#$%&%'%&"#> 2CHH> (") d+,-%* &$+/ 17 &#/&;+)+#8+> <+)$"#,- -&#J$ ,#/ <,%+)#,-&$.3
e+("). 0*&$ *,$ 1++# <,)%&8'-,)-7 $%)"#5 &# %*+ U")%*+,$% "(
A),B&- $&#8+ %*+ 1+5&##&#5 "( %*&$ 8+#%')7 6*+)+ "-&?
!"#$%&' ()* 5,)8*&8 <"6+)> 1,$+/ "# "6#+)$*&< "( -,)5+ +$%,%+$>
d+,-%* &$ %*+ )&5*% "( +=+)7"#+ ,#/ %*+ /'%7 "( %*+ N%,%+> *,$ 1'&-% '< %*+ $7$%+. J#"6# ,$ #.)."%*!+,. @K+,->
5',),#%++/ =&, $"8&,- ,#/ +8"#".&8 <"-&8&+$ 6*&8* ,&. (") %*+
2CTVR 4,")"> 2CC2I3 W+$<&%+ .'8* %,-J ,1"'% )+$%)'8?
)+/'8%&"# "( %*+ )&$J "( /&$+,$+ ,#/ "%*+) &#f')&+$ ,#/
'#&=+)$,- ,#/ +\',- ,88+$$ %" ,8%&=&%&+$ ,#/ $+)=&8+$ (") &%$
%')&#5 %*+ <"-&%&8,- /".,&# ,#/ "<+#&#5 '< %*+ $<,8+
g*+,-%*^$h <)"."%&"#> <)"%+8%&"# ,#/ )+8"=+)73 (") <"<'-,) <,)%&8&<,%&"#> %*+)+ &$ , =&+6 %*,% %*+ "-&?
5,)8*&8 )+-,%&"#$*&<$ &# %*+ U")%*+,$% )+.,&# '#,(?
!"#$%&' ()+ (+8%+/3 0*&$ -,8J "( &.<,8% &# %*+ U")%*+,$% "( A),B&-
0*+ ,8%&=&%&+$ ,#/ <'1-&8 $+)=&8+$ "( *+,-%* ,)+ &#%+5),%+/ &#%" *,$ 1++# ,%%)&1'%+/ %" <"-&%&8,- &#+)%&, 17 %*+ $%,%+
, )+5&"#,-&$+/ ,#/ *&+),)8*&$+/ #+%6")J ,#/ 8"#$%&%'%+ , -+=+- 5"=+)#.+#%$3 0*&$> *"6+=+)> *,$ #"% 1++# %*+
$&#5-+ $7$%+.> ")5,#&$+/ ,88")/&#5 %" %*+ ("--"6&#5 <)&#8&<-+$ 8,$+ &# %*+ $%,%+ "( !+,),O &# 6*&8* %*+)+ *,$ 1++# 8"#?
2 E /+8+#%),-&$,%&"#> 6&%* , $&#5-+ .,#,5+.+#% &# +,8* $<*+)+ %&#'+/ <"-&%&8,- 8"..&%.+#% %" 8*,#5+ 17 %*+ $%,%+
"( 5"=+)#.+#% 5"=+)#.+#% $&#8+ 2CHD> %*)"'5* %*)++ 8"#$+8'%&=+ 5"=?
F E &#%+5),%+/ ,%%+#/,#8+> 6&%* <)&")&%7 %" <)+=+#%&=+
+)#.+#%$ 8"=+)&#5 %6+-=+ 7+,)$ ,#/ 8"#%&#'&#5 $%&-- (")
,8%&=&%&+$> 6&%*"'% <)+f'/&8&#5 .+/&8,- $+)=&8+$
Z E 8"..'#&%7 <,)%&8&<,%&"#
, (')%*+) ("')3 X# <,)%&8'-,)> %*+7 *,=+ &#$%&%'%&"#,-&$+/
0*+ $&#5-+ *+,-%* $7$%+. 6&-- 1+ S#,#8+/ &# %+).$ "( ,)%&8-+ 5)+,%+) ,'%"#".7 (") -"8,- /&$%)&8% 5"=+)#.+#%$
2CV> 6&%* )+$"')8+$ ()". %*+ $"8&,- $+8')&%7 1'/5+% "( %*+ %*)"'5* /+8+#%),-&$,%&"# @$++ 0+#/-+)> 2CCT (") , ('--
i#&"#> %*+ $%,%+$> %*+ (+/+),- /&$%)&8% ,#/ %*+ .'#&8jO <&"$> "=+) /&$8'$$&"# "( %*+ &$$'+$ "( 5""/ 5"=+)#,#8+ &# !+,),OI3
,#/ ,1"=+ "%*+) $"')8+$3 !+,),O ,/"<%+/ , $%,5+/ ,<<)",8* &# %*,% , $+-+8%+/
#'.1+) "( /&$%)&8%$ 6+)+ .'#&8&<,-&$+/ S)$% @<+)$"#,-
!"#$%&' ()) 8"..'#&8,%&"# N&-=&, Y,.+/+> 2CCZ> %*+# /&)+8%") "(
d+,-%* 8,)+ &$ "<+# %" <)&=,%+ $+8%") &#&%&,%&=+ <-,##&#5> N%,%+ N+8)+%,)&,% "( !+,),OI3 [$ <,)% "( &%$
2 E <)&=,%+ &#$%&%'%&"#$ 8,# <,)%&8&<,%+ &# , 8".<-+.+#%,)7 8"..&%.+#% %" &.<)"=&#5 *+,-%* 8,)+> !+,),O $%,%+ *,$
("). %" %*+ $&#5-+ *+,-%* $7$%+.> ("--"6&#5 &%$ <)&#8&<-+$>
1'&-% ,#/ +\'&<<+/ , #+6 ]$8*""-^ (") &#?$+)=&8+ %),&#?
%*)"'5* 8"#%),8% "( <'1-&8 -,6 ") 8"#=+k#&"> 6&%* <)+(+)+#8+
(") %*"$+ +#%&%&+$ 6*&8* ,)+ <*&-,#%*)"<&8 ,#/ 6&%*"'% ,&.$ "(
&#5 ,#/ )+$+,)8* &# <'1-&8 *+,-%* @%*+ _$8"-, /+ N,'O/+
<)"S% `'O1-&8,I 6&%*&# %*+ $%,%+ $+8)+%,)&,% "( *+,-%*3 0*+
F E &% &$ (")1&//+# (") <'1-&8 )+$"')8+$ %" +#/ '<> %*)"'5* ,&/ N8*""- ,&.$ %" <)"=&/+ )+()+$*+) 8"')$+$ 1"%* "# 8-&#&?
") $'1$&/&+$ @,'9jO -&"$ "' $'1=+#8! "l+$I> 5"&#5 %" <)&=,%+ 8,- ,#/ <)+=+#%&=+ 8,)+ ,#/ "# =,)&"'$ ,$<+8%$ "(
&#$%&%'%&"#$ 6&%* <)"S% .,J&#5 ,&.$ *+,-%* $+)=&8+ <-,##&#5 ,#/ .,#,5+.+#% ,% %*+ /&$%)&8%
Z E %*+ /&)+8% ") &#/&)+8% <,)%&8&<,%&"# "( 1'$&#+$$+$ ") (")+&5# -+=+-3 A7 %*+ %&.+ "( %*+ S+-/ $%'/7 @2CCDI> HGa @" b
('#/$ &$ (")1&//+# &# *+,-%* 8,)+ &# %*+ 8"'#%)7 ,<,)% ()". 2QCI "( %*+ 2HQ /&$%)&8%$ *,/ .'#&8&<,-&$+/ ,% -+,$% %"
%*"$+ ,5)++/ 17 -,6 %*+ !"#!$!%"&% -+=+-3 06+-=+ /&$%)&8%$ &# !+,),O $%,%+ *,/
Q E %*+ -,6 6&-- )+5'-,%+ 8"#/&%&"#$ ,#/ #++/$ (") %*+ )+."=,-
,8*&+=+/ +%,!-$*%"( $%,%'$ ,#/ +,8* 7+,) .")+ ,)+ $'8?
"( ")5,#$> %&$$'+$ ,#/ "%*+) *'.,# $'1$%,#8+$ (") %),#$<-,#%>
8++/&#5 &# %*+&) ,<<-&8,%&"#$3 L( "') %*)++ $%'/7 $&%+$>
)+$+,)8* ,#/ %)+,%.+#%> ,$ 6+-- ,$ %*+ 8"--+8%&"#> %)+,%.+#%
,#/ %),#$('$&"# "( 1-""/ ,#/ &%$ /+)&=,%&=+$> (")1&//&#5 ,#7
"#+ *,/ , +%,!-$*%"( 8"#%),8% @, %),/&%&"#,- ')1,# /&$?
%7<+ "( 8"..+)8&,-&$,%&"# %)&8%I 6*&-+ %*+ "%*+) %6" @, .+%)"<"-&%,# ,#/ , )'),-
/&$%)&8%I "#-7 *,/ !"#!$!%"&% $%,%'$3
170
JAU 78 9(:23'"3 +( %&8 ; 7"*2%& 7*2+3*+ < =+/2*23+ >? @ABBBC D?EFDGD
!"#$% &'"(& !"# ;%,$ &',$%'/$ ', 3) ')&9,$%'31',#& 3%#3 -) $"# +#%6
'+"#%* -. $"# ,$3$# /3+'$31 3)& ', +3%$ -. $"# 5%#3$#%
!"#$%& '(#)*()#+', -+"./+$"-#%012* +342#"3$+3( %3/ 4#$%-+-1'$3) %#5'-)7 !"# 3%#3 ', %#13$'<#1* ,4311 29$
1+%&(1 *%#+ 5)%&2(6 &#),#1* +-+913$#&7 =-%> ', $- 2# .-9)& ') $"# ')&9,6
$%'31 3%#3? ') $"# 13%5# .%9'$ 3)& <#5#$321# 43%>#$? ')
!"# $"%## &'(#%#)$ $*+#, -. &',$%'/$, 0#%# ,#1#/$#& -) 1-/31 /-44#%/'31 -9$1#$, -% 2* $%3<#11')5 ')$- $"# /'$*
$"# 23,', $"3$ $"#* 4'5"$ +%#,#)$ &'(#%')5 ,-/'31 %#31'6 /#)$%#7 !"# 1-/31 "#31$" &',$%'/$ "3& 3 ,'@#321# ,$3( 3)&
$'#,7 83$3 -) #3/" &',$%'/$ 3%# ,9443%',#& ') !321# :7 03, -%53)',#& /1#3%1* ')$- &'(#%#)$ &#+3%$4#)$,7 !"#
!321# A
B3') /"3%3/$#%',$'/, -. $"# /-)$%3/$ $*+#, .-% &#/#)$%31',3$'-)
C-)$%3/$ D#,+-),'2'1'$'#, D#E9'%#4#)$,

$*+#
F)/'+'#)$# D#5',$#% 3)& 39$"-%',# +%-<'&#%,G 8#4-),$%3$# ')$#%#,$ ') $3>')5 -) $"#,# %#,+-),'2'1'$'#,G
H13) 3)& 39$"-%',# ')6+3$'#)$ 3)& 342913$-%'31 D#5913$#? /-)<#)# &',$%'/$ "#31$" /-9)/'1G +%-&9/#
+%-/#&9%#, $- 2# +%-<'&#& 2* #3/" .3/'1'$*G 4')9$#,IJ4,G
B-)'$-% 3)& #<3193$# ')6+3$'#)$ 3)& 342913$-%'31 D#5913$#? 43)35# &',$%'/$ "#31$" .9)&G +%-&9/#
,#%<'/#, ') +921'/ 3)& +%'<3$# )-$6.-%6+%-;$ .3/'1'$'#,G ,$3$#4#)$,IJ4,G
8#4-),$%3$# 0'11')5)#,, 3)& 32'1'$* $- $3># -) $"# 8#<#1-+ +%-+-,31 .-% ')$#5%3$'-) 3)& 43)35#4#)$ -.
43)35#4#)$ -. $"# +921'/ 342913$-%'31 .3/'1'$'#, #K',$')5 $"-,# +921'/ 342913$-%'31 .3/'1'$'#, N,$3$# -% .#&#%31O $-
') $"# 49)'/LM +'-G /-4# 9)&#% $"# &',$%'/$G
F)$#5%3$# $"# )#$0-%> -. 23,'/ ,#%<'/#, 3)& 3/$'<'$'#, P Q93%3)$## 3++-')$## %#,+-),'21# .-% ;)3)/'31 +3*4#)$, ',
)9$%'$'-)? #&9/3$'-)? #+'&#4'-1-5'/31 3)& "*5'#)# )-$ 3 2#)#;/'3%* )-% 1')>#& 0'$" 3)* +%-<'&#%
,9%<#'113)/#G /-)$%3/$#&G
8#<#1-+ 3/$'<'$'#, -. -//9+3$'-)31 ,9%<#'113)/# R#$ 9+ /-)&'$'-), $- +13)? 4-)'$-%? /-)$%-1 +%-<','-) -.
,#%<'/#,G
B3')$3') %#/-%&, 3)& 4-)$"1* %#4'$$3)/# -. &3$3 -) 1'<#
2'%$",? 4-%$31'$*? /-4+91,-%'1* )-$';321# &',#3,#,? %#5',$#%
-. #,$321',"4#)$, 3)& +%-&9/$, -. ')$#%#,$ $- "#31$" #$/7
H3%/'31 !3># -<#% +13))')5? 39$"-%',3$'-) 3)& 9,# -. ')6+3$'#)$ H%#,#)$ 3) 3))931 %#+-%$ -. 3/$'<'$'#, ') %#13$'-) $- $"#
3)& 342913$-%'31 +%-/#&9%#, $- 2# +%-<'&#& 3$ #3/" &',$%'/$ "#31$" +13)G
.3/'1'$*G
!3># -) 43)35#4#)$ -. +921'/ 342913$-%'31 .3/'1'$'#, ') H%#,#)$ 3) 3))931 %#+-%$ %#53%&')5 43)35#4#)$
$"# &',$%'/$G 3/$'<'$'#,G
!3># -) .-%4913$'-)? #K#/9$'-)? 4-)'$-%')5 -. 3/$'<'$'#, 8#4-),$%3$# 3))9311* $"3$ /-9)$#%+3%$ .9)&, .%-4 $"#
-. -//9+3$'-)31 ,9%<#'113)/#G &',$%'/$ $%#3,9%* "3<# 2##) 5'<#) $- "#31$"G
S# +3'& 3)* &'(#%#)/# 2#$0##) $"# ;)3)/'31 /#'1')5 H%#,#)$ #<'&#)/# $"3$ $"# &',$%'/$ ', &#<#1-+')5 -% "3,
#,$321',"#& .-% $"# &',$%'/$ 3)& $"# +3*4#)$, 43&# 2* $"# &#<#1-+#& 3 +13) .-% "943) %#,-9%/#, ') $"# "#31$"
.#&#%31 1#<#1 &'%#/$ $- $"# "-,+'$31, 3)& 342913$-%'31 ,#/$-%
.3/'1'$'#, N+921'/ -% +%'<3$# )-$6.-%6+%-;$O
R#4'6+1#)3 !3># -<#% $-$31 %#,+-),'2'1'$* .-% 43)35')5 $"# +%-<','-) H%#,#)$ 3 +13) -. $"# 3))931 $3%5#$, $- 2# 3/"'#<#& 3,
-. ,#%<'/#, 6 +13))')5? %#5',$%3$'-)? /-)$%3/$')5? 3++%-<#& 2* $"# &',$%'/$ "#31$" /-9)/'1G
4-)'$-%')5 3)& +3*4#)$ -. ')6+3$'#)$ 3)& 342913$-%'31
,#%<'/# +%-<'&#%,? +921'/ 3)& +%'<3$#6)-$6.-%6+%-;$G
!3># -<#% 43)35#4#)$ -. $"# 0"-1# +921'/ "#31$" H%#,#)$ ')&'/3$-%, -. 3/"'#<#4#)$ 2* 0"'/" $- #<3193$#
)#$0-%> ') $"# &',$%'/$? 3+3%$ .%-4 %#.#%%31 "-,+'$31, +13)G
9)&#% ,$3$# 43)35#4#)$G
!3># -<#% 4-)'$-%')5 3/$'<'$'#, ') "#31$"? )9$%'$'-) 3)& T3'19%# $- 4##$ 5-31, -<#% $0- ,9//#,,'<# *#3%, /3) 1#3&
#&9/3$'-) 3)& #+'&#4'-1-5'/31? "*5'#)# 3)& -//9+3$'-)31 $- $"# 1-,, -. ,#4'6+1#)3 ,$3$9,G
,9%<#'113)/# .-% $"# &',$%'/$G
D#/#'<# 3)& /-)$%-1 $"# $-$31 ;)3)/'31 %#,-9%/#, .-% ')6 !"# +%-/#,,#, -. #)$'$1')5 49)'/'+'-, $- $"# ,#4'+1#)3
+3$'#)$ 3)& 342913$-%'31 ,#%<'/#, ') 1')# 0'$" $"# 1#<#1 43&# 3$ R$3$# 1#<#1 0'11 2# %#<'#0#& 3$ T#&#%31 1#<#1
#,$321',"#& ;)3)/'31 /#'1')5,7 .-% ;)31 3++%-<31G
B3')$3') 3) ').-%43$'-) ,*,$#4 -) ')6+3$'#)$ 3)&
342913$-%'31 311-/3$'-), -. ;)3)/'31 +3*4#)$,7
171
"# $%&'!()! *% +,# - ").'+, ".'*!.* / 0*1'.'!* 23 456667 839:8;8 WZY
!"#$%"&$ '(# )*+, "* $-. /%#$ #$(0. )1 23*"&"4(+"#($")* .84.&$($")*# )1 -.(+$- &(%. ?3(+"$, (%. -"0-.% (*! #($"#<
(*! -(! .#$(5+"#-.! 1.' 4(%$"&"4($)%, 2.&-(*"#2#6 1(&$")* $-.%.1)%. #)2.$"2.# +)' &)24(%.! '"$- 4))%.%
7-.%. "# (* .8$.*#"9. *.$')%: )1 +(%0. 435+"& -.(+$- %3%(+ 4)43+($")*# CD(%(* .$ (+6= EFFGH6 D)'.9.%= )3$<
&.*$%.# );.%"*0 ( '"!. %(*0. )1 #.%9"&.# (# '.++ (# *3< 43$ (*! )3$&)2. 2.(#3%.# (+#) "*!"&($. ( 4))% ?3(+"$,
2.%)3# 4%"9($. &+"*"&#= -)#4"$(+# (*! 4-(%2(&".#6 >* )1 &(%. )% ($ +.(#$ ( 4))% +.9.+ )1 "24(&$ )* $-. 4)43<
)5#.%9($")*= $-. -.(+$- &.*$%.# (44.(%.! )1 0))! ?3(+< +($")*I# -.(+$- C#.. 7(5+. JH6
"$, "* $.%2# )1 4-,#"&(+ "*1%(#$%3&$3%.= $-. #)4-"#$"&($.! 7-. #.&)*! !"#$%"&$ "# ( $%(!"$")*(+ 3%5(* !"#$%"&$ "*
%(*0. )1 #.%9"&.# );.%.! (*! ?3(*$"$, (*! ?3(+"$, )1 $-. "*$.%")% )1 $-. #$($.= +)&($.! "* ( -"++, (%.(6 7-.#.
#$(; .24+),.!6 @)*.$-.+.##= $-. %($") )1 5)$- *325.% !"#$%"&$# -(9. ( 2(K)%"$, )1 $-. 4)43+($")* 5(#.! "* (*
)1 -.(+$- 1(&"+"$".# (*! *325.% )1 5.!# $) 4)43+($")* 3%5(* &.*$%.= 53$ (+#) -(9. (* "24)%$(*$ %3%(+ &)2<
#"A. "# &)24(%($"9.+, +)' 5, #$($. +.9.+#6 >4.* "*$.%< 4)*.*$6 7-"# $)'* 0%.' (# ( &.*$%. 1)% &);.. $%(!. "*
9".'# '"$- +)&(+ 2.25.%# )1 $-. 4)43+($")* "*!"&($. ( $-. +(#$ &.*$3%, (*! '(# $-. /%#$ $)'* $) 5. +"*:.! $)
+)' +.9.+ )1 #($"#1(&$")* '"$- $-. -.(+$- #.%9"&.# 4%)< $-. #$($. &(4"$(+ 5, $%("*6 B$ %.2("*# (* "24)%$(*$
9"!.!6 B$ "# *)$.! $-($ )$-.% #$3!".# -(9. #-)'* $-($ &.*$%. 1)% $-. #3%%)3*!"*0 %.0")*6 7-. $)'* -(# ( 4%"<
7(5+. J
L$3!, #"$.#M N(&:0%)3*! "*1)%2($")* )* 1)%2(+ -.(+$- #,#$.2 #$%3&$3%.#= 0.)<!.2)0%(4-"& !($( (*! 4%)9"#")* )1 -.(+$- &(%.(
O.$%)4)+"$(* P%5(* Q3%(+

R.0(+ #$($3# B*&"4".*$. L.2"<4+.*( B*&"4".*$.
!"# $%&'() *+*,#'

>%0(*"#($")* '"$- %.#4.&$ $) $-. @) +)&(+ &)3*&"+ R)&(+ &)3*&"+ R)&(+ &)3*&"+
-.(+$- %.1)%2#
D.(+$- S+(* (*! T3*! D.(+$- S+(* (*! T3*! D.(+$- S+(* (*!
T3*!
@) #$(U*0 4+(* @) #$(U*0 4+(* @) #$(U*0 4+(*
L$%3&$3%.! #.&%.$(%"($ L$%3&$3%.! #.&%.$(%"($ R"2"$.! #.&%.$(%"($
S35+"& %.#)3%&.# JE 1(&"+"$".#M EV 1(&"+"$".#M W 1(&"+"$".#M
X -)#4"$(+#= EW &.*$%.#= X #4.&"(+"#$ E -)#4"$(+= E &.*$%.= Z 5"%$- E 2"*"<-)#4"$(+#= Y
&.*$%.#= Y +(5#= Z %.#)3%&. 5(*:# &.*$%.#= W 4)#$# 4)#$#
JYG 5.!# < V6ZZ[-(5 EVZ 5.!# < V6JY[-(5 JG 5.!# < V6ZG[-(5
\D] ^ EVF \D] ^ YY \D] ^ JV
T(2"+, -.(+$- $.(2 ^ J
T3*&$")*"*0 )1 -.(+$- 1(&"+"$".# _))! _))! S))%
S%"9($. 1(&"+"$".# JV 4-(%2(&".# X 4-(%2(&".# E 4-(%2(&,
2(*, 4%"9($. &+"*"&#= +(5)%($)%".#= X 4%"9($. &+"*"&#= E +(5)%($)%, *) 4%"9($. 1(&"+"$".#
-)#4"$(+#
-#%./#'%0&(1"23 #452&%4'#4,
`&)+)0, P%5(* D"++# L.%$(a)
b%.( C:2ZH Fc6WV JXG6JV GFE6GV
S)43+($")*[:2Z EWZX cY6VJ EG6XE
d 3%5(*"#.! FF6YZ WV6ZX XY6XW
d B++"$.%(&, %($.# F6JF ZY6G Jc6FV
L)3%&.# )1 "*&)2. B*!3#$%,= \)22.%&.= B*1)%2(+ (&$"9"$".# b0%"&3+$3%.= b*"2(+ b*"2(+ -3#5(*!%,=
-3#5(*!%,= \)22.%&. (0%"&3+$3%.
67()2,+ 24/23(,%&*
BOQ JG ZF JZ
d O.(#+.# &)9.%(0. EVV EVV cF6W
Ù&".*&, Ce #4.*$M4+(**.!H b9.%(0. $) 0))! S))% S))%
@)6 &+"*"&(+ &)*#3+$($")*#[&(4 E6JW[-(5 Z6YF[-(5 E6WW[-(5
@)6 &+"*"&(+ 4%)&.!3%.#[&(4 V6VJ[-(5 V6ZV[-(5 V6VJ[-(5
@)6 !.*$(+ &)*#3+$($")*#[&(4 V6XG[-(5 V6cW[-(5 V6JY
@)6 %(!")+)0"&(+ .8(2#[&(4 V V6Ec V
@)6 &+"*"&(+ +(56 7.#$#[&(4 V6XX[-(5 E6VV[-(5 V6WF
S)43+($")* L($"#1(&$")* C! ^ EVV S))% fYVd _))! gYVd S))% fYVd
"* .(&-H
(
L)3%&.#M B4+(*&. CEFFGHh L$($. L.&%.$(%"($ )1 D.(+$-h S%"2(%, !($( 1%)2 $-. #$3!,6
172
UVU 67 8'5"&(-& $' *17 9 6-+"*1 6+"$&+$ : ;$3"+"&$ <= >?@@@A B=CDBEB
!"#$%&'#%(')%*)'+# ,'-*.#"/ "&0 " *12/.3 ,$"/#, 3$&#)$4 */"&&.&<; /'3"/ "33'1&#"2./.#6; B1"/.#6 '( 3")$ "&0 *'*%
5,$ /'3"/ ,$"/#, -6-#$7 ,"- " (1)#,$) &$#8')9 '( 2"-.3 1/") $7*'8$)7$&#4
,$"/#, *'-#- ")'1&0 #,$ 0.-#).3#4 :& '2-$)!"#.'&; #,$ 5,$ #,)$$ -#106 -.#$- *)'!$0 #' 2$ !$)6 0.A$)$&#
!").'1- ,$"/#, ("3./.#.$- (1&3#.'&$0 "33')0.&< #' */"&4 ()'7 '&$ "&'#,$) .& 7"&6 )$-*$3#-4 5,$)$ .- '( 3'1)-$;
5,$ 0.-#).3# ,"0 =1-# "3,.$!$0 (1// 71&.3.*"/.-"#.'& "- &' '&$ 3'))$3# 8"6 '( <)'1*.&< #,$ 06&"7.3- $J*$).%
8$ -#")#$0 '1) -#1064 5,.- 8"- #,$ 0.-#).3# 8,$)$ #,$ $&3$0 "&0 '2-$)!$04 5,$ 3"#$<').$- <.!$& ,$)$ -$$7 #'
/'3"/ ,$"/#, 7"&"<$)- -*'9$ #,$ /"&<1"<$ '( #,$ 1- 1-$(1/ .& #,"# #,$6 <)'1* .&#$)"3#.'&"/ ("3#')- 1&0$)
)$(')7- "&0 2$-# $-#"2/.-,$0 #,$ (')7"/ -#)13#1)$- "&0 ,$"0.&<- #,"# 3"& 2$ )$/"#$0 2'#, #' #,$ )$(')7 *)'3$--
*)'3$01)$- "33')0.&< #' #,$ <1.0$/.&$- .&0.3"#$0 26 #,$ "&0 #' ')<"&.-"#.'&"/ ()"7$- '( )$($)$&3$ "&0 .& #,"#
>.&.-#)6 '( ?$"/#, "&0 #,$ @#"#$ @$3)$#")6 '( ?$"/#,4 #,$6 *)'!.0$ 3'&3$*#1"/ 0.7$&-.'&- (') 8,.3, .&0.%
5,$)$ 8"- "& "3#.!$ /'3"/ ,$"/#, 3'1&3./ "&0 " &$8 3"#')- 7"6 2$ -'1<,# .& ')0$) #' )$/"#$ #,$-$ #' .&0.%
*)'<)"77$ '( ("7./6 ,$"/#, #$"7- #,"# '*$)"#$0 .& #,$ 3"#')- '( ,$"/#, 3")$ B1"/.#64
7'-# &$$06 ")$"-4 5,$ /'3"/ *'*1/"#.'& .&0.3"#$0 "
2$##$) -"#.-("3#.'& 8.#, #,$ ,$"/#, 3")$ 'A$)$0 #,"& .& !"#$%$&' ()*+$( ,-% *.'-&-/0 "& )1*&&"&2 *&3 3$+"("-&4
#,$ '#,$) #8' 0.-#).3#-4 ?$"/#, -#"#.-#.3- .&0.3"#$ " )$/"% /*5"&2
#.!$/6 <''0 B1"/.#6 '( ,$"/#, 3")$ '& #,$ 8,'/$ C-$$
5"2/$ DE4 5,$ /'3"/ 3'&#$J# 0.)$3#/6 "A$3#- #,$ -*"3$ #,"# $J.-#-
5,$ #,.)0 -.#$ .- " *''); )1)"/ 0.-#).3# !$)6 #6*.3"/ '( .& )$"/.#6 (') "1#'&'76 .& */"&&.&< "&0 0$3.-.'&%7"9%
0)6 .&#$).') )$<.'& '( #,$ F')#,$"-# '( G)"H./ 3"//$0 .&< .& "# /$"-# #,)$$ 8"6-4
#,$ -$)#"I'4 5,$ 0.-#).3# ,"- '&$ -7"// 3$&#)"/ #'8& "&0
" 3'1*/$ '( '#,$) -7"// 3$&#)$- /'3"#$0 8.#,.& " /")<$ 6-.%+$( -, "&+-/$
")$" '( -3"##$)$0 )1)"/ -$##/$7$&#-4 @13, )1)"/ #'8&- K/#,'1<, #,$ (')7"/ 3'&#)"3# 8.#, #,$ L$0$)"/
0$!$/'*$0 ,.-#').3"//6 ")'1&0 -.#$- 8,$)$ #,'-$ 2).&<% O&.'& 0$&'#$- " 0.-#).3# "- 71&.3.*"/.-$0 #' " <)$"#$)
.&< #,$.) 3"##/$ #' #,$ 3"*.#"/ (') -"/$ 8'1/0 -#'* #' ') /$--$) $J#$&#; .& )$"/.#6 #,$ 1-$ '( #,$ +&"&3$ "//'%
8"#$) #,$7 $& )'1#$4 5,$ 0.-#).3# 8"- '&/6 "# #,$ +)-# 3"#$0 ()'7 #,$ >.&.-#)6 '( ?$"/#, "- *")# '( #,"# 3'&%
-#"<$ '( 71&.3.*"/.-"#.'&; ,"0 ($8 (')7"/ -#)13#1)$- #)"3# .- "/)$"06 #.<,#/6 0$#$)7.&$0 #,)'1<, (')71/"$4
8.#,.& #,$ /'3"/ ,$"/#, -6-#$74 5,$)$ 8"- " /'3"/ ,$"/#, 5,$ +&"&3$ .- "//'3"#$0 #' *"6 (') "721/"#')."/
3'1&3./ $-#"2/.-,$0 8,.3, 0.0 7$$# )$"-'&"2/6 )$<1% C)$($))$0 #' "- @PKQ@O@E "&0 .&%*"#.$&# 3")$ C)$($))$0
/")/64 5,$ #'8& ,"0 " -7"// ,'-*.#"/ "2/$ #' 3'*$ 8.#, #' "- KP?E "&0 .- 3"/31/"#$0 8.#, )$($)$&3$ #' ("3./.#.$-;
1&3'7*/.3"#$0 2.)#,- "&0 -.7*/$ .//&$--$- */1- "721% -#"R&< "&0 *'*1/"#.'&4 5,1-; )$<")0/$-- '( #,$ 9.&0 '(
/"&3$ #)"&-*')# #' #,$ "0="3$&# 0.-#).3#; "& .7*')#"&# 3'&#)"3# 8.#, #,$ L$0$)"/ O&.'&; .& )$"/.#6 " 0.-#).3#
1)2"& 3$&#)$4 5,$)$ 8$)$ ('1) '#,$) ,$"/#, *'-#- ,"- )$/"#.!$/6 /.##/$ 3'&#)'/ '!$) #,$ 1-$ '( #,$-$ (1&0-4
8.#,.& #,$ 0.-#).3# !.-.#$0 26 *,6-.3."&- '& 3$)#".& 0"6- ?'8$!$); #,$ 0.-#).3# -$3)$#")."# '( ,$"/#, 7"6 "/-'
'( #,$ 8$$9 "&0 "& $J#$&-.!$ &$#8')9 '( 3'771&.#6 <".& .&3'7$ ()'7 #,$ @#"#$; 8,.3, 7"6 *)'!.0$ +&"&3$
,$"/#, 8')9$)-4 :& '2-$)!"#.'&; #,$ ,$"/#, 3$&#)$- (') -*$3.+3 *)'<)"77$- ') *)'=$3#-; "&0 ()'7 #,$ 0.-%
(1&3#.'&$0 2"0/6 8.#, *'') "##$&0"&3$ '( -#"A "&0 #).3# /'3"/ <'!$)&7$&#4 5,$ 0$<)$$ '( "1#'&'76 #,"#
/"39 '( )$-'1)3$- ()'7 #,$ 0.-#).3#4 K--$--7$&#- '( #,$ #,$ 0.-#).3# ,$"/#, -$3)$#")."# ,"- '!$) .#- $J*$&0.#1)$
B1"/.#6 '( 3")$ ()'7 2'#, #,$ *'*1/"#.'& "&0 ()'7 #,$)$(')$ 0$*$&0- 7')$ '& #,$ $J#$&# #' 8,.3, .#-
,$"/#, -#"#.-#.3- .&0.3"#$ " *'') B1"/.#6 '( 3")$ ') .&3'7$ 3'7$- ()'7 #,$ >.&.-#)6 '( ?$"/#, 3'7*")$0
,$"/#, .7*"3#; "/#,'1<, -1)*).-.&</6 3'7*")"2/$ 8.#, 8.#, -#"#$; ') 7'-# .7*')#"&#/6; 0.-#).3# (1&0-4 5,'-$
#,$ 7$#)'*'/.#"& 0.-#).3# C-$$ 5"2/$ DE4 0.-#).3#- 8.&&.&< @#"#$ (1&0.&< '!$) "&0 "2'!$ #,"#
<.!$& (') @#"#$%8.0$ *)'<)"77$- -13, "- #,$ 3'771%
&.#6 ,$"/#, 8')9$)- *)'<)"77$; #$&0 #' 2$ #,'-$
8,'-$ /'3"/ <'!$)&7$&#- ")$ *'/.#.3"//6 "R/."#$0 8.#,
!"#$%&" '( "'%)*+ ',-*.)"*&)'. *./ #'+)&)%*+ %0+&0,$ #,$ @#"#$ <'!$)&7$&#4 5,$ "//'3"#.'& '( 0.-#).3# +&"&3$
#' ,$"/#, 8.// 0$*$&0 &'# '&/6 '& #,$ *'/.#.3"/ 8.// '(
5,$ 2$&$+#- '( 0$3$&#)"/.-"#.'& C-$$ L.<4 ME -,'1/0 /'3"/ <'!$)&7$&# #' -1**')# /'3"/ ,$"/#, )$/"#$0 "3#.!.%
0$).!$ ()'7 #8' *).7")6 -#)"#$<.$-N #,$ -*"3$ 3)$"#$0 #.$-; 21# "/-' '& #,$ "2-'/1#$ )$!$&1$ '( #,$ 0.-#).3# .&
(') "1#'&'7'1- 0$3.-.'&%7"9.&< "&0 #,$ -*"3$ 3)$"#$0 #,$ +)-# */"3$4 S'3"/ )$!$&1$ 3'7$- ()'7 !").'1- #"J$-
(') #,$ !'.3$ '( #,$ /'3"/ *'*1/"#.'& #' 2$ .&3')*')"#$0 '& 3'77$)3."/ "3#.!.#.$-; "&0 #,$-$ "3#.!.#.$- ")$ 7')$
.&#' /'3"/ */"&&.&<4 5,$ -'3."/ ')<"&.-"#.'& "&0 *'/.#.% $J#$&-.!$ .& #,$ 1)2"& "&0 *$).%1)2"& 0.-#).3#- #,"& #,$
3"/ 31/#1)$ 8.#,.& 8,.3, #,$ .7*/$7$&#"#.'& '( #,$-$ )1)"/ '&$-4 5,$ )$"/ -*"3$ (') "1#'&'7'1- */"&&.&< '(
#8' -#)"#$<.$- .- $72$00$0 "A$3#- &'# '&/6 ,'8 #,$-$ $J*$&0.#1)$ .- #,$)$(')$ <)$"#/6 .&T1$&3$0 26 #,$ *'/.#.%
-#)"#$<.$- '*$)"#$ .& )$"/.#6 21# "/-' #,$ *)'3$--$- #,"# 3"/ )$/"#.'&-,.* 8.#, #,$ -#"#$ <'!$)&7$&# "&0 #,$ *'%
7$0."#$ #,$ .&#$&0$0 .7*"3# '( #,$-$ -#)"#$<.$- '& #,$ #$&#."/ (') /'3"/ #"J )".-.&<; 8,.3, .& #1)& .- <$&$)"//6
'1#*1# .& #$)7- '( #,$ )$-*'&-.!$&$-- '( ,$"/#, 3")$ <)$"#$) .& 1)2"& "&0 7$#)'*'/.#"& 0.-#).3#- #,"& #,$
173
45 6,7/*3"* (, $%5 8 4"#/$% 4#/(*#( 9 :(;/#/*( <= >?@@@A B=CDBEB \VC
!"!#$ %&'() *#!+,-"$#!$. +/' 0%!' ,0*%1'!,(/'2 !"!#$ &%+ 2,!'-+$. # !'N'-+,%& %5 $%-#$ *%$,+,-#$ -"$+"!' %!
2,(+!,-+(3 (%-,#$ %!6#&,(#+,%&) +/' -#*#-,+. 5%! ,&5%!0#+,%& /#&27
$,&6 ,( $,F'$. +% !'N'-+ +/' ,0*%!+#&-' +/#+ /#( 8''&
!"#$% &"'()*+(*, 6,1'& +% ,&5%!0#+,%& (.(+'0( ,& +'!0( %5 5"&2,&6 #&2
4/'+/'! # *%+'&+,#$ 5%! #"+%&%0%"( *$#&&,&6 -!'7 +/' -%00,+0'&+ %5 /'#$+/ *!%5'((,%&#$( +% $%-#$ *$#&7
#+'2 8. $%-#$ +#9 !'1'&"' %! (+#+' 5"&2,&6 +!#&(5%!0( &,&63 =/' ,&N"'&-' %5 #& #8(%$"+' #1#,$#8,$,+. %5
,&+% # !'#$ (*#-' 5%! #"+%&%0%"( *$#&&,&6 2'*'&2( %& !'(%"!-'( <#( '1,2'&+O +/' !,-/'! 0'+!%*%$,+#& 2,(+!,-+
+/' $%-#$ 6%1'!&0'&+3 :$+/%"6/ ;&#&-' 5%! /'#$+/ /#2 # !'#(%&#8$. (%*/,(+,-#+'2 ,&5%!0#+,%& 8#(') +/'
-#!') </#+'1'! ,+( (%"!-') 6%'( ,&+% # 2,(+!,-+ /'#$+/ "!8#& 2,(+!,-+ F'*+ !'-%!2( %5 ,+( %<& 8#(,- ,&2,-#+%!()
5"&2) ,+ ,( &%&'+/'$'(( "$+,0#+'$. +/' !'(*%&,8,$,+. %5 8"+ &% $%-#$ ,&5%!0#+,%& <#( #1#,$#8$' ,& +/' !"!#$ 2,(7
+/' $%-#$ 6%1'!&0'&+ #20,&,(+!#+,1' (.(+'03 =/,( +!,-+3
0'#&( +/#+ &%+ %&$. ,( +/' +%+#$ 8"26'+ (,>' 2'*'&2'&+ =/' #8,$,+. +% *$#& $%-#$$. #&2 !'(*%&2 +% $%-#$
%& +/' -%00,+0'&+ %5 $%-#$ 6%1'!&0'&+ +% #$$%-#+,&6 &''2( ,( %&' %5 +/' 0#,& #!6"0'&+( *"+ 5%!<#!2 5%!
!'(%"!-'( +% +/' /'#$+/ (.(+'0) 8"+ #$(% %& +/' ,&-$,7 2'-'&+!#$,('2 0#&#6'0'&+ %5 /'#$+/ -#!' #&2 +/"( ,+ ,(
&#+,%& %5 $%-#$ 6%1'!&0'&+ +% 2'$'6#+' -%&+!%$ %1'! *#!+,-"$#!$. ,0*%!+#&+ +% #(('(( 2,P'!'&-'( ,& +/' (*#-'
+/#+ 8"26'+ +% +/' 2,(+!,-+ /'#$+/ ('-!'+#!,#+3 =/' !'$#7 5%! #"+%&%0.3 E& Q%!+/'#(+ @!#>,$) ,+ ,( '1,2'&+ +/#+ #+
+,1' #21#&+#6'( #&2 2,(#21#&+#6'( %5 +/' $%-#$ /'#$+/ $'#(+ +/!'' 2,0'&(,%&( %5 $%-#$ *%$,+,-#$ -"$+"!' #P'-+
(.(+'0 8',&6 "$+,0#+'$. "&2'! +/' $%-#$ 6%1'!&0'&+ #27 +/' *%+'&+,#$ #&2 !'#$ (*#-' 5%! #"+%&%0.R +/' (%"!-'(
0,&,(+!#+,%& -#& 8' $%&6 2'8#+'2 ?('' @#!&#8#() ABBC %5 ,&-%0') +/' !'$#+,%&(/,* <,+/ +/' $%-#$ 6%1'!&0'&+
5%! # 2,(-"((,%&D) 8"+ </#+ ,( -'!+#,& ,( +/#+ +/'!' ,( #&2 +/' #++'&+,%& 6,1'& +% $%-#$ 2#+#3
(*#-' 5%! 6!'#+ 1#!,#+,%& ,& +/' '9+'&+ +% </,-/ +/'
$%-#$ /'#$+/ (.(+'0 /#( -%&+!%$ %1'! ,+( %<& !'(%"!-'( 1/2()(*, 30$#(3 .") $ %"#$% '"/#( /* 0%$**/*&
,&2'*'&2'&+$. %5 +/' $%-#$ 6%1'!&0'&+3 E& %"! !"!#$
(+"2. (,+') +/' $%-#$ 6%1'!&0'&+ '(('&+,#$$. 0#&#6'2 :5+'! +<'&+.7;1' .'#!( %5 0,$,+#!. 2,-+#+%!(/,*)
+/' /'#$+/ (.(+'0) $'#1,&6 +/' 2,(+!,-+ /'#$+/ ('-!'+#!,#+ *!%8#8$. +/' 0%(+ ,0*%!+#&+ #(*'-+ %5 *%$,+,-#$ !'5%!0
<,+/ -%&+!%$ %1'! $,++$' 0%!' +/#& +/' -%00"&,+. ,& @!#>,$ /#( 8''& +/' *"(/ 5%! 0%!' 2'0%-!#+,('2 ,&7
/'#$+/ <%!F'! *!%6!#00') </,-/ ,( 5"&2'2 5!%0 +/' (+,+"+,%&(3 S#!,%"( +.*'( %5 -%00"&,+.78#('2 %!6#&,(7
G+#+' $'1'$3 :$$ %+/'! 2'-,(,%&( <'!' 0#2' 5!%0 5!%0 #+,%&( '0'!6'2 2"!,&6 +/' 2,-+#+%!(/,*( #( #$+'!&#+,1'
+/' $%-#$ 6%1'!&0'&+ %H-'3 @. -%&+!#(+ ,& +/' %+/'! (%-,#$ 6!%"*,&6( +% +/' +!#2,+,%&#$ ,&(+,+"+,%&( %5 *%$,+,7
+<% (+"2. (,+'() #$+/%"6/ *"!-/#(,&6 #&2 (% 5%!+/ /#2 -#$ *#!+,'( #&2 "&,%&() </,-/ <'!' !'*!'(('2 2"!,&6
+% 8' *#(('2 +/!%"6/ +/' $%-#$ 6%1'!&0'&+ #20,&,(+!#7 +/,( +,0'3 =/'(' #$+'!&#+,1'( *!%1,2'2 # 0'#&( +/!%"6/
+,1' (.(+'0) +/'!' <#( &% -%&+!%$ 5!%0 +/' $%-#$ 6%17 </,-/ *!'(("!' 5%! ,0*!%1'2 #--'(( +% !'(%"!-'( #&2
'!&0'&+ %1'! /%< +/' /'#$+/ (.(+'0 (*'&+ ,+( 0%&'.3 ('!1,-'( -%"$2 8' '9*!'(('2 ?5%! '9#0*$'( ('' T(-%8#!
U :$1#!'>) ABBVD3 W%$,+,-#$ !'5%!0 ,& 0"-/ %5 X#+,&
-*.")+$,/"* .") 0%$**/*& :0'!,-# '9*$,-,+$. #++'0*+( +% 8",$2 "*%& +/'(' (%-,#$
I%$$%<,&6 %& 5!%0 +/,() </'+/'! # (*#-' 5%! $%-#$ 0%1'0'&+( +%<#!2( &'< 5%!0( %5 *%$,+,-#$ ,&1%$1'0'&+
#"+%&%0. <,$$ 8' "('2 +% *$#& /'#$+/ -#!' +/#+ ,( 0%!' ,& </,-/ #$$ 6!%"*( ,& (%-,'+. 0#. *%+'&+,#$$. 8' !'*7
!'(*%&(,1' +% $%-#$ &''2( ,& *#!+ 2'*'&2( %& +/' 2,(7 !'('&+'23 E& +/' @!#>,$,#& /'#$+/ ('-+%!) *#!+,-,*#+%!.
+!,-+ /'#$+/ (.(+'0J( %<& -#*#-,+. +% ,2'&+,5. ,+( %<& #&2 2'-,(,%&70#F,&6 /'#$+/ -%"&-,$( /#1' 8''& '(+#87
/'#$+/ *!%8$'0( ,& %!2'! +% *$#& +/' #$$%-#+,%& %5 $,(/'2 #+ #$$ $'1'$( %5 +/' /'#$+/ (.(+'0 ?2,(+!,-+) (+#+'
!'(%"!-'(3 :$$ 2,(+!,-+( /#1' +% -%$$'-+ ,&5%!0#+,%& #&2 5'2'!#$D3 =/' '9,(+'&-' %5 # -%"&-,$ ,( # 5%!0#$
0%&+/$. </,-/ ,( ("80,++'2 +% +/' G+#+'3 =/,( ,&-$"2'( !'Y",!'0'&+ 5%! 2'-'&+!#$,(#+,%&) <,+/ # ;5+.Z;5+. !'*7
+/' *!%2"-+,1,+. 8"$$'+,&( %5 *#+,'&+( (''& %& </,-/ !'('&+#+,%& %5 /'#$+/ *!%5'((,%&#$( #&2 $#. 0'08'!(3
*#.0'&+ ,( 2'*'&2'&+) -#('( %5 &%+,;#8$' 2,('#('() @"+) 8'.%&2 +/,( !'Y",!'0'&+ +/' (,>') 2'+#,$'2 -%0*%7
2'#+/( #&2 8,!+/() #( <'$$ #( 2#+# %& +/' #-+,1,+,'( %5 (,+,%& #&2 #-+"#$ 5"&-+,%&,&6 -#& 1#!. ,00'&('$.3 E&
+/' -%00"&,+. /'#$+/ <%!F'!(3 E&5%!0#+,%& %& 0%!7 #22,+,%&) +/' *!'1,%"( /,(+%!. %5 (%-,#$ 0%8,$,(#+,%&
+#$,+. *!%;$'( %5 +/' 2,(+!,-+() *#!+,-"$#!$. %& E&5#&+ 0#. 8' -!,+,-#$ +% %"+-%0'(3 : $%&67+'!0 (+"2. %5
K%!+#$,+. L#+' ?EKLD #&2 K#+'!&#$ K%!+#$,+. L#+' 2'-'&+!#$,(#+,%& <,+/ *#!+,-,*#+,%& ,& E+#$. 5%"&2 +/#+
?KKLD) #!' #1#,$#8$' #+ +/' G+#+' $'1'$3 M%<'1'!) +/' +/%(' 2,(+!,-+( </,-/ #$!'#2. /#2 # &"08'! %5 -,1,$)
'9+'&+ +% </,-/ +/' 2,(+!,-+ /'#$+/ ('-!'+#!,#+ ,+('$5 -%00"&,+.78#('2 %!6#&,(#+,%&( 5"&-+,%&'2 8'++'! +/#&
0#,&+#,&( #& ,&5%!0#+,%& (.(+'0) ,( #<#!' %5 '1'& +/%(' $'(( %!6#&,('2) +/"( %1'! +,0' ,&-!'#(,&6 ,&'Y",7
8#(,- ,&2,-#+%!( ("-/ #( #&+'7&#+#$ -#!' %! ,00"&,(7 +,'( 8'+<''& 2,(+!,-+( !#+/'! +/#& +/' %**%(,+' ?W"+&#0)
#+,%& -%1'!#6' #&2 ,( #8$' +% ,2'&+,5. *!,%!,+. /'#$+/ ABB[D3
,(("'( 5%! +/' 2,(+!,-+ #( # </%$' #&2 5%! (*'-,;- +#!6'+ E& %"! (+"2.) <' '9*'!,'&-'2 +/!'' 1'!. 2,P'!'&+ '97
#!'#( %! *%*"$#+,%& 6!%"*( <,+/,& +/' 2,(+!,-+ 1#!,'2 *!'((,%&( %5 +/' *%$,-. 2,!'-+,1' 5%! *#!+,-,*#+%!. 2,(7
'&%!0%"($. 8'+<''& %"! +/!'' (+"2. (,+'(3 :$+/%"6/ +!,-+ /'#$+/ -%"&-,$(3 =/' "!8#& (+"2. (,+' <#( +/'
174
NOP 45 6+7)&$%& "+ '(5 8 4%.)'( 4.)"&." 9 2"*).)&" :; <=>>>? @;AB@C@
!"#$%"&$ '$ $('$ $")* )+#$ &,*'%,- &+))"$$*! $+ $(* ,*## #"02"B&'2$ '# %*0'%!# =+4# +. "2.+%)'$"+2 .%+)
%*.+%) /%+0%'))* 1+$( "2 $(* *3$*2$ $+ 4("&( !"#$%"&$ $(* /+/6,'$"+2 $+ $(* !"#$%"&$ (*',$( #*&%*$'%"'$ $('2 '$
#*&%*$'%"'$ #$'5 6#*! $(* %*.+%) !"#&+6%#* '2! "2 $(* B%#$ '//*'%#9 I2 *3$*2#"+2 +. $("# /+"2$ "# $(* +1#*%7
')+62$ +. %*.+%)7%*,'$*! '&$"8"$- 0+"20 +29 :*%* $(* 8'$"+2 $('$ $(* !"#$%"&$ ,*'#$ &+2&*%2*! 4"$( &',,"20
!"#$%"&$ (*',$( &+62&", ('! 1**2 #*$ 6/ $+ +/*%'$* "2 (*',$( &+62&", )**$"20# 4'# ',)+#$ &*%$'"2,- $('$ 4"$(
$(* #/"%"$ +. $(* (*',$( %*.+%) "2$*2$"+2#9 ;(* &+62&", $(* 0%*'$*#$ 26)1*% '2! 8'%"*$- +. *3"#$"20 &"8", '2!
)*$ *8*%- )+2$( %*06,'%,-< 4*%* &+2#6,$*! '1+6$ 6#* &+))62"$-71'#*! +%0'2"#'$"+2# ',%*'!- "2 *3"#$*2&*<
+. .62!#< 4*%* &+2#6,$*! '1+6$ (*',$( /%+0%'))*# '2! #600*#$"20 +$(*% .+%' .+% *3/%*##"20 (*',$(7%*,'$*!
',$(+60( &,'## '2! 0*2!*% !"5*%*2&*# 4*%* *8"!*2$ '# 2**!# )"0($ ',#+ ('8* 1**2 +/*%'$"209
"2=6*2&*# +2 /'%$"&"/'$"+2< )'2- +. $(* ,'- %*/%*#*2$'7 ;(* /%+/+#', $('$ !*&*2$%',"#*! )'2'0*)*2$ +.
$"8*# !*1'$*! "##6*# >**2,- '2! '&$"8*,-9 ?'- )*)1*%# (*',$( &'%* 4",, 1* )+%* %*#/+2#"8* $+ ,+&', 2**!# "2
4*%* #*,*&$*! $+ %*/%*#*2$ 0*+0%'/("&', #617!"8"#"+2# +. /'%$ !*/*2!# +2 !*&*2$%',"#'$"+2 1*"20 '&&+)/'2"*!
$(* !"#$%"&$ '2! &"8", +%0'2"#'$"+2# 4"$("2 $(* !"#$%"&$9 1- "2&%*'#*! "28+,8*)*2$ 1- $(* &'$&()*2$ /+/6,'$"+2
;(*%* 4'# ' %*'#+2'1,* ,*8*, +. &"8", +%0'2"#'$"+2 "2 #+)* 4'- "2 +%!*% $+ !*B2* $(+#* 2**!#9 ;(* 8'%"7
$(%+60( %*#"!*2$#@ '##+&"'$"+2#< '0%"&6,$6%', 62"+2# '2! '$"+2 1*$4**2 !"#$%"&$# "2 $(* &+)/+#"$"+2< .%*J6*2&-
&(6%&( 0%+6/#9 +. )**$"20# '2! .62&$"+2# +. $(*"% (*',$( &+62&",# '2!
A2 $(* %6%', !"#$%"&$< ' &+62&", *3"#$*! 4"$( ' .+%)', $(* *3$*2$ +. +$(*% *3"#$"20 &+))62"$-71'#*! +%0'2"#7
B.$-CB.$- /%+.*##"+2',7,'- )*)1*%#("/ '2! )*$ %*'#+27 '$"+2# "# *8"!*2$< 16$ $(* *5*&$ +. $("#< ". '2-< +2 /,'27
'1,- %*06,'%,- *'&( )+2$(9 D"8", +%0'2"#'$"+2# "2 $(* 2"20 +% +2 (*',$( #*%8"&* J6',"$- 2**!# $+ 1* *3/,+%*!9
!"#$%"&$ 4*%* .*4 '2! 2+$ )6&( %*/%*#*2$*! $(%+60( ;(6#< $(* %*,'$"+2#("/ +. $(* !"#$%"&$ (*',$( &+62&",# $+
$(* &+62&",9 ;(* )+#$ ")/+%$'2$ 4'# $(* '0%"&6,$6%', (*',$( #*%8"&*# '2! $+ $(* &'$&()*2$ /+/6,'$"+2 2**!#
%*.+%) )+8*)*2$ 4("&( !"! 2+$ ,"2> )6&( !"%*&$,- $+ 1* '##*##*! "2 $*%)# +. 4("&( 2*"0(1+6%(++!# '2!
4"$( $(* (*',$( &+62&",9 E*#"!*2$#@ '##+&"'$"+2# 4*%* &+))62"$-71'#*! +%0'2"#'$"+2# '%* %*/%*#*2$*! '2!
2+$ ")/+%$'2$9 ?'- )*)1*%# +$(*% $('2 $(* &+))67 4('$ /%+/+%$"+2 +. $(* /+/6,'$"+2 $("# %*/%*#*2$#< $(*
2"$- (*',$( 4+%>*%# 4*%* ,+&', &+62&",,+%# .%+) 0*+7 *3$*2$ $(* /+/6,'$"+2 "# '4'%* +. $(* &+62&", '2! "$#
0%'/("&', #617!"8"#"+2# +. $(* !"#$%"&$9 :+4*8*%< #"2&* '&$"8"$"*#< '2! 4('$ &('22*,# /*+/,* $("2> $(*- 4+6,!
$(* ,+&', 0+8*%2)*2$ ('! $'>*2 &+2$%+, +. )'2'0"20 !%'4 6/+2 "2 +%!*% $+ &+)/,'"2 '0'"2#$ +% ,+11- .+%
$(* (*',$( #-#$*) '/'%$ .%+) $(* &+))62"$- (*',$( $(* !*,"8*%- +. (*',$( #*%8"&*#9
4+%>*%#< $(*#* 2+%)',,- &')* ',+20 $+ ',, $(* )**$"20#
',#+ '2! "2 *5*&$ $(* &+62&", .62&$"+2*! '# ' .+%6) "2 !"#$%&'()$"* '&* )&$+)+,+)%&'()$"* )&-,"&."$ %&
4("&( $+ $*,, $(* &+))62"$- (*',$( 4+%>*%# 4('$ $(*"% ',+%&%/0 '&* (%.'( 1%)."
'&$"8"$"*# #(+6,! 1* .+% $(* .+%$(&+)"20 )+2$(9
F- &+2$%'#$< $(* )*$%+/+,"$'2 !"#$%"&$ ('! ' 4(+,* 2'&'3"/"&+ $+0("
)+#'"& +. !"5*%*2$ &"8", +%0'2"#'$"+2# "2&,6!"20 #$%+20 I2- .+%)', +%0'2"#'$"+2', #$%6&$6%* #6&( '# "2!"7
%*#"!*2$#@ '##+&"'$"+2#9 :+4*8*%< (*%* $(* !"#$%"&$ &'$*! 1- '2 +%0'2+0%') 4",, #(+4 !"#&%*/'2&"*# 4"$(
(*',$( &+62&", 4'# 2+ ,+20*% )**$"209 ;(* !"#$%"&$ 4(+ "# %*',,- &+2$%+,,"20 +% )'>"20 &%"$"&', !*&"#"+2#9
(*',$( #*&%*$'%- $++> $(* 8"*4 $('$ $(* '&&+62$# '2! #+ I >*- .'&$+% "# $(* )'2'0*)*2$ #$-,* $(* !"#$%"&$ (*',$(
.+%$( 4*%* #+ &+)/,"&'$*! $('$ *8*2 $(* '&&+62$'2$# #*&%*$'%- '!+/$# "2 $*%)# +. $(* *3$*2$ +. !*,*0'$"+2
('! !"G&6,$"*# 62!*%#$'2!"20 $(*) '2! $(6# $(* "!*' $+< &+2#6,$'$"+2 4"$( +% /'%$"&"/'$"+2 +. +$(*%# "2 !*7
$('$ *"$(*% $(* (*',$( /%+.*##"+2',# +% $(* ,'- )*)1*%# &"#"+27)'>"20 '2! /,'22"20 "2 $(* !"#$%"&$9 ;(* )*$%+7
&+6,! )'>* !*&"#"+2 '1+6$ $(*) 4'# %"!"&6,+6#9 ;(* /+,"$'2 !"#$%"&$ &+6,! 1* $*%)*! /'%$"&"/'$"8* 4"$(
#*&%*$'%- .*,$ $(* "!*' +. ' !"#$%"&$ (*',$( &+62&", 4'# %*0'%! $+ $(* (*',$( #$'5 K$(* *3$*2$ +. /'%$"&"/'$"+2 +.
H6#$ ' 16%*'6&%'$"& "28*2$"+2 '2! (* 4'# 2+$ /%*/'%*! $(* /+/6,'$"+2 "# !"#&6##*! "2 $(* 2*3$ #*&$"+2L9 ;(*
$+ 4'#$* /*+/,*@# $")* +2 "$9 I .6%$(*% %*=*&$"+2 +. $("# (*',$( #-#$*) 4'# .+%)',,- #$%6&$6%*! "2$+ !"5*%*2$
8"*4 +. $(* &+62&", '# ' 4'#$* +. /*+/,*@# $")* "# $('$ !*/'%$)*2$# '2! "2 %*',"$- $(* )'2'0*)*2$ +. $(+#*
4(*2 $(* (*',$( &+62&", ('! *3"#$*! *'%,"*%< "$ ('! !*/'%$)*2$# 4'# !*,*0'$*! $+ $(*"% (*'!#9 M**$"20#
&+)/%"#*! +2,- (*',$( /%+.*##"+2',#< 2+$ ,'- %*/%*#*27 4*%* (*,! %*06,'%,- ')+20#$ (*',$( #-#$*) #$'5 $+ !"#7
$'$"8*#< !*#/"$* ,*0', %*J6"%*)*2$# $+ $(* &+2$%'%-9 &6## /%+0%*## '2! /%+1,*)# '2! #+ .+%$(< 1+$( '$ $(*
;(6# +2,- "2 $(* 6%1'2 #$6!- #"$* !"! $(* &+62&", ,*8*, +. $(* (*',$( #*&%*$'%"'$ '2! 1*$4**2 $(* (*',$(
'//*'% $+ )**$ "2 ,"2* 4"$( "$# '0%**! .+%)', &+)/+7 #*&%*$'%"'$ '2! $(* (*',$( .'&","$- #$'59 ;(* 6%1'2 !"#7
#"$"+2 '2! .62&$"+2#9 I4'%*2*## +. $(* *3"#$*2&* +. $(* $%"&$ &+6,! 1* $*%)*! &+2#6,$'$"8* "2 $('$ $(* (*',$(
&+62&",# 4'# /++% "2 ',, !"#$%"&$#9 :+4*8*%< 2')*! /*%7 #*&%*$'%- )'"2$'"2*! )6&( +. $(* !*&"#"+27)'>"20
#+2# $+ 4(+) /*+/,* 4+6,! $6%2 .+% '!8"&* +% '##"#$7 /+4*% 16$ &+2#6,$*! +$(*% #$'5 )*)1*%# '2! $(* !"#7
'2&* .+% (*',$(7%*,'$*! /%+1,*)# 4*%* +.$*2 ,+&', $%"&$ (*',$( &+62&", +2 )'2- )'$$*%#9 A2 $(* %6%', !"#7
,*'!*%# 4(+ 4*%* ',#+ )*)1*%# +. $(* (*',$( &+62&",#< $%"&$ $(* /+4*% .+% !*&"#"+27)'>"20 4'# >*/$ B%),- 1-
#+ $(* ,'&> +. *3/,"&"$ '4'%*2*## +. $(* &+62&", )'- 1* $(* /%*.*&$ +8*% )+#$ )'$$*%#< '# ',%*'!- "2!"&'$*! "2
175
78 91:)&$%& "1 '(8 ; 7%0)'( 70)"&0" < ="*)0)&" >? @ABBBC D?EFDGD LWE
!"# $%#&'()* *#+!'(,- ."# /#01#%*"'$ *!2/# (3 !"# $%#3#+! *!2/# 1(#* ,(! $%(7(!# 0 1#/#50!'&#: +(,*)/!0!'&# *!2/#
40* !")* "'5"/2 +#,!%0/'*#1 4"'/# !"# *!2/# (3 !"# *#+6 0,1 '* ),/';#/2 !( $%(7(!# *)*!0',0>/# '7$%(&#7#,!*
%#!0%2 (3 "#0/!" 40* /0%5#/2 '%%#/#&0,! 0! !"'* !'7#- ', *#%&'+# $%(&'*'(,- .")* !"# %#/0!'(,*"'$ >#!4##, /#06
1#%*"'$ *!2/#*: +(,*)/!0!'(, 0,1 !"# P)0/'!2 (3 !"#
!"#$%&'()$"* ("'*"#$+), "#0/!" *#%&'+#* ,##1* !( ># #&0/)0!#1-
8%09'/'0, *(+'#!2 !%01'!'(,0//2: /';# 7(*! ', <0!',
=7#%'+0: "0* >##, *!%)+!)%#1 0/(,5 &#%!'+0/ /',#* (3 +/'6 -&*).)*/'( '&* 0%(("01)." 2"+'.)%/# ,'11"#&$
#,!#/'*7 ', 4"'+" ),1#%$%'&'/#5#1 7#7>#%* (3 *(+'#!2 ."# $%#&'()* *#+!'(, 1'*+)**#1 !"# #O#+! (3 $#%*(,0/6
0/'5, !"#7*#/&#* 4'!" !"(*# ', $(4#% (% 0*$'%',5 !( '*#1 /#01#%*"'$ (, !"# 4(%;',5* (3 !"# "#0/!" *#+6
$(4#% 4"( !"#2 4'// *)$$(%! $(/'!'+0//2 ', #?+"0,5# %#!0%'0!- ."# (!"#% *'1# (3 !"# *07# 3#0!)%# '* !"# #O#+!
3(% $#%*(,0/ 30&()%* @A'*#,*!01! B C(,'5#%: DEFGH- (, "#0/!"6*##;',5 >#"0&'()% (3 !"# +0!+"7#,! $($)6
."#*# ,#!4(%;* (3 $0!%(,I+/'#,! %#/0!'(,*"'$* ($#%0!# /0!'(,- Q0,2 #?07$/#* #7#%5#1 1)%',5 !"# J#/14(%;
*$#+'J+0//2 0! !"# /(+0/ /#&#/: *)+" !"0! 0 $%#3#+!: /(+0/ (3 $#($/# %#*(%!',5 !( R$0!%(,*S ', (%1#% !( %#*(/&#
+(),+'//(% (% (!"#% /(+0/ $(/'!'+0/ J5)%# (% 0,2 +0,1'6 !"#'% '77#1'0!# "#0/!" $%(>/#7*- = $0!%(, '* *(7#(,#
10!# !( *)+" $(*'!'(, 4'// ># #?$#+!#1 !( 0**'*! $#($/# 4"( "0* /(+0/ ',T)#,+#: (3!#, ',&(/&#1 ', /(+0/ $(/'!'+*:
!( %#*(/&# 1026!(6102 $%(>/#7* !"%()5" 0 $#%*(,0/'*#1 0,1 ', #?+"0,5# 3(% 30&()%*: !"# R+/'#,!S 4'// 5'&# $(/'!'6
%#/0!'(,*"'$- +0/ *)$$(%! ', #/#+!'(,*- ."# $0!%(, 702 "#/$ 4'!"
<(+0/ #/#+!'(,* !((; $/0+# 1)%',5 !"# J#/14(%; ', !%0,*$(%! #'!"#% !( !"# "#0/!" +#,!%# (% 3)%!"#% 0J#/1 !(
K+!(>#%: DEEL- =, '**)# %0'*#1 40* !"0! 70,2 (3 !"# 0 "(*$'!0/ ', 0,(!"#% 1'*!%'+!: 4'!" 7(,#2 !( $02 3(% 0
$(/'!'+0/ $0%!'#* ', !"(*# 1'*!%'+!* +(,*'1#%#1 7(*! $%'&0!# +(,*)/!0!'(,: #?07',0!'(, (% !%#0!7#,! ,(!
01&0,+#1 ', ($#%0!'(,0/'*',5 !"# "#0/!" %#3(%7* "01 0&0'/0>/# 0! !"# /(+0/ "#0/!" *#%&'+#* (% >2 )*',5 ',T)6
30'/#1 !( 5#! !"#'% +0,1'10!# 3(% $%#3#+! %#6#/#+!#1- #,+# 0! !"# "#0/!" +#,!%# !( #,*)%# 0 +(,*)/!0!'(,
M/#0%/2: (,# #?$/0,0!'(, '* !"0! '! !0;#* !'7# 3(% >#,6 *!%0'5"! 0402- =/!"()5" !"# )*# (3 $0!%(,I+/'#,! ,#!6
#J!* 3%(7 !"# "#0/!" %#3(%7* !( ># $#%+#$!'>/# !( !"# 4(%;* 40* 3(),1 ', 0// !"%## *!)12 *'!#*: !"'* 40* 7(%#
$($)/0!'(,: 4"'/# 0,(!"#% '* !"0! !"# %#3(%7* 702 ,(! 3%#P)#,! ', !"# %)%0/ 1'*!%'+! 4'!" %#50%1 !( !%0,*$(%!
', 30+! ># >#,#J!',5 !"# $($)/0!'(,- N(4#&#%: (,# ,##1* 0,1 P)#)# "($$',5: 0,1 7(%# +(77(, ',
',3(%70,! 1%#4 0!!#,!'(, !( !"# '7$(%!0,+# (3 !"'* !%06 7#!%($(/'!0, 0,1 )%>0, 1'*!%'+!* 4'!" %#50%1 !( $026
1'!'(, (3 $#%*(,0/'*#1 /#01#%*"'$- ."# $($)/0!'(, "0* 7#,!* 3(% $%'&0!# +(,*)/!0!'(,: #?07',0!'(, (% !%#0!6
#?$#+!0!'(,* !"0! !"# $%#3#+! 0,1 *#+%#!0%'#* (3 !"# 7#,!- K&#%0//: %#*(%! !( 0 $0!%(,I+/'#,! ,#!4(%; 40*
1'O#%#,! *#+!(%* 4'// ># $#%*(,0//2 ',&(/&#1 ', ',1'&'16 /#0*! +(77(, ', !"# )%>0, 1'*!%'+!-
)0/ $%(>/#7* 0,1 4'// $#%*(,0//2 ),1#%!0;# !( %#*(/&# ."# '7$(%!0,+# (3 )*',5 ',1'&'1)0/ *!%0!#5'#* !(
!"#7- ."# ',3(%70,! 0%5)#1 !"0! 4"#%# /(+0/ 5(&#%,6 %#*(/&# '77#1'0!# "#0/!" $%(>/#7* '* !"0! !"'* 702
7#,!: ',+/)1',5 !"# "#0/!" *2*!#7: "01 $)! 7(%# 4#// 1'7','*" !"# ,#+#**'!2 !( #,505# ', +(//#+!'&#
#7$"0*'* (, !%2',5 !( ',*!'!)!'(,0/'*# ,#4 $%(+#1)%#* $%#**)%# !( 5#! !"# /(+0/ "#0/!" *#%&'+#* '7$%(&#1-
!( '7$%(&# *#%&'+# P)0/'!2: !"'* "0* ,(! >##, *( "'5"/2 .")*: !"# #?!#,! !( 4"'+" $#($/# 1%04 (, $0!%(,I+/'6
&'*'>/# ,(% *( "'5"/2 &0/)#1- = 1%'&#% #?$/0',#1 !( )* #,! %#/0!'(,*"'$* ', (%1#% !( 0++#** "#0/!" +0%# +0,
"(4 5((1 "# !"()5"! (,# $%#3#+! 40* >#+0)*# "# $#%6 "0&# 0, ',T)#,+# (, !"# P)0/'!2 (3 !"# "#0/!" *#%&'+#*
*(,0//2 ;,#4 4"'+" 1%'&#% *"()/1 ># ()! ', 4"'+" +0% $%(&'1#1- =, 011'!'(,0/ '7$(%!0,! $(',! '* !"0! 4"#%#
(, 0,2 *$#+'J+ 102- A?07$/#* 4#%# 5'&#, !'7# 0,1 $#($/# "0&# *)++##1#1 ', )*',5 *)+" 0, ',1'&'1)0/
!'7# 050', >2 /(4#% +01%#* (3 "#0/!" *!0O: 4"#, *!%0!#52 0,1 %#*(/&#1 !"# "#0/!" $%(>/#7: !"#2 702
#?$/0',',5 4"2 !"#2 +(,*'1#%#1 *(7#(,# 0 5((1 "#0/!" 4#// #?$%#** *0!'*30+!'(, 4'!" !"# "#0/!" +0%# *2*!#7
*#+%#!0%2: 0>()! $#%*(,0/ ',!#%&#,!'(, >2 !"0! $#%*(, 0,1 !")* )*#% #&0/)0!'(,* ,##1 !( +(7$0%# !"# /#&#/*
', "#/$',5 !"#7 !( %#*(/&# 0 4(%; $%(>/#7- ."# '76 (3 *0!'*30+!'(, (3 !"'* 5%()$ 4'!" !"(*# 4'!"()! $0!%(,
$(%!0,+# (3 0 $#%*(,0/'*#1 /#01#%*"'$ *!2/# 40* #&'1#,! "#/$-
', 0// !"%## (3 ()% 1'*!%'+!*- N(4#&#%: ', !"# 7#!%($(/'6
!0, 0,1 )%>0, 1'*!%'+!* !"#%# 40* 0 *!%(,5 *#,*# (3 !"# 3)4"#"&0"$ )& ,"#$%&'( )&.%(."5"&1 6)1+ 1+" (%0'(
/#01#% "0&',5 (>/'50!'(,* !( !"# $($)/0!'(,: 4"#%#0* ', *)$1#)01
!"# %)%0/ 1'*!%'+! !"# $#%*(,0/'*#1 ,0!)%# (3 /#01#%*"'$
40* "(4 !"',5* 4#%# 0,1 ',1'+0!#1 !"# (,/2 402 !( 5#! ."# 1#5%## (3 ',&(/&#7#,! !"0! "#0/!" *!0O "0&#
0,2!"',5 1(,#- 4'!" !"# 1'*!%'+! 702 "0&# 0, ',T)#,+# (, 4"#!"#% 0
K,# (3 !"# 0%5)7#,!* 701# 3(% !"# >#,#J!* (3 *$0+# 3(% 0)!(,(72 '* !%0,*3(%7#1 ',!( 0 7(%# %#6
1#+#,!%0/'*0!'(, '* !"0! 0 /(+0/ 70,05#7#,! /02#% ', *$(,*'&# "#0/!" *2*!#7- ."'* (>*#%&0!'(, %#T#+!* 0
!"# "#0/!" *2*!#7 4'// 0//(4 3(% 5%#0!#% +(,*)/!0!'(, >(12 (3 4(%; 3(//(4',5 <'$*;2S* $'(,##%',5 *!)12 (,
4'!" >(!" *!0O 0,1 (!"#% *#+!(%* 4"'+" ', !)%, 30+'/'6 U!%##! <#&#/ 8)%#0)+%0!* @DEFVH 4"'+" %#+(5,'*#* !"#
!0!#* !"# $/0,,',5 (3 7(%# 0$$%($%'0!# 0,1 %#*$(,*'&# $(4#% (3 0+!(%* 4'!"', *2*!#7* !( %#0/'*#: !%0,*3(%7:
*#%&'+#*- ."# !%01'!'(, (3 0 $#%*(,0/'*#1 /#01#%*"'$ *)>&#%! (% +(7$/#!#/2 >/(+; $(/'+2 ',!#,!'(,*- ."#
176
TUV 12 3%4$'0"' &% +52 6 1"*$+5 1*$&'*& 7 8&.$*$'& 9: ;<===> ?:@A?B?
!"#$% !&"'! &( )*+",'-!" .+-/&0 &($&1-"'$ "2* !"+&3&(4 1*("&(#&"% *? ",' !"-6 2*+3&(4 &( - $&!"+&1" &("'+0&(3!
-($ +'0-"'$ -!5'1"! "* !"-6 &(7*07'8'(" 2&", ",' $&!9 2&", ",'&+ 1*88&"8'(" "* ",' $&!"+&1"K! ,'-0", !%!"'8
"+&1": 1*88&"8'(" -($ 1*("&(#&"%; $'7'0*58'("; B,'+' -+' ",+'' -1"*+! *+ 4+*#5! *?
-1"*+! 2,* 1-( -6'1" ",' 1*("&(#&"% *? ,'-0", 1-+' 5+*9
!"##$%#&'% 7&!&*( 0*1-00%: 0*1-0 4*7'+(8'("@ ",' ,'-0", !'1+'"-+%
<'1'("+-0&!'$ 8-(-4'8'(" &! &( '!!'(1' -=*#" 5+*9 -($ ",' ,'-0", 5+*?'!!&*(-0!;
7&$&(4 ,'-0", 1-+' 2&",&( - $'>('$ 4'*4+-5,&1-0 -+'- L&+!"@ ",' 0*1-0 4*7'+(8'(" &! '0'1"'$ '7'+% ?*#+
2&",&( 2,&1, &" &! -!!#8'$ 0*1-0 &("'++'0-"&*(!,&5! 2&00 %'-+!; G'1"*+ ,'-$! !#1, -! ",' !'1+'"-+% *? ,'-0", -+'
5+*7&$' ",' $+&7&(4 ?*+1' ?*+ +'!5*(!&7'('!!@ -55+*5+&9 -55*&("'$ $&+'1"0% =% ",' 5+'?'1" -($ -+' '&",'+ 5*0&"&9
-"'('!! -($ -11*#("-=&0&"%; A*2'7'+@ ,'-0", 5+*9 1-0 *+ 5'+!*(-0 -00&'! *+ =*",; B,&! "%5' *? -55*&("9
?'!!&*(-0! 8-% ,-7' 0&""0' 0*(49"'+8 &("'+'!" &( ",' 8'(" &! 1-00'$ - M5*!" *? 1*(>$'(1'K H*+,-" .&
!5'1&>1 $&!"+&1"; B,&! 7-+&'$ 4+'-"0% ='"2''( ",' ",+'' *"'/+'*! - I -($ 1-( =' "'+8&(-"'$ -" -(% "&8' 2&",*#"
!"#$% !&"'!; -(% (''$ ?*+ (*"&1'; B,#!@ -00 !'1"*+ ,'-$! -+' 10*!'0%
C00 ",+'' !&"'! 2'+' 2&",&( - ?'2 ,*#+! $+&7' *? ",' -0&4('$ "* ",' 0*1-0 4*7'+(8'(" -($ -00 2&00 1,-(4'
!"-"' 1-5&"-0@ 2,&1, &( &"!'0? 'D'+"! -( &85*+"-(" &(E#9 -?"'+ ?*#+ %'-+! &? ",-" 5-+"% &! (*" +'9'0'1"'$; F( "2*
'(1'; F( ",' 5**+ +#+-0 $&!"+&1"@ (*(' *? ",' ,'-0", 5+*9 *? ",' !"#$% !&"'! H#+=-( -($ 8'"+*5*0&"-(I@ ",' $&!"+&1"
?'!!&*(-0! 0&7'$ &( ",' $&!"+&1" &"!'0?; B,' ,'-0", ,'-0", !'1+'"-+% ,-$ 1,-(4'$ $#+&(4 ",' ?*#+ %'-+! -($
!'1+'"-+%@ 2,* ,-$ =''( &( 5*!" ?*+ '&4," %'-+!@ $&$ 1,-(4'$ -4-&( -?"'+ ",' ?*#+ %'-+!;
,'+!'0? 1*8' ?+*8 ",' $&!"+&1" -($ ?+*8 -( &85*+"-(" G'1*($0%@ -( '0'1"'$ 5+'?'1" -" ",&! "&8' 1*#0$ *(0%
0*1-0 ?-8&0%@ -($ -! !#1, ,-! &("'+'!"! -($ 1*88&"8'(" ,*0$ *N1' ?*+ *(' "'+8 -" - "&8'@ !* '7'( &? ",' !-8'
"* ",' $&!"+&1" =*", &( "'+8! *? 5+*5'+"% *2('+!,&5@ 5-+"% &! +'9'0'1"'$@ ",' 5'+!*( 2,* &! ",' 5+'?'1" 2&00 ='
5*0&"&1-0 -8=&"&*(! ?*+ !'0? -($ +'0-"&7'! -($ &( "'+8! $&6'+'(" -($ ",#!@ 4&7'( ",' 1#0"#+' *? 5'+!*(-0&!'$
*? ?-8&0% "+-$&"&*(!; )*('",'0'!!@ ",' !'1+'"-+% 0&7'$ &( 8-(-4'8'(" $&!1#!!'$ -=*7'@ ",' !'1"*+ ,'-$! 8-%
",' !"-"' 1-5&"-0 -($ 1-8' "* ",' $&!"+&1" *(0% "2* -0!* 1,-(4'; O( ",' *",'+ ,-($@ &( 5**+'+ $&!"+&1"!
"&8'! - 2''3 "* 2*+3 &( ",' !8-00 $&!"+&1" ,*!5&"-0; 2,'+' 5+*?'!!&*(-0! -+' 0&8&"'$@ - !'1+'"-+% 8-% 1*(9
G&8&0-+0%@ ",' 5+'?'1" -" ",' "&8' *? ",' !"#$% ,-$ "&(#' ",+*#4, $&6'+'(" 0*1-0 4*7'+(8'("!; F( *#+ +#+-0
5+*5'+"% &( ",' $&!"+&1"@ =#" - 8*+' &85*+"-(" =#!&('!! !"#$% !&"'@ ",' !'1+'"-+% *? ,'-0", ,-$ -0+'-$% =''( &(
&( ",' !"-"' 1-5&"-0 -($ 2-! ,-+$0% '7'+ !''( &( ",' $&!9 5*!" ?*+ '&4," %'-+! 1*("&(#*#!0% -($ 2-! -55*&("'$
"+&1"; B,' 5+'?'1" '0'1" 2-! 5'+1'&7'$ 8#1, 8*+' -! - ?*+ - ?#+",'+ ?*#+; C! - 5,%!&1&-( ?+*8 -( &85*+"-("
0*1-0 8-( 2,* 2-! 1*88&""'$ "* ",' $&!"+&1" -! ,&! 0*1-0 ?-8&0%@ ",' !'1+'"-+% =+&(4! 5*0&"&1-0 !#55*+" ?*+
,*8'; .% 1*("+-!"@ ",' #+=-( $&!"+&1" ,-$ - 1*+' 4+*#5 2,*'7'+ &! &( 5*2'+ -($ ",#! ,-! =''( -=0' "* ('4*9
*? 5,%!&1&-(! +'!&$'(" &( ",' $&!"+&1" 2,* 7%'$ 2&", *(' "&-"' ",' 5*!" 2&", $&6'+'(" 5-+"&'!;
-(*",'+ ?*+ 5*0&"&1-0 5*!&"&*(!@ 2,* 2*+3'$ &( ",' B,' ",&+$ 4+*#5 *? -1"*+! -+' ",' 5,%!&1&-(! -($
0*1-0 $&!"+&1" ,*!5&"-0 H- 5+&7-"' (*"9?*+95+*>" ?-1&0&"%I (#+!'! 2*+3&(4 &( ",' ,'-0", ?-1&0&"&'!; P-(% ,'-0",
-($ !*8' *? 2,*8 ,-$ ",'&+ *2( 5+&7-"' 10&(&1! &( ",' 5+*?'!!&*(-0! &( .+-/&0 2&00 ,-7' 8*+' ",-( *(' J*=@ -
$&!"+&1" -0!*; )#+!'! -($ *",'+ ,'-0", !"-6 -0!* 0&7'$ &( "%5&1-0 1*("+-1" ='&(4 ?*+ "2'("% ,*#+! - 2''3;
",' $&!"+&1" 5'+8-('("0%; C ?-+ 4+'-"'+ 0*(49"'+8 1*89 C0",*#4, 8*!" 2*+3 2&",&( ",' 5#=0&1 !'1"*+ ?*+ 5-+"
8&"8'(" "* ",' $&!"+&1" 2-! -55-+'(" -8*(4!" ",' *? ",' "&8'@ ",' !'1*($ 1*("+-1" 8-% (*" =' &( ",'
,'-0", !"-6 &( ",&! $&!"+&1"; F( ",' 8'"+*5*0&"-( $&!"+&1"@ !-8' $&!"+&1" -($ 8-% (*" =' &( ",' 5#=0&1 !'1"*+;
8-(% *? ",' ,'-0", !"-6 ,-$ =''( 2*+3&(4 ",'+' ?*+ P*=&0&"% ='"2''( J*=! '7'( 2&",&( ",' 5#=0&1 !'1"*+
8-(% %'-+! -0",*#4, (*" ",-" 8-(% 2'+' -1"#-00% +'!&9 1-( =' ,&4,@ 5-+"&1#0-+0% &( $&!"+&1"! ('-+ 0-+4' #+=-(
$'(" &( ",' $&!"+&1"@ ='&(4 -! &" 2-! -$J-1'(" "* ",' 1'("+'!; B,' 1*("&(#&"% *? ",' ,'-0", 5+*?'!!&*(-0! &(
!"-"' 1-5&"-0; B,' !"-6 ",#! ?'00 &("* "2* 1-85!@ ",*!' -(% *(' $&!"+&1" 8-% ",'+'?*+' 7-+% '(*+8*#!0%; B,'+'
2,* ,-$ 2*+3'$ ,'+' 0*(49"'+8 -($ ",*!' (*"; C !&8&9 -+' -0!* $&6'+'(" "%5'! *? 1*("+-1"! ?*+ ,'-0", 5+*9
0-+ 5-""'+( 2-! !''( ='"2''( ",' "2* $&!"+&1" !'1+'"-+&'! ?'!!&*(-0!@ 2,&1, ,-7' $&6'+'(" &850&1-"&*(! ?*+ 0&3'0%
*? ,'-0", -55*&("'$ $#+&(4 ",' +'!'-+1,; B,' !'1+'"-+% 1*("&(#&"% &( -(% *(' $&!"+&1"; Q*("+-1"! 8-% =' ,'0$
*? ,'-0", ?*+ 8*!" *? ",' !"#$% 2-! ('&",'+ ?+*8 ",' -" L'$'+-0@ G"-"' *+ <&!"+&1" 0'7'0; R("+% &("* ",' 5#=0&1
$&!"+&1" (*+ +'!&$'(" &( ",' $&!"+&1"@ -0",*#4, 2-! +'4#9 !5,'+' &! *N1&-00% 7&- - 3&($ *? 1&7&0 !'+7&1' 'D-8&(9
0-+0% ",'+' -($ 1*(!1&'("&*#! &( ,&! 2*+3@ 2,'+'-! ",' -"&*( H1-00'$ - *"'*(,0" I 2,&1, *(1' 5-!!'$ -!!#+'! -
!'1+'"-+% -55*&("'$ "* "-3' *7'+ 2&", ",' &(91*8&(4 J*= ?*+ 0&?'@ -0",*#4, ",' !5'1&>1 5*!"&(4 -($ -!!*1&9
0*1-0 4*7'+(8'(" ,-$ - 0*(4 ,&!"*+% *? 2*+3&(4 &( ",' -"'$ 5'+3! -($ =*(#!'! 8-% 7-+%; A*2'7'+@ 7'+% ?'2
$&!"+&1" -($ ,-$ -0+'-$% =''( ",' ,'-0", !'1+'"-+% $&!"+&1"! ,-7' '!"-=0&!,'$ - *"'*(,0" 5+*1'$#+' -($
#($'+ - 5+'7&*#! 0*1-0 4*7'+(8'("; ",#! !"-6 '850*%'$ =% ",' $&!"+&1" $* (*" ,-7' !#1,
J*= !'1#+&"%; S,'+' ",'+' &! $+-8-"&1 1,-(4' *? -00 "*5
!"'%$'($%) -$8&(&!"+-"&7' !"-6 '7'+% ?*#+ %'-+!@ 2,'+' 5+-19
F" &! '7&$'(" ?+*8 ",' *=!'+7-"&*(! -=*7' ",-" ",' "&"&*('+! -+' $'5'($'(" *( ",' 4**$2&00 *? ",' ,'-0",
177
34 5'6"&(,& $' */4 7 3,+"*/ 3+"$&+$ 8 9$0"+"&$ :; <=>>>? @;AB@C@ C\A
!"#$"%&$' ()$ %*"+$ #),%$&#% #),%+,-+,. &,/ 0*"$" %*" &! & 3"34"$ )( %*" #)33-,+%'7 Q"&1%* !%&9 ), %*"
*"&1%* 2"$!),,"1 &$" *+.*1' 3)4+1"5 %*" 2$)4&4+1+%' )( 0*)1" /+/ ,)% %*+,F %*+! &##"2%&41"7 8*" #)33-,+%'
.))/ 2$&#%+#"! +, *"&1%* #&$" /"1+6"$' #),%+,-+,. )6"$ +,()$3&,%! &,/ %*" #)33-,+%' *"&1%* 0)$F"$! *&/
%+3" 0+11 4" /"2",/",% ), %*" &#%+), )( +,/+6+/-&1! 3+I"/ (""1+,.! &4)-% +%7 8*" +!!-" 0&! 3&+,1' $&+!"/ +,
$&%*"$ %*&, ), 2$)#"/-$"! 4"+,. +,!%+%-%+),&1+!"/ +,%) %*" $-$&1 /+!%$+#%7 WI&321"! &4)-,/ )( #"$%&+, +,/+6+/:
%*" 1)#&1 !'!%"37 8*" (&#%)$! &9"#%+,. 0*"%*"$ & 1)#&1 -&1! F""2+,. 3"/+#+,"! &% *)3" 0*+#* %*"' /+!%$+4-%"
!'!%"3 %*&% 2$)3)%"! $"!2),!+6" &,/ .))/ 2$&#%+#" %) 1)#&1 2")21" 0*) F,)0 %) #)3" &,/ !""F *"12 %*"$"7
1)#&11' 0+11 4" !-!%&+,&41" &#$)!! %*" 6&.&$+"! )( 1)#&1 >-#* 2$&#%+#" +! )(%", #&$$+"/ )-% 0+%* & 2)1+%+#&1 &+3
2)1+%+#! *&! $"#"+6"/ 1+%%1" &%%",%+), +, $"!"&$#* %):/&%"7 %) .&+, (&6)-$ 1)#&11'5 %*" /$-.! 3&' #)3" ($)3 %*"
2-41+# *"&1%* 2)!%! &,/ %*" 2")21" /+!%$+4-%+,. %*"3
!"#$%$&' ()*+$( ,- *++$)'*./$ )%*+'"+$ *&0 &$" -,%$&+,"/5 &,/ '"% +, $-$&1 &$"&! 0*"$" %*"$" &$"
*++,1&'*."/"'2 ("0 *"&1%* 2)!%! &,/ 1+3+%"/ %$&,!2)$%5 %*+! 2$&#%+#"
#&, 4" *+.*1' &/6&,%&.")-! ()$ %*" 1)#&1 2)2-1&%+),7
; <,&1 #),!+/"$&%+), 0+%* $"!2"#% %) !)#+&1 )$.&,+!: 8*+! 2$&#%+#" 0&! #)33), +, %*" $-$&1 &$"& &,/ ,)
&%+), &,/ 2)1+%+#&1 #-1%-$" +! %) /"<," 0*"$" /+9"$",% 3"34"$! )( %*" 2)2-1&%+), %*)-.*% %*+! -,&##"2%&41"7
.$)-2! )( &#%)$! /"1+,"&%" %*" 4)-,/&$+"! )( -,&##"2: ; ,-34"$ )( %*" *"&1%* !%&9 )4V"#%"/ 2&$%+#-1&$1'
%&41" 2$&#%+#" 4' 2-41+# "321)'""! 0+%*+, %*" *"&1%* !+,#" %*" *"&1%* 2)!%! %*"3!"16"! *&/ ("0 /$-.! +,
!'!%"37 =, %*" >%&%" )( ?"&$&@5 &(%"$ %*" ABBC 1)#&1 .)6: !%)#F7
"$,3",% "1"#%+),!5 <(%' )-% )( %*" ADE /+!%$+#%! 0"$" ;1%*)-.* & 6"$' #)321"I &$"& 3"$+%+,. /"%&+1"/
%&F", %) %*" >%&%" 8$+4-,&1 -,/"$ !-!2+#+), )( 3&1: !%-/' +, +%! )0, $+.*%5 +% +! 0)$%* %$'+,. %) &##"!! !)3"
2$&#%+#" %) &,!0"$ G-"$+"! $".&$/+,. %*"+$ &##)-,%!7 +,/+#&%+), )( 6&$+&%+), 4"%0"", /+!%$+#%! +, %*" ,)$3!
?&!"! !-#* &! %*"!" 4"#)3" *+.*1' 6+!+41" 4-% )%*"$: )( &##"2%&4+1+%' )( !-#* 2$&#%+#"! &,/ +%! $"1&%+),!*+2
0+!" +/",%+('+,. 1)#&1 #)$$-2%+), +, $"!"&$#* !%-/+"! +! %) *"&1%* #&$" 2$)6+!+),7 =, %*+! !%-/'5 %*"$" 0&! &
&13)!% +32)!!+41" %) /) 0+%*+, & 2$&#%+#&1 %+3" 1+3+%7 6"$' 4$)&/ %)1"$&,#" %)0&$/! &11 F+,/! )( 2$&#%+#"!
H, %*" )%*"$ *&,/5 6&$+&%+),! +, ,)$3! &,/ 6&1-"! 4"'),/ )T#+&1 2$)#"/-$"! +, %*" $-$&1 /+!%$+#%7 8*"
$".&$/+,. 3)$" #)%+/+&, 2$&#%+#"! #&, 4" "6&1-&%"/ &,/ 3"%$)2)1+%&, &$"& 0&! !)3"0*&% %)1"$&,% &,/ %*"
%*"+$ $"1&%+), %) *"&1%* 2$)6+!+), "I21)$"/ J!"" >*"&9 -$4&, /+!%$+#% 2$)/-#"/ $"1&%+6"1' ("0 #&!"!7
K L"!%5 ABBM ()$ &, "I&321" )( & !%-/' 3&/" +, %*"
NO P&%+),&1 Q"&1%* >"$6+#"R7 S-$+,. %*" <"1/0)$F 0"
$"#)$/"/ 6&$+)-! 2$&#%+#"! +, 1)#&1 *"&1%* !'!%"3! %*&% !"#$% &"'$ (" ()* +",-+
+, 2$+,#+21" 6+)1&%" )T#+&1 2$)#"/-$"!7 ;1%*)-.* #&!"!
#&, &10&'! 4" ()-,/ )( 41&%&,%1' #)$$-2% 2$&#%+#"!5 !"$: 8*" /"!#$+2%+),! &4)6" +11-!%$&%" 0&'! +, 0*+#*
6+,. ),1' %*" +,%"$"!%! )( #"$%&+, 3"34"$! )( %*" !'!: &!2"#%! )( 1)#&1 !)#+&1 )$.&,+!&%+), &,/ 2)1+%+#&1 #-1%-$"
%"35 3)!% (&11 +,%) & (&$ .$"'"$ &$"& "%*+#&11' $".&$/+,. #&, /+9"$",%1' &9"#% %*" +321"3",%&%+), )( *"&1%*
&##"2%&4+1+%'7 $"()$3! &,/ %*" 2$)#"!!"! 4' 0*+#* !-#* $"()$3!
U$&#%+#"! !-#* &! & *"&1%* 2$)("!!+),&1 $"("$$+,. & +32$)6" *"&1%* #&$" G-&1+%' &#$)!! /+!%$+#%!7 =, %*" #&!"
2&%+",% %) %*"+$ )0, 2$+6&%" #1+,+# #&, 4" +, %*" +,%"$: )( X$&Y+15 %*" %0) 3&+, $"()$3 !%$&%".+"! &$" %)
"!%! )( %*" 2&%+",% &! 0"11 &! %*" *"&1%* 2$)("!!+),&17 +,#$"&!" !2&#" ()$ 1)#&1 &-%),)3' &,/ 1)#&1 6)+#"7 8*"
8*+! )( #)-$!" #&, ),1' )##-$ 0*"$" *"&1%* 2$): "I%",% %*&% !-#* !2&#"! &$" #$"&%"/ +! #1"&$1' /"2",/",%
("!!+),&1! *&6" & 2$+6&%" #1+,+# +, %*" /+!%$+#%7 8*+! 0&! ,)% ),1' ), %*" ()$3&1 #),%$&#% 0+%* %*" Z"/"$&1
%*" #&!" ()$ %*" -$4&, &,/ 3"%$)2)1+%&, /+!%$+#%! &,/ N,+), 4-% &1!) ), & $&,." )( 3)$" 1)#&11' /"%"$3+,"/
,):)," ($)3 "+%*"$ %*" !%&9 )$ %*" 2)2-1&%+), #$+%+#+!"/ (&#%)$!7 8*" 2$)#"!!"! 4' 0*+#* %*"!" !%$&%".+"! &$"
%*+! 2$&#%+#"7 L" *&6" &1$"&/' #)33",%"/ &4)6" ), &!!-3"/ %) %$&,!()$3 +,%) .$"&%"$ "32)0"$3",%5
%*" 0+/"!2$"&/ 2$&#%+#" 0*"$"4' %*)!" 0+%* 2)1+%+#&1 &##)-,%&4+1+%'5 $"!2),!+6","!! &,/ G-&1+%' )( *"&1%*
&!2+$&%+),! *"12 2)%",%+&1 !-22)$%"$! .&+, &##"!! %) #&$" &$" 3)/"$&%"/ &% "6"$' 2)+,% 4' %*" 1)#&1 !)#+&1
*"&1%* !"$6+#"! &,/ *"&1%* 2$)("!!+),&1!7 8*" &##"2: )$.&,+!&%+), &,/ 2)1+%+#&1 #-1%-$" +, 0*+#* %*" 1)#&1
%&,#" )( %*+! 0&! 3-#* 3)$" &34+6&1",%5 /"2",/+,. )( *"&1%* !'!%"3 +! "34"//"/7 8*" 1+,"! )( +,[-",#" &$"
#)-$!" ), 0*"%*"$ ')- *&/ &##"!! %) !-#* & 2&%$), )$ !*)0, +, Z+.7 \7
0*"%*"$ ')- 3+!!"/ ."%%+,. & #),!-1%&%+), 4"#&-!" =, )$/"$ %) 2$)6+/" & !+321" !-33&$' )( %*" %*+#F
!)3")," V-32"/ +, &*"&/ )( ')-7 8*" 3&V)$ #)3: /"!#$+2%+), )( !)#+&1 )$.&,+!&%+), &,/ 2)1+%+#&1 #-1%-$"
21&+,%! #&3" ($)3 %*" $-$&1 /+!%$+#% &,/ !)3" ($)3 %*" &,/ %*" &!2"#%! )( ()$3&1 )$.&,+!&%+), &,/ *"&1%* #&$"
3"%$)2)1+%&,7 8*"$" 0"$" ("0 ($)3 %*" -$4&, /+!%$+#%7 G-&1+%' &1$"&/' 2$"!",%"/ +, 8&41" \5 %*" %*$"" /+!%$+#%!
8*" -!" )( #)33-,+%' *"&1%* 0)$F"$! 4' 2)1+%+#+&,! +, &$" $&,F"/ ()$ %*"+$ 2"$()$3&,#" ), &11 )( %*"!" &,/
2)1+%+#&1 #&32&+.,! +! ,)% &11)0"/ )T#+&11'5 '"% %*)!" 2$"!",%"/ +, 8&41" E7 8*" !"1"#%+), )( +,/+#&%)$! %)
0)$F"$! 3&' *&6" 2)1+%+#&1 6+"0! )( %*"+$ )0, &,/ $"2$"!",% %*" ()$3&1 )$.&,+!&%+), &,/ %) "6&1-&%" G-&1:
0&,% %) #&32&+., ()$ & 2)1+%+#&1 2&$%' +, %*"+$ #&2&#+%' +%' )( %*" #&$" +! %) & 1&$." 2&$% /"%"$3+,"/ 4' 0*&%
178
VPW !" #$%&'()' *$ +," - !).&+, !.&*'.* / 0*1&.&'* 23 456667 839:8;8
! E !'.#$'/# #-"# 86#. #-% *+$,"( +$F"4'."#'+4"( .#$6/2

#6$%. '4#+ 8("/% ,"D '4@6%4/% #-% (+/"( .+/'"( +$F"42
'."#'+4 "4! 8+('#'/"( /6(#6$% #+?"$!. " ,+$%
%4")('4F %4&'$+4,%4# *+$ ',8(%,%4#"#'+4=
! J%'#-%$ +4% '. #-% 8$',"$D !%#%$,'4"4# +* #-% +#-%$K
#-%$% '. " /(+.% !'"(+F6% )%#?%%4 #-%, #-"# +8%$"#%.
". " *%%!)"/L (++8 ?-'/- /"4 %./"("#% #-% F++! +$
)"! %I%/#. +* (+/"( -%"(#- .D.#%, "/#'&'#'%.=
>-% B$.# +* #-%.% #-$%% +8#'+4. '. -'F-(D 8%..','.#'/=
J+2+4% ?+$L'4F ?'#-'4 -%"(#- $%*+$,. ?+6(! /"$% #+
$%"/- #-'. /+4/(6.'+4 "4!0 ". M+..%$# 79::N< 4+#%!
7/'#%! ")+&%<0 .6/- " /+4/(6.'+4 ?+6(! "(.+ )% 8+('#'2
/"((D 64"//%8#")(%= >-% 64"//%8#")'('#D +* .6/- " /+42
/(6.'+4 #+ $%.%"$/-%$.0 8$"/#'#'+4%$. "4! 8+('#'/'"4. '.
4+#0 -+?%&%$0 " .6H/'%4# $%".+4 #+ !%4D #-"# '# '. "
8+..')'('#D= >-% $%.6(#. *$+, +6$ #-$%% /".% .#6!'%. /"4
)% '4#%$8$%#%! '4 #-'. ?"D= >-% .%/+4! +* #-% +8#'+4. '.
-+? #-% -%"(#- $%*+$,. "$% 8$%.%4#%! '4 8+('/D $-%#+$'/
"4! '4 8$"/#'/% #-'. '. #-% ,+!%( ,+.# 8$"/#'#'+4%$. "4!
-%"(#- .D.#%,. $%.%"$/-%$. ?+$L ?'#-'4 ',8('/'#(D= >-%
$%.6(#. +* #-'. .#6!D .6FF%.#0 4+# .6$8$'.'4F(D0 #-"# #-'.
'. +&%$(D +8#','.#'/= >-% %G8%$'%4/%. +* #-% (+/"( -%"(#-
.D.#%, '4 #-% $6$"( "$%" '4 8"$#'/6("$ !%,+4.#$"#% -+?
Q'F= P= >-% '4@6%4/% +* .%(%/#%! ".8%/#. +* .+/'"( +$F"4'."#'+4 !%%8(D %4#$%4/-%! ".8%/#. +* .+/'"( +$F"4'."#'+4 "4!
"4! 8+('#'/"( /6(#6$% +4 #-% )%4%B#. +* !%/%4#$"('.%! ,"4"F%2 8+('#'/"( /6(#6$% /"4 !+,'4"#% -+? #-% -%"(#- $%*+$,.
,%4# +* -%"(#- /"$%= "$% ',8(%,%4#%!= >-% #-'$! +8#'+4 ,"D +I%$ #-% ,+.#
8$+,'.% '4 '#. +8#','.#'/ /"6#'+4= 3%$% #-% ',8+$#"4/%
+* (+/"( *"/#+$. '. $%/+F4'.%! "4! #-% ('L%('-++! #-"#0
?'#-+6# /"$%0 !%/%4#$"('."#'+4 ,"D .%$&% #+ '4/$%".%
!'I%$%4/%. )%#?%%4 !'.#$'/#. $"#-%$ #-"4 #-% +88+.'#%0
!"#" "$% "&"'(")(% *$+, #-% !'.#$'/#.0 #-% 1#"#% 1%/2
". !+/6,%4#%! )D O6#4", '4 5#"(D 79::P<= 54 #-'.
$%#"$'"# +* 3%"(#- "4! #-% 1#"#% 1#"#'.#'/"( 54.#'#6#%
.#6!D0 #-%$% "$% '4!'/"#'+4. #-"# #-% "/#'&'#'%. +* #-%
758("4/%0 9::;<= >-%.% !"#" !+ -+?%&%$ $%@%/# #-%
-%"(#- .D.#%, '4 #6$4 %,8+?%$ #-% (+/"( 8+86("#'+4 #+
.#"#%A. +?4 !%B4'#'+4. +* )+#- C6"('#D "4! #-% ',8+$#2
%G8%/# " F'&%4 C6"('#D +* -%"(#- /"$%0 )6# +4(D '4 #-%
"4# ".8%/#. #+ #-% *+$,"( .D.#%,= E /(%"$ 8"##%$4
6$)"4 !'.#$'/# ?-%$% #-% .+/'"( +$F"4'."#'+4 "4! 8+('#'2
%,%$F%. "/$+.. #-%.% #-$%% !'.#$'/#.=
/"( /6(#6$% +I%$ " *"/'('#"#'4F %4&'$+4,%4#= 3+?%&%$0
>-% 6$)"4 "4! $6$"( !'.#$'/#. /+,% +6# /+4.'.#%4#(D
#-% !'"(+F'/ 4"#6$% +* #-% "..+/'"#'+4 !+%. #-%+$%#'/"((D
?%(( "4! )"!(D $%.8%/#'&%(D "/$+.. "(( #-$%% /"#%F+$'%.=
(%"&% $++, *+$ '4@6%4/% #+ F+ '4 )+#- !'$%/#'+4.= 54
>-% ,%#$+8+('#"4 !'.#$'/# '. $"#-%$ ,+$% '4#%$%.#'4F '4
#-'. +8#'+40 #-%$%*+$%0 #-%$% /"4 )% .8"/% *+$ #-% *+$2
#-"# #-% *+$,"( +$F"4'."#'+4 '. /+,8"$")(% ?'#- #-%
,"( -%"(#- .D.#%, #+ '4@6%4/% (+/"( .+/'"( +$F"4'."#'+4
6$)"4 !'.#$'/# )6# #-% C6"('#D +* /"$% '. .6$8$'.'4F(D
"4! 8+('#'/"( /6(#6$% "4! +I%$ " 8+#%4#'"( *+$ /-"4F%=
8++$= >-% %G#%4# #+ ?-'/- ".8%/#. +* .+/'"( +$F"4'.2
54 +$!%$ #+ #"L% "4"(D.%. +4 #-%.% L'4!. +* C6%.#'+4.
"#'+4 "4! 8+('#'/"( /6(#6$% %4")(% +$ -'4!%$ ',8(%,%42
*+$?"$!0 #?+ L'4!. +* *6$#-%$ $%.%"$/- "$% 4%%!%!=
#"#'+4 '4!'/"#%. " ,'G%! '4@6%4/% )6# +4% ?-'/- '.
Q'$.#0 ?-%$% .+,% '!%" +* #-% ',8+$#"4# ".8%/#. +*
.6H/'%4#(D 4%F"#'&% #+ .6FF%.# #-"# (+/"( .+/'"( +$F"4'.2
.+/'"( +$F"4'."#'+4 "4! 8+('#'/"( /6(#6$% '. L4+?40
"#'+4 "4! 8+('#'/"( /6(#6$% ,%!'"#% #-% F"8 )%#?%%4
%G#%4.'&% .6$&%D. '4/+$8+$"#'4F '4!'/"#+$. *+$ #-%.% /"4
*+$,"( +$F"4'."#'+4 "4! $%.6(#'4F C6"('#D=
'!%4#'*D *"/#+$. "..+/'"#%! ?'#- F++! -%"(#- /"$% 8$+2
>-%.% +).%$&"#'+4. )%#?%%4 #-% #-$%% /"#%F+$'%. +*
&'.'+4 "/$+.. " ?'!% $"4F% +* !'.#$'/#.= E ('.# +* '4!'2
"/#'+4 $"'.% /"6.% "4! %I%/# C6%.#'+4. ")+6# #-% 4"#6$%
/"#+$. #-"# -"&% )%%4 6.%! '4 J+$#-%".# M$"R'( )".%!
+* #-%.% 8"##%$4.=
+4 #-'. .#6!D '. F'&%4 '4 >")(% S= E .#6!D +* #-'.
! E4 %4&'$+4,%4# '4 #%$,. +* .+/'"( +$F"4'."#'+4 "4! 4"#6$% ?+6(! "', #+ /(6.#%$ !'.#$'/#. )D #-% F%+!%,+2
8+('#'/"( /6(#6$% #-"# *"&+6$. -%"(#- $%*+$, ',2 F$"8-'/ "4! .+/'"( +$F"4'."#'+4T8+('#'/"( /6(#6$% '4!'2
8(%,%4#"#'+4 '. #-% ,"'4 !%#%$,'4"4# +* #-% *+$,"( /"#+$. "4! %G8(+$% ?-%#-%$ /%$#"'4 /+4.#%(("#'+4. +*
+$F"4'."#'+4 +* #-% (+/"( -%"(#- .D.#%,0 .6//%..*6( *+$,"( +$F"4'."#'+4"( "$$"4F%,%4#. %,%$F% ". "..+/'2
8$"/#'/% "4! $%.6(#'4F C6"('#D +* /"$%= "#%! ?'#- F++! C6"('#D +* /"$%= U%.%"$/- +6#86#. '4
179
"% &'()!*+! ,' -.% / "+0)-. "0),!0, 1 $,2)0)!, 34 567778 94:;9<9 eYY
;*,9- U
V*+:#+3 &% !"- !"'-- /#$!'#)!$ %&' W.*9#!12 %&'(*9 &'3*+#$*!#&+ *+/ $&)#*9 &'3*+#$*!#&+ *+/ 8&9#!#)*9 ).9!.'-
G-!'&8&9#!*+ X',*+ V.'*9
!"#$%&' ("&)"&*
7GV Y H Z
G-*$9-$ )&5-'*3- H H Y
[\)#-+)1 #+ .$- &% %.+/$ H Y Z
L9#+#)*9 )&+$.9!*!#&+$ Y H Z
L9#+#)*9 8'&)-/.'-$ Z H Z
F-+!*9 )&+$.9!*!#&+$ Z H Y
V*/#&9&31 Z H Z
]&8.9*!#&+ $*!#$^-/ D! _ H`` #+ -*)"O Y H Z
"#$ HN H` HT
+(,-#$ *'*&.-
a-3*9 $!*!.$ Z H Z
Q'3*+#$*!#&+b E-*9!" )&.+)#9 Y H H
@-)'-!*'#*! $!'.)!.'- H Z Y
V-$&.')-$
?*)#9#!#-$b +&c$#C-c$-'5#)-$ H Z Y
E.(*+ '-$&.')- )*9#,'- H Z Y
d&6 B-/$c)*8 Z H Z
d&6 LEM H Z Y
d&6 ?*(#91 "-*9!" !-*($ Z H Z
E-*9!" %*)#9#!1 %.+)!#&+#+3 H H Z
"#$ HU HY ZH
/(0%#$ (,1#2%*#&%(2 #23 )($%&%0#$ 0"$&",.
X+!#-/ %.+/#+3 H Z Y
a&)*9 3&5-'+(-+! /-9-3*!#&+ H H Y
a&)*9 #+%&'(*!#&+ $1$!-( H H Y
LG@ %.+)!#&+ Y H Z
L9#-+!-9#$( %&' "-*9!" )*'- Z H Y
G*+*3-(-+! $!19- Z H Y
]-'$&+*9#$-/ 9-*/-'$"#8 H H Z
L&((#!(-+! Z H Y
L&+!#+.#!1 Z H Y
B&.+/*'#-$ &+ 8'*)!#)- Z H Y
"#$ HN HH ZT
!"#$ %&'( )*+ ,- .$-/ #+ !0& 0*1$2 -)"&#+3 !"- /#$).$4 &'3*+#$*!#&+ *+/ 8&9#!#)*9 ).9!.'-6 ?&'(*9 *)!#5#!#-$ &%
$#&+ *,&5-6 7+ * (&'- 8-$$#(#$!#) *88'&*)"2 !"- '-$.9!$ !"- "-*9!" $1$!-( (#3"! #+ !.'+ 89*1 *+ #(8&'!*+! 8*'!
)*+ 8'&5#/- * 9#$! &% &8!#&+$ %&' &'3*+#$*!#&+*9 #+ $.)" )"*+3-$6
*''*+3-(-+!$ !"*! $--( !& 0&': 0-99 #+ )-'!*#+ !18-$ @-)&+/912 #+!-+$#5- )*$- $!./#-$ $#(#9*' !& !"-$- #+
&% $-!!#+3$ %&' &!"-' /#$!'#)!$ !& /'*0 .8&+6 ;"#$ /#A-'-+! )&.+!'#-$ )*+ #/-+!#%1 * )&(8*'*!#5- 9#$! &%
*$$.(-$ !"*! &+91 !"- %&'(*9 *$8-)!$ &% "-*9!" $1$!-( *$8-)!$ &% $&)#*9 &'3*+#$*!#&+ *+/ 8&9#!#)*9 ).9!.'- !"*!
(*+*3-(-+! *'- '-*991 *(-+*,9- !& )&+!'&99-/ )"*+3- -(-'3- *$ #(8&'!*+! %&' "-*9!" $1$!-($6 B'*C#92 #! "*$
*+/ !"*! $&)#*9 &'3*+#$*!#&+ *+/ 8&9#!#)*9 ).9!.'- "*5- ,--+ +&!-/ ,-%&'- ,1 &!"-' 0'#!-'$2 (*1 ,- 8*'!#).9*'91
!& ,- !'-*!-/ *$ 8*'! &% *+ .+*9!-'*,9- 3#5-+ )&+!-<!6 7! */-8! *! (&/#%1#+3 *+/ !'*+$%&'(#+3 %&'(*9 &'3*+#$4
#$ 5#!*92 *! !"- 5-'1 9-*$!2 !& (*:- -<89#)#! '-)&3+#!#&+ *!#&+*9 $!'.)!.'-$ *+/ 8'&)-/.'-$ 0#!"#+ #!$ 8&9#!#)*9
&% %*)!&'$ !"*! *'- +&! *(-+*,9- !& (*+*3-/ )"*+3- ).9!.'- DE-$$ *+/ F*G*!!*2 HIIJK L*9/*$ *+/ M&&/2
,1 !"- "-*9!" $-)!&'6 = (&'- )*.!#&.$91 &8!#(#$!#) HIINO6
*88'&*)" #$ !& .$- !"- '-$.9!$ *$ *+ &88&'!.+#!1 !& PG*+1 )*!-3&'#-$ *+/ *$$.(8!#&+$ +&'(*991
&8-+ * /#*9&3.- 9&)*991 0#!" !"- 8&8.9*!#&+2 0#!" 8&94 -(89&1-/ #+ Q'3*+#C*!#&+ @!./#-$ (*1 ,- &% 9#!!9-
#!#)*9 9-*/-'$ *+/ $& %&'!" *$ !& 0"*! #$ *+/ #$ +&! .$- &' -5-+ +&! *889#)*,9- 0#!"#+ !"- B'*C#9#*+ )&+4
*(-+*,9- !& (*+*3-/ )"*+3-2 *+/ 0"-!"-' *+/ 0"*! !-<!6RDL*9/*$ S M&&/2 HIIN2 86JHTO
*)!#5#!#-$ &.!$#/- !"- >.'#$/#)!#&+ &% !"- "-*9!" $1$!-(
(#3"! ,- #+$!#3*!-/ !& 8'&5&:- )"*+3- #+ 9&)*9 $&)#*9 E&0-5-'2 L*9/*$ *+/ M&&/ *9$& +&!- !"*! !"-1 )*+
18 0
\T] !" #$%&'()' *$ +," - !).&+, !.&*'.* / 0*1&.&'* 23 456667 839:8;8
!"#$% &
'()*+",-./ -0 "/1%+,/ -0 /-+*"$ -.2"(*/",*-( "() 1-$*,*+"$ +3$,3.% *( ,4% +-(,%5, -0 ,4% 6."7*$*"( 4%"$,4 /8/,%9 .%0-.9/
:%(%."$ "/1%+, ;1%+*<+ "/1%+, '()*+",-./
;1"+% 0-. "3,-(-98 *( 1$"((*(2 "() )%+*/*-(= ;-3.+% -0 *(+-9% ? -0 )*/,.*+, 4%"$,4 #3)2%, 0.-9 @-AB;'CD
9">*(2 ;E;F
? -0 4%"$,4 #3)2%, 0.-9 )*/,.*+, *,/%$0
G-+"$ 2-H%.(9%(, !-,"$ #3)2%, -0 )*/,.*+,
? -0 ,-,"$ )*/,.*+, #3)2%, "$$-+",%) ,- 4%"$,4
I4- )%+*)%/ 3/% -0 4%"$,4 #3)2%,
I4- "11.-H%/ 3/% -0 4%"$,4 #3)2%,
I4- ".% ,4% /*2(",-.*%/ ,- ,4% 4%"$,4 03()
'(0-.9",*-( $-+"$$8 0-. C2.%%9%(, -( '@JK @@JK +-H%."2% $%H%$/
1$"((*(2
C2.%%9%(, -( L4%.% 1.-#$%9 ".%"/ ".%
M5,%(, -0 )"," 3/% *( $-+"$ 1.*-.*,8 /%,,*(2
;1"+% 0-. " $-+"$ H-*+% *( 1$"((*(2 N".,*+*1",*-( M5*/,%(+% -0 +-993(*,8=#"/%) -.2"(*/",*-(/
M5,%(, -0 +-993(*,8 1".,*+*1",*-( *( ,4-/%
-.2"(*/",*-(/
J%1.%/%(,",*-( -0 +-993(*,8 -.2"(*/",*-(/ -(
,4% )*/,.*+, 4%"$,4 +-3(+*$
CL".%(%// -0 +-993(*,8 9%9#%./ -0 ,4%
)*/,.*+, 4%"$,4 +-3(+*$
N%.+%*H%) 03(+,*-(/ -0 ,4% +-3(+*$
O.%P3%(+8 -0 +-3(+*$ 9%%,*(2/
C,,%()"(+% ", +-3(+*$ 9%%,*(2/
I4-DL4%.% 1%-1$% L-3$) 2- *0 L"(,%) ,-
+-91$"*( "#-3, /%.H*+%/
N%./-("$*/%) "() *(/,*,3,*-("$*/%) *(Q3%(+%/ -( @"("2%9%(, /,8$% N%.+%1,*-( -0 4%"$,4 /,"R -0 /%+.%,".8S/ /,8$%
"3,-(-98 "() $-+"$ H-*+%
I4- 9">%/ )%+*/*-(/ "#-3, BTUV >%8 *//3%/F
I4- /,"R 2- ,- *0 L"(, ,- +-91$"*( "#-3,
+-()*,*-(/ -0 L-.> -. +-$$%"23%/
O.%P3%(+8 -0 /%+.%,".*", 9%%,*(2/
O.%P3%(+8 -0 9%%,*(2/ -0 /%+.%,".*", L*,4
4%"$,4 0"+*$*,8 9"("2%./
W-993(*,8 1%.+%1,*-(/ -0 /%+.%,".8S/ /,8$%
'()*H*)3"$D+-$$%+,*H% #%4"H*-3. M5,%(, -0 .%/-., ,- 1",.-( ,- .%/-$H% 4%"$,4
1",,%.(/ +".% (%%)/
N%./-("$ *(H-$H%9%(, L*,4 ,4% $-+"$ W-99*,9%(, 6*.,4 1$"+% "() .%/*)%(+% -0 1.%0%+,
6*.,4 1$"+% "() .%/*)%(+% -0 4%"$,4 /%+.%,".8
'0 1.%0%+, -. 4%"$,4 /%+.%,".8 ".% 148/*+*"(/K )-
,4%8 L-.> *( ,4% )*/,.*+, B13#$*+ XD-. 1.*H",%F
G%(2,4 -0 ,*9% /,"R 4"H% L-.>%) *( ,4% )*/,.*+,
M5,%(, /,"R ".% .%/*)%(, *( ,4% )*/,.*+,
W-(,*(3*,8 O-.9/ -0 /,"R +-(,."+, Y 4-3./K L4- L*,4
M5*/,%(+% -0 )*/,.*+, 13#$*+ /%.H*+% %5"9
N-$*,*+"$ 1".,*%/ *( 1-L%. /*(+% ZV[[
A%"$,4 /%+.%,".*%/ /*(+% ZV[[
;1"+% -0 "++%1,"#$% 1."+,*+% "() "++-3(,"#*$*,8 C,,*,3)%/ ,- /1%+*<%) 1."+,*+%/ *( ,%.9/ -0

"++%1,"#*$*,8D(-(="++%1,"#*$*,8
181
)V G=P/&('& $= ".V X )'0/". )0/$&0$ H I$#/0/&$ Z9 TMWWWY ^91_^`^ ZRV
!"" !#$#%&' (")*")+#"! #) ,(-"' .&(#) /$"'#+&) &(#3& *, G'&!#%X *" V ,0(05', *" PQTT 8R& "*#+! &f,'* "*B=B
+,0)('#"! &)* 1',1,!" 2-"'"3"' & $&)&4"'#&% *#!+,0'!" K&f, M&0%,X :*#(,'& /(%&!B
#! 5"#)4 #)(',*0+"* 6',$ & *#7"'")( +0%(0'" 80!0&%%9 G'0+"> cB 8PQQ[=B ]0)*&$")(&% "%"$")(! ,6 (-" D0&%#(9 ,6 +&'"X
),'(-"') :0',1"&);<,'(- /$"'#+&)=> & '01(0'" & !#$1%" 6'&$"2,'AB )=*#/$( /& L"4/.8 ,."&&/&2> M98S=> ZPU
QPB
5"(2"") *#!+,0'!" &)* 1'&?#!> ,' &11"&'&)+" &)* !05@
G'0+"> cB> _ c&#)> /B dB 8PQQP=B J$1',3#)4 (-" D0&%#(9 ,6 +&'"
!(&)+"> #! %#A"%9 (, 5" 1',3,A"*B C-" )&(0'" ,6 (-"!"
(-',04- ,1"'&(#,)! '"!"&'+-B J) aB K"#*$&)> _ aB `,')>
'01(0'"! #) *#7"'")( !"((#)4! #! !,$"(-#)4 #)("')&(#,)&% N6$%"=/'&( %$($"%07F 7$.6/&2 ;"4/.8 6."&&/&2 6%'2%"4( O'%P
'"!"&'+-"'! &)* 1'&+(#(#,)"'! #) !9!("$! *"3"%,1$")( -$==$%B <"2 g,'AX E#%"9U.#!!B
)""* (, A),2 $0+- $,'" &5,0(B /%(-,04- (-" '"D0#'"@ G'9+"> cB> C,,%"> aB> E&%*$&)> WB> _ h,#4(> /B 8PQQS=B
$")(! ,6 "(-),4'&1-9 &'" *"$&)*#)4 5,(- #) +,!(! /!!"!!#)4 (-" D0&%#(9 ,6 6&+#%#(9@5&!"* +-#%* !0'3#3&% !"'@
&)* (#$"> !0+- !(0*#"! &'" ,6 3#(&% #$1,'(&)+" (, 50#%* 3#+"!B 3$".=7 ,'./08 "&# ,."&&/&2> K8S=> PVVUPZRB
01 & 5&!" ,6 A),2%"*4" &5,0( (-" +,)("?(! ,6 -"&%(- N&%*&!> aB MB> _ E,,* c'> CB 8PQQ\=B i],' (-" :)4%#!- (, !""jX
!9!("$! *"3"%,1$")( &! & +,$1%"$")( (, (-" #)("'@ C-" #$1,'(&(#,) ,6 $&)&4"'#&% ("+-),%,49 #) %&(" S[(-@
)&(#,)&% 2,'A ,) -,2 (, &!!"!! D0&%#(9 ,6 -"&%(- !"'@ N")(0'9 G'&O#%B N%2"&/E"=/'&> J8Y=> VP\UVRYB
3#+"!B E#(-,0( & 5,*9 ,6 2,'A *,+0$")(#)4 (-" *" N&'3&%-,> HB JB> _ K&)(,!> .B 8PQQS=B )/(=$4" Q+&/0' #$
)"*+#$F 0'4$&="+%/'( "R @$/ N%2"S&/0" #" )"*+#$ T@$/( UVWUWX
'"&%#(#"! ,6 1,%#+9 #$1%"$")(&(#,) #) +,)("?(> 2" -&3"
1W "&# UV9JMX1WYB K&f, M&0%,X :*#(,'& `0+#("+B
), '"+,'* ,6 2-&( #! '"&%%9 5"#)4 +-&)4"*> 2" *, ),(
N,%%#)!> NB 8PQQV=B I"+")('&%#O&(#,)B J) dB c&),3!A9> 3$".=7
A),2 2-&( 2" &'" $,)#(,'#)4 (-',04- D0&%#(9 #)*#@ 6'./08 "&# (8(=$4( #$?$.'64$&= 811B PZPUP\T=B H")"3&X
+&(,'! &)* 2" +&)),( &!!"!! '"%"3&)+" ,6 '"4#,)&% &)* Eè 8+-&1("' P[=B
)&(#,)&% "?1"'#")+"! 6',$ ,)" +,)("?( (, &),(-"'B N,%%#)!> NB> _ H'"")> /B 8PQQY=B I"+")('&%#O&(#,) &)* 1'#$&'9
-"&%(- +&'"X !,$" )"4&(#3" #$1%#+&(#,)! #) *"3"%,1#)4
+,0)('#"!B <&=$%&"=/'&". :'*%&". '; 3$".=7 )$%?/0$(> MJ>
YVQUY\VB
!"#$%&'()*(+($,- N,%%#)!> NB> _ `0)("'> IB 8PQQ\=B N',!!@+,0)('9 +,$1&'#!,)!
,6 *"+")('&%#!&(#,)> K-,'( 1'"!")(&(#,) &( eI/;J`KI>
C-" '"!"&'+- #! 1&'( ,6 & 1',F"+( 60)*"* 59 (-" G'#(@ `"&%(- K"+(,' W"6,'$> .,)*,)> /1'#%B
#!- H,3"')$")( I"1&'($")( 6,' J)("')&(#,)&% I"3"%@ N,)9"'!> IB 8PQTZ=B I"+")('&%#O&(#,) &)* *"3"%,1$")(X &
,1$")( &)* -,!("* &( (-" :!+,%& *" K&0L*" M0L5%#+&> 6'&$"2,'A 6,' &)&%9!#!B !'44*&/=8 >$?$.'64$&= :'*%&".>
N"&'&L> G'&O#%B M9> TTUP[[B
:#!")!(&*(> KB _ W,)#4"'> .B 8PQTY=B ,"=%'&(A 0./$&=( "&#
;%/$&#(F /&=$%6$%('&". %$."=/'&( "&# =7$ (=%*0=*%$ '; =%*(= /&
('0/$=8B N&$5'#*4" b)#3"'!#(9 M'"!!B
.(/(0($"(- :)4"%A"!> :B 8PQQ[=B M',+"!! "3&%0&(#,) #) & N,%,$5#&) 1'#@
$&'9 -"&%(- +&'" 1',4'&$$"B 3$".=7 ,'./08 "&# ,."&&/&2>
/(A#)!,)> KB 8PQQR=B /)(-',1,%,49 #) '"!"&'+- ,) (-" D0&%#(9 Z8Y=> RS\URRVB
,6 -"&%(- !"3#+"!B !"#$%&'( #$ )"*+#$ ,*+-./0"> 18R=> STRU :!+,5&'> /B> _ /%3&'"O> KB 8PQQS=B [7$ 4"P/&2 '; ('0/". 4'?$\
SQQB 4$&=( /& @"=/& G4$%/0"F /#$&=/=8A (=%"=$28 "&# #$4'0%"08B
/(A#)!,)> KB 8PQQV=B W"!('0+(0'#)4 -"&%(- +&'"X ('&+A#)4 (-" G,0%*"'> NeX E"!(3#"2 M'"!!B
*"+")('&%#O&(#,) *"5&("B ,%'2%$(( /& 3*4"& 5$'2%"678> ]&,',> WB 8PQQP=B N( #'&'( #' 6'#$%F ;'%4"0! &f, *, 1&(',)&(,
918Y=> YTZUV[RB 1,%k(#+, 5'&!#%"#', 8Q(- "*B=B K&f, M&0%,X H%,5,B
/(A#)!,)> KB 8PQQ\=B ]',$ 3#!#,) (, '"&%#(9X J$1%"$")(#)4 ]%9))> WB 8PQQR=B W"!('0+(0'#)4 -"&%(- !9!("$!X & +,$1&'&(#3"
-"&%(- '"6,'$! #) .0!&A&> ^&$5#&B :'*%&". '; <&=$%&"=/'&". &)&%9!#! ,6 :)4%&)* &)* (-" <"(-"'%&)*!B J) aB `#%%> ]$O
>$?$.'64$&=> 18Y=> ZRPUZRQB "2$&#"( /& =7$ (=*#8 '; =7$ 6'./08 6%'0$(( 811B V\UT\=B
/(A#)!,)> KB> _ /50 :% `&F> aB 8PQQZ=B I,$&#) &)&%9!#! 6,' `"$"% `"$1!("&*X `&'3"!("' E-"&(!-"&6B
D0&%#(&(#3" 105%#+ -"&%(- *&(&B 3$".=7 ,'./08 "&# ,."&&/&2> ],'!5"'4> GB> G&'',!> ]B> _ h#+(,'&> NB 8PQQS=B I"3"%,1#)4
998Y=> YRTUYYSB +,0)('#"! )""* $,'" D0&%#(9 &!!0'&)+"X -,2 -"&%(- 6&+#%#(9
G&')&5&!> HB /B 8PQQ\=B @'0". 2'?$%&4$&=A $B*/=8 "&# 6%/4"%8 !0'3"9! +&) +,)('#50("B 3$".=7 ,'./08 "&# ,."&&/&2> K8S=>
7$".=7 0"%$ /& C=7/'6/"B M-I C-"!#!> b)#3"'!#(9 ,6 .,)*,) PQRUPQZB
8.K`Ca=B H&')"'> MB> C-,$&!,)> cB> _ I,)&%*!,)> IB 8PQQ[=B l0&%#(9
G,%$&)> .B HB> _ I"&%> CB :B 8PQQP=B D$;%"4/&2 '%2"&/E"=/'&(F &!!"!!$")( ,6 -"&%(- 6&+#%#(#"! #) '0'&% M&10& <"2 H0#)"&B
G%=/(=%8A 07'/0$ "&# .$"#$%(7/6B K&) ]'&)+#!+,X c,!!"9@G&!!B 3$".=7 ,'./08 "&# ,."&&/&2> Z8P=> YQUVQB
G,!!"'(> CB 8PQQV=B I"+")('&%#O&(#,)B J) dB c&),3!A9> 3$".=7 H&((#)&'&> GB NB> J5&+&+-"> cB> M0")("> NB CB> H#&+,)#> cB> _
6'./08 "&# (8(=$4( #$?$.'64$&= 811B PY\UPZ[=B H")"3&X N&1'&'&> /B 8PQQV=B M"'+"1+#,L) *" %& +,$0)#*&* &+"'+& *"
Eè 8N-&1("' Q=B %& +&%#*&* *" %,! !"'3#+#,! *" !&%0* 10L5%#+,! ") %,! *#!('#(,!
G,!!"'(> CB 8PQQT=B /)&%9O#)4 (-" *"+")('&%#O&(#,) ,6 -"&%(- <,'(" " J+-#%, @ G,%#3#&B !"#$%&'( #$ )"*+#$ ,*+-./0"> 998R=>
!9!("$! #) *"3"%,1#)4 +,0)('#"!X *"+#!#,) !1&+"> #)),3&(#,) YSVUYRTB
&)* 1"'6,'$&)+"B )'0/". )0/$&0$ H I$#/0/&$> JK8P[=> PVPRU H',!!> WB "( &%B 8PQQS=B N%#)#+ *"+")('&%#O&(#,)X &) J!'&"%#
PVS\B $,*"% &)* "3&%0&(#,) !('&("49B :'*%&". '; I"&"2$4$&= /&
G'&O#%#&) N,)!(#(0(#,) 8PQTT=B !'&(=/=*/0! &f, *& '"10L5%#+& 6"*"'@ I$#/0/&$> K8Z=> PPUS[B
182
\a\ JV P/8(*30* #/ &)V A J0:(&) J:(#*:# ^ C#2(:(*# Y> _\XXX` @>?a@b@
!"#"$% &'% &()*(% &'% + ,-./% 0' 123345' 67. 89"*/:; (< ="#. Q#.>>G"$% H'% + X/*@")>M;% 0' 1234a5' T9<)#9#*/&/(0*- "0N
/$ 7(>?/:"*> "$@ 7."*:7 =.$:.#> /$ A">:.#$ B.C/($% D7"$"E 6%#&/ "0<#3 (* 5&3"(*6/0* &%# 2&3"#2 (* U&8)&*2' e.#M.*.;%
0 )/.F <#(G :7. ?":/.$:>' B.>."#=7 B.?(#:% A">:.#$ N0E P$/).#>/:; (< N"*/<(#$/" Q#.>>'
B.C/($% D7"$"' QBSN,B 123__5' +%(9&%E "#&)/" :&%# /"#3&,%,3- M0),9# T &*2
!.>>% &' H'% + &"I"::"% B' 0' 1233J5' !"# $%&'()(&* +,'')#- TTV +%(9&%E Q#&)/" .&%# U<#%&/(0*3 7#3#&%:"' P[0E
.,)/,%# 0* /"# 10%2#%)&*23 04 /"# 5#3/#%* 50%)2' K.F N.$:.# <(# !9G"$ [.#)/=.>'
L(#ME N(*9GO/" P$/).#>/:; Q#.>>' Q9:$"G% B' &' 1233a5' C&8(*6 2#90:%&:E N0%8- :(H(: /%&F
!(<>:.@.% D' 123325' .,)/,%#3 &*2 0%6&*('&/(0*3' I"/@.$7."@E 2(/(0*3 (* 902#%* T/&)E' Q#/$=.:($% KHE Q#/$=.:($
I=D#"FR!/**' P$/).#>/:; Q#.>>'
S?*"$=. 123345' 7&*8(*6 203 9,*(:;<(03= >??@A>??B' T(#:"*.U"E B(.G.#% I'% + I($:(;"R0C9/*"#% N' 123__5' h9"*/:; ">>.>>R
S?*"$=.' G.$: "$@ ">>9#"$=. /$ ?#/G"#; 7."*:7 ="#.' S$ UW3#/ <,1)(F
H"/$% 0' V' 1233W5' C&*&6(*6 D,&)(/E 04 :&%# (* <0<,)&/(0* <%0F :&/(0* *0V >XY' D.$.)"E X!,'
6%&93' X.>: !"#:<(#@E V9G"#/"$ Q#.>>' [.=#.:"i#/" @. ["9i@.% N."#"i 1233W5' U .#&%&L # & 9,*(:(<&)('&F
V"$Y/% K'% V/*/G"% Q'% + I9$/>7/% D' I' 1233W5' G,&)(/E 04 :! "^( @" >"9i@.E .$=($:#( =(G (> $()(> ?#.<./:(> @( N."#"i'
<%(9&%E :,%&/(H# :&%# (* I&%F#3FJ&)&&9' P$/).#>/:; (<
D().#$( @( A>:"@( @( N."#"i'
Z($@($E Z($@($ [=7((* (< !;C/.$. "$@ 6#(?/="*
[7."-% B'% + X.>:% I' 123345' I"#M.:/U":/($% G"$"C.#> "$@
I.@/=/$.'
G(#"* >:#"/$E N7"/#G.$% @/#.=:(#> "$@ ?9O*/= >.#)/=. .:7(>
Z"F#.$=.% Q' B'% + Z(#>=7% H' X' 123\35' I#H#)0<(*6 0%6&*('F
/$ :7. K":/($"* !."*:7 [.#)/=.' +,1)(: P29(*(3/%&/(0*% BY%
&/(0*3- 2(&6*03(3 &*2 &:/(0*' B."@/$C I">>E 0@@/>($R
2_3fW`\'
X.>*.;'
[?#"@*.;% H' Q' 123435' !"# #/"*06%&<"(: (*/#%H(#N' ,#*"$@(E
Z."*% ]' K' 1234J5' .0%0*#)(390= #*K&2& # H0/0- 0 9,*(:;L<(0 #
!(*:% B/$.7"#: + X/$>:($'
0 %#6(9# %#<%#3#*/&/(H0 *0 $%&3()' ["^( Q"9*(E 0*<"R,G.C"'
Z/?>M;% I' 123_`5' J/%##/ )#H#) 1,%#&,:%&:E' K.F L(#ME 6.$@*.#% H' 123345' Z002 60H#%*9#*/ (* /"# !%0<(:3' e"*:/G(#.E
I"=G/**"$' H(7$ !(?M/$> P$/).#>/:; Q#.>>'
I"=7"@(% Z' I' ]' 1233a5' bX. *."#$.@ :( :7/$M ?(*/:/="**;cE 67(G">($% H' 0'% K.FO#"$@.#% X' N'% + V(*.7G"/$.$R
:7. /$d9.$=. (< :7. N":7(*/= =79#=7 "$@ :7. <.G/$/>: 0/:M.$% B' Z' 123325' &.=.$:#"*/U":/($ /$ " @.).*(?/$C
G().G.$: ($ :7. .G.#C.$=. (< :7. 7."*:7 G().G.$: (< =(9$:#;E :7. .j?.#/.$=. (< Q"?9" K.F D9/$." "$@ /:>
:7. H"#@/G K(#@.>:. "#." /$ ["^( Q"9*(% e#"U/*' S$ [' 0' 7."*:7 >.#)/=.' +&:([: 7#3#&%:" C0*06%&<" \Y' N"$O.##"E
B"@=*/-.% + [' X.>:F((@% M(H&- N09#* &*2 <0<,)&% <%0F K":/($"* N.$:#. <(# &.).*(?G.$: [:9@/.>% :7. 09>:#"*/"$
/#3/ (* O&/(* P9#%(:& 1??' __f2225 1=7"?:.# g5' P$/).#>/:;'
I/**>% 0'% ]"9C7"$% H' Q'% [G/:7% &' Z'% + 6"O/OU"@.7% S' X."M*/"G% &' 1233g5' I#H#)0<9#*/ 04 D,&)(/E (*2(:&/0%3 1&3#2
1233`5' Q#&)/" 3E3/#93 2#:#*/%&)('&/(0*- :0*:#</3= (33,#3 &*2 0* <&/(#*/3] <#%:#</(0*3 04 D,&)(/E 40% "#&)/" 3#%H(:# 90*(/0%F
:0,*/%E #K<#%(#*:#3' D.$.)"E X!,' (*6 &/ "#&)/" :#*/%#3 (* Z"&*&' Z/).#?((* [=7((* (<
Q"/$.% Q'% + X#/C7:% I' @"D' 123__5' [":/><"=:/($ F/:7 G.@/R 6#(?/="* I.@/=/$.'
="* >.#)/=.> /$ " *(FR/$=(G. e#"U/*/"$ =(GG9$/:;' X!, 1233J5' I#:#*/%&)('&/(0* &*2 "#&)/" 3E3/#93 :"&*6#- P
RH&),&/(0* &*2 /"# Q#&)/" +%04#33(0*3% >>% a__fa3_' 4%&9#N0%8 40% &*&)E3(3' D.$.)"E K":/($"* 7."*:7 >;>:.G>
Q<.-.#% H' 1233W5' C&*&6(*6 N(/" <0N#%- <0)(/(:3 &*2 (*S,#*:# "$@ ?(*/=; 9$/:% &/)/>/($ (< >:#.$C:7.$/$C 7."*:7 >.#)/=.>%
(* 0%6&*('&/(0*3' e(>:($E !"#)"#@ e9>/$.>> [=7((* Q#.>>' X!,'
183
! The Author 2005. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved.
doi:10.1093/heapol/czj003 Advance Access publication 30 November 2005
HMIS and decision-making in Zambia: re-thinking

information solutions for district health management
in decentralized health systems
RICHARD I MUTEMWA
Centre for AIDS Research, University of Southampton, UK
At the onset of health system decentralization as a primary health care strategy, which constituted
a key feature of health sector reforms across the developing world, efficient and effective health
management information systems (HMIS) were widely acknowledged and adopted as a critical
element of district health management strengthening programmes. The focal concern was about the
performance and long-term sustainability of decentralized district health systems. The underlying logic
was that effective and efficient HMIS would provide district health managers with the information
required to make effective strategic decisions that are the vehicle for district performance and
sustainability in these decentralized health systems.
However, this argument is rooted in normative management and decision theory without significant
unequivocal empirical corroboration. Indeed, extensive empirical evidence continues to indicate that
managers’ decision-making behaviour and the existence of other forms of information outside the
HMIS, within the organizational environment, suggest a far more tenuous relationship between the
presence of organizational management information systems (such as HMIS) and effective strategic
decision-making. This qualitative comparative case-study conducted in two districts of Zambia focused
on investigating the presence and behaviour of five formally identified, different information forms,
including that from HMIS, in the strategic decision-making process. The aim was to determine the
validity of current arguments for HMIS, and establish implications for current HMIS policies.
Evidence from the eight strategic decision-making processes traced in the study confirmed the
existence of different forms of information in the organizational environment, including that provided
by the conventional HMIS. These information forms attach themselves to various organizational
management processes and key aspects of organizational routine. The study results point to the need
for a radical re-think of district health management information solutions in ways that account for the
existence of other information forms outside the formal HMIS in the district health system.
Key words: HMIS, information forms, decentralization, strategic decision making, district health systems
Introduction authority and play its role in the reformed health

structure. Thus, there has been a deliberate movement
Since Alma-Ata in 1978, most developing countries have to strengthen the management capacity of district health
implemented health sector reforms. In almost every case, systems (for instance, Cassels and Janovsky 1996).
a central feature of the reform strategy has been a process
of structural decentralization: the aim being to vest greater One area of focus in district health management
decision-making responsibility in the district health strengthening programmes has been health management
systems. The underpinning primary health care notion is information systems (HMIS) at the district level (for
that decentralization thrives on the essential involvement instance, Acquah 1994; Ankrah and Djan 1996; Lippeveld
of primary-level health management units in the delivery et al. 1997; Bodart and Sapirie 1998). There are challenges
of health services (WHO 1978). Although the geo-politics in clearly defining what is meant by HMIS (Lippeveld
vary from country to country, the district tends to be the and Sauerborn 2000). In this study, HMIS is used to refer
last formal unit of local government and administration to the predominant concept of a formal and structured
(Mills 1990). Across the variations of decentralization in health information system set up to support and facilitate
developing health systems (Mills 1990; Vaughan 1990), health management decision-making at different levels
the success of decentralization has predominantly been of any health system (for instance, Ankrah and Djan 1996;
considered to rely significantly on the capability of the Danish Bilharziasis Laboratory 2002; Gladwin et al.
district health system to effectively exercise its assigned 2003). In that light, HMIS is designed to carry both
184
Re-thinking information solutions 41
epidemiological information (health prevalence, incidence, This problem has to some extent been acknowledged in
mortality, morbidity statistics) and administrative existing literature. For instance, Liebenau and Backhouse
information (resource inputs and service utilization). (1990) have pointed out how little we understand about
what information is and how it affects us in organizations.
The rationale for HMIS has been that the availability of More fundamentally, March (1988) and Mintzberg (1975)
operational, effective and efficient health management noted the general gap that exists between findings of
information systems is an essential component of the research on decision-making and the assertions of classical
required district management capacity. The logic is that normative decision-making theory that underpins the
effective and efficient HMIS will provide district health current argument for information in organizations.
managers with the information required to make effective March (1988) argues that this gap is ‘partly attributable
strategic decisions that support district performance and to limitations in the theories, rather than limitations in the
sustainability in these decentralized health systems. (decision-maker) behaviour’.
However, the arguments for HMIS are not based on The implications of these critical observations for devel-
unequivocal empirical evidence, or tested theory, that the oping health systems ought to be appreciated sensitively.
information carried in HMIS makes a difference, but These are resource-poor economies where new technolo-
rather represents a normative view of management capac- gies should be continuously and rigorously evaluated
ity. A review of empirical literature reveals a prevalence in terms of value creation for the health system, for each
of HMIS failure problems across a range of country dollar invested. Yet, the theory-practice gap being flagged
situations in the developing world (Lippeveld et al. 1997), up by empirical literature on information and decision-
as well as in developed health systems (for instance, making presents potential problems for cost-benefit
Southon et al. 1999; Snyder-Halpern 2001). analysis in these developing health systems. With diver-
gent trajectories or outcome-projection functions, between
Other specific difficulties with far more conceptual theory and actual practice, there is an absence of the
implications pertain to the widely recognized problems necessary agreement on the measurement of benefits,
with the decision-making behaviour of managers in success or indeed failure. The result has been a landscape
organizations in general, at least when that behaviour replete with a plethora of frameworks for measuring
is set against normative theories of management and information system failure or success (Skok et al. 2001).
decision-making practice. For instance, empirical studies This condition has not been helpful to practitioners in
suggest managers use information for political capital, developing health systems. Developing health systems
using information to seek legitimacy for their decisions often set out to strengthen their HMIS based on
rather than to make or clarify those decisions (Feldman normative decision theory principles (Acquah 1994;
and March 1988; Guldner and Rifkin 1993). More Gladwin et al. 2003), but later have to deal with measuring
crucially, it is widely acknowledged that managers use theoretically unanticipated informational phenomena in
information other than that provided by formal organiza- evaluation stages of their HMIS programmes.
tional information systems such as HMIS; and this other
information may take verbal and observational forms, The theory-practice gap that constitutes the root of this
or may be embedded in the training and experiential problem is essentially defined by the way in which
background of managers (for instance, Mintzberg 1975). information is ‘problematic’ in the organizational environ-
ment. An expeditious review of literature on information
This paper, therefore, addresses the challenge of reconcil- and decision-making reveals three major forms of this
ing the rhetoric for HMIS in district health systems with ‘problematic’ presentation of information. These three
observed problems that contradict it, threaten its very forms of presentation are briefly outlined here.
integrity, or, at minimum, recognize its limitations in
relation to management tasks. The paper describes a
Functional versus symbolic use of information
comparative study of two district health systems in
Zambia, and its main intention is to highlight one major The principles of normative decision theory are predicated
implication of the study findings. The paper describes on the functional use of information by decision-makers
the core research problem, key objectives of the study, where, since the onset of Frederick Taylor’s (1911)
the methods and key findings. It then concludes with a ‘scientific management’ paradigm, decision-makers use
discussion of the major practical implication of the study information objectively in making rational decisions.
findings, for HMIS design in developing health systems. Yet, such works as those by Feldman and March (1981),
Feldman (1988), and Dean and Sharfman (1993) represent
now common knowledge that people distort and manip-
ulate information for their own goals, and that this is a
Background
pervasive phenomenon in organizational life. Information
The key research problem confronted by the study was is often used as a symbol of competence, or merely as a
that the interaction between theory and empirical evidence signal of appropriate decision-making to secure legitimacy
so far indicates that organizations, public or private, for decisions made. Guldner and Rifkin (1993) observed
still understand little about the nature and behaviour from their field observations in Vietnam that data were
of information within the organizational environment. being widely used to justify rather than clarify decisions.
185
42 Richard I Mutemwa
Thus, the symbolic use of information directly defies the been insignificant. This study focused particularly on this
traditional logic of the functional value of information third problematique, with a fairly confident theoretical
to the production process. From the perspective of health hunch that the informational phenomena presenting the
systems, information is hence manipulated for goals first two problems would still be explainable from this
not necessarily compatible with the explicit aspirations perspective that recognizes the existence of other forms of
of decentralization. information outside the formal HMIS.
Use versus non-use of information Study objectives

Embedded within the logic of normative decision theory The aim of the study was, first, to establish the presence
is the presumption that decision-makers actually do use of written, verbal, observational, experiential and training
information when it is made available, and they behave information forms in the strategic decision-making
that consistently towards it. However, for decades now process. The focus on the strategic decision-making process
it has been well acknowledged, from observations, that represents a major concern for the management capacity
decision-makers gather information and ignore it; they of decentralized district health management systems and
make decisions first and look for the relevant information their sustainability. Local strategic decisions are central to
afterwards (for instance, March 1982). A study by Finau the definition of district health management capacity and
(1994) in the South Pacific highlighted similar problematic the determination of district health system sustainability,
behavioural tendencies, that local decision-makers in decentralized health systems (Mutemwa 2001).
ignored installed formal health information systems and,
instead, preferred ‘ ‘‘gut feeling’’, hearsay and ad hocry’. Secondly, the study aimed to establish the nature of the
micro-processes through which the above five information
Again, this is a condition that poses HMIS evaluation forms influence the strategic decision-making process.
problems for the health system planner. How credible The third and final aim was to determine the implications
would any form of systemic performance attributions to of these findings for HMIS design and operational
the installed HMIS? considerations. However, this paper will not cover the
second objective due simply to the complexity of the
dynamics involved in the micro-processes. The subject of
Formal HMIS versus other forms of information micro-processes should be better examined in a dedicated,
Contemporary philosophy of organizational management separate paper. Yet, such exclusion does not at all
information systems (including HMIS) is centred on undermine the visibility of overall policy implications
formal structured information systems with, among others, from the study, in the findings presented in this paper.
specified formal encoding, transmission and decoding
rules that govern those structures (Liebenau and
Method
Backhouse 1990; Ward and Griffiths 1996; Boman et al.
1997). As Simon (1957) pointed out, formal information
Study design
systems are based on formal channels of information
which may be characterized by ‘hard’/paper and/or The study was exploratory. The study did not exclusively
electronic forms of transmission in the organization. set out to only search for the different forms of infor-
In the study, these forms were collectively referred to as mation identified in the objectives above, but rather the
the written form of transmission or information, which researcher set out with an ‘open mind’. The basic
includes HMIS. reasoning was that there was still the possibility of finding
other forms of information not yet identified in existing
However, other forms of information have been identified literature, or indeed discovering new interesting insights
in empirical literature as being present in the organiza- into the strategic decision-making process.
tional environment. In his study of managers, Mintzberg
(1975) found that apart from formal management The study was designed as a multi-level, qualitative
information systems, managers used ‘soft’ information comparative case study and was conducted in Zambia,
and favoured verbal over written information. The above- where health sector reforms have involved a significant
mentioned study by Finau (1994) points to similar delegation of decision-making responsibility to district
observations. Mintzberg’s study further indicated that health systems (Mutemwa 2001). In Zambia’s decentral-
managers also use observational information in their ized health system structure, there is separation of policy
work. Experiential and training forms of information are and executive functions in health service provision.
widely acknowledged in the literature as well (for instance, At decentralization, the Ministry of Health retained the
Simon 1976; Melone 1996). All these forms of information national-level sectoral strategic functions of health
are significantly recognized in naturalistic decision theory policy and planning, finance and budget, legislation,
(a perspective on how decision-making occurs in real advocacy and international co-operation (Bergman 1996;
world situations). Yet, there still remains conspicuous MOH 1996).
ignorance of how these information forms operate within
the organizational environment. Hence, presently, their The government then created a parastatal, the Central
practical recognition in HMIS design considerations has Board of Health (CBOH), and delegated to it all the
186
executive functions of service provision: commissioning The second and primary level of comparative cases
health services in the sector, performance support, was the strategic decision-making processes sampled
monitoring and evaluation, national human resource from within the two districts selected for fieldwork.
development, and national health facilities planning The strategic decision-making processes or cases were
(Bergman 1996). Responsibility over actual delivery of compared within each district to establish the degree of
services was further delegated to district health systems, intra-district consistency, and across the two districts
which were re-constituted into District Health Boards to determine the degree of inter-district variation in the
(DHBs). DHBs are legal entities established under the behaviour of information.
Zambia National Health Services Act of 1995 (MOH
1995). They operate on an annual contractual relationship Data collection
with the CBOH, and annual service delivery benchmarks
are evaluated and reviewed each year-end, against which Ethical clearance
funding is negotiated and allocated (MOH 1992, 1996). Ethical clearance was obtained from the national ethical
DHBs have extensive strategic and operational decision- clearance committee, and administrative clearance
making discretion at that primary level, including the obtained from the Central Board of Health acting on
legal mandate to raise and manage their own resources. behalf of the Ministry of Health in Zambia, to conduct
A district can engage in profitable investment activities the study. Consent was also sought and granted by the
that it may deem beneficial; plan, recruit and manage selected districts to conduct the study and access written,
its human resources; and engage in any activities that verbal and observational data sources. Consent to access
may aid the sustainability and prosperity of the district data sources was also a continuous part of the research
health system. process, and was obtained both institutionally, whenever
necessary, and from individuals whose personal insight
For the study, the first level of comparative cases was the on specific issues was sought through interviews from
district health system context. Zambia’s district health time to time.
system profile consists of two main types of district
groups: rural district health systems, and urban district Selecting district cases
health systems. A rural district health system in Zambia One urban district, Lusaka, and one rural district, Monze,
has a district health service structure that serves were purposively sampled from the national sampling
a considerable urban population of the district town, frame of 72 districts in Zambia. The selection process
and further extends to rural village communities situated involved several progressive rounds of scoring all the
outside the town but still falling within the geo-political districts in the country on the basis of: whether a district
boundary of the district. A rural district health service had a functional District Health Management Team
will typically comprise a district health office, a referral (DHMT) and DHB; whether the district was willing to
hospital, at least one urban clinic, and a considerable be hospitable to the study; the final two districts had to
number of rural health centres and community be located in different provinces to control for regional
health posts distributed among the village and farming cultural bias; and a district could not have more than one
communities. donor-funded project running during the time scheduled
for the study, to control for interference from artificial
Conversely, an urban district health system in Zambia human and financial resource capacities that accompany
carries a district health service structure that serves an such health programmes. Donor programmes were
urban community only. An urban district health service considered not a reliable indicator for long-term district
will typically comprise a district health office, one or health system sustainability for two main reasons: first,
more referral hospitals, and a significant number of the short-term and definite life-span nature of inter-
urban health centres distributed among the urban and national development aid; and secondly, the characteristi-
peri-urban communities. cally indeterminate nature of outcome possibilities of
development assistance.
These two groups of district health systems experience
distinct epidemiological and health management problems On the basis of these four criteria, the list was eventually
and challenges, set within their equally varied respective reduced to the two districts. Lusaka is the capital city of
local socio-economies. Based on the understanding that Zambia; while Monze is a rural district in Southern
a number of strategic decision-making processes were to Province, about 200 miles south of Lusaka.
be studied from each district case selected, the researcher
estimated that two district health system cases would be Selecting strategic decision cases
sufficiently representative for the study: one rural district The strategic decision cases were also purposively sampled
and one urban district. These, it was felt, were sufficient to in a process that was closely guided by the methodology
provide empirical insights into how the rural and urban chosen for collecting data on the decision processes.
contexts differentially affect managerial decision-making By design, it had been decided that data on the strategic
and decision-making processes, particularly in terms of decision-making processes were to be collected using the
information variety and volume, and decision-making tracer methodology (Mutemwa 2001). Tracers are con-
activity. cerned with the elucidation of processes and are generally
187
associated with the description of activities over time Prospective tracing of on-going or concurrent strategic
(Barnard et al. 1980; Hornby and Symon 1994). Basically, decision processes was done through unstructured
all the strategic decision processes selected for study in-depth interviews, review of organizational documenta-
were going to be traced, from beginning to end, for each tion, and direct observation of decision-making business
decision-making process. Tracing can be done retro- in the district health office. Observation notes were
spectively on decision cases that have already occurred, recorded in field notebooks and a diary. Direct observa-
or prospectively on decision cases that are concurrent tion took the form of participant observation, the
with the study. In retrospective tracing the researcher is researcher attending and witnessing decision-making
often guaranteed complete decision processes that have sessions without taking active part, but with his status
beginnings and ends, while in prospective tracing it is as a researcher known to the actors. To facilitate
never assured that a decision process being traced will participant observation, the researcher negotiated for
have resolved before the research project winds up its office space within the district health office and focused
fieldwork. The particular advantage with prospective data collection in each district for 6 months each; that is
tracing is that the researcher is able to witness the decision 12 months in all.
process as it unfolds, evolves and develops, which offers
a different and more intimate experience of decision Data analysis
process reality from that of recalled eye-witness reports
or experiential accounts in retrospective tracing. Thus, Data analysis was multi-stage. In the initial stage, data
to optimize the richness of data collected in each district, on each traced strategic decision were brought together
it was felt some of the strategic decision cases selected to reconstruct the story of the strategic decision-making
for the study were to be historical, for retrospective process, bringing out, as much as the data could allow,
tracing; while others were to be concurrent with the study, the reality and chronology of its mechanics. The process
for prospective tracing. of data interrogation to reconstruct decision process
stories started as part of data collection, in many instances
Three criteria were invoked for selection of strategic shaping follow-up interviews, documentary reviews and
decision cases in the two districts. A decision process observations. These reconstructed decision process stories
case had to have evidence of availability and reliability of were then verified with key informants for validation, and
information sources on it; in the case of historical decision any inconsistencies or misrepresentations corrected.
processes, there had to be evidence of the process having
reached some form of end or resolution; and, the district In the second stage, the eight constructed decision process
health office had to give full consent to the study of stories were structured. The search for structure was
a selected decision case. To succeed on these criteria, the a search for a common regularity in the decision process
exercise of selecting strategic decision cases for study in cases, which would enable cross-case comparison and
the two districts was deliberately participatory. DHMT meaningful subsequent abstraction. To educe a common
members, as executive custodians of strategic decision- structure of the decision process from the eight decision
making at district level, were involved in the discursive case stories, the emergent theme approach (Mintzberg
process of recalling, suggesting and listing strategic et al. 1976; Nutt 1984) and critical events principle
decision-making processes, historical and on-going, (Poole and Baldwin 1996) were deployed. Decision
which would be traced in each district. The three selection process stories were examined using the emergent theme
criteria served as a backdrop to the participatory process. approach, with intuition used to organize the stories
A total of eight strategic decision-making processes were into patterns that describe the nature and sequence of
selected for tracing in the study, four from each district. key phases and within-phase steps. The critical events
In each district, two of the decision cases were historical, principle helped identify key milestones or turning points
the other two current or concurrent with the study. in the decision case stories, which were used for
constructing the frame of the structure.
Collecting the data The last stage of data analysis involved individually
Retrospective tracing of historical strategic decision breaking down the structured decision stories for, among
processes was done through unstructured in-depth inter- other aspects: the presence of written, verbal, observa-
views and review of organizational documentation. tional, experiential, training and any other information
Unstructured in-depth interviews were conducted with forms; the source of the present information forms and
key informants on each strategic decision case traced. channels through which the information forms entered
Key informants were mainly those members of the the decision process.
DHMT or of the broader district health office that had
participated in the process. In addition, organizational
Results
documentation relating directly and indirectly to the
decision process was requested and reviewed. This involved
Strategic decision-making processes selected for study
meeting minutes, memos, letters, personal notes, strategic
and operational plans, reports and policies. Validation of As Table 1 shows, a total of eight strategic decision-
data was achieved through multi-informant and methodo- making processes were traced in the two district health
logical triangulation (Pettigrew 1990; Mutemwa 2001). systems. All the four decision processes from Monze were
188
Table 1. The eight strategic decision-making processes in the study, presenting problem situation and identifying the problem
by type and district that they then adopted for targeting.
District General management Health programmes Total Investigation
Monze 4 0 4 An investigation stage emerged as the second stage of
Lusaka 2 2 4 the decision-making process. It covers activities through
Total 6 2 8 which the managers get to understand the root cause
of the problem, and how much the problem may have
impacted on their organization or other aspect of their
service. Here, managers or their assigned proxies actively
about addressing administrative or general management searched for information relating to those aspects of the
strategic problems, and none were about addressing problem. The investigation stage typically ended at the
(epidemiological) health problems. Whereas, in Lusaka, point where the managers had gained full or part answers
two of the decision processes were about addressing on those aspects and they had some general conceptual
general management problems, and the other two were ideas about the attributes of the ideal solution to the
about (epidemiological) health problems. problem. These ideal-solution attributes then provided
a reference ‘blueprint’ for the next and final stage of
There appears to be no other immediate explanation for ‘solution development’.
this distribution in the nature of traced decision processes,
apart from the apparent inter-play between the tight Solution development
decision process selection criteria, coincidence and the
timing of the study. For instance, the decision process Solution development is the third and last stage of the
distribution for Monze does not necessarily imply that decision-making process. It covers activities about the
there were no health problems in the district at the time development of a solution, which in some decision cases
of the study. Major decisions worthy of study had been came in the form of a relatively complex programme
made and health programmes were already running at the design in bound hardcopy print. In other cases, the
time of the study, but these decisions did not exactly solution was nothing more than a simple list of inter-
satisfy the selection criteria. Moreover, it is significant related intervention activities on a one-page internal
that, in fact, the administrative or general management memo on file (or even listed in meeting minutes as recom-
decision processes traced in both districts either directly mendations for action). It is significant that, according
or indirectly, as would be reasonably expected, pertained to the study findings, solution development does not
to health programmes set up to address (epidemiological) include implementation of the solution because it was felt
health problems. ‘implementation’ posed a different set of questions.
Structure of the strategic decision-making process A few empirical observations should be made about the
three stages of the decision-making process delineated
All the eight strategic decision-making processes traced in above. First, the structure also recognizes the transitional
the study exhibited an identical developmental structure. linkages between the stages, and the activities that
However, detailed elucidation of how the decision-making constitute these linkages. These transitional activities
process structure operates can only be the subject of perform specific functions that ensure the relationships
a dedicated, separate paper. What will be attempted between the stages, and hence provide continuity to the
here is to delineate the form of the decision process structure.
structure only as far as it serves as the basic backdrop
to the subsequent presentation of data on presence of Secondly, each of the stages is amenable to analysis as
information, the focal subject of this paper. an episode with a distinctive set of activities that differ-
entiate it from the other stages in the process. This was
Thus, basically, the strategic decision-making process particularly useful to the task of breaking down the
structure that emerged from the data has three stages individual decision-making processes in the search for
around which activities in each decision-making process information in its various forms.
seemed to cluster, from beginning to end. The stages
are: problem recognition, investigation and solution
development. Presence of information in the strategic
decision-making process
Problem recognition
Firstly, all the five forms of information discussed earlier
Problem recognition emerged as the first stage of the were found to exist in the strategic decision-making
decision-making process. It covers dynamics by which process: written, verbal, observational, experiential and
the decision process is triggered, including the ensuing training. District health managers referred to a variety
activities up to the point where the managers arrive at of information forms in the course of strategic decision-
some definition or understanding of what the real or basic making. Table 2 shows, in a comprehensive manner,
problem behind the indicators is. In this stage, managers the information profile across the three decision-
tended to be pre-occupied with making sense of the making process stages for each of the eight strategic
189
Table 2. Information profiles of the eight strategic decision-making processes across the three stages of the decision-making process
Decision process case Problem recognition Investigation Solution development
Transport policy 1. Written (HIS, AIS) 1. Written (HIS, AIS) 1. Written (HIS, AIS)
2. Verbal 2. Observational 2. Verbal
3. Experiential 3. Experiential
4. Training 4. Training
SEATS 1. Written (HIS 1) 1. Verbal (1) 1. Written (pilot)
2. Written (HIS 2) 2. Verbal (2) 2. Experiential
3. Training
De-linkage of outpatients department 1. Written (AIS) 1. Experiential (1) still in process
2. Experiential 2. Experiential (2)
3. Written (AIS)
Fuel 1. Written (AIS) (corrupted) (corrupted)
2. Verbal:
formal
informal
Health centre in-charge post 1. Observational 1. Experiential (1) 1. Written (AIS)
2. Verbal 2. Experiential (2) 2. Training
3. Written (AIS) 3. Experiential
4. Intuition
Strategic environmental health plan 1. Written (HIS) 1. Written (Research 1) 1. Written (Research)
2. Training 2. Written (Research 2) 2. Verbal
Health centre staff recruitment programme 1. Verbal 1. Experiential 1. Written (AIS)
2. Observational 2. Written (AIS) 2. Experiential
3. Written (AIS)
4. Experiential
Human resource policy 1. Observational 1. Experiential still in process
Notes: AIS ¼ administrative information system; HIS ¼ health information system; SEATS ¼ Service Expansion and Technical Support;
still in process ¼ unresolved by end of data collection.
decision-making processes traced in the study. For experience and professional expertize were pooled and
instance, in the ‘transport policy’ decision case, the district shared, and then applied to understand the transport
managers used verbal, written, experiential and training problem being discussed. This information in manage-
information to recognize the ‘transport’ problem in the ment meetings was pooled and shared in verbal form.
district. The managers then used written and observa- Note that, in all decision cases, the exact dynamics of this
tional information to investigate the problem and arrive at pooling and sharing of information was a subject beyond
some understanding of what the ideal solution to the the remit of the study.
problem would be. Finally, to develop the ‘transport
policy’ as the solution to the problem, the managers again Similarly, Tables 4 and 5 present the sources of informa-
used written, verbal, experiential and training information identified in Table 2 under, respectively, the
tion. Note that although the set of information forms used ‘investigation’ and ‘solution development’ stages of the
in the ‘problem recognition’ and ‘solution development’ decision process, for each decision case.
stages seem identical, their particular contents were
different due to the different goals targeted at these Secondly, there was no regular pattern in the presence of
stages. For instance, the written information used in the these information forms, either across decision-making
problem recognition stage was different in content to the processes or across the stages within each strategic
written information used in the solution development decision-making process, as illustrated in Table 2. Each
stage. Both are identified as ‘written’ for the reason that decision-making process was informationally distinct;
both of the information pieces were obtained from written as was each decision-making stage within a process.
paper and electronic documents. Thus, as the decision-making process progressed, infor-
mation in its various forms entered the process for a
Tables 3, 4 and 5 present the sources of the information specific purpose, and exited the process as soon as the
forms identified at each of the three decision process purpose was achieved.
stages in Table 2, for each decision case. For the ‘transport
policy’ decision case, Table 3 indicates that, in the problem In Table 2, the ‘Fuel’ decision process case is listed
recognition stage, the managers obtained written informa- as having ‘corrupted ’ following its first stage, to illustrate
tion from the HMIS, whereas verbal, training and the fact that the decision process lost its initial formal
experiential information were obtained through manage- focus in the subsequent stages due to political conflict that
ment meetings. What this simply means is that manage- emerged and preoccupied the process. Thus, the original
ment meetings served as arenas in which previous problem which the ‘Fuel’ decision case set out to address
19 0
Table 3. Routines associated with information types in the problem recognition stage, across the eight studied strategic decision cases
Decision case Information type
Experiential Intuition Observational Training Verbal Written
Transport policy Shared by DHMT – – Shared by Admin. Supervisory visits: HMIS: HIS and AIS
members in meetings Man. in meetings meetings with
health centre staff
SEATS – – – – – HMIS: HIS

De-linkage of outpatients Shared by DHMT – – – – HMIS: AIS
department members in meetings
Fuel – – – – 1. Formal reports HMIS: AIS
2. Informal reports
Health centre in-charge – By DHMT members Supervisory visits: – Supervisory visits: HMIS: AIS
programme in sense-making observation of meetings with local routine quarterly reports,
meetings staff behaviour community members special reports
in health centres directly from
communities
Strategic environmental Shared by DHMT – – Shared by some – HMIS: HIS
health plan members in meetings DHMT members
in meetings
Health centre staff Shared by DHMT – Supervisory visits: – Supervisory visits: HMIS: AIS
Re-thinking information solutions
recruitment programme members in meetings observation of meetings with

service provision health centre staff
in health centres
Human resource policy Shared by DHMT – DHMT observation – – –
members in meetings of administrative
practice
Notes: HMIS ¼ health management information system; AIS ¼ administrative information system; HIS ¼ health information system; DHMT ¼ District Health Management Team;
SEATS ¼ Service Expansion and Technical Support.
47
191
192
48
Table 4. Routines associated with information types in the investigation stage, across the eight studied strategic decision cases
Transport policy – – Observational Shared by Admin. – HMIS: HIS and AIS

investigation Man. in meetings
conducted by
DHMT
SEATS – – – – Consultative –
meetings of
‘response team’
with NGOs and
youth
De-linkage of outpatients department Shared by DHMT members in meetings – – – – HMIS: AIS
Fuel Corrupted Corrupted Corrupted Corrupted Corrupted Corrupted
Health centre in-charge programme Shared by DHMT members in meetings – – – – –
Strategic environmental health plan – – – – – HMIS: AIS (research)
Health centre staff recruitment Shared by DHMT members in meetings – – – – HMIS: AIS
programme
Human resource policy Shared by task team in meetings – – – – –
Notes: HMIS ¼ health management information system; AIS ¼ administrative information system; HIS ¼ health information system; SEATS ¼ Service Expansion and Technical Support;
DHMT ¼ District Health Management Team.
Table 5. Routines associated with information types in the solution development stage, across the eight studied strategic decision cases

Richard I Mutemwa
Transport policy Shared by DHMT – – Shared by Admin. Consultative meetings HMIS: AIS
members in meetings Man. in meetings with health centres,
WaterAid, filling
station, other staff at
district health office
SEATS Shared by FHSTF – – Shared by FHSTF in – HMIS: HIS and AIS (pilot)
in meetings meetings
De-linkage of outpatients department – – – – – –
Fuel Corrupted Corrupted Corrupted Corrupted Corrupted Corrupted
Health centre in-charge programme Shared by DHMT – – Shared by DHMT – HMIS: AIS
members in meetings members in meetings
Strategic environmental health plan Shared by task team in – – – Shared by task team in HMIS: AIS (research)
planning workshop planning workshop
Health centre staff recruitment Shared by DHMT – – – Supervisory visits: HMIS: AIS
programme members in meetings meetings with health
centre staff
Human resource policy – – – – – –
Notes: HMIS ¼ health management information system; AIS ¼ administrative information system; HIS ¼ health information system; SEATS ¼ Service Expansion and Technical Support;
DHMT ¼ District Health Management Team; FHSTF ¼ Friendly Health Services Task Force.
Table 6. Contribution to the eight decision process cases, per information type
Written Written Verbal Observational Experiential Training Intuition Total of

HMIS Non-HMIS information
types
Transport policy 3 – 2 1 2 2 – 10
SEATS 2 1 2 – 1 1 – 7
De-linkage of outpatients department 2 – – – 3 – – 5
Fuel 1 – 1 – – – – 2
Health centre in-charge programme 2 – 1 1 3 1 1 9
Strategic environmental health plan 1 3 1 – 2 1 – 8
Health centre staff recruitment programme 3 – 1 1 3 – – 8
Human resource policy – – – 1 3 – – 4
Total per information type 14 4 8 4 17 5 1 53
Notes: HMIS ¼ health management information system; SEATS ¼ Service Expansion and Technical Support.
remained unresolved by end of the decision-making between the researcher and the District Administrative
process. Manager during an interview:
Thirdly, as Tables 2, 3, 4 and 5 indicate, written informa- Administrative Manager: ‘‘. . . sometimes we used to get
tion was either from routine HMIS or occasionally the information from (junior) health centre staff that
commissioned formal investigative research or enquiry transport is being misused.’’
report documents that are in circulation within the
district health office. For instance, in the ‘Strategic Researcher: ‘‘Verbal reports?’’
Environmental Health Plan’ decision case, managers
engaged investigative research in the investigation and Administrative Manager: ‘‘Yeah. Verbal reports. Some
solution development stages to gain required information. of them personal reports to me, that I should consider
The solution development stage of the ‘SEATS’ case private and in confidence.’’
involved a pilot study. In some of the traced decision
processes, written information took the form of formal Experiential and training information existed in the
one-off letters or informal anonymous notes, as in the memory stores of the district managers making the
following quote from an interview with the District decisions. This information was typically ‘downloaded’
Administrative Manager on the ‘Transport Policy’ and shared in management meetings, during moments
decision case: of reflecting upon or analyzing the problem at hand.
Whereas, observational information reached the managers
‘‘. . .Sometimes somebody would just come and push through direct or vicarious observation or witnessing
a note under the door to say transport is not being used of organizational activity related to the problem being
as meant for. In fact, not only from the health centre addressed. In vicarious observation, management typically
staff but sometimes also from the community. They used assigned a member of staff within the district health office
to come with a letter to say he (EHT) takes it to to conduct the observation on their behalf.
Mapanza where he comes from . . . So we had to decide to
put up a measure.’’ Fourthly, Table 6 shows the number of times each
information form was used in each of the strategic
Again, in this study, routine HMIS was taken to con- decision-making processes for a specific process activity
stitute two components: routine epidemiological health or purpose. At the bottom of the table are the
information and routine administrative information. corresponding crude totals of the information types,
indicating their respective contributions to the combined
Verbal information equally had formal and informal information profile of all the decision cases in the study.
attributes. Verbal information tended to be shared in Note that, although these figures cannot be taken beyond
formal gatherings, mostly as spoken reports to managers the eight decision cases in the study, information from
during formal supervisory visits to health centres and formal HMIS was not the top contributor to the traced
visits to local communities. Other formal verbal informa- decision process cases. HMIS was certainly a commonly
tion reached district managers through consultative used source of information, but the most common basis
meetings with affected constituencies and/or stakeholder for a decision was experiential information.
organizations during the process of decision making.
Informal verbal information was reported to be mostly Further, there was no ‘observed’ or detected difference,
in the form of informal intimacies about the problem between the two studied districts in the way information
situation; for instance, consider the following interchange behaved in the strategic decision-making process.
193
Finally, no new significant information form was written, verbal, observational, experiential and training
discovered in the study. information.
It is worth noting that the labels of written and verbal

information relate to the formats in which information
Discussion and conclusion was delivered or exchanged, while observational, experi-
From the perspective of health system decentralization, ential and training pertain to the method or way by which
this study has shown that decentralized district health information was gained. Yet, these hints represent some
systems do engage in decisional activity on matters that of the fundamental aspects of any information system:
affect their long-term survival or performance as health collection and delivery of information to the users
system organizations at that primary level. The study has (Finlay 1994; Ward and Griffiths 1996; Boman et al. 1997).
also demonstrated that different forms of information Here, then, it becomes evident that the actual health
are brought to bear, in district decision-making, through management information system for a decentralized
different channels and from a variety of sources in the district health system is by far more integrated and
district health system. HMIS is only one of those channels complex than the formal HMIS, and carries organization-
or sources. The study has confirmed the presence of wide implications. The study results suggest that the
written, verbal, observational, experiential and training actual health management information system involves all
information forms in managerial decision-making, just aspects of organization: human resources, management/
as extant decision-making literature has insisted for organizational processes, organizational structure, and
decades. Yet, this study has gone further to locate these organizational systems. The HMIS is only one of the
various information forms within the process of decision systems in a typical organization (Hardy 1996). In this
making, and establish how they tend to be distributed study, therefore, the realistic informational status of the
over the decision process space and time. formal HMIS within the district health office has been
revealed.
Probably of most significance for policy is the indication
from the study that information in the district health One immediate practical implication is that when deciding
system exists not only in formal HMIS, but is also on installing a new HMIS, diagnosing problems in
embedded in and is brought into the decision-making a troubled existing HMIS, or indeed merely evaluating
process through the whole process of management and key the performance of an established HMIS, practitioners
aspects of organizational routine. In both Lusaka ought to take into account the informational contribution
and Monze districts, information also flowed through of existing human resources, management/organizational
other channels apart from the HMIS. For instance, routine processes and the organizational structure to the total
and other management decision-making meetings were information profile in circulation within the district health
fora for recalling and sharing experiential and training office or system. The study findings suggest that each of
information. Routine supervisory visits to health centres these three organizational elements must be appreciated
provided a channel for gaining verbal and written as a source and/or conveyor of information. HMIS will
information. Routine and specially commissioned moni- not likely succeed in supporting district performance,
toring of activity provided the channel for observational irrespective of success in adoption rate, if these other
information in the district. In addition, task forces specially components of the organization are not strengthened
convened for the decision process also became channels for and aligned for their informational contribution. The very
not only pooling information from various stakeholder design of HMIS must take into consideration the nature
experiences and expert knowledge, but also served as entry of the information ‘gap’ it is coming to fill in the district
points for that information into the decision-making health organization, and not only future interactions
process. Commissioned investigative research and pilot with prospective users – as predominant practice currently
testing of prototype solution designs were channels stands. Thus, sponsors of HMIS in district health systems
for more written information. For some of the should be concerned not only about technology adoption,
decision processes, consultative meetings with as has been the tradition, but also about successful
stakeholders and routine communication activities with technology ‘docking’ into the complex system that the
local communities also provided channels for verbal district health organization is, informationally.
information.
Note that the notion of technology ‘docking’ should not
Thus, information entered the strategic decision-making be confused with the systems approach to technology
process through people (district health managers/staff adoption already argued in the literature (for instance,
directly participating in the decision process); management/ Gladwin et al. 2003). Technology ‘docking’ relies on the
organizational processes (management meetings, super- identified information gap to be filled by the HMIS in
visory visits, task forces, consultation and communication the district health system, and hence necessarily views the
with local communities); organizational structure other key aspects of the organization as components of
(which legitimizes informational contributions); and the broader management information system. This in
the HMIS (as currently conceptualized). From itself suggests a need for a radical re-think of the concept
this collective of aspects of organization emerged and practice of ‘HMIS’.
194
References Mills A. 1990. Decentralization concepts and issues: a review. Part 1.

In: Mills A, Vaughan JP, Smith DL, Tabibzadeh I (eds). Health
Acquah SD. 1994. What do peripheral health care managers expect system decentralization: concepts, issues and country experience.
from the introduction of HMIS? A study into expectations Geneva: World Health Organization.
of peripheral health care managers from the introduction of Mintzberg H. 1975. The manager’s job: folklore or fact. Harvard
HMIS in the Mbeya region of Tanzania. MCH Dissertation. Business Review 53: 49–61.
Liverpool: Liverpool University. Mintzberg H, Raisinghani D, Theoret A. 1976. The structure
Ankrah V, Djan K. 1996. Report on current health management of ‘‘unstructured’’ decision processes. Administrative Science
information system and planning mechanisms in the Eastern Quarterly 21: 246–75.
Region of Ghana. London: Save the Children Fund. MOH. 1992. National Health Policies and Strategies (Health
Barnard K, Lee K, Reynolds J. 1980. Tracing decisions in the NHS. Reforms). Lusaka: Government of the Republic of Zambia.
King’s Fund Project Paper RC 16. London: The King’s Fund. MOH. 1995. National Health Services Act No 22. Lusaka:
Bergman S. 1996. Central Board of Health: Key issues for the Government of the Republic of Zambia.
establishing process. Lusaka: MOH-Government of the MOH. 1996. National Strategic Health Plan. Lusaka: Government
Republic of Zambia. of the Republic of Zambia.
Bodart C, Sapirie S. 1998. Defining essential information needs and Mutemwa RI. 2001. Information and decision-making in district
indicators. World Health Forum 19: 303–9. health systems in Zambia. PhD thesis. Newcastle-under-Lyme,
Boman M, Bubenko JA Jr, Johannesson P, Wangler B. 1997. UK: Keele University, unpublished.
Conceptual modelling. London: Prentice Hall. Nutt PC. 1984. Types of organizational decision processes.
Cassels A, Janovsky K. 1996. Reform of the health sector in Ghana Administrative Science Quarterly 29: 414–50.
and Zambia: commonalities and contrasts. Geneva: World Pettigrew AM. 1990. Longitudinal field research on change: theory
Health Organization. and practice. Organization Science 1: 267–92.
Danish Bilharziasis Laboratory. 2002. Report on the Seminar on Poole MS, Baldwin CL. 1996. Developmental processes in group
Health Management Information Systems and Health Sector decision making. Chapter 8 in: Hirokawa RY, Poole MS (eds).
Monitoring in the Context of Health Sector Reforms, SWAp and Communication and group decision making. London: Sage.
PRSP, Mombasa, Kenya, 2–5 December 2002. Charlottenlund, Simon HA. 1957. Administrative behavior: a study of decision-making
Denmark: Danish Bilharziasis Laboratory. processes in administrative organization. Second edition.
Dean JW Jr, Sharfman MP. 1993. Procedural rationality in the New York: Macmillan.
strategic decision-making process. Journal of Management Simon HA. 1976. Administrative behavior: a study of decision-making
Studies 30: 587–610. processes in administrative organization. Third edition.
Feldman SP. 1988. Secrecy, information, and politics: an essay on New York: The Free Press.
organizational decision making. Human Relations 41: 73–90. Skok W, Kophamel A, Richardson I. 2001. Diagnosing information
Feldman MS, March JG. 1981. Information in organizations as systems success: importance-performance maps in the health
signal and symbol. Administrative Science Quarterly 26: 171–86. club industry. Information and Management 38: 409–19.
Finau SA. 1994. National health information systems in the Pacific Snyder-Halpern R. 2001. Indicators of organizational readiness
Islands: in search of a future. Health Policy and Planning 9: for clinical information technology/systems innovation:
161–70. a Delphi study. International Journal of Medical Informatics
Finlay P. 1994. Introducing decision support systems. Oxford: Blackwell. 63: 179–204.
Gladwin J, Dixon RA, Wilson TD. 2003. Implementing a new health Southon G, Sauer C, Dampney K. 1999. Lessons from a failed
management information system in Uganda. Health Policy and information systems initiative: issues for complex organizations.
Planning 18: 214–24. International Journal of Medical Informatics 55: 33–46.
Guldner M, Rifkin S. 1993. Sustainability in the health sector. Part 1: Taylor FW. 1911. Scientific management. Part One. In: Grusky O,
Vietnam case study. London: Save the Children Fund. Miller GA (eds). 1970. The sociology of organizations: basic
Hardy C. 1996. Understanding power: bringing about strategic studies. New York: The Free Press, pp. 45–52.
change. British Journal of Management 7 (Suppl.): S3–S16. Vaughan JP. 1990. Lessons from experience. Part 3. In: Mills A,
Hornby P, Symon G. 1994. Tracer studies. Chapter 10 in: Cassell C, Vaughan JP, Smith DL, Tabibzadeh I (eds). Health system
Symon G (eds). Qualitative methods in organizational research: decentralization: concepts, issues and country experience.
a practical guide. London: Sage. Geneva: World Health Organization.
Liebenau J, Backhouse J. 1990. Understanding information: an Ward J, Griffiths P. 1996. Strategic planning for information systems.
introduction. London: Macmillan. Chichester: John Wiley.
Lippeveld T, Sauerborn R. 2000. A framework for designing health WHO. 1978. Primary health care: Report of the International
information systems. In: Lippeveld T, Sauerborn R, Bodart C Conference on Primary Health Care, Alma-Ata, USSR, 6–12
(eds). Design and implementation of health information systems. September 1978. Geneva: World Health Organization.
Geneva: World Health Organization, pp.15–32.
Lippeveld T, Sauerborn R, Sapirie S. 1997. Health information
systems–making them work. World Health Forum 18: 176–84. Acknowledgements
March JG. 1982. Theories of choice and decision making.
In: Reader. 1990. General and Strategic Management – 6. The author/researcher would like to express his deepest gratitude to
Glasgow: Strathclyde University Open Learning. the Ministry of Health and Central Board of Health in Zambia for
March JG. 1988. Ambiguity and accounting: the elusive link the permission granted to conduct the study in the two districts. The
between information and decision-making. Chapter 17 in: author/researcher would further like to sincerely thank the District
March JG (ed). Decisions and organizations. Oxford: Blackwell. Health Management Teams and their staff in Lusaka and Monze
Melone NP. 1996. Reasoning in the Executive Suite: the influence districts, for their hospitality, support and time during the data
of role/experience-based expertise on decision processes of collection period. Last but not least, the author’s deep gratitude goes
corporate executives. Chapter 2 in: Meindl JR, Stubbart C, to Dr Steve Cropper, Centre for Health Planning and Management,
Porac JF (eds). Cognition within and between organizations. Keele University, for all the support and academic guidance during
London: Sage. the period of study.
195
The opinions expressed in this paper are those of the author alone in the Centre for AIDS Research. He is also Co-ordinator for the
and do not necessarily reflect formal views of the institutions Community Involvement Group on the Microbicide Development
mentioned. Programme Phase III Clinical Trial (in four African countries)
which is funded by DfID through the UK Medical Research Council
(MRC), and administered by the MRC Clinical Trials Unit (CTU)
and Imperial College in London. His research interests further
Biography include health systems, health economics and organizations.
Richard I Mutemwa is a Research Fellow, Southampton Social
Statistics Research Institute, University of Southampton, United Correspondence: Richard I Mutemwa, Centre for AIDS Research,
Kingdom. He holds a health management MBA and PhD, and has University of Southampton, Highfield, Southampton, SO17 1BJ,
additional previous experience in health and nutrition education and UK. Tel: þ44 2380 597988; Fax: þ44 2380 593846; E-mail:
communication mainly in Zambia. Dr Mutemwa is currently based R.I.Mutemwa@soton.ac.uk
196
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University
Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its
entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org
Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine Health Policy and Planning 2008;23:137–149
! The Author 2008; all rights reserved. Advance Access publication 29 January 2008 doi:10.1093/heapol/czm049
The crisis in human resources for health

care and the potential of a ‘retired’ workforce:
case study of the independent midwifery
sector in Tanzania
Ben Rolfe,1 Sebalda Leshabari,2 Fredrik Rutta3 and Susan F Murray4*
Accepted 27 November 2007
The human resource crisis in health care is an important obstacle to attainment

of the health-related targets for the Millennium Development Goals. One
suggested strategy to alleviate the strain upon government services is to
encourage new forms of non-government provision. Detail on implementation
and consequences is often lacking, however. This article examines one new
element of non-government provision in Tanzania: small-scale independent
midwifery practices. A multiple case study analysis over nine districts explored
their characteristics, and the drivers and inhibitors acting upon their develop-
ment since permitted by legislative change.
Private midwifery practices were found concentrated in a ‘new’ workforce of
‘later life entrepreneurs’: retired, or approaching retirement, government-
employed nursing officers. Provision was entirely facility-based due to regulatory
requirements, with approximately 60 ‘maternity homes’ located mainly in rural
or peri-urban areas. Motivational drivers included fear of poverty, desire to
maintain professional status, and an ethos of community service. However,
inhibitors to success were multiple. Start-up loans were scarce, business training
lacking and registration processes bureaucratic. Cost of set-up and maintenance
were prohibitively high, registration required levels of construction and
equipping similar to government sector dispensaries. Communities were
reluctant to pay for services that they expected from government. Thus, despite
offering a quality of basic maternity care comparable to that in government
facilities, often in poorly-served areas, most private maternity homes were
under-utilized and struggling for sustainability.
Because of their location and emphasis on personalized care, small-scale
independent practices run by retired midwives could potentially increase rates of
skilled attendance at delivery at peripheral level. The model also extends the
working life of members of a professional group at a time of shortage. However,
the potential remains unrealized. Successful multiplication of this model in
1
Research and Reproductive Health Specialist, Options Consultancy Services
London, UK.
2
Assistant Lecturer, Muhimbili University of Health and Allied Sciences
(MUHAS), Dar es Salaam, Tanzania.
3
Research Associate, Maternity Services Research Team, MUHAS, Dar es
Salaam, Tanzania.
4
Reader in International Healthcare, King’s College London, UK.
* Corresponding author. King’s College London, 5.25 Waterloo Bridge Wing,
Franklin Wilkins Building, 150 Stamford Street, London, SE1 9NH, UK.
E-mail: susan_fairley.murray@kcl.ac.uk
137
197
resource-poor communities requires more than just deregulation of private

ownership. Prohibitive start-up expenses need to be reduced by less emphasis on
facility-based provision. On-going financing arrangements such as micro-credit,
contracting, vouchers and franchising models require consideration.
Keywords Human resources, health policy, skilled attendant, retirement, Tanzania,
private sector, qualitative, multiple case study
KEY MESSAGES
! Detail on implementation and consequences of non-government health care provision in specific contexts is important
for guiding policy on human resources for health.
! Following deregulation in Tanzania, independent midwifery practices began to be established by a ‘new’ workforce of
retired Nursing Officers offering personalized care in under-served areas, but delivery coverage is low.
! Sustainability and utilization in poor communities requires supportive measures such as reform of the costly
registration procedures and consideration of on-going financing arrangements such as micro-credit, contracting or
vouchers.
Introduction (Price and Broomberg 1990; Murray 2000). There are a handful of
studies on the attitudes and motivations of doctors in relation to
The human resource crisis in health care means that many
these rates, principally from Latin America (De Mello e Souza
countries are far from reaching the health-related Millennium
1994; Murray and Elston 2005).
Development Goals (MDGs). Factors contributing to this crisis
This article presents findings on the drivers and inhibitors
include mal-distribution and low workforce productivity together
acting upon the development of one new element of non-
with an acute shortage of skilled workers in the government
government provision in Tanzania—the small-scale indepen-
health sector. Losses to other health and non-health sectors can
dent midwifery practice—and considers what contribution this
be as much as 15–40% per year according to estimates from
sector may be expected to make to the MDG target of
Zambia, Ghana and Zimbabwe (High Level Forum on the Health
increasing skilled attendance at delivery. Such independent
MDGs 2004). In sub-Saharan Africa these problems exacerbate
midwifery practices have yet to be the subject of much research
an absolute shortage of health workers. The result is chronic
or evaluation, although they exist in many African settings
under-provision, impacting disproportionately on vulnerable
(Ghana: McGinn et al. 1990, Obuobo et al. 1999; Uganda: Seiber
groups such as women and the rural poor (WHO 2006).
and Robinson-Miller 2004, Agha 2004; Kenya: Yumkella and
One strategy to alleviate the strain upon government services
Githiori 2000), and Southeast Asia (Philippines: John Snow
has been to encourage differing forms of non-government
Inc. 2005; Indonesia: Geefhuysen 1999, Suryanigsih 2005). They
provision (Harding and Preker 2003; Marek et al. 2005), but
have become an explicit element in Safe Motherhood policy to
there are concerns that this may contribute to the further drain of
increase coverage of skilled attendance in Indonesia and have
scarce expertise from public services (Van Lerberghe et al. 2002),
attracted some ‘donor’ attention in Uganda, Kenya and
to inequity of access (Wyss et al. 1996; Benson 2001; Brugha and
elsewhere (see http://www.psp-one.com).
Pritze-Aliassime 2003), and to difficulties of stewardship in
increasingly fragmented systems (Saltman 2000; Sharma 2001).
Certainly, careful analysis of both anticipated and unanticipated Deregulation to permit private provision
consequences of shifts in the balance of mixed economies of in Tanzania
health care are required (Hanson et al. 2001; Brugha and Zwi As yet little consideration has been given to the possible
2002; McKee and McPake 2004). Detailed studies of the positive and negative effects for the workforce, or for public
contextual dynamics and constraints in specific settings can health, of the expanding1 private sector in Tanzania. There has
help develop an understanding of what role non-government been a long tradition of policy focused on creation of a unified
forms of provision will have within the achievement or frustra- health care system provided by government, voluntary faith-
tion of public health goals. based organizations and parastatals with oversight from the
The consequences for maternity care coverage and outcomes Ministry of Health. Facilities run by voluntary faith-based
of the general rise in private sector provision are unclear organizations play an important role as ‘designated district
(Brugha and Pritze-Aliassime 2003), but there are areas of hospitals’, in rural areas. Private for-profit ownership of health
concern. In many countries, private obstetrician-led services are facilities was banned in 1977,2 but reinstated in 1991,3 and by
associated with inappropriately high levels of technological 2001 it accounted for just under 20% of health care facilities
interventions such as induction of labour and Caesarean section in Tanzania. The greatest private for-profit activity is at
198
INDEPENDENT MIDWIFERY SECTOR IN TANZANIA 139
dispensary level, 21% of which were privately owned in 2001 Macro 2005). As rural areas are largely served by low-level cadres
(Ministry of Health 2002). Significant spatial inequalities have (Dominick and Kurowski 2004), many women were probably
emerged with this process, with a tendency for for-profit actually attended by nursing assistants with one year of formal
providers to congregate in the urban areas with existing training (Maestad 2006). Delivery care by family members and by
government provision (Benson 2001). Seventy-eight per cent traditional birth attendants (TBAs) is widespread at 26% and 11%
of the facilities in Dar es Salaam are provided by the for-profit of births, respectively (National Bureau of Statistics [Tanzania]
sector (Ministry of Health 2002). and ORC Macro 2005).
The specific contribution of non-government provision to
maternity care coverage is seldom documented and in Tanzania
The challenge of delivery care coverage
the information is fragmentary. A 2003 estimate of coverage of
Sub-Saharan Africa currently accounts for 47% of all maternal births in Ilala municipality, Dar es Salaam, indicates that one in
deaths (UN Millennium Project 2005). There are ambitions to six deliveries there takes place in private facilities ranging
dramatically increase (to 90%) the proportion of births assisted from large private for-profit and foundation hospitals to small-
by a skilled attendant by 2015 in line with targets set for the scale private and NGO-run clinics (Murray and Nyambo 2003).
MDGs. However, the reality is that levels of skilled attendance In 1997 legislation specifically permitted the establishment of
at delivery increased by only 1% between 1990 and 2003 (UNDP private nursing and maternity homes by Nursing Officers
2005). Increasing rates of skilled attendance at delivery4 in the (Nurses and Midwives Registration Act 1997). Information
context of poorly functioning health systems presents an drawn from the Nursing Council, Ministry of Health, Regional
enormous challenge. It is widely recognized that innovative and District Health Offices, and from the Private Nurses’ and
models of service delivery are urgently needed. Midwives’ Association (PRINMAT) suggests that there are
In Tanzania, the lifetime risk of maternal death is estimated approximately 60 independent midwifery practices, commonly
to be one in ten (WHO 2004). The economic crises of previous known as ‘maternity homes’. Below we describe and con-
decades (Commission for Africa 2005), compounded by some textualize this nascent independent midwifery sector, and use
out-migration of skilled staff (McKinsey & Co 2005) and by these findings to consider its potential within a strategy to
multiple impacts of HIV/AIDS on the workforce (Beckmann increase overall skilled attendance at delivery.
and Rai 2004), are reflected in the deterioration of health care
provision. In 2005 the Joint Annual Health Sector Review
stated that the health worker crisis in Tanzania had reached
emergency proportions. The overall nurse-to-population ratio Methods
was estimated to be 160:100 000 and declining. In some rural Research clearance was obtained from the Tanzania Commission
districts it was just 6:100 000 (High Level Forum on the Health for Science and Technology, and from Muhimbili University
MDGs 2004; Maestad 2006). Accurate data on current work- College of Health Sciences Ethics Committee. An initial national
force composition has been lacking, but the 2001–2 Human situation analysis included 20 key informant interviews with
Resources for Health (HRH) Census indicated that there were senior health planners and representatives from relevant profes-
approximately 13 300 active nursing staff across government sional organizations, plus a review of relevant documentary
and non-government sectors in Tanzania. From this, Kurowski evidence. From these we generated initial hypotheses about the
et al. (2007) estimate that 8940 fulltime equivalent of nurses current social context, organization and delivery of independent
and midwives are engaged in Safe Motherhood interventions. midwifery care in Tanzania. From mid-2003 to mid-2004,
The HRH census also highlighted an ageing health care we tested and extended these hypotheses in a multiple case
workforce, with half over the age of 40. Owing to the employment study. The methodology was chosen for its ability to embrace
freeze in much of the 1990s, the average age of employed health complexity, and to generate and test hypotheses in real
workers increased significantly and high losses due to retirement world settings, where boundaries between phenomenon and
are anticipated over the next decade (Kurowski et al. 2004). context are not clearly evident (Yin 2003). In order to place the
Recently there has been some increasing momentum around midwife-owned practices within their community and health
workforce issues, including establishment of a high level Human system, we used local Council districts that included maternity
Resources Working Group in 2003 and plans that include homes within the range of health care provision as the contextual
increased zonal training for a range of health cadres, but the ‘cases’.
challenges are formidable (Dominick 2004; HERA 2006). Nine case districts (see Tables 1–3) were selected, using a
Rates of skilled attendance at birth (those attended by doctors, purposive sampling strategy. We aimed to include the breadth
nurses, midwives, clinical officers and assistant clinical officers) of geographical, organizational and socio-economic contexts in
fell in Tanzania during the 1990s from an estimated 46% in which private small-scale midwifery practices were thought
1992 to 36% by 1999 (Bureau of Statistics Planning Commission to be operating. Information from the incomplete national
1993, and National Bureau of Statistics 1999, respectively). register was supplemented with information from key infor-
Approximately 84% of health workers are employed in mants such as PRINMAT. Case districts contained between
rural areas serving 80% of the population (Dominick and 1 and 6 maternity homes each. Overall they included 23 such
Kurowski 2004) but this statistic hides geographical disparities practices, some 40% of those operating in Tanzania at that time.
in service coverage and utilization. According to the most recent This range was important in order to build confidence that the
official survey, over 80% of urban women but only 35% of hypotheses might hold in a variety of contexts and therefore be
rural women reported having a skilled attendant for their relevant for informing future policy development for the larger
delivery (National Bureau of Statistics [Tanzania] and ORC workforce.
199
20 0
Table 1 Hypotheses tested over multiple case studies: motivation and supply-side financial issues
Region Kilimanjaro Dodoma Dodoma Dodoma Mbeya Mbeya Coast Mara Mara 140
Moshi Dodoma Mbeya Musoma
District Rural Municipal Kondoa Mpwapwa Municipal Rungwe Kinondoni Municipal Tarime
Code Urban/Rural Rural Urban Peri-Urban Rural Urban Rural Urban Urban Rural Summary of findings
A Maternity home estab- Yes Yes No Yes Yes Yes Yes No Yes Hypothesis supported,
lished at or in anticipation exceptions followed pre-
of retirement from gov- dicted pattern of a minority
ernment sector of mid-career
entrepreneurs.
B Have little or no previous Yes Yes Yes Yes Yes Yes Yes with Yes Yes with Hypothesis well supported.
experience of micro explained explained Exceptions had previously
business exceptions exceptions owned a small pharmacy.
C Maternity home is the sole Yes Yes Limited Yes Yes Yes Limited Yes Limited Hypothesis supported. Most
midwifery activity of the dual dual dual facilities staffed by retired
owner/staff practice practice practice or otherwise unemployed
staff.
HEALTH POLICY AND PLANNING
D Experienced significant Yes Yes Yes Yes Yes Yes Yes Yes Yes Financial barriers in excess
financial barriers to start of original hypotheses.
up
E Empathetic motivation in Yes Yes Yes Yes Yes Yes Yes Yes Yes All owners reported an
response to need in unmet need for MCH ser-
community vices in their communities.
Many also noted lack of
demand.
F Economic motivation to Yes Yes No Yes Yes Yes Yes No Yes Hypothesis supported,
supplement retirement exceptions as predicted for
income a minority of mid-career
entrepreneurs.
G Professional vocation/iden- Yes Yes Yes Yes Yes Yes Yes No Yes Hypothesis well supported.
tity is a motivator
H No recent training received Yes Yes Yes Yes Yes Yes Yes Yes Yes Hypothesis well supported.
in business skills
I Significant administrative Yes Yes Yes Yes Yes Yes Yes Yes Yes Hypothesis well supported,
barriers to start up procedures widely perceived
as obstructive. In most
cases registration incom-
plete 2þ years after
application.
J Maternity homes inte- Limited Limited Limited Limited Limited Limited Limited Yes Limited Hypothesis not well sup-
grated into district health ported. Integrated for
system referral purposes, but lim-
ited support for day-to day
operation & supplies.
K Income shortfall including Yes Yes Yes Yes Yes No Yes No Yes Hypothesis well supported.
difficulties paying staff Exceptions were clinics on
main roads into large
towns.
L Homes facing sustainabil- Yes Yes Yes Yes Yes Yes Yes No Yes Hypothesis well supported.
ity difficulties One exception operating on
significantly different busi-
ness model, comparable to
a doctor-led dispensary.
Table 2 Hypotheses tested over multiple case studies: location, range and quality of independent midwifery services
Region Kilimanjaro Dodoma Dodoma Dodoma Mbeya Mbeya Coast Mara Mara
Moshi Dodoma Mbeya Musoma
District Rural Municipal Kondoa Mpwapwa Municipal Rungwe Kinondoni Municipal Tarime
Urban/Rural Rural Urban Peri-Urban Rural Urban Rural Urban Urban Rural Summary of findings
M Average distance from Yes Yes Yes Yes Yes Yes Yes Yes Yes Hypothesis well supported.
urban centre greater for
midwife-owned facilities
than for doctor-owned
dispensaries
N Underserved location near Not tested Not tested Yes Yes Yes Yes Unclear Yes Unclear Hypothesis supported by
transport corridor asso- theoretical replication. Dar
ciated with self-reported es Salaam not tested due
clinic sustainability to geographical
complexity.
O Basic MCH services Yes Yes Yes Yes Yes Yes Yes Yes Mostly Hypothesis well supported.
provided
P Integrated into district Limited Limited Limited Limited Limited Limited Limited Yes Limited Hypothesis not well sup-
health system ported. Integrated for
referral purposes, but lim-
ited support for day-to day
operation (registration,
supplies etc.).
Q Absence of suitable quality Yes Yes Yes Yes Yes Yes Partial Yes Yes Hypothesis well supported.
assurance mechanism Plans for PEER supervision
by Private Nurses and
Midwives Association may
partially remediate.
R Quality of personal care Yes Yes Yes Yes Yes Yes Yes Yes Yes Hypothesis well supported.
reported by community to
be superior to that in
government facilities
S Quality of maternity ser- No Yes No Yes Yes Yes Yes Yes Yes Hypothesis well supported
vices provided as good as but complex. Highly vari-
or better than in nearest able quality in both sec-
government dispensaries tors. Many government
facilities either do not
conduct deliveries or are
open short hours.
T Few clinical training Yes Yes Yes Yes Yes Yes Yes Yes Partial Hypothesis well supported.
opportunities available to
extend skills
U Cell phone technology Yes Yes Yes No Yes No Yes with two Yes No Hypothesis supported.
available for emergency exceptions Situation recently much
communications improved. Limited trans-
INDEPENDENT MIDWIFERY SECTOR IN TANZANIA
port available however.

V Local community benefits No No No Yes No No Yes Yes No Hypothesis not well sup-
from improved access to ported. Transport facilities
141
transport for obstetric and similarly poor across sec-

other emergencies tors. Emergency finance
available in private sector.
201
Hypothesis broadly supported.
those very close to an equiva-
Tanzania, exacerbated by cost
vate services was widespread.

Exceptions were predicted as
Whilst not universal, opposi-

For each case study, health care provision was mapped.
tion to the expansion of pri-
largely based on the doctor-

Hypothesis supported, parti-
Reflects general situation in
owned facility-based model.

cularly for delivery services.
Public perception, however,

and negative perception by
Hypothesis well supported.
Hypothesis not supported.

Qualitative and quantitative data were collected from the range
lent government facility.

Summary of findings of sources listed in Box 1. In total 125 in-depth interviews and 58
focus group discussions (FGDs) were conducted, in English or
wider community.
Kiswahili according to respondent preferences. Case studies were
conducted consecutively; the iterative research design allowed
further hypotheses to be generated and tested as data collection
progressed. Tables 1–3 present the key hypotheses relevant to this
article within a data matrix and these are cross-referenced in the
text. Hypotheses fell into three broad groupings: those concerned
Tarime
with motivations and relationships to the wider health care

Rural
Mara
system; those concerned with location, range and quality of

Yes
Yes
Yes
No
services; and those concerned with demand-side issues of
acceptability and utilization. Each hypothesis was tested against
Municipal
Musoma
Table 3 Hypotheses tested over multiple case studies: demand-side financial issues, acceptability and utilization of independent midwifery services
the triangulated data for each case derived from the sources listed
Urban
Mara
in Box 1, and supported or modified through pattern replication

Yes
Yes
No
No
over the multiple case studies. Hypotheses could thus be

supported by literal replication across multiple case studies (in
Kinondoni
acceptance
which theoretically predicted replications occur consistently in

limited
Urban
Mostly
data, as in hypothesis D in Table 1) or by theoretical replication

Coast
Some
(in which contrasting results are theoretically predicted, as in

Yes
No
hypothesis F in Table 1).

All respondents gave informed consent to be interviewed,
Rungwe
Mbeya
Rural
consistent with guidelines for interviewing literate and non-

Yes
Yes
Yes
No
literate subjects (Nuffield Council on Bioethics 2002). Consent of

local leaders for conduct of the study in each district and village
Municipal
was also obtained. Instruments were drafted in English and

Mbeya
Mbeya
translated and adapted for conceptual equivalence in Kiswahili by

Urban
bilingual members of the research team. A sample was

Yes
Yes
Yes
No
independently back-translated and checked.5 Interviews and

FGDs were tape-recorded, transcribed and, where applicable,
Mpwapwa
Dodoma
translated into English by experienced Tanzanian social research-

Rural
ers. In the few interviews where tape recording was not possible6
Yes
Yes
No
No
detailed notes were taken. The 220 primary documents were

analysed in English using Atlas Ti 5 by Scientific Software.
Peri-Urban
Dodoma
Kondoa
Yes
Yes
Findings
No
No
At the time of the data collection the formal independent

Municipal
Dodoma
Dodoma
midwifery sector in Tanzania consisted of about 60 small-scale

Urban
facility-based practices providing antenatal and childbirth care

Yes
Yes
Yes
No
within a range of primary care services. Private midwifery

practices were found concentrated in a ‘new’ workforce: retired,
Kilimanjaro
or approaching retirement, government-employed Nursing

Not tested
Officers (Table 1: A) who made the switch’ to self-employment

Moshi
Rural
Rural
late in life and could be characterized as ‘elder’ or ‘later-life

Yes
Yes
Yes
entrepreneurs’ (Spoonley et al. 2002; Weber and Schaper 2003).

The median age of the 23 owners we interviewed was 54 years.
Concept of private professional
waiting time) cited as barrier
Opportunity costs (travel and
Low demand in communities
Of these the vast majority were business novices, new to self-

for private maternity services
midwifery provision accepta-

to utilization of government
employment and entrepreneurship (Table 1: B). A small

Private maternity homes
number had a background of ‘serial’ micro-businesses run in

parallel with their main government employment. This is not
maternity services
ble in community
uncommon in Tanzania where nursing salaries are often

Urban/Rural
supplemented with other petty income-generation activities.

under-used
Here they were used to generate the capital necessary to set up

District
Region
the independent practice:
I was shown the Ministry of Health guidelines and saw that they
W
had so many requirements. So I started a small business of keeping

X
202
Box 1 Data sources and sampling used in regulations did not permit them to live in the premises, a
the analysis location close to their residence was required for provision of
24-hour ‘cover’. Some therefore made compromises on the
optimal location, with eventual implications for ease of
National and Regional level health management
financial sustainability (Table 2: N).
information system data, interviews with senior managers
Government regulations stipulate that services provided by
and documentary review.
nursing and maternity homes must focus around maternal and
At District level in each of nine case study districts: child health (Table 2: O). All homes provided antenatal care
and were equipped at least in basic fashion to attend deliveries,
! District Health Management Team Members but we found that most practices actually attended only a few
interviewed (District Medical and Nursing Officers) births per month (range 0 to 26; median of 3 births/month;
! All owner-managers of existing and recently closed Table 3: W). Most practices also provided more remunerative
private maternity and nursing homes, in-depth interview. minor curative care; some employed Clinical Officers. A signifi-
! All clinical staff members employed at active cant part of income came from selling non-prescription drugs
private maternity homes, in-depth interview. for malaria and minor illnesses. Some also provided home-
! Clinic inspection checklist for basic equipment based care for HIV/AIDS, ‘youth-friendly’ reproductive health
and other physical attributes completed at all active services, and child growth monitoring.
private maternity homes and nearest equivalent
public sector facility.
! Public sector staff working at the nearest ‘equivalent
Determinants of individual engagement in
public facility’: two oldest midwives on shift
independent midwifery
at first visit. The motivational aspirations of these independent providers
! Public sector nurses near retirement, interview: encompassed economic, caring and professional goals. Reported
two oldest midwives aged over 55 years on shift ‘push’ factors without exception centred around financial
at first visit. insecurity: extremely poor government sector salaries, inade-
! Retired Nursing Officers, FGD: snowball sampled quacy of pensions and fear of a decline into poverty after
from older nurses at District Hospitals. retirement (Table 1: F).
! Public and private users, FGD: approximately eight users Reported ‘pull’ factors often focused on financial rewards
with youngest children recruited at immunization or expressed as a stable income source rather than significant
growth monitoring clinics. profits. Additionally cited were flexible working hours and what
! Separate female and male community members, Kendall et al. (2002) call ‘mercantile motivation’—the sense of
FGD: participants recruited using ‘ten cell leader’ autonomy and achievement to be gained from running one’s
nearest the private clinic, where possible one participant own small business venture. Motives also included concern for
from each ‘cell’ or street. the health and well-being of women in labour and satisfaction
! ‘Younger’ nurses, FGD: all available Nursing Officers in meeting the needs of under-served communities (Table 1: E).
under 30 years at district hospital. Sometimes activity had been initiated in response to a perceived
need, other times because of repeated requests for services.
Respondents frequently expressed the desire to ‘use one’s
talents’, not to ‘sit idle’ after retirement from government
pigs, started with one male and two female pigs during the rainy
employment. Linked to this was a desire to maintain social
season where it was easy to obtain food to keep them. I got eight
standing through a professional identity (Table 1: G).
piglets from those two females; I sold the first eight piglets and got
Focus groups with soon-to-retire public sector nursing officers
money to make a local delivery bed; I kept the other eight together
and nurse-midwives in all nine districts confirmed the general
with their mothers. Those two females gave birth again, and as the
applicability of these various push and pull factors, and
dry season was getting near I decided to sell them all. I went to the
suggested that opening an independent practice may be an
mission hospital where I worked before, they sold me some used
attractive idea to many. However, successful multiplication of
equipment. (Midwife, Mbeya Region)
the small-scale midwifery practice model is dependent also on
the dynamics of the social and institutional environment, and
Some home owners continued to be involved in micro-business
here we found there were considerable barriers in spite of the
activities such as keeping chickens, but their maternity
legislated deregulation.
practices, often with pharmacies attached, were their core
work activity and represented a significant investment of scarce
financial resources (Table 1: C,D). A strong service ethos was Low levels of demand
also consistently represented in their accounts of their activity, The case studies indicate that individual users valued the
particularly amongst those falling into the ‘later life entrepre- proximity of the maternity homes. They would trade off the
neur’ typology (Table 1: E). costs of user-fees against the opportunity and financial costs of
Most of the maternity homes were in rural or peri-urban transport to government services further afield (Table 3: X) and
areas, distinguishing them from doctor-run clinics (Benson against the unpredictable ‘under the table charges’ (Abel-Smith
2001; and Table 2: M). Most owners of maternity homes in our and Rawal 1992) often encountered there. However, most of
study had attempted to locate these in previously under-served the private maternity practices still suffered from chronic
areas adjacent to key transport corridors. However, as under-utilization, in relation to their capacity and to the local
203
need for midwifery and other health care (Table 3: Y). This was (Rutashobya 1998; Chijoriga et al. 2002; Stevenson and St-Onge
due to low interest in professionally attended childbirth in 2005). The midwives consistently reported credit to be
facilities amongst rural communities, and to seasonally variable expensive and hard to access (Table 1: D). None of those
incomes and scarcity of cash to spend on health care. It also interviewed had the business plan that would normally be
reflected some antagonism on ideological grounds from com- required to demonstrate project viability prior to a loan being
munities to private sector expansion (Table 3: Z). Community granted.
focus groups indicated that notions of citizen rights to health
care are still strong. Where local people had contributed to the
Unrealistic specifications
building of local public dispensaries, for example, they expected
These difficulties were compounded by the high start-up costs
to continue to be provided with government services. Even
of a home (US$5000–10 000), which represented a large
where extended kinship and tribal networks might seem to
financial risk even to those with access to capital. Most
provide a natural client base for midwives returning home to
maternity home owners in the study had invested their entire
their village, the reality can be more complex because of
savings and pensions into the venture. These high costs were
expectations that such neighbourly services be provided without
due to infrastructural specifications required by the Ministry of
charge.
Health which mirrored the physical and human resource
Such demand-side inhibitors caused demoralization and
criteria specified for public sector dispensaries (an eight-room
discouragement among the majority of these private midwifery
facility with generator, oxygen and various staff). Such
providers, who were unable to actualize their aspirations for
‘minimum requirements’ were unrealistic for independent
their practices. This was compounded by the lack of business
providers working at peripheral level and too expensive to be
skills (Table 1: H) that might have helped them to adapt their
easily sustainable given the prevailing economic conditions in
approach to accommodate a relatively hostile environment.
rural areas. None of the owners reported making a profit
comparable to the salary that they previously received in the
Restrictions on ownership public sector.7 Some homes did provide employment and
Legislation restricted ownership of these facilities to Nursing informal in-service training for nursing staff, but these certainly
Officers who are a key cadre and compose the most senior third posed little threat to the government sector with respect to
of professional nurse-midwives in the country (http:// poaching of staff, as such staff were being paid irregularly
www.nbs.go.tz/health.htm, accessed 6 July 2006). Other less (Table 1: K).
senior midwives who may have many years of recent ‘hands on’
experience of maternity care had no approved route to self-
Further constraints on profitability
employment within their profession. While probably serving to
Inconsistent and unclear policy relating to charging structures
contain early- to mid-career ‘leakage’ from the government
and taxation compounded difficulties for the maternity home
workforce, this limited the size of the post-retirement pool of
owners. Government pronouncements on exemption from user
self-employed midwives.
fees for maternal and child health services were widely
understood by the population and some district health
Bureaucratic constraints managers to imply free services in all sectors, although there
The complex registration procedures for nursing and maternity was no mechanism to reimburse small-scale providers of care
homes tended to be poorly understood by local health managers such as the maternity homes. Additionally, small health care
whose role was to inspect the facilities (Table 1: I; Table 2: P). facilities were charged for tax and business licences in the same
They also required coordination and communication between way as profitable commercial businesses. In the context of high
different levels of the system that was unrealistic for a start-up costs and low demand from poor communities, such
struggling health care bureaucracy. Many practices reported institutional behaviours served to further limit the financial
that they had been unable to complete the registration viability of the sector (Table 1: L).
procedure over a number of years. Tanzanian territory covers
some 945 000 km2, but to comply with rules for national
Weak integration in the local health system
registration of homes after approval, midwives needed to
Management systems for the regulation of private facilities
travel personally to Dar es Salaam to pay the fees, incurring
were extremely weak at all levels. These private practices were
significant travel and opportunity costs. These barriers were
less well integrated into referral and administrative networks
compounded by a generally difficult environment for commer-
than equivalent level public facilities (Table 1: J). This was
cial activity. The banking, business licensing and taxation
reflected in generally poorer access to on-going training,
systems all present obstacles to such small-scale entrepreneurs.
supplies and supervision. Regulation and support of private
facilities was highly dependent on the inclinations of individual
Barriers to accessing set-up finance District and Regional Medical Officers. Some maternity homes
Shortage of capital and lack of appropriate and unsecured credit were actively supported and given vaccines, drug fridges and
represents a major obstacle to the expansion of female delivery registers from district stores, but many received no
enterprises in many settings (Epstein 1993; Mayoux 2001). In support. District supervision of private facilities existed in
Tanzania, women face socio-cultural and institutional barriers, theory, but it was limited, as it is in the public sector, by lack of
often lacking the ownership title to land and property needed to vehicles or fuel (Table 2: Q). Reports of experiences from
meet prescribed collateral requirements for commercial loans countries such as Ghana (Obuobo et al. 1999) had led to an
20 4
initial hypothesis of resistance from government health workers Limited skills in the early detection of obstetric complications
to receiving such referrals from private care. This was not are compounded when facilities are geographically isolated. The
supported in our case studies, often because the maternity median distance from the maternity home to the nearest
home owners could draw upon their long government sector district referral hospital was 9 km. The furthest in the case
careers for credibility. study districts was 65 km. Half were over 30 km away and on
very poor roads. Communication and transportation in an
obstetric emergency was therefore an important issue. Recent
Using an independent midwifery practice—what advances in communications technology have been important
quality of care? in reducing some of the isolation of small clinics and most
Concerns about obstacles to the maternity homes’ sustainability practices surveyed did have telephone communication, usually
rest upon an implicit assumption that they can, under current a cellphone (Table 2: U). None of the maternity homes had a
or more favourable circumstances, offer women a good quality formal emergency transport plan (Table 2: V), but all facilities
service. We used an equipment and services checklist to assess reported some established method for emergency referral.
quality of care in the maternity homes, and triangulated the Transport was much more readily available for those in peri-
findings with narratives from users. Quality was similarly urban settings—in most cases using public transport (taxi or
assessed at the nearest comparable government facility. bus)—and far more limited in rural conditions. The costs of
Quality of personal care was reported by community members referral were significant and in all cases borne by the client,
and by providers to be far superior in these private practices to although some maternity homes reported lending money in
that in government facilities. As a ‘relational good’ (Kendall emergency cases. Whilst referral for complications was often
et al. 2002), personal interactions have important implications difficult to accomplish quickly, it was just as difficult for
for quality of care in pregnancy and particularly in childbirth, equivalent local government facilities which also lacked their
but these are often neglected in government facilities. Verbal own motor transport, and expected the referred patient to bear
and sometimes physical abuse by midwives in the public sector the costs of transfer.
featured frequently and consistently in women’s accounts of
their care from all the case districts, and it was reported in user Public health implications—what does the
focus groups to be a major deterrent to seeking care at the independent midwifery sector offer for increasing
government facilities (Table 2: R). coverage of skilled attendance for childbirth in
One study of antenatal care in Dar es Salaam (Boller et al. countries like Tanzania?
2003) highlighted that technical quality of care is related to the The findings presented here suggest that small-scale indepen-
cadre of staff providing the care, and found that 80% of dent midwifery practices may have potential to contribute to
antenatal care in their sample of public facilities was provided rates of ‘skilled attendance’8 for delivery at peripheral level.
by MCH Aides, with only a two-year basic training. This can be These ‘nursing homes’ or ‘maternity homes’ do not possess
compounded in rural areas by high vacancy rates and low some of the negative attributes associated with doctor-owned
motivation in staff. We found that some of the private private for-profit services, such as concentration in better-off
maternity homes also were staffed by lower cadres of staff urban areas and over-intervention (Mackintosh and
such as MCH Aides when the owner was absent. Such Tibandebage 2002). Doctors owning dispensaries often practice
situations tended to occur in the cases where the owner- multiple job-holding or ‘dual practice’ in public and private
manager had other professional commitments elsewhere. The sectors (Van Lerberghe et al. 2002; Harrington 2003), but we
technical quality of care was basic at the maternity homes, but found little dual practice among these independent sector
it was similar to that offered by equivalent government facilities midwives. Independent practice is currently seen primarily as a
(Table 2: S). Shortages of basic drugs and equipment were post-retirement option,9 so there is little drain on, and more
common to both. In the public sector, these were caused by complementarity with, the government sector maternity
irregular supplies from medical stores, in the private sector by workforce.
insufficient capital to pre-purchase from commercial sources However, this form of provision has yet to make any signi-
and lack of access to discounted supplies from government ficant contribution to rates of skilled attendance at delivery. To
medical stores. make a contribution of 1% to national coverage of deliveries,10
On-going professional development was extremely limited for example, all the existing independent midwifery practices
amongst independent sector midwives, the exception being would each need to be providing, every week of the year,
clinical updates offered by PRINMAT as part of their annual delivery care to 4–5 women who would not otherwise have
conference events (Table 2: T). Private sector midwives reported obtained professional care from the health care system, and this
that they were almost never invited to update-training arranged level of activity is not currently being met. The average volume
by the government sector. However, such resources are in short of deliveries attended in the maternity homes is not high or
supply and many public sector midwives also receive little in- sustained, for all the reasons already outlined. Furthermore,
service training. At the time of data collection, for example, some of those women using private maternity homes are indi-
none of the practising midwives interviewed in either sector in viduals who are substituting delivery care in the public sector
the study districts had received any specialized in-service for private sector treatment, representing little net gain in
training in managing obstetric emergencies, and we found overall rates of skilled attendance.
only erratic use of the partograph to monitor well-being in Structural changes in the health sector labour market, includ-
labour in both sectors. ing a public sector employment freeze in 1993 and an increase
20 5
in the retirement age from 55 to 60 in 1999 (Kyejo 2001), and prevented other interested midwives from engaging in such
contributed to an ageing health care workforce (Kurowski et al. activity.
2004). There will be a large cohort of retiring midwives over the As a result of this research a special working group of the
next few years, and our data, derived from case studies in a Nursing Council, including private practitioners, drafted new
variety of districts, suggests that returning to home villages may guidelines in late 2004. These are based on the intended care
be quite a common practice at retirement. However, to harness rather than the current blueprint facility-based specification.
this resource, and indeed for any significant expansion of this They will reduce start-up costs and should allow private
sector, reduction of the legal and institutional barriers will be practitioners to tailor their services according to their skills
needed. and local needs, and open up a future possibility of domiciliary
midwifery care. The Registrar’s office is also considering
revising legislation to allow Enrolled Nurse-Midwives and
Nursing Officers to set up these practices, thus expanding the
Discussion potential private midwifery workforce.
Brugha and Zwi (2002), in their review on the evidence for Increasing the size of the maternity workforce can only be part
global approaches to private sector provision, end with a strong of the solution. ‘Skilled attendance’ requires at least two key
plea for caution in implementation of policies to enhance the components: a skilled attendant and an enabling environment
private sector’s role in delivering health care, and a call for that includes equipment, supplies, drugs and transport for
more detailed research to inform this. Health systems need to referral, and backup emergency obstetric care (EmOC) (Bell
be understood within their local social and political contexts, et al. 2003). This requires lifting of current restrictions that
and such case studies using multiple sources and methods of prohibit midwives from dispensing the full list of drugs suggested
data collection are labour intensive, but as Keen and Packwood for routine delivery and basic EmOC (WHO 2003). Health services
(2000) argue, they prove valuable in situations where policy in Tanzania are currently undergoing a process of decentraliza-
change is occurring in ‘messy real world settings’. tion and the responsibility to ensure facilities have affordable
One advantage of the approach is its ability to start with access to essential drugs and equipment falls to district managers.
generic questions and to become more focused and specific as Whilst providing free or discounted supplies to facilities operating
knowledge of the subject matter increases. We did not know on a market model may seem generally counter-intuitive, the
when we set out to map this sector that we would find the supply of basic equipment to self-employed midwives operating
post-retirement model of maternity home ownership to be so on a subsistence basis in under-served areas may keep their
predominant in Tanzania at the current time. When we then practice afloat and affordable.
conducted a search for documentation on the mobilization of If retiring nurse-midwives take up the possibility now
‘mature’ or ‘retired’ workforces to compensate for shortages in theoretically open to them, and devise more tailored low cost
health care resources, we found recent reference to ‘flexible services that do not simply attempt to replicate government
retirement’ and ‘retire and return’ policies in industrialized facilities, then their potential to create ‘something new and
countries such as the UK (DoH 2006; Nursing Research Unit different’ (Drucker 1985; Faugier 2005) and be more truly
2007) and Australia (NSW DADHC 2000). We did not, however, ‘entrepreneurial’ may be realized. For example, developing new
identify any research studies on the implementation of such services and extending into domiciliary clinical practice may be
strategies in low-income countries. The potential of the a greater possibility. It remains to be seen whether this is
untapped pool of skilled health workers represented by retired attractive to the midwives themselves. Despite the problems
workers is beginning to be recognized in Africa as one within a faced, many of the owners we met were immensely proud of
series of measures to increase inflows of human resources their clinics, which represented personal achievement and
(High-Level Forum on the Health MDGs 2004; Maslin 2005; social standing. It may be that ownership of one’s own clinic
Global Health Workforce Alliance 2006). Our findings do will remain a powerful motivator.
suggest that there may be scope, in this Tanzanian context at If sustainability and the needs of poor communities are to be
least, for encouraging retired nurse-midwives to develop properly addressed then on-going financing needs to be
independent practices in under-served areas within a network considered. There would seem to be some real benefits in
of coordinated and supported health services, although it is combining public finance with private provision in this scenario
necessary to be cautious about extrapolations from a small because of the potential to draw in a ‘new’ workforce, rather
group of early adopters to the wider workforce. What does seem than simply to replace public with private provision. Other
clear is that for any such scenario to function optimally, countries offer some examples of targeted micro financing:
changes are needed at several levels and that supply and micro-credit lending to users increased the use of trained TBAs
demand-side barriers need to be taken into consideration. in Bangladesh; micro-loans to private midwives in Uganda
contributed to improved quality of services (Walker et al. 2001);
and targeted performance-based contracts have been combined
Moving forward with vouchers distributed to potential users in Indonesia
Our study demonstrates the real life complexity of enactment of (Institute for Health Sector Development 2004). Franchising
a policy ‘good idea’. The proprietors of the private practices we models piloted in the Philippines (John Snow Inc. 2005) may
studied aspired to combine financial, caring and professional be possible via a private midwives’ association such as
aims, but despite the legislative change, they faced institutional PRINMAT. Such approaches would merit pilot studies in
barriers that systematically failed to support these aspirations Tanzania.
20 6
Because of their location and emphasis on personalized care, Beckmann S, Rai B. 2004. HIV/AIDS, work and development in the United
small-scale independent practices run by retired midwives could Republic of Tanzania. Geneva: International Labour Organization.
potentially—with the right support—increase rates of skilled Bell J, Hussein J, Jentsch B et al. 2003. Improving skilled attendance at
attendance at peripheral-level delivery. They cannot be seen as delivery: a preliminary report of the SAFE strategy development
more than one strand in the human resources strategy required tool. Birth 30: 227–34.
to bring skilled attendance at delivery to the majority of Benson JS. 2001. The impact of privatization on access in Tanzania.
Tanzanian mothers, but this model may represent an opportu- Social Science and Medicine 52: 1903–15.
nity to harness currently under-utilized human resources in the Boller C, Wyss K, Mtasiwa D, Tanner M. 2003. Quality and comparison
push towards Safe Motherhood and the MDGs. of antenatal care in public and private providers in the United
Republic of Tanzania. Bulletin of the World Health Organization 81:
116–22.
Brugha R, Pritze-Aliassime S. 2003. Promoting safe motherhood through
Endnotes the private sector in low and middle income countries. Bulletin of
1
the World Health Organisation 81: 616–23.
Estimated to account for 6% in 1995, rising to 14% in 2002 (Dominick
2004). Brugha R, Zwi A. 2002. Global approaches to private sector provision:
2
Private Hospital (Regulation) Act 1977. where is the evidence? In: Lee K, Buse K, Fustukian S, (eds).
3 Health policy in a globalising world. Cambridge: Cambridge University
Private Hospitals (Regulation) (Amendment) Act 1991.
4
A ‘skilled attendant’ is defined by the World Health Organization and Press, pp. 63–78.
the professional confederations (WHO, ICM, FIGO 2004) as ‘an Bureau of Statistics Planning Commission, Tanzania. 1993. Tanzania
accredited health professional . . . who has been educated and
demographic and health survey 1991/1992. Dar es Salaam and
trained to proficiency in the skills necessary to manage normal
pregnancy, childbirth and the immediate post partum period, and Columbia MD: Macro International Inc.
in the identification and management and referral of complications Chijoriga M, Olomi DR, Nchimbi M. 2002. Jobs, gender and small
in women and newborns’. entrepreneurship in Tanzania: factors affecting women entrepre-
5
English and Kiswahili versions of the research tools are available neurs in the MSE sector. Primary research synthesis, Nov. Dar es
online at http://www.kcl.ac.uk/teares/nmvc/research/project/more Salaam/Geneva: UDEC/International Labour Organization.
info.php?id ¼ 76&the_group ¼ 1.
6
A small number of interviews and FGDs were not recorded, usually Commission for Africa. 2005. Our common interest: Report of the
where a respondent expressed such a preference due to excessive Commission for Africa. London: Commission for Africa.
background noise or technical failure. De Mello e Souza C. 1994. C-sections as ideal births: the cultural
7
For Nursing Officer grade, the salary is about US$80 a month construction of beneficence and patients’ rights in Brazil. Cambridge
8
The broader question of ensuring skilled attendance requires a well-
Quarterly of Healthcare Ethics 3: 358–66.
functioning health system overall. This will require improving
training, drugs and equipment supplies, and transportation net- Department of Health (DoH). 2006. Recruitment and retention of
works not only for the maternity homes but also across the public mental health nurses. The Chief Nursing Officer’s review. London:
sector as already indicated. Department of Health.
9
The age profile of employed midwives has changed in recent years Dominick A, Kurowski C. 2004. Human resources for health – an
partly due to a public employment freeze in 1993 and partly due to
appraisal of the status quo in Tanzania mainland. Washington, DC:
mortality in younger age groups. Fifty per cent of health staff are
estimated to be over 40 years old (Kurowski et al. 2004). World Bank. Online at: http://www.ilo.org/public/english/
10
Figure based upon 2004 population estimate of 36 588 225 and protection/trav/aids/publ/cp_2_tanzania.pdf
39 births per 1000 population, from http://www.odci.gov/cia/ Drucker P. 1985. Innovation and entrepreneurship. New York: Harper & Row.
publications/factbook/geos/tz.html, accessed 3 February 2005.
Epstein TS. 1993. Female petty entrepreneurs and their multiple roles.
In: Allen S, Truman C, (eds). Women in business: perspectives on
women entrepreneurs. London: Routledge Press, pp. 14–27.
Acknowledgements Faugier J. 2005. Developing a new generation of nurse entrepreneurs.
Nursing Standard 19: 49–53.
This research was funded with a grant from the Health
Foundation (Ref 1824/1506). The authors thank M Joyce Geefhuysen CF. 1999. Safe motherhood in Indonesia: a task for the next
century. In: Berer M, Ravindran TKS, (eds). Safe Motherhood
Safe, Mr Godfrey Mohamella (GM) and Mr Gustaf Moyo,
initiatives: critical issues. London: Reproductive Health Matters.
Ministry of Health, United Republic of Tanzania; Dr Thecla
Global Health Workforce Alliance. 2006. Sharing Experiences Meeting
Kohi and Dr Helen Lugina, Muhimbili University College of
Report, Lusaka Zambia, 4–6 April 2006. Geneva: World Health
Health Sciences, Tanzania; and Ms Pauline Khng, King’s
Organization.
College London, for their assistance, and the communities,
Hanson K, Archard B, McPake B. 2001. Creating markets in hospital
practitioners and managers who took part.
care: the adoption of developed country health sector reforms by
developing countries. Is it appropriate? Tropical Medicine and
International Health 6: 747.
Harding A, Preker AS. 2003. Private participation in health services.
References Washington, DC: World Bank.
Abel-Smith B, Rawal P. 1992. Can the poor afford free health-services – Harrington JA. 2003. Law and the commodification of health care in
a case-study of Tanzania. Health Policy and Planning 7: 329–41. Tanzania. Law, Social Justice & Global Development 2: [e-journal].
Agha S, Balal A, Ogojo-Okello F. 2004. The impact of a microfinance Health Research for Action (HERA). 2006. District health services
program on client perceptions of the quality of care provided by delivery in Tanzania, where we are in terms of quantity and quality
private sector midwives in Uganda. Health Services Research 39: of health care provision? Online at: http://www.herabelgium.com,
2081–100. accessed 5 May 2007.
207
High Level Forum on the Health MDGs. 2004. Health workforce Tanzania. Country Report. AG: 3128. London: Options Consultancy
challenges: lessons from country experiences. Online at: http:// Services and King’s College London.
www.hlfhealthmdgs.org/Documents/HealthWorkforceChallenges- Murray SF. 2000. Relation between private health insurance and high
Final.pdf, accessed 1 July 2006. rates of casearean section in Chile: qualitative and quantitative
Institute for Health Sector Development. 2004. Private sector participation study. British Medical Journal 321: 1501–5.
in health. London: Institute for Health Sector Development. National Bureau of Statistics, Tanzania. 1999. Tanzania – reproductive and
John Snow Inc. 2005. Social franchising health services; a Philippines child health survey. Dar es Salaam and Columbia, MD: Macro
case study and review of experience. Boston: JSI. International Inc.
Keen J, Packwood T. 2000. Using case studies in health services and National Bureau of Statistics, Tanzania and ORC Macro. 2005. Tanzania
policy research. In: Pope C, Mays N, (eds). Qualitative research in demographic and health survey 2004–05: key findings. Calverton, MA:
health care. London: BMJ Books, pp. 50–8. ORC Macro.
Kendall J, Matosevic T, Forder J et al. 2002. The motivations of NSW DADHC. 2000. From awareness to action. Mature Workforce Retention
domiciliary care providers in England: building a typology. LSE Project. Sydney: Department of Ageing, Disability and Home Care.
Health and Social Care Discussion Paper No. 8. London: London Nuffield Council on Bioethics. 2002. The ethics of research related to health-
School of Economics and Political Science.
care in developing countries. London: Nuffield Council on Bioethics.
Kurowski C, Wyss K, Abdulla S, Mills A. 2007. Scaling up priority health
Nursing Research Unit. 2007. Will an ageing nursing workforce work?
interventions in Tanzania: the human resources challenge. Health
Policyþ 2:1–2. London: King’s College London.
Policy and Planning 22: 113–27.
Obuobo AAD, Pappoe M, Ofosu-Amaah S, Boni DY. 1999. Private health
Kurowski C, Wyss K, Abdulla S et al. 2004. Human resources for health:
care provision: Greater Accra Region of Ghana. Bethesda MD:
requirements and availability in the context of scaling-up priority
Partnerships for Health Policy Reform Project Associates Inc.
interventions in low-income countries. Case studies from Tanzania
and Chad. HEFP Working Paper 1/4. London: London School of Price M, Broomberg J. 1990. The impact of the fee-for-service
Hygiene and Tropical Medicine. reimbursement system on the utilisation of heatlh services.
A comparison of caesarean section rates in white nulliparous
Kyejo KP. 2001. Tanzania. In: Technical assistance in fiscal policy and
women in the private and public sectors. South African Medical
tax administration in developing countries. Keio University
Journal 78: 136–38.
Research Papers, Tokyo. Online at: http://www.ic.keio.ac.jp/
jjwbgsp/researchpapers.html, accessed 1 July 2006. Rutashobya LK. 1998. Women entrepreneurship in Tanzania: entry and
performance barriers. GIRR No. 9. Addis Ababa: Organization for
Mackintosh M, Tibandebage P. 2002. Inclusion by design? Rethinking
Social Science Research in Eastern and Southern Africa.
health care market regulation in the Tanzanian context. Journal of
Development Studies 39: 1–20. Saltman R. 2000. The concept of stewardship in health policy. Bulletin of
the World Health Organization 78: 732–39.
Maestad O. 2006. Human resources for health in Tanzania: challenges,
policy options and knowledge gaps. CMI Report. Bergen: Chr. Seiber E, Robinson-Miller AR. 2004. The impact of a microfinance
Michelsen Institute. program on private health care providers in Uganda. Washington,
DC: USAID/Commercial Marketing Strategies Project.
Mamdani M, Bangser M. 2004. Poor people’s experiences of health
services in Tanzania: a literature review. Reproductive Health Matters Sharma S. 2001. The private sector and child health care. Chapel Hill, NC:
12: 138–53. Carolina Consulting Corporation.
Marek T, O’Farrell C, Yamamoto C, Zable I. 2005. Trends and opportunities Spoonley P, de Bruin A, Firkin P. 2002. Managing non-standard work
in public-private partnerships to improve health service delivery in Africa. arrangements: choices and constraints. Journal of Sociology 38: 425–41.
Washington, DC: World Bank. Stevenson L, St-Onge A. 2005. Support for growth-orientated women
Maslin A. 2005. Pre-WHA meeting of Commonwealth Health Ministers, entrepreneurs in Tanzania. Geneva: International Labour
May 2005. Report from the Commonwealth Steering Committee for Organization.
Nursing and Midwifery. London: Commonwealth Nurses Federation. Suryanigsih D, Caiola N, Febriani E et al. 2005. Bidan Delima: Improving
Mayoux L. 2001. Jobs, gender and small enterprises: getting the policy environment quality of care of private sector midwives. Paper presented at the
right. SEED Working Paper 15. Geneva: International Labour Office. International Society for Quality in Health Care conference,
McGinn T, Aboagye-Owusu H, Dugan R et al. 1990. Private midwives: a Vancouver.
new approach to family planning service in Ghana. Midwifery 6: UNDP. 2001. International/Millennium Declaration goals. Tanzania progress
117–24. report. (Overview). Geneva: United Nations Development Program.
McKee M, McPake B. 2004. Commentary: the devil is in the detail. UNDP. 2005. Human Development Report 2005. New York: United Nations
British Medical Journal 329: 1175–76. Development Program.
McKinsey & Company. 2005. Acting now to overcome Tanzania’s United Republic of Tanzania. 1997. Nurses and Midwives Registration Act
greatest health challenge. Addressing the gap in human resources 1997. Dar es Salaam.
for health. New York: Touch Foundation. UN Millennium Project. 2005. Who’s got the power? Transforming health
Ministry of Health, Tanzania. 1997. Health sector reform programme of systems for women and children. UN Millennium Project Task Force
work. Dar es Salaam: Ministry of Health. on Child Health and Maternal Health. London/New York:
Ministry of Health, Tanzania. 2002. Health statistical abstract 2002. Earthscan/UNDP.
Inventory statistics 2. Dar es Salaam: Ministry of Health. Van Lerberghe W, Conceiçao C, Van Damme W, Ferrinho P. 2002. When
Murray SF, Elson MA. 2005. The promotion of private health insurance staff is underpaid: dealing with the individual coping strategies of
and the implications for the social organisation of healthcare: a health personnel. Bulletin of the World Health Organisation 80:
case study of private sector obstetric practice in Chile. Sociology of 581–84.
Health and Illness 27: 6. Walker D, Muyinda H, Foster S et al. 2001. The quality of care by private
Murray SF, Nyambo N. 2003. Examining the role of private maternity practitioners for sexually transmitted diseases in Uganda. Health
services in Nepal, India and Tanzania. Private maternity services in Policy and Planning 16: 35–40.
20 8
Weber P, Schaper M. 2003. Understanding the grey entrepreneur: a WHO. 2006. The World Health Report 2006: working together for health.
review of the literature. Paper presented at the16th Annual Geneva: World Health Organization.
Conference of Small Enterprise Association of Australia and Wyss K, Whiting D, Kilima P et al. 1996. Utilisation of government and
New Zealand, University of Ballarant, Victoria, Australia. private health services in Dar es Salaam. East African Medical Journal
WHO, ICM, FIGO. 2004. Making pregnancy safer: the critical role of the skilled 73: 357–63.
attendant. A joint statement by WHO, ICM and FIGO. Geneva: World Yin R. 2003. Case study research: design and methods. 3rd edn. London:
Health Organization. Sage.
WHO. 2003. Integrated management of pregnancy and childbirth. Pregnancy, Yumkella F, Githiori F. 2000. Technical Report 21: Expanding opportu-
childbirth, postpartum and newborn care: a guide for essential practice. nities for postabortion care at the community level through
Geneva: World Health Organization. private nurse-midwives in Kenya. Chapel Hill, NC: University of
WHO. 2004. Maternal mortality in 2000: Estimates developed by WHO, North Carolina, Program for International Training in Health
UNICEF and UNFPA. Geneva: World Health Organization. (INTRAH).
20 9
ARTICLE IN PRESS

Are health services protecting the livelihoods of the urban poor

in Sri Lanka? Findings from two low-income areas of Colombo
Steven Russella,!, Lucy Gilsonb
a
University of East Anglia, Norwich, UK
b
London School of Hygiene and Tropical Medicine, UK
Available online 12 June 2006
Abstract
Investing in pro-poor health services is central to poverty reduction and achievement of the Millennium Development
Goals. As health care financing mechanisms have an important influence over access and treatment costs they are central
to the debates over health systems and their impact on poverty. This paper examines people’s utilisation of health care
services and illness cost burdens in a setting of free public provision, Sri Lanka. It assesses whether and how free health
care protected poor and vulnerable households from illness costs and illness-induced impoverishment, using data from a
cross-sectional survey (423 households) and longitudinal case study household research (16 households). The findings
inform policy debates about how to improve protection levels, including the contribution of free health care services to
poverty reduction. Assessment of policy options that can improve health system performance must start from a better
understanding of the demand-side influences over performance.
Keywords: Poverty; Vulnerability; Illness cost; Coping strategy; Sri Lanka
Introduction Bank, 2004; Xu et al., 2003). Calls for the removal

of user fees at public health care facilities have, thus,
Illness can cause impoverishment through a once again come to the international policy agenda
downward spiral of income loss, treatment costs (Commission for Africa, 2005). Out of pocket
and asset depletion. Investing in pro-poor health payments add to the other barriers that poor people
services is therefore central to poverty reduction and face when seeking health care, and contribute to
achievement of the Millennium Development Goals their experience of social exclusion. Even relatively
(WHO, 2002; World Bank, 2004). As health care small health care payments might push vulnerable
financing mechanisms have an important influence households into absolute poverty or deepen their
over access and treatment costs they are central to poverty (Gilson & McIntyre, 2005).
debates over health systems and their impact on Discussion of financing policy options to improve
poverty (Kawabata, Xu, & Carrin, 2002; World health system performance in resource poor settings
must, however, start from better understanding of
!Corresponding author. Tel.: +44 1603 593373. demand-side factors and the consequences of out of
E-mail addresses: s.russell@uea.ac.uk (S. Russell), pocket payments for household poverty and liveli-
lgilson@iafrica.com (L. Gilson). hoods. This paper uses a household livelihood
doi:10.1016/j.socscimed.2006.04.017
210
ARTICLE IN PRESS
S. Russell, L. Gilson / Social Science & Medicine 63 (2006) 1732–1744 1733
framework to examine people’s utilisation of health contributed to income poverty and vulnerability to
care services, illness costs and their implications for wage losses caused by incapacitating illness. The
impoverishment in a setting of free public provision, settlements lie a few miles from the centre of
Sri Lanka. It examines whether and how free health Colombo and close to many health care providers:
care protected poor and vulnerable households a local municipal dispensary where a GP can be
from illness costs and illness-induced impoverish- consulted with no charge; several Ministry of
ment, informing national policy measures to im- Health tertiary hospitals where services are free to
prove protection levels and international debates on the user; and a large number of private GPs,
the contribution of free health care services to pharmacies and several private hospitals. Although,
poverty reduction. the research was conducted 8 years ago health
Sri Lanka provides a particularly relevant case service financing and delivery arrangements remain
study with which to examine these issues. Histori- the same at the time of writing. The two case study
cally the country has been relatively successful in urban areas were selected because they were typical
‘making services work for poor people’ (Rannan- of many deprived settlements in Colombo.
Eliya, 2001), benefiting from the ‘long route’ to The research objectives were to record treatment
government and provider accountability to the poor seeking behaviour, measure the household costs of
(World Bank, 2004). Since democratisation in the illness, and assess coping strategies and their
1930s, competitive politics, left-wing political par- consequences for the household economy. The
ties, trade unions and public pressure have con- conceptual framework that guided the research
structed a strong policy discourse that makes it the (see Russell, 2004) was based on inter-disciplinary
state’s responsibility to deliver free health care as a approaches that have analysed the numerous
basic right for all citizens (Sen, 1988). After resources people draw on to promote health or
Independence government invested in a network cope with illness costs (Berman, Kendall, &
of accessible and free health care services and well Bhattacharyya, 1994; Wallman & Baker, 1996) as
trained nurses and doctors. Effective use of this well as a livelihood framework (Scoones, 1998).
network by a well-educated population, notably Direct illness costs and indirect costs are defined,
literate women, helped bring about ‘good health at respectively, as expenditure linked with seeking
low cost’ in Sri Lanka (Halstead, Walsh, & Warren, treatment and income losses caused by illness. The
1985). term ‘cost burden’ refers to direct or indirect costs
The health care market, however, has been expressed as a percentage of household income.
changing, with private sector expansion since the Health care spending and income losses will reduce
1980s and a slow public sector response to changing household budgets and threaten members’ mini-
disease burdens and patient preferences. Detailed mum basic needs such as food consumption or
demand analysis is therefore appropriate to assess education, triggering coping strategies such as
current patient utilisation patterns in this more borrowing or asset sales. The resource strategies
complex market, as well as the levels of protection, used to cope with illness costs were also recorded
and gaps in coverage, offered to poor households by because such strategies can mitigate or exacerbate
free public health services. the overall economic impact of illness for the
household. Together illness costs and coping
A household livelihood framework to inform pro-poor strategies have implications for household income-
health services poverty and livelihood outcomes, assessed using
indicators such as changes to income, working days,
Study setting and the conceptual framework used in assets, consumption levels and food security.
the research Household vulnerability or resilience to illness
costs is defined as the capacity to cope with illness
The research on illness and its livelihood impact costs without long-term damage to assets and
was conducted between 1998 and 1999 in two low- impoverishment. It is linked, first, to illness severity,
income settlements of Colombo, the capital of Sri with higher costs and less sustainable coping
Lanka. The urban sites were characterised by strategies likely as severity and duration of illness
overcrowded housing, poor sanitation, drug abuse increase. Second, capacity to cope is influenced
problems and low incomes due to uncertain and by household asset portfolios (physical and finan-
daily employment opportunities. These livelihoods cial capital, human capital, social networks) and
211
ARTICLE IN PRESS
1734 S. Russell, L. Gilson / Social Science & Medicine 63 (2006) 1732–1744
policy-related resources that include health services month to be calculated and analysis of the effect of
as well as other public policy measures (e.g. health care spending on the monthly household
education services) or community-based initiatives budget. Patient and caregiver days off work due to
(e.g. micro-credit institutions) that contribute to illness were converted to a lost income figure using
resilience. These policy and community-based re- an average daily wage derived from the local setting
sources represent entry points for health and other (Rs. 150 or US$2.30 per day). Only days lost by
social policy interventions that may protect house- economically active members were included in the
holds. indirect cost calculations because valuing unpaid
activities is both fraught with difficulties and less
Research methods immediately relevant to understanding the econom-
ic burden of illness.
The research design had three phases spanning 18 The third phase of the research was an in-depth
months. First, individual and group interviews were longitudinal study of 16 case study households over
conducted to generate qualitative data on treatment 8 months conducted to allow detailed investigation
behaviour and livelihood difficulties. Second, a of illness costs and livelihood impacts over time.
cross-sectional survey of 423 households and 2197 Using the survey data as a sampling frame the
individuals produced a statistical profile of house- households were selected purposefully to be ‘typical’
hold income and assets, illness episodes, treatment of four per capita income quartile groups and,
actions, illness costs and coping strategies. The within each, a range of illness, treatment and cost
households were selected by systematic random experiences. Finally, the selection process ensured
sampling and the sample covered 20% of the 2100 that households with varying vulnerability or
households in both settlements. resilience to these costs were included within the
The survey collected data on three categories of case studies. As assets, like income, reflect ability to
illness expected to cause different treatment, cost cope with illness costs and livelihood change,
and coping patterns: assessment of household vulnerability or resilience
was based on a simple audit of assets: the number of
! Acute illness episodes in the previous 2 weeks workers and security of work; physical capital
(except hospital admission). including house construction; education; and finan-
! Chronic illness in the previous month, cate- cial capital.
gorised as such if the condition had persisted Each household was visited at least every 2 weeks.
for over 1 month or the respondent knew the Structured interviews were used for more quantita-
diagnosis and the name of the chronic condition tive variables (expenditure, illness costs, borrowing),
(e.g. diabetes, high blood pressure); the recall and semi-structured interviews and observation to
period allowed the survey to capture patients’ generate qualitative data. The intensive study of a
regular monthly visits to providers. small number of families over time was necessary to
! Hospital inpatient (IP) treatment in the previous explore the ways that people took action in their
year, with the recall period designed to maximise every day lives to treat illness and cope with its
hospitalisation events recorded. costs, a well as the multiple factors that mediated
the impact of illness on livelihood outcomes.
The survey estimated household income using The knowledge claims from case studies are often
detailed consumption and expenditure questions. criticised on the grounds that the evidence is
There is limited seasonality of casual labour or wage ‘anecdotal’ or ‘unrepresentative’. But just as clinical
levels in Colombo so expenditure or income levels science uses cases to understand disease causation,
were not influenced by the timing of the survey. In so social science can use cases to understand illness-
most cases either the household head (usually male) induced poverty causation. Such understanding
or their partner (usually wife) was interviewed, and must go beyond the identification of vulnerable
sometimes more than one adult was present. Where groups’ characteristics to consider the social pro-
possible the wife or mother was asked questions cesses that cause vulnerability to illness costs and
concerning illness and treatment among family how these operate within households to ‘filter’
members. policy effects. As case study data are not statistically
All illness cost data were converted to a cost per representative but aim to strengthen understanding
month figure to allow a total illness cost burden per of social processes, sample size is of less concern
212
ARTICLE IN PRESS
than the depth of understanding generated. Gen- cost was higher at US$7.50 (Rs. 487) per month
eralisation is possible in terms of the concepts or because a minority of households experienced a
frameworks (e.g. vulnerability) developed from case high direct cost. A mix of public and private
study analysis that can be applied to other providers was used (see Section ‘Protecting the
individuals, households and settings (Coast, 1999; poor? Universal coverage and its limitations in Sri
Mitchell, 1983). Lanka’). The main direct cost components from
The policy relevance of case study material does, private sector use were consultation fees and
however, rely on it being ‘typical’ for a larger group medicine and the main cost item from public sector
of households, requiring the careful selection of use was transport. No ‘under the table’ payments
cases from different population groups of relevance were recorded.
to the study. Here, use of the survey data enabled The majority of households (77% or 250/323)
the selection of cases that were typical of different that experienced illness incurred a low or moderate
household types in the two settlements, providing direct cost burden of 5% or less of monthly income
the basis for the conceptual generalisation of their (Fig. 1), either because the illness was mild or
experiences to other households with similar char- because free public services protected against high
acteristics in the same communities, such as the or catastrophic cost burdens associated with serious
income- or asset-rich and poor. illness. Low direct costs were not caused by people
failing to seek the medical care that they needed.
Illness costs and livelihood change: an overview However, a considerable minority of households
experienced what some analysts have called a
Cost burdens ‘catastrophic’ direct cost burden in terms of its
potential consequences for poverty (Prescott, 1999;
Among the 323 households (out of 423) that Ranson, 2002): 10% of households (n ¼ 32) in-
experienced illness and self-treated or sought treat- curred a direct cost burden above 10% of monthly
ment, the median direct cost of illness was US$2.10 income (Fig. 1).
(Rs. 138) per household per month or equivalent to The majority of households incurred no or low
just under an average daily wage. The mean direct indirect cost burdens (Fig. 1). Many illnesses did not
80
70
60
% of households (n=323)
50
40
30
20
10
0
0
.0
+
1.
2.
3.
4.
5.
6.
7.
8.
9.
.1
10
1-
1-
1-
1-
1-
1-
1-
1-
10
1
1-
8.
0.
1.
2.
3.
4.
5.
6.
7.
9.
Cost burden (cost as % of monthly income)
direct cost burden indirect cost burden total cost burden
Fig. 1. Distribution of illness cost burdens across households.
213
ARTICLE IN PRESS
cause income loss because children of school age months and usually in 1 or 2 months direct (and
disproportionately suffered from acute illnesses, a indirect) cost burdens were particularly high (see
large proportion of acute illnesses experienced by Fig. 3). This ‘lumpy’ feature of illness costs made
economically active adults were not serious enough them harder to manage. Even the peaks in Fig. 3
to affect work, and the majority experiencing were average cost burdens over 30 days that smooth
chronic illness and hospital admission were eco- higher daily cost burdens, often exceeding 100% of
nomically inactive. However, a minority (11%, the daily wage. The poorest and most vulnerable
n ¼ 35) incurred an indirect cost burden above households dependent on a low daily wage found it
10% of normal monthly income (Fig. 1). difficult to manage any cost associated with illness,
Combined (total) cost burdens were relatively low let alone these peaks, and had to borrow or pawn
for the majority of families surveyed (Fig. 1). jewellery to cope.
However, a fifth (19.2%, n ¼ 62) incurred a total A cost burden figure only indicates the potential
cost burden above 10% and most of this group or likely consequences of an illness cost for house-
incurred a total cost burden between 10.1% and hold impoverishment. The actual impact will
40.0%. depend on household income (for a poor household
Households in the poorest income quartile were a relatively low cost burden may cause impoverish-
disproportionately affected by a catastrophic direct ment but for a non-poor household a burden above
cost burden above 10% because of their particularly 10% may not be ‘catastrophic’) and household
low income. However, there was no statistically capacity to mobilise additional resources (vulner-
significant difference in mean direct cost burdens ability or resilience). However, it is still a useful
across income groups (Table 1). Low median direct indicator of the extent of protection provided by
cost burdens reflect the public health system’s public health services.
coverage of the majority.
Case study households’ average direct (and Illness-related poverty and livelihood change
indirect) cost burdens per month over 8 months
are plotted in Fig. 2, with the households grouped The survey data were analysed to estimate the
into the three vulnerability categories determined short-term poverty implications of health care
from asset portfolios (see Section ‘Illness-related spending, using two indicators: the poverty count
poverty and livelihood change’). The majority (incidence) and the poverty gap. The first calcula-
experienced a low to moderate direct cost burden tion estimates the proportion of households pushed
per month under 5% of income, but a minority below a US$30.00 per capita per month (US$1.00 a
(n ¼ 3), all in the middle (vulnerable) group day) absolute poverty line by health care spending.
experienced a higher direct cost burden over 5% Household health expenditure was subtracted from
and one (Geetha) over 10%. Highly vulnerable household income and a new household per capita
households’ low average direct cost burden income level calculated. As a result of health care
stemmed from their greater use of free public payments the poverty incidence rose from 54.1%
providers. Resilient households’ low direct cost (n ¼ 229) to 57.0% (n ¼ 241): 12 households were
burdens per month were because of their higher pushed below the poverty line. Health care spending
incomes and use of public hospitals for IP treatment therefore added 2.9% to the poverty incidence, a
(see Section ‘Protecting the poor? Universal cover- level comparable to estimates from India and
age and its limitations in Sri Lanka’). Vietnam (Wagstaff, 2002). This analysis assumes
For all case study households the average cost that the money spent on health care was no longer
burden per month conceals fluctuations over 8 available to spend on other essential goods and
Table 1
Average household direct illness cost burden per month by income quartile
Direct cost burden Household income per capita quartile group
1 (poorest) (n ¼ 82) 2 (n ¼ 90) 3 (n ¼ 74) 4 (n ¼ 77) All (n ¼ 323)
Mean (95% confidence interval) 15.2 ("6.6–37.0) 3.0 (2.1–3.9) 3.5 (2.4–4.5) 4.1 (2.8–5.5) 6.5 (1.0–12.0)
Median 1.2 1.4 1.8 1.6 1.3
214
ARTICLE IN PRESS
Dilani
Mary
Resilient
Rani
Mayori
Pushpa
Raja
Nishanthi
Households
Kumudu Vulnerable
Amali
Renuka
Geetha
Nimal $$
Selvaraja
Jayasinghe $$ Highly
Valli vulnerable
Sumithra
0 5 10 15 20 25 30 35
Average total cost burden per month (% of income)
direct cost burden indirect cost burden
Fig. 2. Average illness cost burden per month among case study households. $$Nimal and Jayasinghe were the main breadwinners in their
households but had been forced to stop work due to serious illness before research started. Indirect cost burdens were therefore high but
incalculable. Nimal experienced high direct costs of illness over 8 months but his extended family paid these costs.
services and so pushed households into absolute These indicators of changes to poverty derived
poverty. It might also be argued that without free from a cross-sectional survey should be interpreted
health care there would have been higher levels of cautiously. A fall below the poverty line for example
spending and the potential for more households to may be very short-term, households may have
have fallen below the poverty line. coped by mobilising other resources, and the health
The second calculation uses the poverty gap care spending may not have involved any damaging
indicator (the average income shortfall from the cuts to consumption or assets. The advantage of the
poverty line) to estimate the deepening of household longitudinal case study methodology was the ability
poverty caused by health care spending. Among the to track the actual implications of illness costs for
229 households below the US30.00 poverty line the income-poverty, assets and livelihood change in
mean income shortfall was US$8.90 (Rs. 577) per some detail over 8 months.
capita per month, or a daily shortfall of US$0.30 The 16 households were chosen to represent four
below the US$1.00 a day poverty line (Table 2). household income groups derived from the survey
After health care spending among the same house- data, but the other selection criteria (illness and
holds the poverty gap rose to US$9.30 per month, a vulnerability) meant they were not equally distrib-
5.2% rise in the depth of poverty. If the 12 uted across quartiles (Table 3). Households in the
additional households that fell below the poverty lowest two quartiles earned less than US$1.00 per
line are included in the calculation the depth of capita per day (less than US$30.00 per month). In
poverty rises by only 0.82%, from US$8.90 to the poorest quartile households struggled to meet
US$9.00 per capita per month (Table 2). food and fuel needs on a daily basis. Even in the
215
ARTICLE IN PRESS
35
30
Cost burden per month (% of income)
25
20
15
10
0
1 2 3 4 5 6 7 8
Month
Pushpa - direct cost Selvaraja - direct cost

Raja - indirect cost Valli - indirect cost
Fig. 3. Monthly fluctuations of direct and indirect illness cost burdens: selected cases.
Table 2
Changes in the depth of household poverty (poverty gapa) due to health care payments
Mean poverty gap US$ Standard deviation US$

(Rs.)b (Rs.)
Poverty gap before health spending (n ¼ 229) 8.90 (577) 5.6 (367)
Poverty gap after health spending (n ¼ 227)c 9.30 (607) 5.6 (361)
Poverty gap including 12 new households below the 9.00 (582) 5.7 (370)
poverty line after health spending (n ¼ 239)
a
Average shortfall from a US$30.00 per capita per month poverty line.
b
At the time of research US$1.00 ¼ Sri Lankan Rupees (Rs.) 65.00.
c
Two outlier households excluded: very high health expenditure had pushed the households’ income into a negative income value that
prevented analysis of the poverty gap.
upper quartile most households earned only available for health care. Income insecurity due to
US$40–50.00 per capita per month (US$1.00–2.00 lack of available work or illness was a great source
per capita per day). So despite their relatively high of vulnerability:
cash income in these poor areas many families
classified as ‘better-off’ were only marginally above Illness is something we are all scared of here.
the poverty line. How can we live without working? If my
Seven of the 16 case study households were in the husband is ill we have to get money from
poorest quartile and an additional expense such as somewhere for food and for the medicine, we
health care usually triggered coping strategies that have to borrow.
pushed them deeper into poverty. The two house- (Selvaraja, woman from poorest income quartile,
holds in the second quartile also had little money most vulnerable).
216
ARTICLE IN PRESS
Table 3
Location of case study households in the community income profilea
Household income per capita quartile group
1 2 3 4
Household per capita income: US$/month $ 0–21 $ 21–29 $ 29–40 $ 40+

(Rs./month) (0–1352) (1353–1880) (1881–2609) (2610+)
Households with illness Nimal Nishanthi Raja Dilani
Jayasinghe Sumithra Rani
Valli
Households without illness Kumudu Pushpa Mary
Selvaraja Mayori Renuka
Amali
Geetha
a
Case study households have been given pseudonyms for confidentiality.
Households in the third and fourth quartiles one or sometimes two workers with insecure jobs.
could to differing degrees meet the costs of Physical capital was limited to a small wooden or
treatment for most acute and chronic illnesses in poorly maintained cement block house with no
months when incomes were maximised. However, in electricity or water connection. Financial capital
months when workers lost earnings due to illness or had been depleted: they had pawned all or most of
the vagaries of the labour market, or when illness their jewellery, in some cases due to previous illness
expenses coincided with other ‘lumpy’ expenses such (Jayasinghe, Valli), and were in debt to money-
as education or clothing, those in the third quartile lenders.
had to adopt strategies to cope with illness costs. Over the 8 months these households experienced
Those in the fourth quartile had to mobilise a decline in at least four of the six livelihood
additional resources when more serious or pro- variables, most commonly the loss of an income
longed illness caused income loss. In other words earner or growing insecurity of work, pawned
household ability to cope with illness costs could jewellery, increased debt and lasting cuts to food
not be seen in isolation from other expenses and consumption. Three of the four households were on
income fluctuations. a path of decline triggered by illness before research
Across income groups, case study households started. For example cancer had forced Jayasinghe
were also selected from three vulnerability—resi- to give up work with damaging economic conse-
lience categories (Fig. 2). Over 8 months, livelihood quences for the household; and Sumithra’s husband
change among the households was evaluated by had experienced a serious accident which, after over
analysing six livelihood outcomes using quantitative a month in (a public) hospital without earning
and qualitative data: the number of workers and job income, had undermined assets and caused high
security; income levels; physical capital; financial levels of debt.
capital (changes to savings or jewellery); debt levels; Three of the highly vulnerable households in-
and consumption (focusing on number of meals per curred low or moderate average cost burdens per
day). Households were placed into three categories month (Fig. 2) but these costs were a persistent
of livelihood change: struggling (impoverishment); attack on the household budget and assets. Valli
coping (stability); investing (improvement). experienced indirect cost burdens of over 20% in
some months (Fig. 3) which could be judged to be
Highly vulnerable households: struggled and became ‘catastrophic’ because they forced her and her
more impoverished husband deeper into poverty: they had to borrow
Three out of four households in this group (see at high interest, cut food consumption and pawn
Fig. 2 for pseudonyms) were located in the poorest last items of jewellery. The group’s low and insecure
income quartile (Table 3) and struggled to eat three incomes meant they had to meet a high propor-
meals a day. They had weak asset portfolios. tion (58%) of direct and indirect illness costs
Members had less formal education and relied on through these types of strategy, but their weak
217
ARTICLE IN PRESS
asset portfolios meant they struggled to cope. this group was distinguished from the highly
Multiple asset weaknesses made health service vulnerable households by their stronger asset
protection particularly important for this group of portfolios and capacity to cope, particularly the
households (see Section ‘Protecting the poor? Uni- strength of their social networks. Although income
versal coverage and its limitations in Sri Lanka’). poverty meant the group could not cover a
considerable proportion (45%) of their total illness
Vulnerable households: coped to different degrees costs through usual income sources, they mobilised
This group spanned the full range of per capita low cost asset and borrowing strategies, which
income quartiles and to differing degrees had contributed to livelihood stability. Access to free
stronger asset portfolios than the highly vulnerable services for more serious illness contributed to this
group, even among the income-poorest (Amali, resilience (see Section ‘Protecting the poor? Uni-
Nimal, Geetha). Renuka’s household was located in versal coverage and its limitations in Sri Lanka’).
the top income quartile but was vulnerable because
her husband used the income to fund a heroin Resilient households: invested and improved
addiction and the rest of the family (Renuka and These four households had higher and more
four children) were left with barely enough income secure incomes derived from a household member
for food and few assets. Compared to the highly with a secure government job, or several workers in
vulnerable group, adults were in general better the family, or a successful small business. They had
educated (Nimal, Geetha, Amali, Pushpa) or the the strongest asset portfolios including better
household had more workers (Raja, Pushpa, education, a larger house made from bricks and
Nishanthi). Some had more financial capital with mortar, and a range of physical assets in the
women participating in rotating savings (seetu) household (electrical goods, furniture).
groups or credit societies (Nishanthi, Amali, Ku- Over 8 months the group experienced improve-
mudu, Raja, Pushpa), although Nimal’s wife Sita, ment in at least four livelihood variables, and nearly
Geetha, and Renuka were not involved due to their all borrowing was for investment purposes. Histori-
income-poverty. Some had jewellery available to cally they were on steady trajectories of improve-
pawn (Geetha, Amali, Kumudu, Raja, Pushpa), but ment even though they had originally started from
others had depleted these financial assets due to socio-economic positions similar to the other house-
previous illness (Nimal, Nishanthi). holds. Notably no breadwinners had been affected
These households experienced little change to at by serious illness.
least four dimensions of livelihood. Debt levels had Although household members used private pro-
not increased and if people had borrowed it was viders more often than public providers for treat-
from low cost and flexible sources such as family, ment of acute and chronic illnesses, the group
friends and local credit societies. Historically they experienced relatively low or moderate cost burdens
were on steady livelihood paths characterised by because of their higher income (Fig. 2). However,
vulnerability but fewer shocks including fewer they relied on the safety net or ‘insurance’ of the
serious illness events. Gradual improvements were public sector for IP treatment, which protected
sustained by Kumudu and Puspha despite high assets, kept debts low and also allowed them to
illness costs (Fig. 2) and the others were coping to divert resources to investment strategies. As a result,
differing degrees. As Nimal’s household had already they only had to cover a small proportion (9%) of
suffered dramatic decline due to serious illness total illness costs through asset strategies, usually
before the start of the study, it could have been low cost borrowing from strong social networks.
placed in the struggling group. However, when Across these three groups of household there was,
research started he and his wife were coping (at a not surprisingly, a strong link between vulnerability
lower level) and not suffering further impoverish- at the start of research and livelihood change
ment because free health services enabled him to category at the end. Fig. 2 also suggests there was
make regular visits to the hospital for consultations no clear link between illness cost burden and
and blood tests, and strong family networks livelihood change. Highly vulnerable households
provided funds for nearly all their daily and health with direct cost burdens less than 5% struggled and
care expenses. fell further into poverty. In contrast some of the
Despite the group’s higher direct cost burdens, middle (vulnerable) group incurred a high or ‘cata-
over 100% in some months for Geetha and Nimal, strophic’ burden but managed to cope, although
218
ARTICLE IN PRESS
households experiencing serious illness and a high and 194 sought treatment on a regular basis. Across
cost burden (Geetha and the special case of Nimal) the first three income quartile groups a public
were only just coping. Given the complexity of hospital OP clinic was the main source of regular
livelihoods and the multiple factors influencing treatment, particularly among patients from the
livelihood trajectories, the lack of a clear link poorest households who used public providers far
between cost burden and impoverishment is not more frequently (62%) than private providers
surprising. (18%). Only the ‘better-off’ income group used
private providers more often than public providers
Protecting the poor? Universal coverage and its for their regular treatment. Free treatment was the
limitations in Sri Lanka most common reason cited for using a public
provider but confidence in the technical competence
Inpatient treatment of doctors was again important (Russell, 2005).
Widespread use of public providers meant that out
The household survey found that the vast of the 155 households with a member seeking
majority of people in the two communities, from regular treatment 50% incurred a direct cost burden
all income groups, used one of the large public of 1% of monthly income or less, 87% a burden of
hospitals in the city rather than a private hospital 5% or less and only 3% of households incurred
(98% of admissions, n ¼ 177). Among case study regular monthly burdens over 10%.
households all hospital admissions over the 8- The case study data confirmed that free health
month period were to public hospitals. This utilisa- care offered important protection to livelihoods.
tion pattern was explained by the free IP care For the highly vulnerable group with no surplus
offered by public hospitals compared to the money to pay for health care (even to cover
prohibitively high cost of a private hospital admis- transport costs), free regular treatment of chronic
sion, but in addition a dominant theme from the conditions was vital protection against higher
qualitative data was people’s trust in the technical borrowing or deeper cuts to food consumption.
quality of care at public hospitals, based on the Among the vulnerable (middle) group free treat-
widely held view that they had the best staff and ment was also a vital entitlement that prevented
equipment to deal with serious conditions (Russell, borrowing for health care expenses. A comment by
2005). Geetha, diagnosed with Type 2 diabetes during the
Use of public hospitals meant patients and their 8-month study period, exemplifies the experience of
families incurred a relatively low direct cost burden diabetics from vulnerable households:
for a hospital admission. From the household
If I go private, I pay money, but then if things get
survey 82% of households experiencing one or
worse they refer me to the government and they
more IP admission (n ¼ 134) in the previous year
would have to do all the tests again. So if I have a
faced a direct cost burden of 1% or less per month
big problem, or one that needs continuous
(i.e. a burden of 12% or less in 1 month spread over
treatment like diabetes, I go to the government
12 months), and 95% of households experienced a
hospitalyIt is freeyhow could I pay for the
burden of 5% per month or less, the main cost items
tablets everyday?
being transport, special food and complementary
(Geetha: woman from poorest income quartile,
religious therapies or medicine bought outside the
vulnerable).
hospital. Case study household experiences also
demonstrated how free IP services protected all Free treatment was particularly important to
socio-economic groups against high direct medical livelihood security at times when workers fell ill
costs. causing income levels to drop and a consequent
struggle to pay for a range of essential items. Raja’s
Regular treatment of chronic illness household, for example, experienced high wage
losses in some months (see Fig. 3) because he and
Chronic conditions requiring regular treatment his wife (Ranji) suffered from asthma. In month 1
have the potential to impose high- and long-term Raja had a sore chest and took two days off work,
cost burdens on poor households unless free services losing US$5.40 (Rs. 350) in wages (an indirect
are available. The household survey identified 342 cost burden of 6%). Raja went to a nearby private
people (15.6%) who reported a chronic condition clinic and pharmacy for treatment, which incurred a
219
ARTICLE IN PRESS
direct cost burden of 15%. The high direct cost common acute illnesses (Russell, 2005), including
burden combined with the indirect cost forced the limited opening hours, long waiting times, short
family to borrow from Raja’s workplace. Later in consultations and poor inter-personal quality. As a
the month the chest problems persisted but they had result the majority of the ‘better-off’ and even a
no cash available so Raja resorted to the free considerable minority of the poorest were willing to
municipal dispensary. Without the alternative of pay to get quicker care, secure a longer consultation
cheap public treatment towards the end of the with more patient focus, and build a long-term
month the household’s borrowing would have been doctor–patient relationship with a ‘family’ doctor.
significantly higher. The poorest pawned jewellery and borrowed money
to finance private treatment.
Acute illnesses requiring OP treatment Nonetheless, free public health care of adequate
quality offered important protection to the most
The survey identified 266 out of 2197 individuals vulnerable and income-poorest households with
(12.1%) who reported an acute illness episode in the several small children who experienced frequent
previous 2 weeks, the most frequent being cold, and concurrent acute illness events. Selvaraja’s
cough, fever, flu, headache, injury and diarrhoea. family offers a typical illustration of this protection.
Self-treatment at home was the most frequently In month 5 the three children suffered illness
reported first response (58%), reflecting the mild concurrently (high fever and vomiting) and Selver-
nature of many of the illnesses. aja took them to the National Children’s Hospital
In contrast to the dominant use of public OP department, a visit which incurred a direct cost
providers for IP and regular chronic care, the use burden of only 0.5% (US$0.50/Rs. 30 for trans-
of health care providers outside the home was more port):
equally split between public and private providers
for moderate acute illnesses, with private GPs and
I take the kids to Lady RidgewayyRs. 300 or
pharmacies slightly more dominant. Even among
more would have gone if I had gone priva-
the poorest quartile a considerable minority of
teyand I would need to borrow even more
patients (46%) used private doctors and pharmacies
money for that—maybe with interest.
(Russell, 2005).
(Selvaraja, woman from poorest income quartile,
Widespread use of private providers meant higher
highly vulnerable).
household cost burdens for OP treatment of acute
illness. Out of the 210 households experiencing one
or more acute illness episode, 47% experienced a The livelihood implications of having to pay for
direct cost burden of 1% or less but 20% health care were starkly illustrated in the same
experienced a burden over 5 and 7% a burden month. The family spent an additional Rs. 320
above 10%. (US$5.00) on health care due to private sector use
All case study household respondents, whether by Selvaraja’s husband (for a recurring shoulder
male, female, poor or better-off, stated that they injury; he could not afford to miss work) and
preferred to use a private doctor or pharmacy for Selvaraja’s mother (for a tooth extraction; there was
common illnesses. Income levels and cash avail- a long waiting list at the public hospital). These
ability, however, influenced actual utilisation pat- private visits imposed a direct cost burden of 5%
terns. Members of the seven households in the top which exceeded the household budget after food
income quartiles (with the exception of Renuka) purchases and triggered coping strategies that
consistently used a private GP with whom they were pushed the household deeper into poverty. They
familiar (their ‘family doctor’). In the seven poorest had to borrow from an ex-employer (Rs. 500), delay
households wage-earners used private doctors and payment of the electricity bill, delay debt repayment
pharmacies more frequently than public providers to the local food shop, and could not redeem a ring
to obtain treatment quickly and avoid wage losses, that Selvaraja had pawned in an earlier month to
but members who did not work used a municipal pay for health care. If Selvaraja had taken her
dispensary as frequently as private doctors and children to a private doctor that month the overall
pharmacies. direct cost burden for the family would have been
The research identified several reasons for the over 12%, forcing even more risky borrowing or
lower uptake of free public health services for asset strategies.
220
ARTICLE IN PRESS
Discussion illness, or broken relationships. Previous events and

processes had placed households on longer-term
The household survey and case study data show trajectories of struggling, coping or improving, and
that free health care services in urban Sri Lanka, path dependency continued to influence livelihood
financed through taxation, protected the majority of change over the brief research period. Given the
poor households against high out of pocket pay- strength of these trajectories the impact of illness on
ments for treatment at the time of illness. This impoverishment and livelihood was heavily depen-
protection against even relatively low fees was an dent on its severity, frequency and duration. Low
important poverty reduction measure because, as and infrequent illness costs made little impact. Low
shown by the case study findings, even a small direct or moderate but more frequent illness costs exacer-
cost could cause impoverishment. Nine case study bated vulnerability and livelihood decline. Serious
households in the two poorest income quartiles illness that caused high or catastrophic and persistent
(Table 3), selected to be typical of 50% of total cost burdens could have a major negative impact on
households in these settlements, relied on low paid livelihood paths. In Sri Lanka the availability of free
and insecure work and struggled on less than public health services meant it was the indirect costs
US$1.00 per capita per day. These households had arising from serious illness, rather than direct costs,
little or no ‘ability to pay’ for health care after which were the most obvious cause of illness-induced
meeting basic food, shelter and fuel needs. Other poverty, as the examples of Nimal and Jayasinghe
essential but ‘lumpy’ expenses, on education, rites of demonstrated.
passage, housing or clothing for example, were Other studies have shown that the Sri Lankan
already beyond the household budget. Any health public health system has a pro-poor benefit in-
care expense, even a moderate direct cost burden of cidence and is among the most equitable in Asia
2.5–5% of monthly income, or a loss of income due (Rannan-Eliya & EQUITAP partners, 2005). From
to illness, inevitably triggered borrowing, pawning, the data presented here, free health care’s contribu-
or cuts to food and education. Longitudinal research tion to protecting against illness-induced impover-
showed that when a low or moderate direct cost ishment for the three household livelihood groups
burden affected a poor household only once or twice can be summarised as:
over 8 months then recovery was easier and illness
made little difference to poverty. However, frequent ! Free treatment mitigated further impoverishment
moderate illness costs experienced by poor families of declining households: Already on trajectories of
with small children or a chronically sick member livelihood decline, free health services mitigated
were a persistent attack on already overstretched deepening poverty from illness by reducing direct
budgets that contributed to debt accumulation, asset cost burdens. Low and insecure incomes, asset
depletion and made the household vulnerable to weaknesses and burdens imposed by other
other shocks. Vulnerability to income losses caused expenses, however, meant that free health care
by illness, as well as transport costs, increased the services alone were not enough to prevent
importance of the protection against medical costs livelihood decline. This demonstrates the vital
offered by free health care services. importance of free care for this group, and the
The case study data also showed that the relation- need for other interventions to build resilience
ship between cost burden and livelihood change is against illness costs and other shocks.
complex. Highly vulnerable households that experi- ! Free treatment prevented the decline or impover-
enced low or moderate burdens declined, but less ishment of relatively stable households: Free
vulnerable households that experienced a high or treatment, particularly free IP treatment and
‘catastrophic’ burden coped and remained stable. regular treatment of chronic illnesses, prevented
The longitudinal case study research could explore high cost burdens and contributed greatly to
the processes explaining the links between illness cost lower debts and the prevention of asset depletion
and livelihood change, through retrospective inter- among this group. Free treatment was particu-
views (life histories) and the prospective 8-month larly important when income earners could not
study. Multiple factors affected livelihood and work or at times when the household faced
poverty trajectories over time, including problems combined expenses. By protecting assets free
arising from legal expenses, drink and other drug treatment also made these households more
problems, earlier shocks, the loss of land or an resilient to future shocks.
221
ARTICLE IN PRESS
! Free treatment enabled investment by improving Coast, J. (1999). The appropriate use of qualitative methods in
households: Free hospital IP care acted as health economics. Health Economics, 8, 345–353.
‘insurance’ that allowed households to allocate Commission for Africa. (2005). Our common interest. London:
Report of the Commission for Africa.
resources to saving and investment strategies, Gilson, L., & McIntyre, D. (2005). Removing user fees for
rather than having to save to finance the costs of primary care in Africa: The need for careful action. British
a future hospital admission. Medical Journal, 331, 762–765.
Halstead, S. B., Walsh, J. A., & Warren, K. S. (1985). Good health
Free IP hospital services meant the health system at low cost. New York: Rockefeller Foundation.
Kawabata, K., Xu, K., & Carrin, G. (2002). Preventing
protected the full range of socio-economic groups impoverishment through protection against catastrophic
covered by the study. It also demonstrated that health expenditure. Bulletin of the World Health Organization,
effective protection requires a broad package of 53, 61–67.
curative treatment that is free at the point of Mitchell, J. C. (1983). Case and situation analysis. Sociological
delivery. However, the findings also showed that Review, 31, 187–211.
Prescott, N. (1999). Coping with catastrophic health shocks. In
free public services only protected the poor when Paper presented at a conference on social protection and
they were of a quality acceptable enough to be used. poverty, Inter-American Development Bank, Washington,
Public health care services were less successful in DC.
protecting patients against the direct costs of acute Rannan-Eliya, R. P. (2001). Strategies for improving the health
of the poor: The Sri Lankan experience. In Paper prepared for
illness requiring treatment outside the home because
Health Systems Resource Centre, Department of Interna-
people across income groups, even from the poorest tional Development, UK, by the Institute of Policy Studies,
income quartile, preferred to use private providers Colombo, Sri Lanka.
(Russell, 2005). Rannan-Eliya, R. P., & EQUITAP partners. (2005). Compara-
Overall, these findings can be applied to similar tive experiences of tax-funded health care systems in Asia:
urban settings in Colombo because the study sites Bangladesh, Hong Kong SAR, India, Indonesia, Malaysia
and Sri Lanka. Equity in Asia-Pacific Health Systems:
and households were selected to be typical of such Working Paper # 7.
settings and populations. In rural Sri Lanka income Ranson, K. (2002). Reduction of catastrophic health care
poverty is wider and deeper, and the direct costs of expenditures by a community-based health insurance
illness likely to be higher due to transport costs. scheme in Gujurat, India: Current experiences and
Protection against medical costs is therefore likely to challenges. Bulletin of the World Health Organization, 80(8),
613–621.
be even more important for poverty reduction and Russell, S. (2004). The economic burden of illness for house-
livelihood sustainability in rural areas of the country. holds in developing countries: A review of studies focus-
Although Sri Lanka’s universal provision model ing on malaria, tuberculosis and HIV/AIDS. American
faces financing and quality problems, the govern- Journal of Tropical Medicine and Hygiene, 71(Suppl. 2),
147–155.
ment should not start charging the user to raise
Russell, S. (2005). Treatment seeking behaviour in urban Sri
revenue. Fees would undermine livelihoods and one Lanka: Trusting the state, trusting private providers. Social
of the few pro-poor health systems in the world. Science and Medicine, 61, 1396–1407.
Scoones, I. (1998). Sustainable rural livelihoods: A framework for
Acknowledgements analysis. IDS Working Paper 72, Institute of Development
Studies, Brighton.
Sen, A. (1988). Sri Lanka’s achievements: How and when. In T.
Thanks to Anne Mills for her comments. The Srinivasan, & P. Bardhan (Eds.), Rural poverty in South Asia.
paper was written with financial support from the New York: Columbia University Press.
Health Economics and Financing Programme, Wagstaff, A. (2002). Poverty and health sector inequalities.
London School of Hygiene and Tropical Medicine, Bulletin of the World Health Organization, 80, 97–105.
UK, funded by the UK Department for Interna- Wallman, S., & Baker, M. (1996). Which resources pay for
treatment? A model for estimating the informal economy of
tional Development. The views and opinions health. Social Science and Medicine, 42(5), 671–679.
expressed are those of the authors alone. WHO. (2002). Improving health outcomes of the poor: The report
of working group 5 of the commission on macroeconomics and
health. Geneva: World Health Organization.
References World Bank. (2004). World development report: Making services
work for poor people. Washington, DC: World Bank.
Berman, P., Kendall, C., & Bhattacharyya, K. (1994). The Xu, K., Evans, D. B., Kawabata, K., Zeramdini, R., Klavus, J.,
household production of health: Integrating social science & Murray, C. J. L. (2003). Household catastrophic health
perspectives on micro-level health determinants. Social expenditure: A multi-country analysis. The Lancet, 362,
Science and Medicine, 38(2), 205–215. 111–117.
222
doi:10.1093/heapol/czh053 Health Policy and Planning 19(6),
HEALTH POLICY AND PLANNING; 19(6): 380–390 © Oxford University Press, 2004; all rights reserved.
The emergence of political priority for safe motherhood in

Honduras
JEREMY SHIFFMAN,1 CYNTHIA STANTON2 AND ANA PATRICIA SALAZAR3
1The Maxwell School of Syracuse University, Syracuse, NY, USA, 2Department of Population and Family Health
Sciences, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA and 3Japan International
Cooperation Agency, Tegucigalpa, Honduras
Each year an estimated 500 000 to 600 000 women die due to complications from childbirth, making this one
of the leading causes of death globally for women in their reproductive years. In 1987 a global initiative was
launched to address the problem, but few developing countries since then have experienced a documented
significant decline in maternal mortality levels.
Honduras represents an exception. Between 1990 and 1997 the country’s maternal mortality ratio – the
number of deaths due to complications during pregnancy, childbirth and the postpartum period per 100 000
live births – declined 40% from 182 to 108, one of the largest reductions ever documented in such a short
time span in the developing world.
This paper draws on three political science literatures – constructivist international relations theory, policy
transfer and agenda-setting – to explain how political priority for safe motherhood emerged in Honduras, a
factor that underpinned the decline. Central to the explanation is the unusually cooperative relationship that
developed between international donors and national health officials, resulting in effective transfer of policy
and institutionalization of the cause within the domestic political system. The paper draws out implications
of the case for understanding the political dynamics of health priority generation in developing countries.
Key words: policy transfer, agenda setting, constructivism, safe motherhood, maternal mortality, Honduras
Introduction bureaucrats throughout the decade, and because these inter-

actions resulted in successful policy transfer, implementation
Each year developing world health ministries accept financial and impact. In the 1990s the Honduran state made safe
and technical assistance from dozens of international health motherhood among its foremost priorities, and the country
policy networks promoting causes such as AIDS prevention, experienced one of the most dramatic declines in maternal
polio eradication, reproductive health, safe motherhood and mortality ever documented in such a short time span in the
health sector reform. Despite the resources they offer, these developing world. Between 1990 and 1997 the Honduran
networks must compete for the attention of ministries, since maternal mortality ratio declined from 182 to 108 maternal
limited health systems capacities prevent governments from deaths per 100 000 live births (Castellanos et al. 1990;
giving implementation priority to more than a handful of Meléndez et al. 1999). Both the 1990 and 1997 figures are
causes. highly reliable, as they are based on Reproductive Age
Mortality Surveys (RAMOS), the gold standard in maternal
Scholars of developing world health policy have analyzed the mortality investigations that examine every maternal death
emergence and forms of these networks (Reich 2000; Walt in a country over the course of a year and generate statistics
2001; Ogden et al. 2003; Widdus 2003), and the structure and for the entire population, rather than sample-based estimates
effectiveness of health ministries (Berman 1995; Bossert et with wide confidence intervals. There have been other cases
al. 1998; Olsen 1998). With only a few exceptions (Okuonzi of documented decline in such a short period of time, but
and Macrae 1995; Buse and Gwin 1998; Walt et al. 1999; Walt they are few and far between.1
et al. 2004), they have given little systematic attention to the
interactions between the two. Understanding the nature and Danel (1998) has analyzed the medical and technical inter-
quality of these interactions is crucial since these have ventions associated with the Honduran maternal mortality
bearing on why developing world governments may priori- decline. In this paper, we investigate how political priority
tize some health causes and neglect others. emerged for the cause. We employ concepts from three
political science literatures – constructivist international
This paper investigates network–ministry interactions and relations theory, policy transfer and agenda setting – to
their impact on health priority setting through a study of safe examine why successful policy transfer and implementation
motherhood in Honduras in the 1990s. The case is revealing occurred and to highlight the case’s significance for under-
because international officials concerned with safe standing network–ministry interactions and health priority
motherhood interacted repeatedly with Honduran health formation in developing countries.
223
Political priority for safe motherhood 381
Background ideas from elsewhere and import these of their own accord.
Policies may be transferred with compulsion if powerful
International policy networks
organizations such as the World Bank threaten to withhold
Over the past decade scholars have given increasing atten- lending to countries that do not embrace particular practices.
tion to the role of policy networks as actors in the inter- Policies may be transferred via structural forces when policy-
national system. These vary both in form and level of making elites play no active role, and ideas enter national
institutionalization. Two of the more widely researched systems through processes scholars often refer to as ‘conver-
forms are epistemic communities and transnational advocacy gence’.
networks. Haas (1992b) and colleagues coined the term epis-
temic communities to refer to groups of professionals who, Constructivist theory from the political science sub-field of
by virtue of their knowledge-based authority and shared international relations offers a useful framework for thinking
beliefs about causal processes, are able to influence national about how policy transfer may occur in certain instances.
policies. Among other issues, such groups have been able to Constructivism works from the premise that nation-states,
influence global trade agreements (Drake and Nicolaidis like individuals, are not isolated entities. They exist within
1992), nuclear arms control agreements (Adler 1992), societies of other nation-states and are socialized into
commercial whaling practices (Peterson 1992) and ozone commonly shared norms by their encounters with inter-
protection policy (Haas 1992a). Keck and Sikkink (1998) national actors such as the policy networks just discussed
have examined transnational advocacy networks. These (Wendt 1992; Finnemore 1996). Mainstream international
differ from epistemic communities in that their members relations scholars traditionally have downplayed this form of
consist of multiple organizational types, from labour unions transnational influence, as they have sought to understand
to churches, and are linked not by expertise but by shared the behaviour of nation-states in the international arena by
commitment to particular causes. In the 1990s they have looking inside states, taking state preferences as given
promoted environmental preservation, human rights and (Finnemore 1996). Neo-realism seeks to explain outcomes in
many other causes, and have had significant influence at the international system, such as alliances and warfare, in
global United Nations conferences (Chen 1996). terms of the pursuit by states of power and security. Neo-
liberalism is another version that understands state behav-
Scholars have noted the involvement of these networks in iour largely in terms of the pursuit of wealth. Both assume
international health promotion as well. Ogden, Walt and the nature of state preferences and seek to demonstrate their
Lush have analyzed networks involved in shaping policy for utility by their capacity to predict and explain outcomes in
sexually transmitted infections (Lush et al. 2003) and tuber- the international system (Finnemore 1996). In these frame-
culosis (Ogden et al. 2003). Reich, Widdus and Buse and works international policy networks are viewed as epi-
Walt have investigated the emergence of public-private phenomenal, unable to alter existing state preferences or
partnerships that link governments, pharmaceutical serving only to promote the interests that powerful states
companies and international organizations in legal structures would pursue anyway.
designed to find solutions to particular health problems
(Buse and Walt 2000; Reich 2000; Widdus 2003). Reinicke Constructivist international relations theory challenges
(1999) has identified Roll Back Malaria (a WHO-headquar- mainstream conceptions by raising the issue of how states
tered organization grouping governments, multilateral come to know what they want in the first place. Proponents
agencies, non-governmental organizations (NGOs) and do not necessarily reject neo-realist or neo-liberal ideas.
private sector organizations in a fight against the disease) as However, they argue that the pursuit of power, security and
a ‘global public policy network’ which he defines as ‘loose wealth cannot explain many critical international outcomes.
alliances of government agencies, international organiz- Constructivist theorists argue that on any given policy issue,
ations, corporations, and elements of civil society such as a state may not initially know what it wants but come to hold
nongovernmental organizations, professional associations, or certain preferences as a result of interactions in international
religious groups that join together to achieve what none can society with other state and non-state actors. For instance, a
accomplish on its own’ (p.44). state originally may not prioritize a health cause such as polio
eradication, but come to adopt the cause because domestic
health officials learn at international gatherings that other
Policy transfer and constructivist theory
countries are pursuing this goal and they are likely to be left
While attention has been paid to the emergence and forms behind. Thus, constructivists argue, state preferences cannot
of these health networks, there has been less research on the be taken as given (Wendt 1992; Finnemore 1996), but rather
means by which they influence national priorities. One should be conceived of as created in the process of trans-
concept of value on this subject is that of ‘policy transfer’ national interactions.
which concerns the use of knowledge about policies or
administrative arrangements in one time or place to develop International organizations are critical global actors in
such arrangements in another time or place (Dolowitz and frameworks influenced by constructivism. Organizations
March 1996). Stone (1999) notes that scholars employ such as the World Health Organization (WHO), UNICEF,
multiple terms to speak of the concept, including ‘lesson- the World Bank and the United Nations Population Fund
drawing’, ‘emulation’, ‘external inducement’, ‘convergence’ (UNFPA) are created by a global community of nation-states
and ‘diffusion’. She identifies three modes of transfer. Policy with a view to serving their jointly and individually held inter-
may be transferred voluntarily if elites in one country value ests. However, these organizations may acquire the power to
224
382 Jeremy Shiffman et al.
act as independent, autonomous agents, shaping the policy issues rise to the fore. The problems stream refers to the flow
preferences of the nation-states that created them (Abbot of broad issues facing societies. It is from this stream of issues
and Snidal 1998). International health policy networks, which that agendas are shaped. The policy stream refers to the set
link these actors with other kinds of organizations, may play of alternatives that scholars, politicians, bureaucrats and
similar roles in shaping national policy preferences. other prominent figures propose to address national
problems. This stream contains proposals concerning how
Risse-Kappen (1995) argues that the capacity of inter- problems may be solved. Finally, there is a politics stream.
national networks to influence national priorities depends on National mood, changes in political structure, social upris-
the international and domestic political structures through ings, elections and global political events are among the
which these actors must work. He contends, for instance, constituent elements of the politics stream.
that, other things being equal, transnational actors will find
states with centralized structures harder to penetrate than Kingdon and others have argued that there are systematic
those with fragmented structures; however, once they pene- elements in agenda setting which shape the likelihood that
trate these systems they are more likely to have policy an issue will receive national attention. In one of the earliest
impact. The reason is that unlike democratic, federalist works on agenda setting, Jack Walker (1974), analyzing
political systems (India, Brazil), power in authoritarian, traffic safety policy in the United States, argued that among
unitary systems (China, Vietnam) is concentrated in the the factors that shape whether an issue rises to the attention
hands of a few elites. External networks have fewer points of of policy-makers is the presence of a clear, measurable indi-
access, but if they are able to gain access and convince state cator to mark that issue. Kingdon confirmed Walker’s insight
leaders of the legitimacy of their agenda, these leaders are in his study of health and transportation policy-making in the
able to mobilize much of the political system in service of the United States, from which he developed his streams model.
cause. Agenda setting scholars argue that indicators make a differ-
ence because they have a uniquely powerful effect of giving
Ogden et al. (2003) point to another factor that may shape visibility to that which has remained hidden, serving not just
the level of network influence. Analyzing the case of global monitoring purposes, the way they are traditionally under-
tuberculosis policy, they show that international health advo- stood, but also as catalysts for action.
cates were able to convince many developing world govern-
ments to accept a particular treatment regimen. However, A second factor that researchers have identified is political
the consequence of promoting a uniform solution was its entrepreneurship (Walker 1974; Kingdon 1984; Doig and
insufficient tailoring to local context and a lack of ownership Hargrove 1987; Waddock and Post 1991; Schneider and
by domestic health officials. Policy transfer occurred, they Teske 1992). Whether an issue rises to the attention of policy-
argue, but the policy was not always implemented effectively. makers is not simply a matter of the flow of broad structural
Their study suggests that international health networks that forces that stand beyond the reach of human hands. Much
hand over a measure of control of resources and decisions to depends on the presence of individuals and organizations
domestic officials, and allow for adaptation of policy committed to the cause. As John Kingdon (1984, pp. 190–1)
solutions to local context, may be more effective in insti- puts it, ‘Entrepreneurs do more than push, push, and push
tutionalizing national priority for their causes than networks for their proposals or for their conception of problems. They
that are inflexible in these respects. also lie in wait – for a window to open. In the process of
leaping at their opportunity, they play a central role in
coupling the streams at the window.’
Agenda setting theory
Scholars concerned with policy transfer have focused A third factor is the occurrence of focusing events (Kingdon
primarily, if not exclusively, on the movement of policies 1984; Birkland 1997). These are large-scale happenings such
across national borders. Scholars concerned with agenda as crises, conferences, accidents, disasters and discoveries
setting have considered these processes predominantly inside that attract notice from wide audiences. They function much
domestic political systems. A previous study employed like indicators, bringing visibility to hidden issues. Birkland
agenda setting concepts to explain the emergence of political has demonstrated that disasters, including hurricanes, earth-
priority for safe motherhood in Indonesia (Shiffman 2003). quakes, oil spills and nuclear power plant accidents, lead to
The following discussion draws from that paper. heavy media coverage, interest group mobilization, policy
community interest and policy-maker attention, causing
Agenda setting is that stage in the public policy process shifts in national issue agendas.
during which certain issues rise to prominence and others are
neglected. It is the first stage in the process and precedes
Political science theory and the formation of developing
three others: policy formulation; the enactment of authori-
world health priorities
tative decisions; and policy implementation. The most influ-
ential theory of agenda setting is Kingdon’s streams model These political science literatures offer concepts useful for
(1984). He argues that agenda setting has a random charac- understanding network–ministry interactions and their influ-
ter and is best described as resembling a garbage can in which ence on health priority formation in developing countries.
problems, policies and politics develop and flow along in Constructivist international relations theory offers a way of
independent streams, meeting at random junctures in history understanding how developing world health bureaucracies
and creating windows of opportunity during which particular may come to embrace particular health causes: they may be
225
socialized into preferences through interactions with repre- We used a process-tracing methodology in constructing the
sentatives of international organizations, bilateral develop- case history, seeking to employ multiple sources of infor-
ment agencies and other actors that comprise international mation in order to minimize bias and establish common
health policy networks. The capacity of these networks to patterns of causality. Our aim was to investigate how safe
influence national priorities will be mediated by the inter- motherhood appeared on the Honduran health agenda, the
national and domestic structures through which they must degree to which the cause had been institutionalized in the
work. Also, scholars studying policy transfer and agenda country, and the factors behind the prioritization of the issue.
setting offer a set of propositions concerning the circum- In the language of case study methodology our inquiry was
stances under which these interactions are more likely to holistic in nature and selected based on its revelatory and
result in adoption and institutionalization of particular health unique characteristics (Yin 1994). That is to say, we analyzed
causes: the nation-state of Honduras holistically as a unit rather than
any of its sub-regions; we sought to make use of our access
(1) When networks are willing to share authority and control to policy-makers to reveal insights that may not have been
of resources with domestic officials; available otherwise; and we justified selection of Honduras
(2) When a focusing event gives a cause political visibility; for analysis because of its uniqueness in being one of the few
(3) When a credible indicator exists to mark the severity of developing countries to have experienced a documented
the problem; significant decline in maternal mortality in a short time span.
(4) When domestic political entrepreneurs are available who
embrace the cause and are willing to push from behind We chose a case study design because of the need to recon-
the scenes to promote it. struct holistically the history of the safe motherhood initia-
tive in the country in order to examine the processes at work.
In the sections that follow we draw on these constructivist, The case study approach is better suited than other research
policy transfer and agenda setting ideas to investigate how methodologies, such as a structured survey and statistical
safe motherhood became institutionalized as a priority in analysis of health service utilization, to achieve this objective
Honduras. (Yin 1994). This is true because the defining feature of the
case study is that it considers a phenomenon in its real-life
context and is therefore a research strategy well-suited to
revealing underlying processes.
Methods
We relied on four types of sources to develop the case study: The research design imposes limits on internal and external
interviews with officials involved in Honduran safe mother- validity. In-depth exploration enables us to develop hypoth-
hood policy; government reports and documents; donor eses concerning why political priority may have emerged for
agency reports; and published research on Honduran safe safe motherhood in Honduras, and to suggest general propo-
motherhood. We conducted in-depth semi-structured inter- sitions concerning public health agenda setting and
views with 30 individuals involved in Honduran safe mother- network–ministry interactions. On the other hand, the design
hood, 25 of which occurred in the country. We interviewed creates uncertainty about the conclusions, as they are
each of the five individuals who led the maternal and child grounded in consideration of only a single case. Additional
health division of the Ministry of Public Health between the comparative research on other countries that controls for
years 1986 and 2002; a number of senior officials in the alternative explanations will be necessary in order to assess
Ministry; NGO and private sector consultants; and members the causal power of the factors we identify. Also, any gener-
of the donor community including the Pan American Health alization to other settings must be done with caution given
Organization (PAHO – the Americas branch of the WHO), elements of the sociopolitical and health context that are
USAID, UNFPA, UNICEF and the World Bank. unique to Honduras.
The government reports we consulted included national

The case
health plans, national health surveys, safe motherhood
strategy papers and official documents on safe motherhood
The development of a national health infrastructure
norms. Of particular importance were the 1990 and 1997
RAMOS that provided evidence of a maternal mortality While Honduras’ neighbours – Guatemala, Nicaragua and El
decline in the country and information on its possible causes Salvador – were engulfed in civil war through much of the
(Castellanos et al. 1990; Meléndez et al. 1999). Donor docu- 1980s, Honduras faced no domestic insurrection and enjoyed
ments included regional and national safe motherhood plans United States support as a Cold War ally and bulwark of anti-
of action from PAHO, USAID-Government of Honduras communist resistance in the region. These favourable
health agreements and evaluations, and UNFPA project domestic and geo-political circumstances in part explained a
plans and reports for Honduras. Among the published heavy USAID presence in the country, and the capacity of
studies on Honduras we consulted, a World Bank report the Honduran state to devote a significant portion of its
(Danel 1998) was particularly valuable for historical analysis national budget to health infrastructure development.
on safe motherhood policy development and for evaluation
of interventions. The entire manuscript was reviewed for In 1987 health comprised 11.7% of the national budget
factual accuracy by several Honduran officials centrally (USAID 1988), considerably higher than the regional
involved in safe motherhood policy in the country. average. USAID supplemented this funding with grants of
226
US$54 million for health sector development and rural water motherhood policy meetings. Also, Honduras was listed as
and sanitation projects between 1981 and 1988 (USAID one of the regional priority countries for maternal mortality
1988). The agency cooperated closely with the Inter- reduction, and the government approved of the PAHO initia-
American Development Bank (IDB) (USAID 1988), which tive. Throughout the 1990s government delegations partici-
in 1987 approved a US$27 million loan for the construction pated in global meetings that reaffirmed international goals
and equipping of hospitals across the country (USAID 1988). for maternal mortality reduction, such as the 1994 Inter-
The Ministry of Health used domestic and donor resources national Conference on Population and Development in
to sustain a policy of extending health services throughout Cairo. Officials also joined in follow-up regional meetings,
the country, targeting the rural poor (USAID 1988). including an official Central American launch of the global
Between 1978 and 1987 the number of health centres staffed safe motherhood initiative at a conference in Guatemala in
by auxiliary nurses increased from 379 to 533; the number of 1992 (APROFAM et al. 1992).
health centres with doctors from 76 to 116; and the number
of hospitals from 16 to 21 (USAID 1988).
A 1990 maternal mortality study shocks the political system
Through the 1970s and 1980s, with donor assistance, the
government also prioritized maternal health. In 1968 the The appearance in 1990 of a credible study revealing a high
Honduran government, supported by USAID, established a level of maternal mortality in Honduras spurred national
project for the health of mothers and infants (Almanza-Peek health officials to respond to these global and regional calls
1998a) and in 1974 started an official maternal and child for action. Prior to the study many health leaders believed
health programme, the first stated objective of which was to Honduras did not have a serious maternal mortality problem,
decrease maternal mortality (HMPH et al. 1986, 1989). In the taking for granted a 1983 figure, derived solely from hospital-
1970s the Ministry of Health initiated a training programme based estimates, of 50 maternal deaths per 100 000 live births
for the approximately 10 000 traditional birth attendants (Castellanos et al. 1990).
across the country (Martinez 1994; HMPH 1998). UNFPA
also supported maternal and child health, financing An official from the Honduran office of PAHO, who was
programmes from 1978 through 1991, with technical support formerly with the Ministry of Public Health, played a key role
from PAHO, that had explicit goals of reducing maternal in organizing the study. He suspected the country had a
mortality (Almanza-Peek 1998a). These legacies facilitated maternal mortality problem, knew from his experience in the
the emergence of political priority and gave health leaders Ministry that Honduras had no reliable maternal mortality
the institutional capacity to address safe motherhood in the data, and had internal knowledge from his PAHO position
1990s. that the organization was about to make safe motherhood a
priority and allocate funds for the cause. He believed that
Honduras could secure resources for a national programme,
Safe motherhood emerges as a national priority
but only if it had credible data to prove a problem existed.
The emergence of safe motherhood as a global priority in the He lobbied and successfully generated financial support for
late 1980s raised political attention to maternal mortality the study from several organizations, including PAHO and
reduction in Honduras to a new level. The watershed event UNFPA.
was an international conference on safe motherhood in
Nairobi, Kenya in 1987, sponsored by the World Bank, The 1990 RAMOS study results shocked health officials. The
WHO, UNFPA and the United Nations Development research revealed a maternal mortality ratio of 182 maternal
Program (UNDP). At that time the global dimensions of the deaths per 100 000 live births, nearly four times the previ-
crisis – nearly 600 000 maternal deaths per year – were widely ously accepted figure (Castellanos et al. 1990).2 Furthermore,
publicized, and delegates called for a global reduction of 50% credible data showing haemorrhage as the leading cause of
by the year 2000. The conference officially launched a global maternal death, and twice as many maternal deaths occurring
safe motherhood movement, and solidified an international at home as opposed to in hospital, suggested not only a
safe motherhood network that linked these organizations problem of a much different scale than anticipated, but also
with government bodies, NGOs and safe motherhood advo- a problem of a different nature. Honduran women were not
cates across the globe. reaching public or private obstetric services. In some regions,
between 80 and 90% of deaths occurred at home. Even in the
At the conference advocates promoted risk assessment metropolitan area of Tegucigalpa, nearly one in four
during antenatal care to distinguish between women at high maternal deaths occurred at home. Armed with this infor-
and low risk of suffering obstetric complications at delivery, mation and committed to making maternal mortality reduc-
and the training of traditional birth attendants for low risk tion a political priority, the official and his colleagues actively
women. Responding to this launch, PAHO prioritized the publicized the study’s results. They produced and distributed
cause, in 1990 producing a plan for the reduction of maternal over 1000 copies of the report, presented the study to the
mortality in the Americas and securing its approval from its media, briefed international organizations on the results and
member states (PAHO 2002a). lobbied health officials in the capital and regions of the
country. By the end of 1990 a new health minister had
The government of Honduras participated extensively in commented in the national media on the study, noting that
these global priority-setting initiatives. It was a member of the country had a serious problem with maternal mortality
PAHO and its minister of health participated in safe and that the government was in negotiations with UNFPA to
227
generate funds for a national programme (La Tribuna to make safe motherhood a priority, the official informed
1991a,b). their superior, the minister, who spoke to them directly.
Domestic health officials mobilize the political system for

safe motherhood Donors provide resources for safe motherhood
Public efforts by the study’s authors brought national atten- These advocacy efforts may have had limited impact had they
tion to the issue. Entrepreneurship behind the scenes by mid- not been backed by financial and technical resources. In this
level health officials made the issue an ongoing priority. respect existing donor commitment to safe motherhood and
the participation of their local representatives in the working
The new health minister had longstanding ties with the head group proved crucial.
of one of Honduras’ seven health regions. The minister was
assembling a new team in the capital and asked the regional The only major safe motherhood intervention funded
head to serve as director of the maternal and child division. primarily from the central government health budget was
The official agreed on the condition that he would have direct the training of several thousand traditional birth attendants.
access to the minister, even though several levels of bureau- Local governments provided some additional resources and
cracy stood between the two men. The minister assented to donors many more. USAID supported maternal mortality
the request. reduction through a renewal of a grant to the country,
providing a further US$57.3 million to the health sector
As he took up his new post in September 1990, the official between 1988 to 2000 (USAID 1988) and sponsoring a mid-
paid careful attention to the published study, taking advan- term evaluation of the grant that recommended safe mother-
tage of his access to the minister to convey to him the seri- hood be the country’s top health priority (Population
ousness of the country’s maternal mortality problem and the Technical Assistance Project 1998). UNFPA approved new
need to make safe motherhood a policy priority. He then funding for Honduras for 1991 to 1995, including a sub-
employed his close ties with the minister, other health programme on reproductive health and the health of
officials and donors to lead an effort to mobilize the health mothers (Almanza-Peek 1998a,b), providing nearly half a
system in service of the safe motherhood cause. He formed million dollars for reproductive health projects in two
a working group that devised national strategies, engaged regions of the country. The Honduran office of PAHO
regional health bureaucracies and organized donor resources offered technical expertise, receiving financial backing from
and expertise. the Netherlands and other donors (Martinez 1994). The
World Bank financed a Honduran Social Investment Fund
This working group became the unofficial centre for national that provided financing for safe motherhood (Martinez
safe motherhood efforts. Meeting regularly over several 1994). A Swedish-assisted initiative, termed ‘Project
years and at certain points on a weekly basis, the group Access’, carried out health system decentralization in order
included members of the Ministry’s division of maternal and to increase access to facilities for the poor (Population Tech-
child health, the initiator of the maternal mortality study nical Assistance Project 1998). Other donors that provided
from PAHO, and local representatives of USAID, UNFPA, financial or technical assistance for safe motherhood
UNICEF and other donors and agencies. The group included the Germans, the Canadians, the Spanish, the
produced a national plan of action for maternal mortality European Union, UNICEF and the Latin American Center
reduction for the period 1991 to 1995, adopting many ideas for Perinatalogy in Uruguay.
from PAHO’s 1990 regional plan, while tailoring them to fit
local circumstances (AHPF and HMPH 1991). Donor efforts at the regional level in the Americas also
helped to sustain political priority and the capacity of the
The group also embarked on an effort to mobilize regional Honduran health system to carry out safe motherhood
health bureaucracies in service of safe motherhood. As a programmes. In 1991, PAHO, UNFPA, UNICEF, USAID
former regional health leader, the official was aware of the and the IDB formed an inter-agency committee to work to
many health problems his colleagues had to face with limited institutionalize commitment to safe motherhood and other
resources, and of the challenge he therefore confronted in health initiatives throughout the region (PAHO 1996).
convincing them to prioritize safe motherhood. For this Representatives of the Honduran government participated
reason, members of the working group travelled to each of in a Central American launch of the global safe motherhood
the regions, spending a week or more with leaders, hospital initiative in 1992. Encouraged by PAHO, the spouses of
directors and other officials involved in safe motherhood, heads of state in the Americas region, including the
presenting the results of the study, persuading them of the Honduran first lady, made safe motherhood a central topic
seriousness of the problem, and facilitating the creation of of attention at their annual meetings from 1993 on (PAHO
local action plans. The existence of the 1990 study proved to 2000). With U.S. first lady Hillary Clinton playing a central
be a powerful tool in generating regional attention as it role, the spouses backed a USAID and PAHO regional safe
provided credible evidence that a problem existed. The motherhood initiative begun in 1995 to upgrade emergency
official and colleagues also organized annual, national safe obstetric care facilities in high maternal mortality settings
motherhood evaluation meetings, bringing together officials (PAHO 2002a). Honduras was one of three priority countries
from throughout the country to review progress and develop (PAHO 2002b) and received additional funding for this
future plans. When some regional heads were still reluctant purpose.
228
Outcomes around 1990 (AHPF and HMPH 1991) to 85% in the late
1990s (HMPH et al. 2001). Institutional delivery rose from 45
These Honduran government and donor efforts resulted in to 61% over this same time period (HMPH et al. 2001), with
substantial expansion of the country’s health and safe increases particularly evident in rural areas (HMPH et al.
motherhood infrastructure, with resources concentrated in 1989, 1996, as reported in Danel 1998). Likewise, caesarean
those regions identified by the 1990 report as having the sections, the most common life-saving procedure among
highest levels of maternal mortality. Between 1990 and 1997 emergency obstetric care practices, increased to 8%, with
seven new area hospitals were opened, 13 birthing centres, rural rates reaching nearly 5% in 1998 (Figure 1).
36 medical health centres and 266 rural health centres (Danel
1998, p. 5). The number of doctors rose 19.5%, the number In 1997, a second national RAMOS study was conducted on
of professional nurses 66.4% and the number of auxiliary the country’s maternal mortality levels (Meléndez et al.
nurses 41.9% (from Ministry of Public Health statistics, cited 1999). The same official who organized the first study again
in Danel 1998). In 1993 and 1994 half of the country’s secured donor funding for the second, and once more the
approximately 11 000 traditional birth attendants were results drew the attention of health officials. The investigation
trained in the reproductive risk approach (Martinez 1994). revealed a maternal mortality ratio of 108, indicating a signifi-
Community leaders developed censuses of women of repro- cant decline from the 1997 ratio of 182. The report provided
ductive age (AHPF and HMPH 1991) and health workers strong evidence that increased access to maternal health care
lists of pregnant women (Danel 1998, p. 11). Health centres played a role in this decline (Danel 1998). For example,
organized community groups to support educational whereas a third of maternal deaths occurred in hospitals in
programmes directed at pregnant women (Martinez 1994). 1990, more than half occurred in hospitals in 1997. Dystocia,
The Ministry of Health published the Norms for the Inte- or prolonged labour, for which effective care can often be
grated Care of Women employed at health facilities through- provided within 24 hours or more, basically disappeared as a
out the country (Danel 1998). cause of maternal death (decreasing from 4% in 1990 to less
than 1% of maternal deaths in 1997). In contrast, haemor-
Access and utilization by Honduran women of safe mother- rhage, which requires immediate medical attention, remained
hood services increased markedly over this period. Antenatal the leading cause of maternal death in 1997, but was substan-
care increased and became increasingly professionalized with tially reduced in numbers and a higher percentage of these
smaller proportions of women relying only on traditional birth deaths occurred in hospital. Finally, the reductions in
attendants for care during pregnancy. Use of antenatal care maternal mortality and the percentages of maternal deaths
with a medically trained professional increased from 72% shifting from home to the hospital are apparent in the
100
90 1998
1993
80
1989/90
70
1998
60
1993
50 1989/90
40
30
20
10 1998
1993
1989/90
0
Antenatal care Institutional Delivery Caesarean Section Rate
1989/90 1993 1998
Figure 1. Honduran safe motherhood process indicators: percentage of births in last 5 years to women 15–44 years with at least one ante-
natal care visit with medically trained personnel, percentage with an institutional delivery and caesarean section rate
Sources: AHPF and HMPH (1991), HMPH et al. (1996), HMPH et al. (2001).
229
metropolitan area of Tegucigalpa, as well as in the most network, including the World Bank, WHO, USAID and
disadvantaged regions of the country (Meléndez et al. 1999) UNFPA. The network provided a conduit for the influence
Although disparities in maternal mortality and access to care of international safe motherhood advocacy on the Honduran
remained in 1997, these results suggest that Honduras made state. Also, the Honduran state provided a receptive environ-
important strides in making effective maternal health care ment for such influence. It was not politically fragmented or
available to a broad section of the population. unstable like its neighbours, it had developed a solid health
infrastructure over several decades, and it had prioritized
As noted above, political and health infrastructural develop- maternal health since the late 1960s. Each of these conditions
ments were taking place globally and in Honduras well facilitated transnational influence and increased the likeli-
before 1990, so it is unlikely the decline was solely a function hood of domestic support for the cause.
of activities taken in the time period between the two studies.
Also, there are no reliable data prior to 1990 on the country’s Another facilitating factor was the emergence of shared
maternal mortality levels, so we cannot discern trends in decision-making authority between domestic and inter-
periods prior to that year. This being said a change from 182 national officials. This was not a case of international donors
to 108 represents a decline of 40% in just seven years, a wielding financial resources to push particular policy alterna-
difference rarely seen in the developing world over such a tives on a supplicant, uninterested state. On the contrary, a
short time span, strongly suggesting the impact of activities working group linking domestic health bureaucrats and
undertaken between these years. representatives of international and donor organizations in a
cooperative relationship emerged as the unofficial centre of
national safe motherhood efforts. The group included repre-
Discussion
sentatives from the Ministry of Health, bilateral donors and
United Nations organizations. It shared resources, coordi-
Political science theory and the case of safe motherhood in
nated strategy, worked collectively to promote priority for
Honduras
the cause across the country, and facilitated adaptation of
Between 1990 and 1997 domestic health officials and inter- global safe motherhood policies by encouraging local govern-
national donors cooperated to institutionalize safe mother- ments to develop contextually-relevant implementation
hood as a policy priority in Honduras, resulting in successful strategies.
policy transfer, implementation and impact on maternal
mortality levels. The political science literatures reviewed Three factors identified in agenda setting scholarship also
above – constructivism, policy transfer and agenda setting – were influential. International focusing events, particularly
help to identify the factors behind these successful outcomes. the Nairobi conference, placed safe motherhood on the
global health agenda. Regional focusing events, including a
Constructivist theory suggests that states may be socialized safe motherhood conference in Guatemala, raised the issue
into particular policy preferences by virtue of their partici- onto the Central American agenda. Domestic focusing
pation in international society. The Honduran state was events, including media conferences publicizing results of the
socialized in this way. Beginning in the late 1980s and 1990 Honduran maternal mortality study, facilitated the rise
continuing through the 1990s international organizations of the issue onto the national agenda. Also, this study
prioritized maternal mortality reduction, facilitating the produced a credible indicator – a high maternal mortality
creation of a global norm that maternal death in childbirth is ratio – which revealed levels of maternal death far higher
unacceptable and that states must act to address the issue. than expected, sparking alarm in the political system. In the
The Honduran government was influenced to embrace the absence of such evidence, advocates would have had diffi-
norm through two concurrent processes. First, Honduran culty promoting the cause. Finally, the Honduras PAHO
officials were members of a number of international organiz- representative and his colleagues acted as political entrepre-
ations that prioritized safe motherhood. In particular the neurs, organizing the 1990 RAMOS study, deliberately publi-
Honduran government actively participated in PAHO, which cizing the results to convince key health officials that the
urged its member states to pay attention to the cause. country faced a serious problem, and allying themselves with
Through participation in these and other forums, Honduran donor officials to mobilize the health system in service of the
government officials came to learn of and pay attention to safe motherhood cause. They worked as forces behind the
the issue. Secondly, these same organizations had local scenes pushing to ensure priority was institutionalized in the
presence in the Honduran capital. Their representatives, political system.
many of whom were Honduran nationals, interacted with
Ministry of Health officials, and a number jumped back and In sum, constructivist, policy transfer and agenda setting
forth between positions with the donor agencies and the constructs help us identify the factors that underpinned
Ministry. These individuals served as conduits of priority, successful policy transfer and implementation:
linking transnational and national forces.
(1) the effective socialization of the Honduran state into
Constructivist-influenced scholarship also suggests that global safe motherhood norms;
certain kinds of international and domestic structures will (2) favourable international and domestic mediating struc-
facilitate the capacity of transnational actors to influence tures, particularly a strong international safe mother-
domestic policy priorities. Powerful international institutions hood policy network and domestic political stability, that
concerned with safe motherhood were linked in a tight facilitated policy transfer;
230
(3) shared power by domestic and international officials that shape the behaviour of the Honduran state? Or was he a
facilitated local embrace of the cause and contextually Honduran citizen who utilized his position in PAHO to
relevant policy adaptation; generate resources for an existing national policy priority?
(4) the organization of attention-generating focusing events And what was the status of this working group of which he
that gave visibility to the cause internationally and was a part? It included Honduran nationals, some of whom
domestically; were employed by the government and others by inter-
(5) the existence of a credible indicator to mark the severity national donors, as well as nationals of other countries, all
of the problem; and working together for the objective of reducing maternal
(6) political entrepreneurship by national health officials to mortality in the country. As they engaged in this initiative,
institutionalize domestic priority for the cause. they formed a collectivity defined not so much by nationality
or organizational affiliation but by cause.
Study limitations, further research and implications for
public health strategy The nature and authority of these locally situated nodes of
linkage between international and national forces remain
Our case study design involving a single country and health largely unexplored. These deserve considerably more atten-
policy issue enables us only to raise questions and suggest tion for at least three reasons. First, they may be more
answers, not to provide definitive conclusions. The govern- common than imagined and hold considerable influence over
ments of many nation-states were exposed to and part- national health priorities in many developing countries.
icipated in the creation of a global norm concerning Secondly, their very existence presents a challenge to a basic
the unacceptability of maternal death in childbirth. Only a presumption in constructivist, policy transfer and agenda
handful such as Honduras embraced the norm and acted setting theory that there exists a neat demarcation between
decisively to address the problem. We have explained the the ‘international’ and the ‘national’. In these working groups
divergent reaction by considering a set of trans- these two categories may be fused, and in some instances
national–national linkages and domestic political factors. In meaningless. Thirdly, their emergence may help explain why
the absence of comparative inquiry we cannot be certain that policy transfer and implementation proceed effectively.
the factors we point to were the primary causal forces.3 There
is a need for further research that considers multiple states This latter point may be the most significant lesson that
and health policy issues in order to assess the validity of these emerges from the Honduran case for health agenda setting
causal claims, and to discern systematic features of health in developing countries. Many relationships between inter-
agenda-setting processes. Among the issues that should be national health policy networks and developing world health
investigated are: bureaucracies are fraught with tension. Often donors,
wielding control over resources, have sought to impose their
(1) What kinds of focusing events shape policy attention for priorities upon bureaucracies without considering local inter-
health causes? What are the features of focusing events ests, the capabilities of domestic bureaucrats, the need for
that give them agenda setting power? policy adaptation, and the considerable national political
(2) Under what conditions do indicators have agenda setting manoeuvring that must take place in order to institutionalize
power? Under what conditions do they fail to have a health cause as a domestic priority. It is rare that overseas
impact? donor or health network officials have the legitimacy or
(3) Under what circumstances can/do domestic political expertise to pursue such political manoeuvring successfully;
entrepreneurs make a difference? What is it they do that that capability, if it exists, almost always resides in the hands
makes a difference? of domestic bureaucrats and political officials. While many
(4) What features of international health policy networks factors shape the agenda-setting process, as dozens of inter-
give them the capacity to influence domestic health national health policy networks compete for attention, it may
priorities? In particular, what is the relationship between be those that are willing to hand over a measure of control
network structure and the power to influence? and forge alliances with domestic bureaucrats that stand the
(5) As donor–government relations in health are so best chance of having their causes institutionalized.
frequently contentious, under what circumstances is
productive cooperation likely to emerge? Endnotes
1 Since a 1987 global safe motherhood initiative was launched,
The latter question is particularly important and little inves-
tigated. The authority of the Honduran working group high- the only other case of a documented major decline in a poor country
confirmed by two Reproductive Age Mortality Studies (RAMOS)
lights the fact that the forces shaping priority for safe is Egypt, which had a maternal mortality ratio of 174 in 1992 and 84
motherhood in Honduras were not unidirectional, flowing in 2000 (Ministry of Health and Population, Egypt 2001). Histori-
from international to domestic actors alone. Influence moved cally, there are a handful of developing countries that have experi-
in both directions, merging as members acted collectively to enced documented declines, including Sri Lanka and Malaysia
address the country’s safe motherhood problems. Moreover, (Pathmanathan et al. 2003), and China (Koblinsky 2003). More
in some instances the boundaries between the international recently, a number of countries with moderate levels of maternal
mortality around 1990 have documented further declines over the
and national were indistinct. How should we characterize the following decade, including Uzbekistan, Azerbaijan, Argentina,
official who organized the first maternal mortality study? Was Cuba, Costa Rica and Chile (WHO et al. 2001).
he a representative of the international organization, PAHO, 2 Other publications have reported a maternal mortality ratio
who employed his organizationally derived authority to of 220 for 1990. The figure 220 came from the 1990 study, but was
231
the pregnancy-related mortality ratio: the number of deaths per Haas PM. 1992a. Banning chlorofluorocarbons: epistemic
100 000 live births occurring to women during pregnancy, childbirth community efforts to protect stratospheric ozone. International
or the postpartum period, but not necessarily causally related to the Organization 46: 187–224.
pregnant state. The maternal mortality ratio based on the definition Haas PM. 1992b. Introduction: epistemic communities and inter-
of maternal death in the International Classification of Diseases national policy coordination. International Organization 46:
(Revision 10), also reported in the study, is 182. 1–35.
3 It should be noted that a number of the factors identified here Honduran Ministry of Public Health (HMPH). 1991. Maternal
were also influential in Indonesia (Shiffman 2003), providing mortality: the problem and how to solve it. Tegucigalpa,
additional evidence for their causal power. These include the avail- Honduras: Honduran Ministry of Public Health.
ability of a credible indicator showing that a problem existed, Honduran Ministry of Public Health (HMPH). 1998. Manual para
effective political entrepreneurship and the organization of atten- capacitadores de parteras tradicionales en Honduras.
tion-generating focusing events. Also, there as in Honduras, a rela- Tegucigalpa, Honduras: Honduran Ministry of Public Health.
tively stable political system and the development of a national Honduran Ministry of Public Health (HMPH), Ashonplafa.
health infrastructure made it possible for international and domestic 1996. Encuesta nacional de epidemiologia y salud familiar –
safe motherhood advocates to promote the cause. Honduras 1996 (ENESF), informe final. Tegucigalpa,
Honduras: Honduran Ministry of Public Health and
References ASHONPLAFA.
Honduran Ministry of Public Health (HMPH), Ashonplafa, Family
Abbott KW, Snidal D. 1998. Why states act through formal inter- Health International (FHI), Management Sciences for Health
national organizations. Journal of Conflict Resolution 42: 3–32. (MSH). 1986. Encuesta nacional de salud materno infantile de
Adler E. 1992. The emergence of cooperation: national epistemic Honduras, 1984. Tegucigalpa, Honduras: Honduran Ministry of
communities and the international evolution of the idea of Public Health and Ashonplafa.
nuclear arms control. International Organization 46: 101–45. Honduran Ministry of Public Health (HMPH), Ashonplafa,
Almanza-Peek L. 1998a. El Fondo de Población de las Naciones Management Sciences for Health (MSH), Family Health Inter-
Unidas y la evolución de la salud reproductiva en Honduras. national (FHI). 1989. Epidemiology and family health survey
Tegucigalpa, Honduras: UNFPA. Honduras, 1987. Tegucigalpa, Honduras: Honduran Ministry of
Almanza-Peek L. 1998b. Evaluación cualitativa: Proyectos del Public Health, Ashonplafa, Management Sciences for Health
subprograma de salud reproductiva, 1994–1997. Tegucigalpa, and Family Health International.
Honduras: UNFPA. Honduran Ministry of Public Health (HMPH), Ashonplafa,
Asociación Hondureña de Planificación de Familia (AHPF), USAID, CDC, Management Sciences for Health (MSH). 2001.
Honduran Ministry of Public Health (HMPH). 1991. Encuesta Honduras: Encuesta nacional de epidemiologia y salud familiar
nacional de epidemiología y salud familiar 1991. Tegucigalpa, 2001. Tegucigalpa, Honduras: Honduran Ministry of Public
Honduras: Honduran Ministry of Public Health. Health, Ashonplafa, USAID, CDC and Management Sciences
Asociación Pro-Bienestar de la Familia de Guatemala for Health.
(APROFAM), Family Care International (FCI), Inter- Keck ME, Sikkink K. 1998. Activists beyond borders: advocacy
American Parliamentary Group on Population and Develop- networks in international politics. Ithaca, NY: Cornell.
ment (IAPG). 1992. Guatemala safe motherhood declaration. Kingdon JW. 1984. Agendas, alternatives and public policies. Boston
Guatemala City, Guatemala: APROFAM, FCI and IAPG. and Toronto: Little, Brown and Company.
Berman P. 1995. Introduction: special issue on health sector reform Koblinsky MA. 2003. Reducing maternal mortality: learning from
in developing countries. Health Policy 32: 3–11. Bolivia, China, Egypt, Honduras, Indonesia, Jamaica, and
Birkland TA. 1997. After disaster: agenda setting, public policy, and Zimbabwe. Washington, DC: The World Bank.
focusing events. Washington, DC: Georgetown University La Tribuna. 1991a. Salud intentara reducir embarazos de Hondure-
Press. nas. Newspaper article, 31 January 1991. Tegucigalpa,
Bossert T, Hsiao W, Barrera M et al. 1998. Transformation of Honduras.
ministries of health in the era of health reform: the case of La Tribuna. 1991b. Salud promete reducir defuncion por embarazo.
Colombia. Health Policy and Planning 13: 59–77. Newspaper article, 12 January 1991. Tegucigalpa, Honduras.
Buse K, Gwin C. 1998. The World Bank and global cooperation in Lush L, Walt G, Ogden J. 2003. Transferring policies for treating
health: The case of Bangladesh. The Lancet 351: 665–69. sexually transmitted infections: what’s wrong with global guide-
Buse K, Walt G. 2000. Global public-private partnerships: Part I – a lines. Health Policy and Planning 18: 18–30.
new development in health? Bulletin of the World Health Martinez L. 1994. Evaluation of the regional plan of action for the
Organization 78: 549–561. reduction of maternal mortality in the Americas: Honduras 1994.
Castellanos M, Ochoa JC, David V. 1990. Investigation of the mortal- PAHO, WHO and UNFPA.
ity of woman of reproductive age with an emphasis on maternal Meléndez JH, Ochoa JC, Villanueva Y. 1999. Investigación sobre
mortality. Washington, DC: PAHO [in Spanish]. mortalidad materna y de mujeres en edad reproductiva en
Chen MA. 1996. Engendering world conferences: the international Honduras: informe final correspondiente al año 1997. Teguci-
women’s movement and the UN. In: Weiss T, Gordenker L galpa, Honduras: PAHO/WHO.
(eds). NGOs, the UN and Global Governance. Boulder, CO: Ministry of Health and Population, Egypt. 2001. National maternal
Lynne Rienner Press, pp. 139–53. mortality study, Egypt 2000, report of findings and conclusions.
Danel I. 1998. Maternal mortality reduction, Honduras, 1990–1997: Cairo, Egypt: Directorate of Maternal and Child Health Care
A case study. Washington, DC: World Bank. and the Healthy Mother/Healthy Child Project.
Dolowitz D, March D. 1996. Who learns from whom: a review of the Ogden J, Walt G, Lush L. 2003. The politics of ‘branding’ in policy
policy transfer literature. Political Studies 44: 343–57. transfer: the case of DOTS for tuberculosis control. Social
Doig JW, Hargrove EC (eds.) 1987. Leadership and innovation: a Science and Medicine 57: 179–88.
biographical perspective on entrepreneurs in government. Balti- Okuonzi S, Macrae J. 1995. Whose policy is it anyway? International
more: Johns Hopkins University Press. and national influences on health policy development in
Drake WJ, Nicolaidis K. 1992. Ideas, interests, and institutionaliz- Uganda. Health Policy and Planning 10: 122–32.
ation: ‘trade in services’ and the Uruguay Round. International Olsen I. 1998. Sustainability of healthcare: a framework for analysis.
Organization 46: 37–100. Health Policy and Planning 13: 287–95.
Finnemore M. 1996. National interests in international society. PAHO. 1996. Intensive implementation of the basic emergency obstet-
Ithaca, NY: Cornell University Press. ric care (BEOC) aspects of the regional plan for reduction of
232
maternal mortality [Proposal of PAHO to USAID, 1997–2001]. resources in the health sector: are there lessons for SWAPs?
Washington, DC: PAHO. Health Policy and Planning 14: 273–84.
PAHO. 2000. Program on the reduction of maternal mortality Wendt A. 1992. Anarchy is what states make of it: The social
[Report for the eighth conference for wives of heads of states construction of power politics. International Organization 46:
of the governments of the Americas, Santiago, Chile, 1998]. 391–425.
Washington, DC: PAHO. Widdus R. 2003. Public-private partnerships for health require
PAHO. 2002a. Regional strategy for maternal mortality and morbid- thoughtful evaluation. Bulletin of the World Health Organiz-
ity reduction [Provisional agenda item for 130th session of the ation 81: 235.
executive committee] Washington, DC: PAHO. World Health Organization (WHO), United Nations Children’s
PAHO. 2002b. The Latin America and Caribbean (LAC) regional Fund (UNICEF), United Nations Population Fund (UNFPA).
initiative to reduce maternal mortality. Washington, DC: PAHO 2001. Maternal mortality in 1995: estimates developed by WHO,
[available at http: www.paho.org/English/HPP/HPF/REP/ UNICEF, UNFPA. WHO/RHR/01.9. Geneva: WHO.
SMI_lac.htm]. Yin RK. 1994. Case study research: design and methods (2nd ed.).
Pathmanathan I, Liljestrand J, Martins J et al. 2003. Investing in Thousand Oaks, CA: Sage Publications.
maternal health, learning from Malaysia and Sri Lanka.
Washington, DC: The World Bank.
Peterson MJ. 1992. Whalers, cetologists, environmentalists, and the Acknowledgements
international management of whaling. International Organiz-
Research for this article was made possible through a grant from the
ation 46: 147–86.
Bill and Melinda Gates Institute for Population and Reproductive
Population Technical Assistance Project. 1998. Mid-term evaluation
Health of the Johns Hopkins Bloomberg School of Public Health.
of the health sector II project [Report prepared for U.S. Agency
Funding from the Institute is gratefully acknowledged. The authors
for International Development]. Arlington, VA: Population
would like to thank Jessica Hughes for her valuable research and
Technical Assistance Project.
translation assistance. The authors would also like to thank the
Reich MR. 2000. Public-private partnerships for public health.
many donor and government officials who were generous with their
Nature Medicine 6: 617–20.
time and insights, and without whom it would have been impossible
Reinicke WH. 1999. The other world wide web: global public policy
to evaluate the Honduran safe motherhood experience.
networks. Foreign Policy 78: 44–57.
Risse-Kappen T (ed). 1995. Bringing transnational relations back in:
non-state actors, domestic structures and international insti- Biographies
tutions. New York: Cambridge University Press.
Schneider M, Teske P. 1992. Toward a theory of the political entre- Jeremy Shiffman is assistant professor of public administration at
preneur: Evidence from local government. American Political the Maxwell School of Syracuse University. His research focuses on
Science Review 86: 737–47. the political dimensions of public health policy-making in develop-
Shiffman J. 2003. Generating political will for safe motherhood in ing countries. He has published on the subjects of maternal mortal-
Indonesia. Social Science and Medicine 56: 1197–207. ity, family planning, reproductive rights, infectious disease control
Stone D. 1999. Learning lessons and transferring policy across time, and health sector reform.
space and disciplines. Politics 19: 51–9.
USAID. 1988. Honduras project paper: health sector II. Washington, Cynthia Stanton is an assistant public health professor in the Depart-
DC: USAID. ment of Population and Family Health Sciences at the Johns
Waddock SA, Post JE. 1991. Social entrepreneurs and catalytic Hopkins Bloomberg School of Public Health, Baltimore, USA. Her
change. Public Administration Review 51: 393–401. research primarily concerns measurement issues related to maternal
Walker JL. 1974. Performance gaps, policy research, and political and perinatal mortality and morbidity in the developing world.
entrepreneurs: toward a theory of agenda setting. Policy Studies
Journal 3: 112–16. Ana Patricia Salazar works for the Japanese International Cooper-
Walt G. 2001. Global cooperation in international public health. In: ation Agency on maternal mortality and other reproductive health
Milsen M, Black R, Mills A (eds). International public health: issues in Honduras. She was formerly employed by the Honduran
diseases, programs, systems and policies. Gaithersberg, MD: office of the United Nations Population Fund.
Aspen Publishers, pp. 667–97.
Walt G, Lush L, Ogden J. 2004. International organizations in Correspondence: Jeremy Shiffman, Assistant Professor of Public
transfer of infectious diseases: iterative loops of adoption, Administration, The Maxwell School of Syracuse University,
adaptation and marketing. Governance 17: 189–210. Syracuse, New York 13244–1090 USA. Tel: +1 315 443-4928 ; fax: +1
Walt G, Pavignani E, Gilson L, Buse K. 1999. Managing external 315 443-9734 ; email: jrshiffm@maxwell.syr.edu
233
3. The ethnographic First, ethnographies that have followed the life, or lives, of
individuals and groups affected by a particular health
lens condition have developed our understanding of how and
why people are enabled (or hindered) in their efforts to
Karina Kielmann make effective use of services and manage their
ueen Margaret niversity, Scotland, nited ingdom of conditions. For example, recent work has examined how
reat ritain and orthern reland people living with a condition draw on a collective
‘biosocial’ identity to formulate claims to treatment,
While ‘ethnographic’ has mistakenly come to be used as
compensation, and other social resources. In the case of
a blanket term to refer to various qualitative methods,
HIV, some have argued that this form of ‘therapeutic
ethnography is more accurately seen as a particular
citizenship’ has directly affected policies around access to
methodology. The term refers to both a research
treatment as well as the delivery of HIV care (see Nguyen,
approach (literally, ‘writing about people’) as well as the
2008).
written product of the research (such as a text, report or
book). It represents a defining moment in anthropology, Second, ethnographies that have explicitly focused on
the point at which scholars abandoned the ‘armchair’ in practitioners and their professional socialization within
favour of fieldwork to capture the totality of social life in health systems provide important insights into the
an alien setting. The classical approach to ethnography feasibility of health systems interventions that assume (or
generally involves lengthy periods of fieldwork, immer- introduce shifts in) particular professional hierarchies or
sion in the ‘everyday life’ of a chosen setting through working arrangements. One focus has been to examine
observation, interaction, talking to members of the how working environments and workplace dynamics
particular social world being studied, and looking at shape provider identities and interprofessional collabo-
documents or artefacts. The written account is a synthesis ration. For example, attention has been paid to the often
of the researcher’s impressions recorded as fieldnotes, complex working relations between nurses and clinicians
observations or interview data – sometimes handwritten, (Fitzgerald, 2008) as well as to the working ‘cultures’ of
but increasingly captured with the help of recording less visible cadres of health staff, such as ‘peons’ (Justice,
devices. Perhaps because of the tensions involved in 1986).
being a participant-observer, as well as the open
Third, a number of classical ethnographies have focused
approach to what constitutes ‘legitimate knowledge’ on organizations. Stemming from the work of a feminist
(Savage, 2000:1401), ethnography has raised more sociologist (Smith, 1987), such studies aim to examine how
concerns than any other form of social research regar- work activities shape and maintain the institution, analy-
ding the problem of ‘representation’, i.e. the way in sing the ideological procedures that make these work
which researchers choose to depict the ‘reality’ of processes accountable and exploring how work pro-
people’s lives and give voice to their subjects. cesses are connected to other social processes. Here, the
Classical ethnographic approaches are rare in applied ethnographic lens allows a nuanced analysis of organi-
health research not only because of the constraints on zational culture and dynamics, a means of identifying, for
time, and practical feasibility, but also because they do example, how “ the organization’s formal structure
not resonate with the positivist framing of most health- (its rules and decision-making hierarchies) are influ-
related study designs. However, the various genres of enced by an informal system created by individuals or
‘traditional’ ethnography that have been conducted by groups within the organization” (Savage, 2000:1402).
medical anthropologists and sociologists offer important Examples include hospital ethnographies (for example,
insights for understanding health policy and systems Van der Geest & Finkler, 2004) and project ethnographies
issues. (for example, Evans & Lambert, 2008) that examine the

contexts within which policies formulated at a national Health policy and systems researchers can benefit from
or international level play out in the context of local reading classical ethnographies to better understand the
institutional codes of practice. theoretical framing, social, political and historical contexts
of policy formulation and critical assessments of how
Finally, ethnographies have also focused on controver-
policies translate in local health systems. At the same
sies or debates in order to bring to light the tensions
between rhetoric and practice in health systems relation- time, an ethnographic approach can be used in time-
ships. Taylor, for example, one of the first to undertake an limited studies to allow for a more in-depth, rich, and
ethnography of a health system, uses a controversy over nuanced analysis of the relationships between power,
resource allocation in a Scottish archipelago to make knowledge, and practice in health systems – and how
visible the ways in which “ different groups formulate the introduction of changes (in the form of interventions
and pursue their interests both within and outside of the and initiatives) may “ generate different and often
formal structure of the local health care system” (Taylor, unexpected results in different circumstances helping
1977:583). to identify system dynamics and their key outcomes,
which may not be apparent at the outset.” (Huby et al.
Although there are very few extensive ethnographies of
2007:193). An ethnographic lens is, therefore, useful in
biomedical practice and health systems in low- and
studies seeking to explore and explain health policy and
middle-income (LMIC) settings, anthropologists have
systems experiences.
explored ‘biomedical’ or ‘Western’ health care ideology
and practices within discussions of medical pluralism, for
example in South Asia (see, for example, Leslie & Young Rigour in adopting
1992). Classic ethnographies have also examined the an ethnographic lens
ideas of the ‘great’ traditions of institutionalized non-
allopathic medical traditions (for example Ayurveda and Three key methodological characteristics underpin the
Chinese medicine) or the realm of ‘traditional healing’, as rigour of the ethnographic lens as applied within HPSR
opposed to everyday ‘practiced medicine’ (Khare, 1996). studies. First, such studies adopt methods that are open-
Important insights regarding the historical and structural ended, in-depth and flexible in order to capture multiple
bases of Western medical policies and systems and dimensions of how things work (or don’t work) in ‘real
health care organization in LMIC settings can be gleaned time’ and with privileged attention given to the pers-
from ethnographies of colonial medical systems (for pective and experiences of those being interviewed
example see Allen, 2002). Additionally, there are a or studied. Some researchers specifically triangulate
limited number of ethnographies on global health methods to improve validity, but also to explore diverse
policies as introduced in local contexts. An excellent early perspectives in the data. Second, their analysis is inter-
example is provided by Judith Justice’s (1986) ethno- pretive, seeking to situate the meaning of particular
graphy on international health bureaucracy in Nepal that health policy and systems ‘practices’ in social, political
examines the context of policy-making and imple- and/or historical context. Third, to address the challenge
mentation for an initiative known as the Integrated of representation, they adopt a reflexive position vis-a-vis
Community Health Programme. Whiteford & Mander- their areas of inquiry, that is they explain how their own
son’s edited volume (2000) also provides a good range position as researchers and participant-observers (in
of rich case studies of the gaps between the world of some cases) help to shape their areas of interest, the
global health policy-making and local implementation questions they posed and their interpretive lens.
within specific social, political and health systems
contexts.
Allen DR (2002). Managing motherhood, managing risk:
The selected papers provide examples of work conducted
fertility and danger in est Central an ania. Ann Arbor,
by social researchers who have adopted ethnographic
University of Michigan Press.
approaches and methods in their work on policy-making,
Evans C, Lambert H (2008). Implementing community disease control programmes, ‘routine’ health systems
interventions for HIV prevention: Insights from project practices and provider dynamics in low-income settings.
ethnography. Social Science Medicine, 66(2):467–478.
Aitken (1994) examines the implementation of
n
Fitzgerald R (2008). Rural Nurse Specialists: Clinical provider training activities in Nepal and shows how
Practice and the Politics of Care. Medical nthropology: the values providers demonstrate in their daily
actions (values in use) shape their engagement with
Cross-Cultural Studies in Health and llness, 27(3):257–
these activities and undermine the performance
282.
improvements that they are expected to achieve.
Huby G et al. (2007). Addressing the complexity of health Behague & Storeng (2008) examine global policy
n
care: the practical potential of ethnography. ournal of debates around vertical and horizontal approaches to
Health Services Research and Policy, 12:193–194. maternal health care provision and evidence-based
policy-making, teasing out the underlying episte-
Justice J (1986). Policies, plans, and people: culture and miological positions and relevance for policy and
health development in epal. Berkeley, University of advocacy.
California Press.
George (2009) examines routine human resource
n
Khare RS (1996). Dava, Daktar, and Dua: anthropology management and accountability practices in Koppal
of practiced medicine in India. Social Science Medicine state, India, showing how a complex web of social
43(5):837–848. and political relations among different actors in
primary health care influences local understandings
Leslie C, Young A (1992). Paths to sian medical know- and channels of accountability.
ledge. Berkeley, University of California Press. Lewin & Green (2009) explore two sets of common
n
Nguyen V-K (2008). Antiretroviral Globalism, Biopolitics, rituals in South African primary health care clinics –
and Therapeutic Citizenship. In: Ong A, Collier SJ, eds. Directly Observed Therapy for tuberculosis and
morning prayers – in both of which nurses and
lobal assemblages: technology, governmentality, ethics.
patients participate, showing how these different
Oxford, Blackwell Publishing. rituals serve to reinforce traditional power relation-
Savage J (2000). Ethnography and health care. ritish ships between providers and patients.
Medical ournal, 321(7273):1400–1402.
Smith DE (1987). he everyday world as problematic:

a feminist sociology. Toronto, University of Toronto Press.
Taylor R (1977). The local health system: An ethnography

of interest-groups and decision-making. Social Science
Medicine, 11(11-13):583–592.
Van der Geest S, Finkler K (2004). Hospital ethnography:

an introduction. Social Science Medicine 59(10):
1995–2001.
Whiteford LM, Manderson L (2000). lobal health policy,

local realities: the fallacy of the level playing field.
Boulder, Colorado, Lynne Rienner Publishers.

Aitken JM (1994). Voices from the inside: Managing
district health services in Nepal. nternational ournal of
Health Planning and Management, 9(4):309–340.
http://dx.doi.org/10.1002/hpm.4740090405
n
Behague DT, Storeng KT (2008). Collapsing the vertical-

horizontal divide: an ethnographic study of evidence-
based policymaking in maternal health. merican ournal
of Public Health 98(4):644–649.
http://dx.doi.org/10.2105/AJPH.2007.123117
Reproduced by permission of the American Journal
n
of Public Health. Copyright American Journal of
Public Health, 2008.
George A (2009). By papers and pens, you can only do so

much: views about accountability and human resource
management from Indian government health admini-
strators and workers. nternational ournal of Health
Planning and Management. 24(3):205–224.
n
Lewin S, Green J (2009). Ritual and the organisation of

care in primary care clinics in Cape Town, South Africa.
Social Science Medicine, 68:1464–1471.
n
Elsevier, 2009.
 FRAMING HEALTH MATTERS 
Collapsing the Vertical–Horizontal Divide: An Ethnographic

Study of Evidence-Based Policymaking in Maternal Health
| Dominique P. Béhague, PhD, and Katerini T. Storeng, MSc
field has emerged as a coherent and recog-

Using the international maternal health field as a case study, we draw on
nizable network of specialist actors, technolo-
ethnographic research to investigate how public health researchers and policy ex-
gies, and ideas.29,30
perts are responding to tensions between vertical and horizontal approaches to
health improvement. Despite nominal support for an integrative health system ap- We triangulated 3 methods—open-ended,
proach, we found that competition for funds and international recognition pushes in-depth interviews; participant observation;
professionals toward vertical initiatives. We also highlight how research practices and review of published and gray literature
contribute to the dominance of vertical strategies and limit the success of evidence- documents—to improve the validity of find-
based policymaking for strengthening health systems. Rather than support disease- ings and explore diverse perspectives.31
and subfield-specific advocacy, the public health community urgently needs to en- We interviewed a total of 67 professionals
gage in open dialogue regarding the international, academic, and donor-driven (Table 1), identified opportunistically through
forces that drive professionals toward an exclusive interest in vertical programs. professional networking, publications, and
(Am J Public Health. 2008;98:644–649. doi:10.2105/AJPH.2007.123117)
conference proceedings. Of 67 informants,
19 were from developing countries. Many
informants had experience working in multi-
Debates about vertical versus horizontal ap- measurable outcome, is unsuitable for investi- ple domains of public health.
proaches to health improvement have a gating the population-level, nonclinical, and Interviews followed a semistructured guide
lengthy unresolved history in public health.1 context-specific health system domains.9–12 In to explore definitions of evidence and evi-
Vertical approaches are generally disease response, authors have begun calling for non- dence-based policymaking, professional expe-
specific and promote targeted clinical inter- experimental epidemiological methods and riences with production and use of evidence
ventions delivered by a specialized service. an interdisciplinary approach.13–17 for policymaking, historical shifts in policy,
Horizontal approaches, by contrast, tackle Professionals in the international maternal debates around integration and health sys-
several interrelated health issues by strength- health subfield are currently grappling with tems strengthening, and relationships with
ening health systems and developing inte- how to improve vertical and horizontal syn- donors. Using an inductive process, we modi-
grated delivery systems.1–3 ergy.18–20 This has lead to lively debate on fied the interview guide to reflect concerns
Despite prolonged efforts to combine verti- the role of evidence production. Whereas that emerged during data collection.31
cal and horizontal approaches, vertical pro- some call for the scientific rigor of random- We conducted participant observation
grams have dominated and are often found ized controlled trails,21–24 others claim that within academic settings, as well as at 15 re-
competing with one another for funds and using randomized controlled trials is mis- search meetings, academic conferences, and
professional recognition.1,4,5 Authors have placed because of the complex health systems policy meetings. Of these, 8 were not specific
warned that disproportionately concentrating needs of maternal health interventions.25–28 to maternal health but focused on general
funds into disease-based initiatives in develop- Using the subfield of maternal health as a public health or child, neonatal, or reproduc-
ing countries may compromise health systems case study, we explore 2 questions: What are tive health. All informants were interviewed
and fragment complex interventions.6,7 An in- the main challenges faced in bringing vertical formally once, although participant observa-
creasingly popular compromise approach is and horizontal approaches together? What tion enabled repeated contact with many of
to devote general health resources to a lim- are the social and epistemological factors that the informants. Because key players in the
ited package of interventions prioritized on constrain researchers from producing evi- field often represent their views via publica-
the basis of cost-effectiveness. This approach dence for synergistic vertical and horizontal tions, we used published and gray literature
has thus far failed to enable synergy between policymaking? as forms of ethnographic data.
vertical and horizontal approaches.1 We transcribed interviews verbatim and
One reason for continuing tensions relates METHODS stored and organized them in NVivo7 version
to the difficulty of producing evidence of ef- 7.0 (QSR International, Cambridge, Massachu-
fectiveness for evidence-based policymaking Our research focused on debate at the in- setts). Both authors read the interviews and
in a wide array of social, political, and health ternational level because of our interest in notes taken during participant observation and
system contexts.8 Critics claim experimental developments that are critical to the field’s carried out thematic analysis independently.
research, originally developed to assess the overarching reputation and identity. Concep- No significant discrepancies in identified
effect of targeted clinical interventions on a tually, we explored how the maternal health themes and analytic conclusions were found.
644 | Framing Health Matters | Peer Reviewed | Béhague and Storeng American Journal of Public Health | April 2008, Vol 98, No. 4
239
TABLE 1—International Interview was the pressure to support vertical approaches By contrast, a minority of informants felt
Participants: October 2004–June 2007 because of an intense sense of competition that the “attention-seeking strategies of verti-
between subfields. As one informant stated: cal initiatives such as [Global Alliance for
No. of Vaccines and Immunisation and] Roll Back
Participants The maternal health field really competes Malaria” were disempowering because they
against other fields for money. And other
International academic researchers 19 alienated subfields from one another and
fields, like the big spenders—malaria, HIV/
UN agency representatives 10 AIDS, even child health—have a better record fragmented initiatives within each subfield.
Donor body representatives 8 of promoting evidence-based interventions. Ma- These informants explained that maternal
ternal health might be at risk of being left be-
International NGO representatives 16 health experts have attempted to bolster the
hind, because if you miss the target too often,
National-level researchers 8 with traditional birth attendant training, then field’s reputation by searching for a single tar-
National-level policy experts and 6 risk screening, you create donor fatigue. geted vertical intervention, or “magic bullet,”
program managers that would appear to be globally applicable
Total 67 Anxieties around how donors view the and feasible to donors and governments.
relative importance of health problems were The search for a single intervention was
Note. UN = United Nations; NGO = nongovernmental
paramount. “All fields have that anxiety,” said not only reductionistic, some argued, but con-
organization.
one policy expert. “Maternal health had its tributed to infighting and the constant shifting
heyday, and newborn health is now having of proposed vertical interventions, from train-
its heyday. They’re all scared they won’t get ing traditional birth attendants to antenatal
RESULTS the attention and money they had before.” care to emergency obstetric care, each vying
Several informants claimed that integration, for policy attention. Such dynamics resulted
Horizontal Versus Vertical Programmatic although theoretically sensible, would in actu- in the splintering of what could be a compre-
Approaches ality divert funds and policy attention from hensive community and facility-based health
Over the past 20 years, the maternal maternal health. As one academic stated, “I systems approach into specific targeted sub-
health field has undergone 2 significant con- think the jury is out on whether [the fields components, or, as one policy expert described,
ceptual shifts, first toward and then away will integrate] or whether one will get sucked isolated “bits of the jigsaw puzzle.” These in-
from vertical approaches. In 1987, the into the other’s agenda and get lost.” The lack formants claimed that the search for new,
launch of the Safe Motherhood Initiative of funds for strengthening comprehensive targeted vertical solutions ironically had the
aimed to separate maternal health from health systems added to the view that strate- opposite effect than originally intended. Rather
child health to highlight the much neglected gies being promoted in related subfields such than boost the field’s reputation, the picture
issue of maternal mortality.32 Although the as child and neonatal health counter those that emerged to donors and governments was
field of maternal health is still considered by needed in maternal health. that of an uncoordinated and divisive group.
some to be weak, this initiative has suc- In general, the greater the sense of compe-
ceeded in rallying support for maternal mor- tition and threat, the more liable informants The Role of Researchers and the
tality and in garnering support for vertical were not only to reject integration but also Limitations of Current Evidence-Based
interventions, such as antenatal risk screen- to endorse the view that a focused vertical Models
ing, training traditional birth attendants, and approach is more effective in capturing the Polarization of academic researchers and policy
providing emergency obstetric care.19,33 attention of funders and policymakers. experts. Our second research question examined
Maternal health specialists have also in- Informants demonstrated academic support the factors that constrain researchers from pro-
creasingly recognized that vertical interven- for this position by making reference to policy ducing evidence that enables synergistic vertical
tions cannot be delivered without a function- studies (in particular, Schiffman37) that have and horizontal policymaking. Our results indi-
ing health system. By definition, this implies highlighted the importance of maternal cate that researchers were hindered by a detri-
integration of vertical interventions used health-specific “focusing events” and “political mental polarization that positions the academic
within maternal health and greater collabora- champions.” Making comparisons with global community in stark opposition to a group we
tion with other subfields.14,20,27,33 Support for programs such as Integrated Management of termed policy experts. In broad terms, this
this position has resulted in widespread inter- Childhood Illnesses, these informants high- group includes professionals from UN agencies,
est in coordinating initiatives, reflected most lighted the need to establish a simple and uni- international nongovernmental organizations,
recently in the merging of 3 separate partner- fying set of policies that is easy to market to and developing country governments.
ships into the Partnership for Maternal, Neo- politicians and donors. One such policy ex- Our informants’ attention to this polarization
natal, and Child Health.13,34–36 pert argued that maternal health’s “very sad reinforced opposing views regarding the rela-
In interviews, we asked informants to re- history” could be attributed to “a failure of tive importance of advocacy and program de-
flect on the implications of this history for strategy” and that “the question [now] is velopment versus research for ensuring the
improving vertical–horizontal synergy. The whether this constituency can get its act to- field’s survival. In general, researchers felt pol-
most prominent issue informants mentioned gether and push more effectively.” icy experts were more deeply involved in the
April 2008, Vol 98, No. 4 | American Journal of Public Health Béhague and Storeng | Peer Reviewed | Framing Health Matters | 645
24 0
process of advocating for political and financial At the same time that informants put forth als. If you want to deviate from this rule you
investment in maternal health. Researchers re- such dichotomizing statements, several re- should have very good reasons.
luctantly accepted the need for such advocates, searchers were well aware that tensions be-
even if what they espoused was empirically tween research, advocacy, and policymaking It was only after prompting informants to
unfounded. One informant claimed, needed to be assuaged for the sake of the describe specific instances involving the use
field’s professional coherence and future and interpretation of evidence for policy that
There would not be a penny of funding if peo- success. In response, some researchers explic- issues relating to the limitations of experimen-
ple listened to me. . . . I’m too negative. Some itly devoted considerable attention to what tal research emerged. Core to these discus-
people are good spokespersons for Safe Moth- sions were the logistical, ethical, and analytic
erhood. [They] will stand up and say things; they termed “advocacy research,” such as
they know there is no data behind it, but they estimating the global magnitude of maternal difficulties of conducting effectiveness re-
will keep saying it. And it gets the work done. health problems compared with other dis- search on horizontal approaches. As one in-
eases. Researchers highlighted the political formant explained,
Other researchers went further, claiming importance of this work, even if some
Designing a study for skilled attendance at de-
that the field’s failures relate directly to an claimed this type of research does not answer livery is [very difficult] because how the hell
insufficient “evidence-based approach” that analytic questions relating to programmatic do you do a trial of a midwife versus no mid-
wife or a midwife versus a traditional birth at-
was partially caused by advocates’ “militant” development and evaluation.
tendant? It becomes a very difficult medical
style. As one interviewee noted, Policy experts and researchers are clearly and organizational dilemma. Do you get
in a mutually interdependent, if tumultuous, women to deliver at home and women to de-
When people became aware of the M in MCH liver at hospital?
relationship. When asked to reflect critically
[Maternal and Child Health], the field was
on this relationship, informants often made
dominated by people on a mission, and while Informants interested in health systems
it is good to have such people, because they reference to the rapidly expanding body of
questions (e.g., budget support and human re-
are the ones who attract attention and bring literature on communication problems be-
money, if it is too exclusive, you will miss the source strengthening) expressed frustration at
tween academic researchers and policy ex-
scientific rigor and skepticism. the scientific method’s inability to adequately
perts.38 Indeed, several respondents felt that
research these topics. As one informant argued,
these difficulties were at the core of failed
In contrast to this critique, policy experts
effectiveness for evidence-based policymak- It’s really hard to measure the impact, you
frequently held researchers responsible for
ing and argued for improved communica- know, what are you measuring? And the line
paralyzing action and political will by empha- of attribution [from budget support] down to
tion channels, more effectively disseminat-
sizing the scientific uncertainty of the current improvements in maternal health outcomes is
ing new evidence, and capacity building for also difficult.
evidence base. These informants claimed that
each respective group.
research often directly contradicted policy ex-
Diverting attention from questions of episte- According to many informants, resistance
perts’ need to persuade donors of the impor-
mology. Although important to elucidate, the to the Partnership for Maternal, Neonatal, and
tance of maternal mortality and suitability of
intensive focus on improving communication Child Health’s promotion of the continuum of
a particular programmatic approach. As one
diverted our informants’ attention from en- care approach is based largely on difficulties
informant highlighted,
gaging with epistemological questions relating relating to affect evaluation:
to evidence-based health system policymak-
The big challenge is that there’s uncertainty no If you want to say the continuum of care is the
matter what. And policymakers have to deal
ing. Despite growing debates regarding the
answer, how do we validate and monitor that?
with uncertainty. When it’s uncertain, the ret- limitations of current epidemiological meth- How do we say it was proven to work, what
rospectoscope is going to prove that you were ods for health systems questions, few inform- are the outcomes, how many lives are saved?
wrong in your efforts to be certain. Policymak-
ers can’t sit on the fence. Researchers can.
ants spontaneously engaged in discussions
about research models. Rather, several re- Despite such frank discussions, most in-
peatedly espoused the superiority of the ran- formants rarely questioned their own episte-
Some policy experts even claimed that in-
domized controlled trials design for providing mological positions or ventured into new meth-
vesting resources in effectiveness research
definitive proof of the causal relationship be- odological and disciplinary arenas. Rather, they
would undermine the field by diverting atten-
tween intervention and outcome, irrespective modified their research questions—specifically,
tion and funds from much-needed programs.
of the type of intervention being evaluated. the types of interventions being tested and
As one senior policy expert described,
With the randomized controlled trial, said one the units of analysis used—to suit an experi-
statistician, “you don’t need to understand mental or quasi-experimental design. Most
This field has been so contentious because
there hasn’t been enough money. If [only] how the interventions work” to establish its often this meant avoiding questions relating
there had been money to do both research relative advantage. Another claimed that to health systems strategies and focusing on
and [develop] programs in the way that child vertical clinical interventions, such as the ef-
health has had money. . . . This contentious- no design can [control confounding] as the ran-
ness causes donors to turn around and run in domized controlled trials. One should probably fect of calcium supplementation or oxytocin
the opposite direction, so it’s a vicious cycle. always aim at doing randomized controlled tri- administration. These informants explained
241
that clinical research will always be relevant relative cost-effectiveness of different subcom- As this comment suggests, informants
to policy and that such research allows them ponents to help governments in developing sometimes felt that reducing the focus of the
to carve out their own area of expertise and countries with resource allocation. research question to conduct randomized
publish successfully. A less explicit reason for informants’ reti- controlled trials was scientifically unnecessary
Other informants more committed to study- cence to deviate from experimental designs but politically and professionally indispensa-
ing health systems issues attempted to over- relates to the field’s low status and to the is- ble. At the same time, informants also frus-
come the limitations of experimental study de- sues of competition reviewed in the “Results” tratingly acknowledged that this approach
signs by testing only a single subcomponent of section. Referencing a recent publication,23 a reinforced the dominance of vertical ap-
a larger health systems package. Examples in- number of informants claimed that because proaches and compromised a health systems
clude the effect of road construction or intro- the lack of an evidence-based approach in approach. As one researcher aptly summa-
ducing mobile phones and ambulances on maternal health has compromised the field’s rized, the scientific attempt to discern if a par-
health utilization rates. As one informant ex- standing, only the highest research standard ticular community- or facility-based strategy
plained, conclusively evaluating complex multi- should now be accepted. Contextual, observa- has a greater effect on mortality over another
component interventions is such a challenge tional epidemiology, and multidisciplinary re- “is just trappings, and feels like a waste of
that “people are avoiding those kinds of studies search were not viewed as proper academic time and money. . . . I wouldn’t say one is bet-
and instead proposing studies like ‘what if we research and were often relegated to the less ter than the other, I would say if you neglect
put an ambulance in the villages? Will that do scientific realm of operations research. As one the nuts and bolts of the system, you risk get-
it?’” However, as another informant aptly sum- international policy expert described, “Health ting nothing done.”
marized, the practical implications of using the systems research can’t really ever tell us much,
randomized controlled trials for multicompo- other than at a highly contextualized level.” DISCUSSION
nent interventions are tremendously complex: One researcher stated that only those in well-
established subfields who are “starting from Many policy experts support the agenda
To do a [sic] good randomized controlled trials,
the top” can afford to take on the profession- to integrate subfields and wish to work to-
you have to ask a very narrow question. There
isn’t enough money in the world to answer all ally risky activity of pushing the limits of epi- ward health systems strengthening. In prac-
the questions with randomized controlled tri- demiological theory and methods. Maternal tice, however, the competitive playing field
als. So people say, ‘we’ll put three things to-
health, by contrast, is starting from the bottom pressures policy experts to support subfield-
gether that we think work and then we’ll test
that against no change.’ But it’s highly unlikely and, therefore, needs more-rigorous experi- specific initiatives and funding in an effort to
that all of [the components] are equally cost mental studies to be able to provide conclu- bolster the field and advocate for resources
effective or that you need all to be synergistic.
sive recommendations and secure its status. and political will. These findings indicate
You could take a few and get the same
amount of change. . . . Your hypothesis could Other informants were more critical of this that a distinction exists between what can
be that it’s any one or the combination of fac- position, stating that the scientific community’s be termed policy-relevant approaches and
tors or even some synergy about using certain
insistence on using randomized controlled tri- advocacy-sensitive approaches. The former
ones together. To test all those combinations is
impossible! als has created a dogmatic and detrimental respond to policymaking and program imple-
donor demand for experimental evidence. As mentation needs, be they vertical or horizon-
Reasons for the predominant research focus a couple researchers stated, the indiscriminate tal. The latter, by contrast, are used to advo-
on vertical interventions. The normative power use of the randomized controlled trials often cate for the survival and status of the
of scientific values surely persuades research- provides very rigorous answers to irrelevant maternal health professional community and
ers to abide by experimental designs. How- questions. However, being bold and diverting tend to be vertical approaches. Under the
ever, informants highlighted other important from experimental designs means opening pressures of subfield competition, our results
reasons for the predominant focus on experi- oneself up to criticism and potentially losing show that key experts are being pushed to-
mental studies of vertical interventions. Re- publications, funds and political credibility. As ward advocacy-sensitive practices, and be-
sults from studies that clearly demonstrate another epidemiologist stated, cause they are more vertical by nature, this is
the effectiveness of a single specific subcom- I am so convinced of the argument. . . . But happening at the expense of practices that
ponent were said to generate consensus, to be what makes policymakers shift? Do we need could more adequately respond to synergistic
easier to disseminate to policy experts, and to another beautiful trial showing that traditional vertical–horizontal policymaking.
birth attendants make no difference? I hope not.
have more straightforward applications in pol- It’s not whether in the perfect circumstances Maternal health researchers, in turn, re-
icy development. Vertical studies were also al- you can train traditional birth attendants and spond to the pressures for financial support
legedly preferred by donors, who demanded supervise them. Of course that can make a dif- and professional prestige by aiming to pro-
ference. But then you’re talking about an expen-
to see a return on their investments by en- sive system; you might as well train skilled produce evidence that is politically expedient,
couraging governments to implement policies viders. . . . Quite a few people are calling for useful for securing their academic reputa-
for which both intervention and outcome trials of community health workers . . . and the tions, and able to ensure the survival of the
donors are taking note. If we’ve gone that far
could easily be monitored. Informants felt . . . what a waste of money. Maybe we have to maternal health community. For many, this
mounting pressure to use evidence about the play the game; I don’t know. means the use of experimental research to
242
evaluate either clinically targeted interven- ensuing from this body of research remain Contributors
tions or vertical subcomponents of larger theoretical and difficult to operationalize. D. P. Béhague led development of the project proposal
for funding and wrote the article. K. T. Storeng con-
health systems packages. These dynamics tributed to developing the project proposal for funding
impede researchers from following the lead CONCLUSIONS and commented on several drafts of the article. Both
of recent literature13–17,25–28 that scrutinizes authors collected and analyzed the data.
the suitability of an experimental clinical re- On the basis of our findings, we suggest
search model for questions relating to com- modifying evidence-based policymaking prac- Acknowledgments
Funding for this research was received from the Eco-
plex health systems interventions. As a re- tices in 2 main ways. First, it is important to nomic and Social Research Council (Small Research
sult, the production of useful evidence for create institutional environments that actively Grant RES-000-22-1039). K. T. Storeng received sup-
horizontal policymaking, as well as for vertical– port from the Research Council of Norway during the
promote the development of new research
preparation of this article.
horizontal synergy, is sorely lacking. models for investigating complex and context- We would like to thank, first and foremost, all our
The issues of rivalries over funding, diverse specific interventions. As we and other au- informants for being such forthright and helpful partici-
donor-driven agendas, and what informants pants in this research. Their willingness to discuss sensi-
thors have shown, context-specific health sys-
tive problems and tensions emerging from within the
describe as the “false and damaging” di- tems research contradicts the need in public field of maternal health should not be taken as an ob-
chotomies between maternal and child health, health for a generalizable and marketable jective indicator of the field’s shortcomings compared
as well as between community versus facility- with other fields. Rather, our informants’ open and
evidence-base of vertical programs that are
frank discussions indicate high levels of critical aware-
based interventions, have received consider- easy to evaluate and show a measurable im- ness and self-reflection and an ability to engage with
able attention in the literature.19,39–41 The re- pact on outcomes.10,50,51 A major challenge the difficult questions that many, if not most, experts in
cent Lancet series19 on maternal survival had diverse public health subfields currently face.
for public health lies in prioritizing context-
Note. The Economic and Social Research Council
as one of its main aims to “provide an oppor- specific horizontal initiatives even where im- had no involvement in determining the study design,
tunity to mark a shift [away] from unhelpful pact cannot be as precisely shown as in the the collection, analysis, and interpretation of data, or in
writing this article.
dichotomies that slow action in countries case of vertical interventions.52,53 This is partic-
[and] stifle funding.”4(p9) Given the results of ularly the case in developing countries, where
our analysis, we must question whether such vertical initiatives to reduce mortality quickly Human Participant Protection
Ethics approval was granted by the London School of
high-profile statements will have the desired are vital and, yet, where progress in general Hygiene and Tropical Medicine ethics board.
effect of joining diverse factions if they do not development requires active intersectoral col-
(or cannot) address the factors that drive ver- laboration and wide-ranging social initiatives.54 References
tically oriented programs and research. Second, it is important to examine the 1. Mills A. Mass campaigns versus general health
services: what have we learnt in 40 years about vertical
Policy researchers have argued that signif- larger international, donor-driven, political,
versus horizontal approaches? Bull World Health Organ.
icant obstacles to a coherent policy agenda and academic factors that persuade policy 2005;83:315–316.
on vertical–horizontal synergy include weak experts and researchers to adhere to current 2. Hill P. Between intent and achievement in sector-
health systems, current priority-setting normative models of vertical programming wide approaches: staking a claim for reproductive
health. Reprod Health Matters. 2002;10:29–37.
mechanisms based on uncritical support and evidence production. Otherwise, competi-
3. Cueto M. The ORIGINS of Primary Health Care
for traditional disease ranking and cost- tion will continue to encourage subfield-
and SELECTIVE Primary Health Care. Am J Public
effectiveness measures, and uncoordinated specific advocacy, give preeminence to verti- Health. 2004;94:1864–1874.
and conflicting donor agendas on which cal and subcomponent interventions, and 4. Walt G, Lush L, Ogden J. International organiza-
many developing countries are reliant.42–45 push researchers toward the uncritical appli- tions in transfer of infectious diseases: iterative loops of
adoption, adaptation, and marketing. Governance. 2004;
In addition to these obstacles is the crucial cation of experimental methods. Continuation
17:189–210.
issue of how to improve evidence-based of the status quo may also lead to the margin-
5. Lawn J, Tinker A, Munjanja S, Cousens S. Where
policymaking practices. alization of one of epidemiology’s primary is maternal and child health now? Lancet. 2006;368
The literature suggests that poor communi- contributions to public health: that of identify- (9546):1474–1477.
cation between researchers and policymakers ing interrelated determinants of disease pat- 6. Travis P, Bennet S, Haines, A. Overcoming health
systems constraints to achieve the Millennium Develop-
is a key constraint to improving evidence- terns and mechanisms of change.
ment Goals. Lancet. 2004;364(9437):900–906.
based policymaking.46,47 Our informants have
7. Daniels N. Toward ethical review of health system
clearly been influenced by this literature, yet transformations. Am J Public Health. 2006;96:447–451.
About the Authors
our findings suggest that the limited ability of At the time of the study, Dominique P. Béhague and 8. Waters E, Doyle J. Evidence-based public health
experimental methods to provide evidence Katerini T. Storeng were with the Department of Epidemi- practice: improving the quality and quantity of the
about integration and complex health systems ology and Population Sciences, London School of Hygiene evidence. J Public Health. 2002;24:227–229.
and Tropical Medicine, London, England. 9. Black N. Evidence-based policy: proceed with
is a more important impediment. This is Requests for reprints should be sent to Dominique P. care. BMJ. 2001;323(7307):275–279.
being increasingly recognized in the public Béhague, Department of Epidemiology and Population
Sciences, London School of Hygiene and Tropical Medicine, 10. Klein R. From evidence-based medicine to
health and sociological literature.17,48,49 As we evidence-based policy? J Health Serv Res Policy. 2000;
Keppel Street, London WC1E 7HT, United Kingdom
have shown, when it comes to actually chang- (email: dominique.behague@lshtm.ac.uk). 5:65–66.
ing evidence-based practices, the messages This article was accepted October 27, 2007. 11. Victora C, Habicht J, Bryce J. Evidence-based public
243
health: moving beyond randomized trials. Am J Public 30. Law J, Hassard J. Actor Network Theory and After. 49. McQueen D. Strengthening the evidence base for
Health. 2004;94:400–405. Oxford, England: Blackwell Publishing; 1999. health promotion. Health Promot Internat. 2001;16:
12. Oliveira-Cruz V, Kurowsku C, Mills A. Delivery of 261–268.
31. Bernard HR. Research Methods in Anthropology.
health interventions: searching for synergies within the London, England: AltaMira Press; 1994. 50. Van Gool K, Lancsar E, Viney R, Hall J, Haywood P.
vertical versus horizontal debate. J Int Dev. 2003;15: Diagnosis and prognosis of Australia’s health informa-
32. Rosenfield A, Maine D. Maternal mortality: a ne-
67–86. tion for evidence-based policy. J Health Serv Res Policy.
glected tragedy. Where is the M in MCH? Lancet. 1985;
13. Task Force on Health Systems Research. Informed 2(8446):83–85. 2002;7(suppl 1):S40–S45.
choices for attaining the Millennium Development 51. Sackett D, Rosengerg W, Muir Gray J, et al.
33. Goodburn E, Campbell O. Reducing maternal
Goals: towards an international cooperative agenda for Evidence-based medicine: what it is and what it isn’t.
mortality in the developing world: sector-wide ap-
health systems research. Lancet. 2004;364(9438): BMJ. 1996;312:71–72.
proaches may be the key. BMJ. 2001;322:917–920.
997–1003.
34. World Health Organization. The Partnership for 52. Atun R, Lennox-Chhugani N, Drobniewski F,
14. Task Force on Child Health and Maternal Health.
Maternal, Newborn, and Child Health Web site. Avail- Samyshkin Y, Coker R. A framework and toolkit for
Who’s Got the Power? Transforming Health Systems for
able at: http://www.who.int/pmnch/en. Accessed Janu- capturing the communicable disease programmes
Women and Children. London, England: UN Millennium
ary 8, 2006. within health systems: tuberculosis control as an
Project; 2005.
illustrative example. Eur J Public Health. 2004;14:
15. Speller V, Learmout A, Harrison D. The search 35. Costello A, Osrin D. The case for a new Global
267–273.
for evidence of effective health promotion. BMJ. 1997; Fund for maternal, neonatal and child survival. Lancet.
2005;366(9485):603–605. 53. Hanson S. Health sector reform and STD/AIDS
315:361–363.
control in resource poor settings: the case of Tanzania.
16. McPake B. Recognizing patterns: health systems 36. Victora C, Hanson K, Bryce J, Vaughan J. Achieving
Int J Health Plan Manage. 2000;15:341–360.
research beyond controlled trials. Paper presented at: universal coverage with health interventions. Lancet.
Meeting of the London School of Hygiene and Tropical 2004;364(9444):1541–1548. 54. McMichael C, Waters E, Volmink J. Evidence-
Medicine Health Systems Discussion Group; November 9, 37. Shiffman J. Generating political will for safe moth- based public health: what does it offer developing
2006; London, England. erhood in Indonesia. Soc Sci Med. 2003;56:1197–1207. countries? J Public Health. 2005;27:215–221.
17. Tang K, Ehsani J, McQueen D. Evidence-based 38. Dobrow MJ, Goel V, Upshur RE. Evidence-based
health promotion: recollections, reflections, and recon- health policy: context and utilisation. Soc Sci Med.
siderations. J Epidemiol Community Health. 2003;57: 2004;58:207–217.
841–843.
39. AbouZahr C. Cautious champions: international
18. AbouZahr C. Safe motherhood: a brief history of agency efforts to get safe motherhood onto the agenda.
the global movement 1947–2002. BMJ. 2003;67: Stud Health Serv Organ Policy. 2001;17:387–414.
13–25.
40. Maine D, Rosenfield A. The Safe Motherhood
19. Starrs AM. Safe Motherhood Initiative: twenty Initiative: why has it stalled? Am J Public Health. 1999;
years and counting. Lancet. 2006;368(9542): 89:480–482.
1130–1132.
41. Storeng K, Béhague D, Campbell OMR. An an-
20. Graham W, Hussein J. Minding the gaps: a re- thropological exploration of the use of historical
assessment of the challenges to safe motherhood. Am J constructions by key players in the maternal health
Public Health. 2007;97:978–983. field. Paper presented at: the Wellcome Centre His-
21. Luck M. Safe motherhood intervention studies in tory in Public Health: Workshop on the interna-
Africa: a review. East Afr Med J. 2000;77:599–607. tional history of maternal health; May 10, 2006;
London, England.
22. Gulmezoglu AM. Promoting standards for quality
of maternal health care. BMJ. 2003;67:73–83. 42. Oliff M, Mayaud P, Brugha R, Semakafu A. Inte-
23. Miller S, Sloan N, Winikoff B, Langer A, Fikree F. grating reproductive health services in a reforming
Where is the “E” in MCH? The need for an evidence- health sector: the case of Tanzania. Reprod Health
based approach in safe motherhood. J Midwifery Wom- Matters. 2003;11:37–48.
ens Health. 2003;48:10–18. 43. Mayhew S, Walt G, Lush L, Cleland J. Donor
24. Ross J, Simkhada L, Smith W. Evaluating effective- agencies’ involvement in reproductive health: saying
ness of complex interventions aimed at reducing mater- one thing and doing another? Int J Health Serv. 2005;
nal mortality in developing countries. J Public Health 35:579–601.
Policy. 2005;27:331–337. 44. Walt G. What sort of international cooperation in
25. Bullough J, Meda N, Makowiecka K, Ronsmans C, health 2055? J R Soc Med. 2005;98:235–237.
Achadi E, Hussein J. Current strategies for the reduc- 45. Sachs J. Health in the developing world: achieving
tion of maternal mortality. BJOG. 2005;112(9): the millennium development goals. Bull World Health
1180–1188. Organ. 2004;82:947–949.
26. Hotchkiss D, Eckert E, Macintyre K. The role of 46. Horby P, Perera H. A development framework for
health services research in the Safe Motherhood Initia- promoting evidence-based policy action: drawing on
tive. Am J Public Health. 2000;90:810–811. experiences in Sri Lanka. Int J Health Plan Manage.
27. Campbell O, Koblinsky M, Taylor P. Off to a rapid 2002;17:165–183.
start: appraising maternal mortality and services. Int J 47. Denis JL, Lehoux P, Hivon M, Champagne F.
Gynecol Obstet. 1995;48(suppl):S33–S52. Creating a new articulation between research and prac-
28. Milne L, Scotland G, Tagiyeva-Milne N, Hussein J. tice through policy? The views and experiences of re-
Safe motherhood program evaluation: theory and prac- searchers and practitioners. J Health Serv Res Policy.
tice. J Midwifery Womens Health. 2004;49:283–292. 2003;8(suppl 2):44–50.
29. Latour B. Reassembling the Social: An Introduction 48. Boaz A, Pawson R. The perilous road from evi-
to Actor-Network-Theory. Oxford, England: Oxford Uni- dence to policy: five journeys compared. J Soc Policy.
versity Press; 2005. 2005;34(2):175–194.
24 4

Ritual and the organisation of care in primary care clinics in Cape Town,
South Africaq
Simon Lewin*, Judith Green
Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom
Article history: Few sociological studies have examined care organisation in primary health settings in low- and middle-
Available online 9 March 2009 income countries. This paper explores the organisation of health care work in primary care clinics in Cape
Town, South Africa, by analysing two elements of clinic organisation as rituals. The first is a formal,
Keywords: policy-driven element of care: directly observed therapy for tuberculosis patients. The second is an
South Africa informal ritual, seemingly separate from the clinical work of the team: morning prayers in the clinic. We
Tuberculosis
draw on data from an ethnography in which seven clinics providing care to people with tuberculosis
Ritual
Organisation of care
were theoretically sampled for study. These data include participant observation of clinic sessions, and
Treatment adherence interviews and group discussions with providers and patients, which were analysed using approaches
drawn from grounded theory. Our findings suggest that rather than seeing the ritualised aspects of clinic
activities as merely traditional elements of care that potentially interfere with the application of good
practice, it is essential to understand their symbolic values if their contribution to health care organi-
sation is to be recognised. While both staff and patients participate in these rituals, these performances
do not demonstrate or facilitate cohesion across these groups but rather embody the conflicting values of
patients and staff in these clinics. As such, rituals act to reinforce asymmetrical relations of power
between different constituencies, and to strengthen conventional modes of provider–patient interaction.
Introduction development and implementation of DOT, and the wider TB control

strategy in which it is embedded, have been described extensively
This paper concerns the organisation of work in primary health elsewhere (Ogden, Walt, & Lush, 2003; Raviglione & Pio, 2002;
clinics in Cape Town, South Africa that deliver care for tuberculosis Volmink, Matchaba, & Garner, 2000; Walt, 1999). Supporters of DOT
(TB). It explores the usefulness of considering two rather different have argued that it is required to ‘protect’ the limited set of TB drugs
elements of organisation – Directly Observed Treatment (DOT) for from the growth of drug resistance,1 and that it needs to be seen as
TB patients and morning prayers in the clinic – as rituals. DOT is the just one component of a larger TB control strategy (Ogden et al.,
practice recommended by the WHO in which TB treatment taking is 2003). By exploring the symbolic content of DOT and morning
observed for the full duration of treatment – usually six months or prayers in the clinic, we aim to contribute to an explanation of
more – by a health care provider or someone nominated by the seemingly ‘non-rational’ behaviours in health care work.
patient and the provider to take on this role (WHO, 2002). The In studying care organisation in this setting, we were struck by
reports from clinic nurses that TB patients, on hearing that their six
months of DOT was complete, sometimes asked whether they
could continue attending the clinic until the week ended. That
q We thank the patients and health care providers who participated in the study;
patients would want to prolong what, for many, was a burdensome
the health service managers who provided support for this work; and the research
assistants on the project. Karen Daniels, Judy Dick, Simon Goudie, Karina Kielmann,
Jessica Ogden, Liz Rose, Carolyn Stephens, Hester van der Walt, Nomsa Yekiso and
Merrick Zwarenstein provided insightful comments and advice. The Common-
1
wealth Programme of the Nuffield Foundation (Project CW28); the UK Department The use of a multi-drug treatment regime helps to ensure that if the tubercu-
for International Development (Project R561); the Health Systems Trust (Project losis bacillus becomes resistant to one of the drugs, it will be eliminated by one of
226/98); and the Medical Research Council of South Africa provided funding. A Phil the other drugs, thereby helping to reduce the overall development of resistance.
Strong Memorial Prize from the British Sociological Association facilitated addi- Resistance is problematic because of the relatively limited number of effective TB
tional fieldwork by SL. drugs; the difficulties, in terms of length of treatment, mode of treatment delivery
* Corresponding author. Tel.: þ47 46400403. (injection), and costs of treating multi-drug resistant cases; and the danger of
E-mail address: simon.lewin@nokc.no (S. Lewin). multi-drug resistant strains spreading in the community.
0277-9536/$ – see front matter ! 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2009.02.013
245
S. Lewin, J. Green / Social Science & Medicine 68 (2009) 1464–1471 1465
daily clinic visit to receive treatment suggested that DOT had 1981). Ritual therefore entrenches, through performance, cate-
become an integral part of their routine. Yet its completion – gories created within biomedicine, such as ‘sick’ and ‘well’,
a significant event – went apparently unmarked. Staff would record ‘adherent’ and ‘non-adherent’.
the patient’s completion of treatment into the TB register but, for In nursing, one functionalist argument draws on psychoanalytic
the patient, there was no ritual marker of their reintegration into theory to identify the functions of ritual for individual health care
the social body (Van Gennep, 1960). Health care settings are replete workers, proposing that it is through unconscious defence mech-
with ritual, from the organisation of surgery in the operating anisms that individuals deal with painful or difficult feelings, such
theatre (Katz, 1981), to ward rounds (Strange, 1996), patient as fear or loathing, that may harm the self (Lupton, 1997). This
clerking and the traditional return of a patient’s ‘normal’ clothes to perspective suggests that providers may experience difficult, even
mark the end of an inpatient stay. Yet, at a point when some ritual conflictual, feelings as a result of patients’ emotional expectations
might be expected, these patients were left seeking a natural end and direct contact with patients’ bodies (Menzies-Lyth, 1988;
point, the weekend, to mark their new status as ‘healthy’. Obholzer & Roberts, 1994; Skogstad, 1997; van der Walt & Swartz,
The puzzle of this ‘missing ritual’ raised a more general question 2000), arousing deep anxieties that may be too difficult to
of the functions served by the unusual procedure of DOT for both TB consciously examine (such as helplessness in the face of inability to
patients and health care providers. Why are TB patients in particular cure). Psychoanalytic approaches go on to note that ritualistic
treated in this way? Although DOT is now commonplace within TB defensive techniques on both individual and collective levels may
programmes, there are clearly other ways in which treatment protect against these anxieties (Chapman, 1983; Skogstad, 1997).
delivery could be organised. Following from McCreery’s study of More sociologically, rituals in nursing work can be seen as
meaning in therapeutic ritual, we address two key questions: ‘‘What having social functions. Turner’s definition of rituals as ‘‘dramas of
are the possible meanings of this [ritual] work? What is the audience social events which emphasize the importance of the event they
to which it is addressed and the situation to which it speaks?’’ symbolize or represent’’ (Turner, 1969, p. 59) emphasises rituals as
(McCreery, 1979 p. 70) Because the care of TB patients includes a set performances that enact and institutionalise culturally constructed
of highly standardised and detailed procedures, some of which are of categories. Thus, in health care, ritual practice is not only used as
unproven efficacy (Volmink & Garner, 2007), this care presents an a defence against anxiety, but also for social effect, creating and
interesting opportunity to examine the role of ritual in the reflecting cultural values regarding the treatment of the sick
management of a common infectious disease. To illuminate the (Chapman, 1983). Rituals are essential to healing itself, especially in
possible meanings of ritualised activity in this context, we also draw terms of reintegrating the ‘sick’ person into the ‘healthy’ social
on data on a more obvious ritual in the clinic – that of morning body. For example, the discharge of a patient from hospital involves
prayers. We suggest that examining the symbolic meanings of these returning their civilian clothes, indicating that they may rejoin the
two contrasting work practices contributes to understanding the world of the ‘healthy’. Ritual may also be used to maintain
ways in which care is achieved (or not) in formal health settings. boundaries between states, such as dangerous or safe, sterile and
non-sterile. This reduces uncertainty and increases the autonomy
Accounting for rituals in health care of actors by indicating clearly which states are operative at any
particular time (Katz, 1981, p. 336).
The term ‘ritual’ has been used in multiple ways (Douglas, 1996; Much work on the social role of rituals assumes that they act to
Katz, 1981; McCreery, 1979; Turner, 1969), and the growing body of unite a homogenous group, with all those participating sharing
literature on the role of ritual or ritualised practice in nursing work values and meanings, as expressed in the enactment and symbols
(Chapman, 1983; Holland, 1993; Strange, 1996; Wolf, 1988) draws of the ritual (Baumann, 1992). For Leach (1976), for instance, the key
on a range of theoretical starting points. First, the term ‘ritual’ has aspect of ritual is that there is no separation between performer
been used atheoretically by some commentators, to merely differ- and audience. Such assumptions of homogeneity are problematic in
entiate those practices that have a good ‘evidence base’ from those modern health care organisations, in which different constituencies
that do not, characterising the latter as ‘traditional’ practices, or (of staff groups, of patients) may not subscribe to the same set of
‘rituals’. Such ‘rituals’ are cited as reducing the effectiveness of meanings. More recent work on ritual has highlighted these
nursing care (Walsh & Ford, 1989). Thomson, for instance, notes: potential conflicts. Drawing on the work of Durkheim, Baumann,
‘‘Ritualistic practices have long stood in the way of effective infec- for instance, argues that rituals may be ‘‘performed by competing
tion control’’ (Thomson, 1990, cited in Strange (1996, p. 106)). constituencies’’ (Baumann, 1992, p. 99) with different relationships
Within this perspective, a ritual has no meaning, being merely an to the performance, symbols and meaning of the rituals. Rather
obstacle to greater efficiency rather than a theoretical tool for than being limited to ‘‘insiders’’, participants in rituals in plural
understanding nursing work. societies may include a range of outsiders with these different
Others have drawn on social science literature to explore the parties ‘‘each using symbolic forms to stake mutual claims’’ (p. 101)
meanings of ritual. From a functionalist perspective, ritual has been through the enactment of the ritual. Ritual, Baumann suggests, is
seen as serving: psychological, social and protective functions; the therefore a platform for defining and negotiating relationships with
identification of values and rules; and the negotiation of power others. This paper takes this approach as a starting point, to explore
(Bell, 1992; Helman, 2000; Strange, 1996). From Van Gennep (1960 how ritualised practices in primary care clinics may embody and
[1909]) onwards, there has been a particular interest in rituals of entrench power relations, being potentially functional for some
transition, and their functions in helping to ameliorate and control constituencies while being dysfunctional for others.
danger and anxiety related to changes of state or to a lack of clarity
in classifying a category or state. This has been of particular interest Methods and setting
in health care, with a focus on how health providers, in their day-
to-day work, cope with uncertainties of diagnosis and management This study formed part of a larger ethnographic study of the
and how patients manage the transition between illness and impact of clinic organisation on professional responses to change in
wellness (Helman, 2000). Rituals provide boundaries to categories primary health care clinics in Cape Town (Lewin, 2004). The setting
in the context of transition, for example, between being ‘well’ and was urban and peri-urban municipal primary health clinics within
being diagnosed with TB, thus allowing social actors, such as health the Cape Town metropolitan area that deliver care to TB patients.
care providers, family and friends, to respond appropriately (Katz, The size, patient load and staff complement of these clinics ranged
246
1466 S. Lewin, J. Green / Social Science & Medicine 68 (2009) 1464–1471
from small ‘satellite’ clinics to very large clinics seeing over one was asked to consent to participation and clinic staff gave written
hundred clients per day. All employed a mix of nursing cadres as or verbal consent for all individual and group interviews.
well as additional nursing assistants, administrative staff and
cleaners. The larger clinics may have a health educator and Ritual in the primary health clinic
sessional doctors, social workers and pharmacists.
A ‘theoretical sampling’ approach was used to select clinics for The DOT ritual: its functions within the clinic
study (Glaser & Strauss, 1967), taking into account issues such as
staff mix, patient load and the ethnic group served by the clinics. International TB policy suggests that the swallowing of all 130
Four clinics were studied in depth (in this paper called Tortoise doses of TB drugs by the patient, in the case of new patients, should
Lane, Old Oak, Hilltop and New Township), while three others were be supervised either by a health worker or by a community or
studied in less detail (Windy Plains, Steady Village and Border’s family member or work colleague nominated by the patient and
Edge). New Township, Hilltop and Windy Plains were large, rela- the health provider for this purpose. For retreatment patients, the
tively new clinics serving extensive ‘black’ townships on the number of doses is higher as treatment is continued for eight
periphery of Cape Town and largely staffed by ‘African’ Xhosa- months. That TB patients, along with substance users and people
speaking nurses.2 Old Oak was a large, well-established clinic suffering from leprosy, have been singled out for the ‘supervised
serving a historically ‘coloured’ or ‘mixed-race’ residential area and consumption’ of medication is unlikely to be coincidental. All are
largely staffed by ‘mixed-race’ nurses. In contrast, Tortoise Lane was stigmatised groups, with little power within the health system or
based in a busy business district, drew clients from a wide range of society at large (Cross, 2006; Munro et al., 2007; Room, 2005).
areas and was staffed by nurses with a wide range of socio-cultural With small local variations in procedure, the key features of DOT
backgrounds. Both Border’s Edge and Steady Village were smaller remained constant across all the clinics studied. Hilltop was typical:
clinics serving the rapidly developing townships of low cost
At 8.40am the staff begin dispensing DOT. There are about eight
housing further from the city centre and staffed by ‘mixed-race’
patients waiting, including one child and one woman. One of the
nurses. For the most part, the clients of all these clinics were poor
patients looks very thin and weak. Most of the [TB] patients look
and many were unemployed.
very poor – even poorer than those in the clinic as a whole. All the
Data were collected over two years, from 1999 to 2001, by the
patients come into the DOT/TB room and sit on chairs against the
first author (SL) and two research assistants. Fieldwork included
wall, facing the desk where one of the nurses is seated.the
around 120 h of participant observation of clinic sessions; formal
nurse.counts out the pills for each patient on DOT, which they
individual in-depth interviews (n ¼ 21) and group discussions
swallow in her presence, having helped themselves to water from
(n ¼ 8) with staff; and informal interviews with staff. All formal
the tap. [Clinic Observation – Hilltop]
interviews and group discussions were taped, transcribed and
translated from Xhosa into English where appropriate. A number of At the start of treatment, the patient and nurse decided where
informal interviews with patients were undertaken alongside clinic the patient should best receive DOT. The majority of patients opted
observations and a large number of provider–client interactions for clinic-based supervision. The procedure for daily clinic-based
were observed. Data were generated using an iterative and induc- DOT was as follows: referring to the patient’s treatment card, the
tive approach, with analysis of the initial data collected informing nurse counted out the daily batch of pills for each patient into a small
later data collection strategies. cup labelled with the patient’s name. Most patients on clinic-based
Analysis was guided by elements from the grounded theory DOT visited the clinic just after it opened in the morning. They went
approach (Strauss & Corbin, 1990). Field notes and interviews straight to the treatment room where the cups of pills were waiting.
were examined as one body of data and indexed into categories. A member of the clinic staff was assigned to observe the pills being
‘Open’ coding resulted in a re-ordering of data as thematic links swallowed. She usually greeted the patients and gave the patient her
between sections of data became evident. Interpretive hypotheses cup of pills. The patient then swallowed the tablets under supervi-
were cross-checked against the data and deviant cases sought. sion and left the clinic. If the observing staff member had time, she
The notion of ritual, as developed in the sociological and spoke briefly to the patient. Often, however, the treatment room was
anthropological literatures, was then used as a conceptual tool to very busy and conversation with the patient was not possible. Direct
link observed micro-level practices with the broader patterns of observation was often conducted by a non-clinical member of staff,
work organisation. So, while the conceptual categories emerged such as the cleaner or a clerk, either because patients arrived before
from the data, they were subsequently framed within relevant the clinical staff (officially) began their shift or because they were
theoretical perspectives and linked to the work of others. For engaged in other tasks. After the patient had taken the pills, the
each key category, data extracts were identified on the basis of attendant recorded this in the patient record.
being representative and/or interesting illustrations of that Patients defaulting from this daily regimen were usually ques-
category. tioned by the professional clinic staff on their next visit. These
Ethical approval for the study was granted by Medical Research discussions often involved the provider chastising the patient for
Council of South Africa and the London School of Hygiene and his or her non-adherence; highlighting the patient’s lack of
Tropical Medicine. Municipal health department managers in Cape responsibility for his or her own health; and demanding better
Town also gave their permission to conduct the study. Each clinic performance. One staff nurse, who had worked at Old Oak clinic for
many years, noted:
I like doing TB, but you can’t do TB for a year. After nine months it
becomes very frustrating. Sometimes you have to give patients
2
Under apartheid, the Population Registration Act classified South Africa’s a scolding because of defaulting.[ ]. Some of the patients really
population as ‘African’, ‘coloured’ (‘mixed race’), ‘Indian’ or ‘white’. These categories
get to you. The same old patients from last year. [staff nurse – Old
are used in this paper as, although the structures of apartheid have been disman-
tled, this classification has had, and continues to have, profound effects on Oak]
socio-economic conditions, the provision of health services and health status. The
use of these categories is not intended in any way to legitimise or support the
Some providers attempted to focus their interactions with patients
notion of genetically distinct race groups. The term ‘black’ is used to refer to those on barriers to treatment completion and how these might be overcome.
designated, under apartheid legislation, as ‘African’, ‘coloured’ or ‘Indian’. However, such exchanges were the exception rather than the rule.
247
Symptoms of TB rapidly improve during the first two to three creates conflict for providers between the need to support patient
weeks of therapy and patients usually report feeling much better choice, which may include requests to self-supervise treatment,
after one month. However, in order to be effective, treatment needs and the need to apply rigorously treatment policies. The DOT
to be continued for the full six month period. The management of strategy, as applied in this setting, does not have the flexibility to
TB therefore parallels that of many chronic illnesses, such as adapt to patient needs as these evolve over the course of treatment.
hypertension, in that patients are expected to continue therapy The same ritual is rigidly maintained, sometimes creating frustra-
despite feeling physically well. The rationale for DOT is that it tion and anger in both patients and staff. The value of DOT here is
ensures patients complete their course of therapy, maximising the therefore primarily symbolic – it indicates to patients, providers
chance of cure and reducing the risk of drug resistance. DOT is also and society at large that the disease is still present and that the
portrayed, on paper, as an opportunity for the health worker to patient is not yet ready to re-enter the world of the well.
provide support to the patient (WHO, 2002). Patient support and If DOT is in part a ritual to sustain the sick role, one might expect
treatment completion are important goals, but considerable it to include an element to indicate the end of illness and the
research evidence suggests that DOT does not fulfil these aims reintegration of the patient into society. Interestingly, this was not
(Ogden, 1999; Volmink & Garner, 2007). If not clinical effectiveness, found. For providers, treatment completion is marked by entering
what, then, is the function of DOT within the primary health clinic? the words ‘cured’ or ‘treatment completed’ into the official TB
Is it meaningful at the symbolic level, and if so, for which of its Register. For patients, treatment is merely stopped from one day to
different constituencies? the next. The absence of any form of closure suggests a flawed
DOT is clearly ‘ritualised activity’, in that there are formalised ritual, which fails to give meaning to the illness or to shape the
arrangements of objects, people, bodies and spaces which ‘‘trigger boundaries between disease and cure.
the perception that these practices are distinct and the associations The second set of symbolic meanings of DOT relates to the ways
that they engender are special’’ (Bell, 1992, p. 220). DOT involves it functions to construct the patient as a passive, non-trustworthy
specific and specialised ‘personnel’ – the health care provider and recipient of care in need of monitoring and control. Health care
the TB patient – both of whom are assigned specific roles and follow providers had ambivalent feelings towards TB patients. On the one
a largely predefined series of steps or tasks during their interactions. hand, they espoused a need for more ‘patient centred’ care, even if
The interactions often involve several patients and take place within this was difficult to enact within the clinic:
a designated space to which public access is restricted. The inter-
.I try to be at the same level as the patient and not talk down to
action also involves a range of objects used largely in this activity
them. Then it’s a pleasure for them to come to the clinic. [profes-
alone, such as the cups of pills labelled with each patient’s name and
sional nurse – Old Oak]
the TB treatment supervision card. We suggest, however, that DOT is
more than ritualised activity. It derives its significance as ‘ritual’ from If there was time to sit down with the patient and ask them, ‘How
its symbolic meanings, evident in the ways it produces and repro- do you feel?’, it would strengthen the nurse–patient relationship.
duces relations of power between patients and providers: first, But unfortunately there is no time for this. [nurse – New Township]
through the medicalisation of the illness to create a sustainable ‘sick On the other hand, providers also viewed patients as child-like
role’, and second, through constructing the patient as a passive, non- and irresponsible, in need of constant surveillance:
trustworthy recipient of care in need of monitoring and control.
Having received a diagnosis of TB, the patient is cast into the sick There’s something about the TB patient – I don’t really trust them.
role, with a responsibility to follow the prescribed treatment of the We watch them swallow their tablets. If not, they can put it in their
health provider and to endeavour to become ‘well’ (Parsons, 1951). pockets. [staff nurse – Old Oak]
In the early stages of treatment, ongoing symptoms reinforce the Must have all eyes on the patient during the TB session. That to me
provider’s message that this is a serious ailment requiring careful is very important. [senior professional nurse – Old Oak]
treatment and supervision. As the patient’s health improves,
however, it becomes less clear that the sick role is appropriate. The Beyond the desire to monitor patients, all cadres of nurses
illness is resolved, and she may now wish to shed the sick role and expressed concern about the problem of ‘control’. This was an
resume a normal life. However, this is not possible as patients are immediate issue for them in terms of controlling the physical
expected to return daily to the clinic for treatment for the full six bodies of patients within the clinic:
months if they are to be cured. This ambiguous period of transition .there were lots of patients [in the clinic] and I am new. I told
between resolution of illness and cure, in which the patient no them finally that they must sit down and I will call them. They were
longer feels ill but still harbours TB bacilli, is dangerous for the all crowding around. You don’t want to come across too aggres-
health care system in that patients may fail to attend the clinic for sively, but you want to show that you are in control. You have to
daily treatment. Indeed, providers’ frequently expressed consider- make them understand that there are boundaries and they have to
able frustration and anger towards patients who did not comply: respect those. I realised that this is my clinic and this has to be safe
to me. [professional nurse – Old Oak]
I’m very much on the defaulter patients. I hate those who default
[laughs]. I would [makes a guillotine noise] them. You have to push It also referred to the more general problem of control of bodies
them. You do become frustrated and moody.[staff nurse – Old Oak within the community:
Clinic]
.some patients should be admitted [to hospital] to ensure
You have to run after [patients] and make sure they take their compliance, but they refuse. As a health worker, you don’t have any
medication. [nurse – Tortoise Lane] control over that. [nurse – Tortoise Lane]
DOT, by labelling people as ill, helps to delineate and maintain Although professing to provide patient support, DOT is, then,
the ambiguous and dangerous boundary between illness, as largely orientated towards control of patients and their bodies, and
perceived by patients, and ‘disease’, relating to biomedical cure. It is the suppression of resistance to the treatment regimen. This desire
a collective attempt by health providers to extend the sick role for control at the clinic level mirrors in many ways the public health
beyond the period of symptomatic illness. The resistance of policies of ‘TB control’ at management level. Others have suggested
patients to inhabiting a ‘sick role’ beyond the symptom period that, in the case of TB care, the desire to control the spread of the TB
248
bacillus has extended to a desire to control the person in whom the However, the ways in which these connections play out in the day-
bacillus is resident (Ogden, 2000; van der Walt, 1998). The more to-day provision of care is little described or analysed from
stringent the control of the patient, the more likely control of the a sociological perspective.
bacillus will be achieved. DOT sanctions the surveillance and Although not a formal part of health care provision, and at first
disciplining of patients who, through ‘ignorance’ and ‘poverty’, sight more inclusive than DOT, prayers too enacted and facilitated
have contracted TB. They are belittled by a ritual that emphasises power relations within the clinic. At Hilltop clinic, staff led patients
their dependency and disempowerment and the need for them to in prayer every morning before starting clinical work:
accept the moral authority of health care professionals. Further-
At 8.25am the clinic staff start morning prayers. They stand at the
more, patients who do not adhere to TB therapy are constructed as
front of the clinic waiting area, facing the rows of clients, and lead
being to blame for their own illness. This scapegoating places the
the waiting clients in a hymn, the Lord’s Prayer, and a second hymn.
accountability for treatment failure on the patient rather than on
Everyone in the clinic stands, and almost all of the clients partici-
the disempowering rituals enacted within the primary care system
pate in the singing. [Observation – Hilltop]
or the deeper structural issues such as poverty and migrancy which
have contributed to disease spread and to poor treatment adher- These prayers lasted approximately 10 minutes and had
ence (Munro et al., 2007). DOT therefore juxtaposes professional a solemnity and restfulness which contrasted with the usual hurly-
authority with patient disempowerment, constructing and main- burly of the clinic. Although the clinic prayers could be led by staff or
taining the micro-level power relations within the provider– patients, they were usually initiated by the staff. Nurses indicated
patient interaction in which patients are subordinated and in which that many gatherings in the Xhosa-speaking community routinely
the patient’s body is the centre of both control and resistance. began with prayers, including meetings at schools and clinics.
Studies from other spheres of health care delivery in South Africa From the staff perspective, prayers served a number of explicit
have identified similar difficulties in relationships between nurses functions. One of the nurses noted that the morning prayers helped
and patients (Jewkes, Abrahams, & Mvo, 1998; Kim & Motsei, 2002), the staff to relax before they started work, so that they could face
perhaps suggesting that poor nurse attitudes to patients extend the day. She also suggested that the prayers drew patients and staff
beyond care for people with TB. together and reinforced the idea that everyone in the clinic was
Patients, of course, do not necessarily subscribe to norms of human and should be treated as such. Another nurse explained that
obedience to medical authority embodied in DOT, seeing instead prayers allowed patients to ‘‘open up’’ to staff as, ‘‘If I [the patient]
the ritual as inappropriate or humiliating (van der Walt et al., 1999). prayed with so-and-so, they can’t be that bad’’ [nurse trainer –7/6/
Turner suggests that, ‘‘By exposing their ill-feeling in a ritual 01]. In the view of some staff, the prayers, as a shared activity,
context to beneficial ritual forces, individuals are purged of rebel- facilitated communication with their clients in a setting in which
lious wishes and emotions and willingly conform once more to the there were few opportunities to release emotion and establish
public mores’’ (Turner, 1970, p. 49–50). However, rather than being rapport (Dick & Pekeur, 1995; Jewkes et al., 1998). Prayers were,
purged by the ritual of ‘dangerous’ emotions related to their non- then, a form of emotional labour (James, 1992) in that they provided
adherent behaviour in this case patients simply resist by not for the sharing and management of feelings between providers and
returning to the clinic. In effect, in terms of adherence, the ritual patients. During this process the clinic staff acted as ‘priests’, albeit
may have the opposite effect from that originally intended. DOT, as unofficially and temporarily. Although their roles as clinicians and
ritual, therefore has potentially conflicting symbolic meanings to managers were temporarily suspended, the hierarchy of provider
the different constituencies who participate in it. The performance (priest)–patient (congregant) was maintained.
of DOT is functional for health care providers and the health care On a symbolic level, however, prayers enacted some key
system as, at the micro-level, it allows control to be asserted over tensions between different ritual constituencies, through the
patients and the medicalisation of illness to be sustained. At the apparent breaching, but in a limited and controlled way, of three
macro-level, it allows providers to adhere to international policies key boundaries. The first was the separation between the clinic –
for TB control. In this, the DOT ritual may speak as much to the local and the cultural ‘world’ of biomedicine – and the broader
constituency – health care providers and managers – as to the more community in which the clinic is located. The presence of prayer
distant constituency of national and international policy makers. introduced elements of the community’s world into the medical
On the other hand, the ritual is dysfunctional for patients in that it world and thereby acted to emphasise the clinic’s location within
does not sufficiently incorporate their values and beliefs, or address a particular socio-cultural setting and to provide a bridge between
their need for empathy and support nor their likely perception of these two worlds. By participating in the prayers, both patients and
their own progress in being ‘cured’. The failure of DOT, as a ritual, to providers exposed and shared a part of their private (non-medical)
engage patients in a way that allows them to manage their tran- selves. Patients were also allowed to ‘see’ an aspect of providers
sition to ‘wellness’, instead of potentially prompting their resis- which they would otherwise not usually be able to observe.
tance to treatment, may be one of the major failings of the current However, this breaching was limited and controlled, in that the
system of TB care. ritual took place in separate time, before the official start of clinical
work. As others have noted in this setting, nurses may attempt to
Morning prayers in the clinic: breaching the boundaries? limit their emotional relationships with clients. This may occur
particularly where nurses’ clients are very similar to themselves
A contrasting ritual activity was morning prayers, which were and therefore act to undermine nurses’ feelings of being safe from
observed in clinics in Xhosa-speaking areas. These clinics, largely these diseases of poverty (van der Walt & Swartz, 2000).
staffed by Xhosa-speaking nurses, served clients who mainly The second partial breach is that of the normal relationship
belonged to a range of Christian denominations. Here staff and between professional and client. Prayers in which both patients and
clients therefore shared more cultural resources than was the case staff actively contribute, rather than one being the recipient of the
in the other settings of this study. Morning prayers are a long ministrations of the other, provided a mechanism for displaying
established part of clinic practice in many parts of South Africa. As unity across the typically hierarchical nurse–client relationship in
Marks and others have noted (Marks, 1994; Stein, Lewin, & Fairall, this and other settings. However, this brief display of unity did not
2007; Sweet, 2004), nursing, and biomedicine more broadly, are significantly challenge the underlying power relations in normal
closely associated with Christian belief systems in the country. ‘clinic time’.
249
Thirdly, introducing a spiritual element into daily work confined to becoming well (Parsons, 1951). That the DOT ritual
breached the boundary of ‘scientific’ biomedical practice, indicating usually fails to include any form of closure for patients – an issue
to both staff and patients that care and healing are not simply that they seem to see as important – illustrates its limitations in
rational processes amenable to health care intervention. The assisting patients in coping with their illness and its focus on the
prayers suggested that there were other forces operating to which needs of the health system.
both patients and staff might appeal, in their respective roles as Much of the work on ritual in nursing and health care has
‘healers’ and the ‘sick’. Morning prayers cast the staff into the role of assumed, at least implicitly, a functionalist perspective. Ritual is
healers in a broader sense, both spiritual and physical, emphasising seen as serving particular social or psychological purposes, with all
that they were responsible for the physical and spiritual health of those participating sharing the values and meanings of these
their clients. In some ways, this resembled more traditional forms performances. We argue here that the constituencies to a ritual
of healing which do not routinely separate the physical from the performance do not necessarily share a common relationship to its
spiritual, as is typically the case in Western biomedicine, and act to symbols and meanings. Indeed, in this case the DOT ritual acts to
both treat illness and reduce the tensions brought on by it. divide health care providers from patients. Unlike providers, who
However, the use of Christian prayer in the clinics also echoed the appear to ‘buy-in’ to the DOT ritual, many patients resist partici-
historical role played by nurses in South Africa in bringing pation. The most obvious form of resistance is a refusal to subject
‘enlightenment’ to what were seen as heathen, diseased commu- their bodies and persons to this form of control. That absenting
nities (Marks, 1994). The use of prayer in conjunction with the themselves from treatment is one of the very few avenues of
delivery of nursing care therefore also reinforced the links between resistance available to patients is unfortunate, given the repercus-
Western biomedicine, and the scientific and moral superiority it sions on individual wellbeing and the ways in which their resis-
claims, and the godliness and order of ‘Christian civilisation’. It acts, tance is seen by providers to confirm their view of patients as
if subtly, to marginalise more traditional health systems. irresponsible and non-compliant. In contrast, participants in the
In the example of prayers there appeared, then, to be more prayer ritual seemed to share a common relationship to its symbols
cohesion and continuity between the different constituencies to and meanings. Here too, however, ritual was used to circumscribe
the ritual – that is, patients and providers – than was the case for the limits to this common bond, separating it from ‘clinical’ rela-
DOT. Prayers spoke to a single congregation of both clients and tionships and underlining the co-option of religious symbolism to
providers, for whom this ritual had shared symbolic content. The medical power. Prayer rituals may therefore act to maintain the
ritual was not generated from within the world of biomedicine, distancing by providers of clients by reinforcing how any bond that
from which non-professionals are largely excluded, except as nurses and their clients share in the wider social world is con-
patients. Rather, it was generated from the world of religion, strained within the world of the clinic.
where patients and professionals could participate, on apparently The public rituals described here are concerned with shaping
equal terms. However, the separation in time between ‘prayers’ relationships with others – in this case providers and patients – who
and clinical work demonstrates the limits of this equality. As form the constituencies to the ritual performance. Their reasons for
ritual, the prayers reproduce relations of power in the clinic rather participating; the meanings that they bring to these rituals; and
than challenge the roles of providers and clients. Prayers also their responses to them reflect the position of their constituency,
embody the tensions experienced by nursing staff, who are both within biomedicine and the wider community in which these
simultaneously apart from and part of the community they serve clinics are situated. There are, of course, other approaches to
(and of course potentially also patients) and thus anxious to understanding the roles of ritual activities and the meanings that
establish boundaries for control. these communicate within health care (Bell, 1997). However, by
considering the meanings that different constituencies bring to
Discussion ritual performances, we show how DOT can be functional for health
care providers, in reinforcing relations of power with patients.
This paper provides insights into two areas that have been Simultaneously, DOT can be dysfunctional for patients, in that it
under-explored – the roles of ritual within TB care, and the more attempts to render them as passive subjects of health care. Rather
general issue of the organisation of nursing work in middle- and than uniting patients and providers in the treatment of disease, DOT
low-income settings. DOT is used worldwide as part of the highlights the different viewpoints of those participating in it.
management strategy for TB, although its form varies from setting There is growing evidence that the asymmetrical power rela-
to setting (Volmink et al., 2000). The use of DOT does not appear, in tions of health care encounters are being challenged by both
Cape Town at least, to achieve its aim of improving adherence to TB consumers and providers (Brown, 1999; Farrell, 2004). In many
treatment and, indeed, research evidence suggests that it is settings health care is shifting towards models of shared decision
unlikely to do so (Munro et al., 2007; Volmink & Garner, 2007). For making and patient centredness (Cline, Granby, & Picton, 2007; van
many health care professionals within the TB programme in South der Weijden, van Veenendaal, & Timmermans, 2007). This is
Africa, the national goal of 85% cure rates may therefore seem reflected to a limited extent within the field of TB care (Macq,
unachievable. Unfeasible goals, it has been suggested, may Torfoss, & Getahun, 2007). In general, however, it remains one of
contribute to work becoming ‘ritualised’, focusing labour on its the few areas of health care where this asymmetry continues to be
physical and measurable aspects rather than on interaction and actively promoted both explicitly in policies and implicitly through
patient care (Coser, 1963). the symbolic content of these policies. This can be contrasted with
While DOT has generally been viewed as a method of ensuring the rollout of highly active anti-retroviral treatment (HAART) in
patient adherence to TB treatment, we suggest that it needs to be South Africa and other low- and middle-income country settings.
seen as both a medical procedure and a ritual. As ritual, DOT has For example, the South African government’s national plan for the
a range of symbolic meanings for patients and staff within the treatment of HIV and AIDS notes that, ‘‘specific education or drug-
world of clinic care, facilitating the medicalisation of TB treatment readiness training is essential to provide the knowledge to enable
and its ongoing control by health care professionals. DOT reinforces individuals to take ownership of their own health’’ (p. 73). It also
traditional modes of interaction in which the patient accepts the suggests that providers need to ‘‘negotiate a treatment plan that the
‘sick role’, and relinquishes responsibility for the management of patient can understand and to which he/she commits’’ (p. 75)
therapy to the health care professional. The patient’s role is (National Department of Health, 2003). Findings from a recent
25 0
study of nurses’ views on HAART implementation in South Africa genuinely embody the wide range of values of the constituencies
appear to reinforce these differences between the TB and HAART involved in health care, foster empowerment and are therapeutic,
programmes, highlighting the strong engagement of nurses with may have to evolve rather than be specifically engineered. More
people living with HIV and AIDS and nurses’ strong desire to be able research in this area, which draws on sociological analyses of work
to offer a patient centred approach to treatment (Stein et al., 2007). organisation, is needed.
Similar views have been described in other studies (Rajaraman &
Palmer, 2008). A number of factors may account for these differ-
ences between the HAART and TB programmes. These include the References
very large number of people who have died from AIDS; the politi- Baumann, G. (1992). Ritual implicates ‘others’: rereading Durkheim in a plural
cisation of the government’s response to the epidemic in South society. In D. de Coppet (Ed.), Understanding rituals. London: Routledge.
Africa; and the fact that most health care providers have experi- Bell, C. (1992). Ritual theory, ritual practice. Oxford: Oxford University Press.
Bell, C. (1997). Ritual: perspectives and dimensions. New York: Oxford University
enced HIV and AIDS in their household or social group, which is not
Press.
the case for TB. Blumenfeld, S., de los Santos, M., Teoxon, M., Cruz, N., & Dizon, F. (1999). Reducing
We would argue that efforts to change professional practice in treatment default among tuberculosis patients in the Philippines. Bethesda,
TB, for example to make it more ‘patient centred’, need to engage Maryland: The Quality Assurance Project, Center for Human Services (CHS).
Available at. http://www.qaproject.org/pubs/PDFs/TBREPORT.PDF.
with the symbolic level of clinic ritualised practices, as well as with Brooks, I. (1996). Using ritual to reduce barriers between sub-cultures. Journal of
the more usual concerns of the clinical and public health evidence Management in Medicine, 10(3), 23–30.
base for practice. Viewing ritual simply as an impediment to Brooks, I., & Brown, R. (2002). The role of ritualistic ceremonial in removing barriers
between subcultures in the National Health Service. Journal of Advanced
evidence based practice in nursing fails to acknowledge these Nursing, 38(4), 341–352.
symbolic functions and their importance in the construction of Brown, S. (1999). Patient-centered communication. Annual Review of Nursing
health and health care. Furthermore, by ignoring ritual in health Research, 17, 85–104.
Chapman, G. (1983). Ritual and rational action in hospitals. Journal of Advanced
care organisations, policy makers may inadvertently doom their Nursing, 8(1), 13–20.
change efforts to failure or, at best, partial success. Their efforts may Cline, W., Granby, T., & Picton, C. (2007). A competency framework for shared deci-
result, as Douglas (1996) has suggested, in one ritual simply being sion-making with patients. Achieving concordance for taking medicines. NPC Plus
and Medicines Partnership Programme. Keele: Keele University.
replaced by another that embodies similar power relations. Coser, R. (1963). Alienation and the social structure. In E. Freidson (Ed.), The hospital
It is not clear how generalisable the findings presented here are in modern society. London: The Free Press of Glencoe.
to TB care in other settings or, indeed, to care for other health Cross, H. (2006). Interventions to address the stigma associated with leprosy:
a perspective on the issues. Psychology, Health and Medicine, 11(3), 367–373.
problems. It seems likely, however, that the findings would be
Dick, J., & Pekeur, P. (1995). An investigation into the workload and workpattern of
widely applicable in South African settings that share a similar clinic nurses at a local authority clinic. CHASA Journal of Comprehensive Health,
organisation of care (Oskowitz, Schneider, & Hlatshwayo, 1997; 6, 4–10.
Petersen, 2000). Further, we acknowledge that patient–provider Douglas, M. (1996). Purity and danger. An analysis of the concepts of pollution and
taboo. London: Routledge.
relations, and the rituals and structures that construct them, are Ezzy, D. (2001). A simulacrum of workplace community: individualism and engi-
only several of many factors impacting on treatment adherence. We neered culture. Sociology, 35(3), 631–650.
do not intend to suggest that addressing the one component dis- Farrell, C. (Ed.). (2004). Patient and public involvement in health: The evidence for
policy implementation. A summary of the results of the health in partnership
cussed here might result in measurable improvements in adher- research programme. London: Department of Health.
ence in TB care. However, turning a sociological lens on the current Glaser, B., & Strauss, A. (1967). The discovery of grounded theory: Strategies for
patterns of care with a view to unpacking the less obvious mean- qualitative research. Chicago: Aldine.
Helman, C. (2000). Culture, health and illness. Oxford: Butterworth-Heinemann.
ings of these organisational behaviours does suggest that inter- Holland, C. K. (1993). An ethnographic study of nursing culture as an exploration for
ventions relating to rituals of closure might be worth considering. determining the existence of a system of ritual. Journal of Advanced Nursing,
We started this paper with the observation that, in the study 18(9), 1461–1470.
James, N. (1992). Care, work and carework: a synthesis? In J. Robinson, A. Gray, &
setting, there was no marker of the reintegration of patients into
R. Elkan (Eds.), Policy issues in nursing Milton Keynes: Open University Press.
the healthy social body on the completion of their TB treatment. Jewkes, R., Abrahams, N., & Mvo, Z. (1998). Why do nurses abuse patients?
Given that ritual may be an important element of healing, is it Reflections from South African obstetric services. Social Science & Medicine,
47(11), 1781–1795.
possible to create rituals within TB care which are more responsive
Katz, P. (1981). Ritual in the operating room. Ethnology, 20, 335–350.
to both patient and staff needs over the course of therapy? Can Kim, J., & Motsei, M. (2002). Women enjoy punishment: attitudes and experiences
rituals be established that create a sense of closure for patients who of gender-based violence among PHC nurses in rural South Africa. Social Science
complete a long course of medical treatment and have to re-adjust & Medicine, 54, 1243–1254.
Leach, D. (1976). Culture and communication: The logic by which symbols are con-
to being well again? This applies not only to TB but also to a wide nected. Cambridge: Cambridge University Press.
range of other chronic illnesses, and might include markers such as Lewin, S. (2004). The organisation of TB nursing work in primary health clinics in Cape
a certificate or a completion ceremony, organised by the health Town, South Africa [PhD]. London: Department of Public Health and Policy,
London School of Hygiene and Tropical Medicine, University of London.
services, for people who complete successfully their course of TB Lupton, D. (1997). Psychoanalytic sociology and the medical encounter: Parsons and
treatment (Blumenfeld, de los Santos, Teoxon, Cruz, & Dizon, 1999), beyond. Sociology of Health and Illness, 19(5), 561–579.
or other acknowledgements of patients’ progress through the McCreery, J. (1979). Potential and effective meaning in therapeutic ritual. Culture,
Medicine and Psychiatry, 3(1), 53–72.
treatment cycle. Such attempts to ‘engineer’ culture through Macq, J., Torfoss, T., & Getahun, H. (2007). Patient empowerment in tuberculosis
creating rituals that speak to disparate constituencies are wide- control: reflecting on past documented experiences. Tropical Medicine & Inter-
spread within the so-called post-Fordist workplace (Ezzy, 2001; national Health, 12(7), 873–885.
Manley, J. E. (1998). Symbol, ritual, and doctrine: the cultural ‘tool kit’ of TQM.
Manley, 1998). Even within health care, it has been suggested that Journal of Quality Management, 3(2), 175–191.
rituals may be manipulated to make care more patient oriented and Marks, S. (1994). Divided sisterhood. Race, class and gender in the South African
flexible (Brooks, 1996; Brooks & Brown, 2002). Our findings could nursing profession. London: The McMillan Press Ltd.
Menzies-Lyth, I. (1988). Containing anxiety in institutions. Selected essays. London:
be read as suggesting that any such interventions would need to
Free Association Books.
have symbolic value to both the provider and patient constituen- Munro, S., Lewin, S., Smith, H., Engel, M., Fretheim, A., & Volmink, J. (2007). Patient
cies and, possibly, to the policy-making constituency. There may, adherence to tuberculosis treatment: a systematic review of qualitative
however, be difficulties with ‘engineering’ rituals that, on the one research. PLoS Medicine, 4(7), e238. doi:10.1371/journal.pmed.0040238.
National Department of Health. (2003). Operational plan for comprehensive HIV and
hand, do not act as a form of normative control and, on the other AIDS care, management and treatment for South Africa. Pretoria, South Africa:
hand, speak symbolically to a wide range of groups. Rituals that National Department of Health.
251
Obholzer, A., & Roberts, V. (1994). The unconscious at work: Individual and organ- Sweet, H. (2004). ‘Wanted: 16 nurses of the better educated type’: provision of
isational stress in the human services. London: Routledge. nurses to South Africa in the late nineteenth and early twentieth centuries.
Ogden, J. (1999). Compliance versus adherence: just a matter of language? The Nursing Inquiry, 11(3), 176–184.
politics and poetics of public health. In J. Porter, & J. Grange (Eds.), Tuberculosis. Turner, V. (1969). The ritual process: Structure and anti-structure. Chicago: Aldine.
An interdisciplinary perspective. London: Imperial College Press. Turner, V. (1970). The forest of symbols. Aspects of Ndembu ritual. Ithaca: Cornell
Ogden, J. (2000). Improving tuberculosis control – social science inputs. University Press.
Transactions of the Royal Society of Tropical Medicine and Hygiene, 94, 135– Van Gennep, A. (1960). The rites of passage. Chicago: Chicago University Press.
140. Volmink, J., & Garner, P. (2007). Directly observed therapy for treating tuberculosis.
Ogden, J., Walt, G., & Lush, L. (2003). The politics of ‘branding’ in policy transfer: the Cochrane Database of Systematic Reviews(4). doi:10.1002/
case of DOTS for tuberculosis control. Social Science & Medicine, 57(1), 179–188. 14651858.CD003343.pub3. Art. No.: CD003343.
Oskowitz, B., Schneider, H., & Hlatshwayo, Z. (1997). Taking care of quality: Volmink, J., Matchaba, P., & Garner, P. (2000). Directly observed therapy and
perspectives of the patient and the providers at an STD clinic. Johannesburg: treatment adherence. Lancet, 355(9212), 1345–1350.
Centre for Health Policy, University of the Witwatersrand. Walsh, M., & Ford, P. (1989). Nursing rituals, research and rational actions. Oxford:
Parsons, T. (1951). The social system. London: Routledge and Kegan Paul. Heinemann Nursing.
Petersen, I. (2000). Comprehensive integrated primary mental health care for South Walt, G. (1999). The politics of tuberculosis: the role of process and power. In
Africa. Pipedream or possibility? Social Science & Medicine, 51(3), 321–334. J. Porter, & J. Grange (Eds.), Tuberculosis. An interdisciplinary perspective. London:
Rajaraman, D., & Palmer, N. (2008). Changing roles and responses of health care Imperial College Press.
workers in HIV treatment and care. Tropical Medicine & International Health, van der Walt, H. (1998). Nurses and their work in tuberculosis control in the Western
13(11), 1357–1363. Cape: too close for comfort [PhD]. Cape Town: Department of Community Health,
Raviglione, M. C., & Pio, A. (2002). Evolution of WHO policies for tuberculosis University of Cape Town.
control, 1948–2001. Lancet, 359(9308), 775–780. van der Walt, H., Schoeman, J., Hoogendoorn, E., Dick, J., Swartz, L., & Dikweni, L.,
Room, R. (2005). Stigma, social inequality and alcohol and drug use. Drug and et al. (1999). Being a TB patient: stories from across the divide. In 30th Inter-
Alcohol Review, 24(2), 143–155. national Union Against Tuberculosis and Lung Disease (IUATLD) World Conference
Skogstad, W. (1997). Working in a world of bodies. Defensive techniques on on Lung Health. Madrid, Spain.
a medical ward – a psychoanalytical observation. Psychoanalytic Psychotherapy, van der Walt, H., & Swartz, L. (2000). Isabel Menzies Lyth revisited: institutional
11(3), 221–241. defences in public health nursing in South Africa during the 1990s. Psychody-
Stein, J., Lewin, S., & Fairall, L. (2007). Hope is the pillar of the universe: health-care namic Counselling, 5(4), 483–495.
providers’ experiences of delivering anti-retroviral therapy in primary health- van der Weijden, T., van Veenendaal, H., & Timmermans, D. (2007). Shared decision-
care clinics in the Free State province of South Africa. Social Science & Medicine, making in the Netherlands – current state and future perspectives. Zeitschrift für
64(4), 954–964. ärztliche Fortbildung und Qualitätssicherung, 101(4), 241–246.
Strange, F. (1996). Handover: an ethnographic study of ritual in nursing practice. WHO. (2002). An expanded DOTS framework for effective tuberculosis control. Geneva,
Intensive and Critical Care Nursing, 12(2), 106–112. Switzerland: World Health Organization.
Strauss, A., & Corbin, J. (1990). The discovery of grounded theory: Strategies for Wolf, A. (1988). Nursing rituals. The Canadian Journal of Nursing Research, 20(3),
qualitative research. Newbury Park: Sage. 56–69.
252
4. Advances in impact without the intervention – known as the ‘counterfactual’
– in order to be able to attribute the observed change to
evaluation the intervention under study. Methodological develop-
ment in this field has focused to a substantial degree on
Kara Hanson different approaches to establishing this counterfactual,
London School of Hygiene and ropical Medicine, and on how best to minimize different forms of selection
nited ingdom of reat ritain and orthern reland bias.
Although there is a rich body of literature on health This body of work also recognizes the importance of
programme evaluation, the work that focuses on system- external validity – the extent to which findings can be
level interventions is smaller. However, recent years have generalized to other settings. This requires understanding
seen a growth of interest in understanding the ‘impact’ the causal mechanism, looking more closely at its causal
of development interventions, including health system pathway and testing the validity of assumptions that are
interventions, in order to guide development practice and made about the route between intervention and impact,
investments using evidence about ‘what works’ and an in order to assess whether those assumptions are likely
understanding of why it works (Evaluation Gap Working to hold in other contexts. It also means paying careful
Group, 2006). New bodies have been established to attention to the implementation setting and how this
promote and finance impact evaluations, such as the mediates the effects of the intervention.
International Initiative for Impact Evaluation (3IE)
Two main types of study design are currently used within
(http://www.3ieimpact.org ), and bilateral donors and
impact evaluations:
other funders have given renewed emphasis to streng-
thening their approaches to evaluation and their capacity n perimental design: This involves a random assign-
ment of the programme to an intervention group and
to use this evidence in their decision-making. At the
a control group, with the effect that potential unobs-
same time, influenced by trends within social pro-
erved confounding factors are also randomly distri-
gramme evaluation in higher-income countries (Harrison, buted between the two groups, minimizing risks of
2001), there is an emerging interest in critical realist bias.
approaches to evaluation (for example see FEMhealth: n uasi-e perimental designs: These can involve
http://www.abdn.ac.uk/femhealth ). Such approaches ‘natural experiments’ which take advantage of a
consider the question: What works for whom in what policy or other change that generates an appropriate
circumstances? All approaches to impact evaluation, thus, control group. Study designs then compare groups or
aim to explain health policy and systems changes and areas with and without the intervention; make
before-and-after comparisons; adopt ‘difference-in-
interventions.
difference’ approaches (before and after with a
control group); or take advantage of a phased
Rigour in impact evaluation implementation that provides variation in the
duration of exposure to the programme. Another
There are different meanings of ‘impact’ in the general approach is to use matching methods (such as
evaluation literature, but in the contemporary literature, propensity score matching) in a cross-sectional
impact is understood to refer to a causal mechanism – design to create a control group that is matched on
as many observable factors as possible.
the change in an outcome that is caused by a particular
programme. This focus on causal mechanisms has meant Health system interventions have some particular
that a lot of attention is paid to methods for arriving at features that influence the choice of evaluation approach.
an unbiased measure of the change that is due to the First, they often work through complex causal pathways
programme or intervention. A starting point to measure and are particularly influenced by features of the policy
such impacts is to consider what would have happened and implementation context. Recent guidance on the

(impact) evaluation of complex public health inter- Another feature of evaluation designs for health system
ventions can also be applied to health system inter- interventions is that it is often difficult to use a ‘control
ventions, and emphasizes the need to: group’ to establish the counterfactual because, for
develop a good theoretical understanding of the
n example, a policy change takes place at national level
change mechanism; (the ‘small n’ (sample size) problem). For instance,
address explicitly the risk of implementation failure by
n changes in regulatory or health financing systems often
including a process evaluation; occur across a whole country at one time so there is no
recognize the higher level influences on individual
n other unit to use as a comparison group.
behaviour, and design studies that take these into
For both reasons – complexity and the need for alter-
account;
native approaches to establish the counterfactual –
adopt multiple measures of outcome, including poten-
n
tial unintended consequences of an intervention; it seems appropriate to recommend that to enhance
their rigour all evaluations of health system interventions
recognize that strict fidelity to a protocol is unlikely,
n
and allow for local adaptation in the intervention should be based on a strong programme theory (White,
model (Craig et al., 2008). 2009).
Writing about interventions from a public health pers- Indeed, theory-based evaluation approaches represent a
pective, Victora, Habicht & Bryce (2004) challenge the third form of study design for impact evaluation. These
primacy of the randomized controlled trial as contri- approaches are based on an explicit programme theory
buting the best evidence for policy-making when causal that sets out the links between inputs, outputs and
pathways are complex. They describe the value of impacts and tests these causal links using a mix of
‘plausibility designs’ in which studies that are non- qualitative and quantitative methods. Realist evaluation,
randomized nonetheless aim at making causal state- meanwhile, focuses attention on the links between
ments using observational designs with a comparison context, mechanisms of change and outcomes, given its
group. This form of causal reasoning can be supported by interest in how the intervention leads to which effects,
evidence that implementation has been adequate, under what circumstances (Pawson & Tilley, 1997). It
demonstrating progress in intermediate steps along the requires that middle range theory, the analysts’ initial
causal pathway, analysing the temporal sequence of ideas about these links is developed prior to, and then
events and using ‘dose-response’ reasoning to link the tested through, the evaluation. Realist evaluation tends
strength of programme implementation to changes in to rely on mixed-methods, with greater use of qualitative
the outcome. de Savigny & Adam (2009) also identify the methods than other impact evaluations, and adopts
need for adaptations to conventional study designs when approaches to generalization which rely more on
evaluating health system interventions, emphasizing the analytic, rather than statistical, generalization. Its rigour
need to measure a wide variety of outcomes (intended or is then safeguarded by the adoption of approaches
unintended) and for a comprehensive analysis of the common in case-study practice (see section on the case
contextual factors that may help to explain the success study approach).
or failure of an intervention.
Craig et al. (2008). eveloping and evaluating comple The papers in this section were chosen because they
interventions: new guidance. Medical Research Council, address system-level interventions and reflect a broad
(http://www.mrc.ac.uk/complexinterventionsguidance, range of approaches to impact evaluation.
accessed 6 September 2011).
Björkman and Svensson (2009) use a randomized
n
de Savigny D, Adam T (2009). Systems thinking for health study design to evaluate the impact of a report-card
approach to improving community accountability. This
system strengthening. Geneva, World Health Organization.
paper was selected because of its focus on a novel
Evaluation Gap Working Group (2006). hen will we health system intervention and its use of an experi-
mental design to measure impact.
ever learn mproving lives through impact evaluation.
Washington DC, Center for Global Development. Macinko et al. (2007) examine a large-scale health
n
system intervention (a national community-based
Harrison S (2001). Policy Analysis. In Fulop et al., eds. primary care programme in Brazil) using a quasi-
Studying the organisation and delivery of health services: experimental design which takes advantage of the
gradual expansion of the programme to generate an
research methods. London, Routledge:90–106.
internal control group to measure impact.
Pawson R, Tilley N (1997). Realistic evaluation. London, Marchal, Dedzo & Kegels (2010) use realist evalu-
n
Sage Publications. ation methods to examine the impact of a particular
human resource management approach within one
Victora C, Habicht JP, Bryce J (2004). Evidence-based hospital in Ghana. It looks at the link between
public health: moving beyond randomized trials. organizational practices and performance, has strong
merican ournal of Public Health, 94(3):400–405. theoretical underpinnings, and uses exclusively
qualitative methods to explore the causal links
White, H (2009). heory-based impact evaluation: between management practice and behaviour within
principles and practice. International Initiative for Impact the organization.
Evaluation Working Paper No. 3 (http://www.3ieimpact.org Wang et al. (2009) look at the impact on health
n
/admin/pdfs_papers/51.pdf, accessed 6 September 2011). status of a community-based health insurance
scheme in China, in which increased financial risk
protection was accompanied by service innovations
including more selective purchasing, changes to
the provider payment mechanism, and changes to
the prescription system. They both adopt a quasi-
experimental approach (before-and-after with a
control group) and employ propensity score matching
to construct a comparison group.

Björkman M, Svensson J (2009). Power to the people:
evidence from a randomized field experiment on
community-based monitoring in Uganda. he uarterly
ournal of conomics, 124(2) :735–769.
http://dx.doi.org/10.1162/qjec.2009.124.2.735
n
Macinko J et al. (2007). Going to scale with community-

based primary care: an analysis of the family health
programme and infant mortality in Brazil. Social Science
Medicine, 65:2070–2080.
n
Elsevier, 2007.
Marchal B, Dedzo M, Kegels G (2010). A realist evalu-

ation of the management of a well-performing regional
hospital in Ghana. MC Health Services Research, 10:24.
http://dx.doi.org/10.1186/1472-6963-10-24
Wang H et al. (2009). The impact of rural mutual health

care on health status: evaluation of a social experiment in
rural China. Health conomics 18(S2):S65–S82
http://dx.doi.org/10.1002/hec.1465
n
POWER TO THE PEOPLE: EVIDENCE FROM A
RANDOMIZED FIELD EXPERIMENT ON
COMMUNITY-BASED MONITORING IN UGANDA∗
MARTINA BJÖRKMAN AND JAKOB SVENSSON
This paper presents a randomized field experiment on community-based mon-

itoring of public primary health care providers in Uganda. Through two rounds of
village meetings, localized nongovernmental organizations encouraged communi-
ties to be more involved with the state of health service provision and strengthened
their capacity to hold their local health providers to account for performance. A
year after the intervention, treatment communities are more involved in moni-
toring the provider, and the health workers appear to exert higher effort to serve
the community. We document large increases in utilization and improved health
outcomes—reduced child mortality and increased child weight—that compare fa-
vorably to some of the more successful community-based intervention trials re-
ported in the medical literature.
I. INTRODUCTION
Approximately eleven million children under five years die
each year and almost half of these deaths occur in sub-Saharan
Africa. More than half of these children will die of diseases (e.g.,
diarrhea, pneumonia, malaria, measles, and neonatal disorders)
that could easily have been prevented or treated if the children
had had access to a small set of proven, inexpensive services
(Black, Morris, and Bryce 2003; Jones et al. 2003).
Why are these services not provided? Anecdotal, and re-
cently more systematic, evidence points to one possible reason—
ineffective systems of monitoring and weak accountability
∗ This project is a collaborative exercise involving many people. Foremost, we

are deeply indebted to Frances Nsonzi and Ritva Reinikka for their contributions
at all stages of the project. We would also like to acknowledge the important
contributions of Gibwa Kajubi, Abel Ojoo, Anthony Wasswa, James Kanyesigye,
Carolyn Winter, Ivo Njosa, Omiat Omongin, Mary Bitekerezo, and the field and
data staff with whom we have worked over the years. We thank the Uganda Min-
istry of Health, Planning Division, the World Bank’s Country Office in Uganda,
and the Social Development Department, the World Bank, for their cooperation.
We are grateful for comments and suggestions by Paul Gertler, Esther Duflo, Ab-
hijit Banerjee, and seminar and conference participants at Stanford, Berkeley,
LSE, Oxford, IGIER, MIT, World Bank, NTNU, Namur, UPF, CEPR/EUDN con-
ference in Paris, and BREAD & CESifo conference in Venice. We also thank three
anonymous referees and the editor, Lawrence Katz, for very constructive sugges-
tions. Financial support from the Bank-Netherlands Partnership Program, the
World Bank Research Committee, the World Bank Africa Region division, and the
Swedish International Development Agency, Department for Research Coopera-
tion is gratefully acknowledged. Björkman also thanks Jan Wallander’s and Tom
Hedelius’ Research Foundation for funding.
C 2009 by the President and Fellows of Harvard College and the Massachusetts Institute of
"
Technology.
The Quarterly Journal of Economics, May 2009
735
257
736 QUARTERLY JOURNAL OF ECONOMICS
relationships.1 This paper focuses on one of these accountability

relationships, citizen-clients’ ability to hold providers accountable,
using primary health care provision in rural Uganda as a testing
ground.
To examine whether community-based monitoring works, we
designed and conducted a randomized field experiment in fifty
communities from nine districts in Uganda. In the experiment, lo-
cal nongovernmental organizations (NGOs) facilitated village and
staff meetings in which members of the communities discussed
baseline information on the status of health service delivery rela-
tive to other providers and the government standard. Community
members were also encouraged to develop a plan identifying key
problems and steps the providers should take to improve health
service provision. The primary objective of the intervention was to
initiate a process of community-based monitoring that was then
up to the community to sustain and lead.
The community-based monitoring project increased the qual-
ity and quantity of primary health care provision. A year after the
first round of meetings, we found a significant difference in the
weight of infants—0.14 z-score increase—and a markedly lower
number of deaths among children under five—33 percent reduc-
tion in under-5 mortality—in the treatment communities. Uti-
lization for general outpatient services was 20 percent higher in
the treatment compared to the control facilities and the overall
effect across a set of utilization measures is large and signifi-
cantly positive. Treatment practices, including immunization of
children, waiting time, examination procedures, and absenteeism,
improved significantly in the treatment communities, thus sug-
gesting that the changes in quality and quantity of health care
provision are due to behavioral changes of the staff. We find evi-
dence that the treatment communities became more engaged and
began to monitor the health unit more extensively. Using varia-
tion in treatment intensity across districts we show that there is
a significant relationship between the degree of community moni-
toring and health utilization and health outcomes, consistent with
the community-based monitoring mechanism.
Community-based, randomized, controlled field trials have
been used extensively in medical research to evaluate the
1. For anecdotal and case study evidence, see World Bank (2003). Chaudhury
et al. (2006) provide evidence on the rates of absenteeism. On misappropriation
of public funds and drugs, see McPake et al. (1999) and Reinikka and Svensson
(2004).
258
COMMUNITY-BASED MONITORING IN UGANDA 737
effectiveness of various health interventions (see footnote 14). Our

paper is related but differs in one important dimension. Whereas
the medical field trials address the question of impact of a biolog-
ical agent or treatment practice when the health workers compe-
tently carry out their tasks, we focus on how to ensure that the
health workers actually carry out their tasks and the impact that
may have on health utilization and health outcomes.
This paper also relates to a small literature on improving gov-
ernance and public service delivery through community partici-
pation. Olken (2007) finds minor effects of an intervention aimed
at increasing community participation in the monitoring of cor-
ruption in Indonesia. Our work differs in several ways. First, the
intervention we evaluate was structured in a way to reduce the
risk of elite capture. Second, unlike corruption, which is not easily
observable, the information discussed in the meetings was basic
facts on utilization and quality of services based on the commu-
nity’s own experience. Finally, the intervention sought to address
two constraints highlighted in the literature on community mon-
itoring: lack of relevant information and inadequate participa-
tion. Banerjee, Deaton, and Duflo (2004) evaluate a project in
Rajasthan in India where a member of the community was paid
to check whether the nurse-midwife assigned to the health center
was present at the center. The intervention had no impact on at-
tendance and the authors speculate that a key reason for this is
that the individual community member did not manage to use his
or her information on absenteeism to invoke community partici-
pation. Here, on the contrary, we explicitly try to address the par-
ticipation constraint by involving a large number of community
members and encouraging them to jointly develop a monitoring
plan.
Finally, the paper links to a growing empirical literature on
the relationship between information dissemination and account-
ability (Besley and Burgess 2002; Strömberg 2004; Ferraz and
Finan 2008). In this paper, however, we focus on mechanisms
through which citizens can make providers, rather than politi-
cians, accountable. Thus, we do not study the design or allocation
of public resources across communities, but rather how these re-
sources are utilized. Second, we use microdata from households
and clinics rather than disaggregated national accounts data. Fi-
nally, we identify impact using an experimental design.
The next section describes the institutional environment.
The community-based monitoring intervention is described in
259
Section III. Section IV lays out the evaluation design and the re-
sults are presented in Section V. Section VI concludes. Details
about the experiment and additional results are reported in the
Online Supplemental Appendix.
II. INSTITUTIONAL SETTING

Uganda, like many newly independent countries in Africa,
had a functioning health care system in the early 1960s. The
1970s and 1980s saw the collapse of government services as the
country underwent political upheaval. Health indicators fell dra-
matically during this period until peace was restored in the late
1980s. Since then, the government has been implementing ma-
jor infrastructure rehabilitation programs in the public health
sector.
The health sector in Uganda is composed of four types of
facilities: hospitals, health centers, dispensaries, and aid posts
or subdispensaries. These facilities can be government-operated
and -owned, private for-profit, or private not-for-profit. The impact
evaluation focuses on public dispensaries. Dispensaries are in the
lowest tier of the health system where a professional interaction
between users and providers takes place. Most dispensaries are
rural. According to the government health sector strategic plan,
the standard for dispensaries includes preventive, promotional,
outpatient care, maternity, general ward, and laboratory services
(Republic of Uganda 2000). As of 2001, public health services are
free. In our sample, on average, a dispensary was staffed by an in-
charge or clinical officer (a trained medical worker), two nurses,
and three nursing aids or other assistants.
The health sector in Uganda is decentralized, and a number
of actors are responsible for supervision and control of the dispen-
saries. At the lowest tier, the Health Unit Management Committee
(HUMC) is supposed to be the main link between the community
and the facility. Each dispensary has an HUMC, which consists
of both health workers and nonpolitical representatives from the
community. The HUMC should monitor the day-to-day running
of the facility but it has no authority to sanction workers. The
next level in the institutional hierarchy is the health subdistrict.
The health subdistrict monitors funds, drugs, and service delivery
at the dispensary. Supervision meetings by the health subdistrict
are supposed to appear quarterly but, in practice, monitoring is in-
frequent. The health subdistrict has the authority to reprimand,
26 0
but not dismiss, staff for indiscipline. Thus in severe cases of

indiscipline, the errand will be referred to the chief administra-
tive officer of the district and the District Service Commission,
which are the appointing authorities for the district. They have
the authority to suspend or dismiss staff.
Various local NGOs, so-called community-based organiza-
tions (CBOs), focusing primarily on health education, are also
active in the sector.
III. EXPERIMENTAL DESIGN AND DATA

III.A. Overview
In response to perceived weak health care delivery at the
primary level, a pilot project (citizen report cards) aimed at en-
hancing community involvement and monitoring in the delivery
of primary health care was initiated in 2004. The project was de-
signed by staff from Stockholm University and the World Bank,
and implemented in cooperation with a number of Ugandan prac-
titioners and eighteen community-based organizations.
The main objective of the intervention was to strengthen
providers’ accountability to citizen-clients by initiating a process,
using trained local actors (CBOs) as facilitators, which the com-
munities themselves could manage and sustain.
Based on a small but rigorous empirical literature on com-
munity participation and oversight, and extensive piloting in the
field, our conjecture was that lack of relevant information on the
status of service delivery and the community’s entitlements, and
failure to agree on, or coordinate expectations of, what is rea-
sonable to demand from the provider, were holding back initia-
tives to pressure and monitor the provider. Although individual
community members have private information—for example, they
know whether their own child has died and whether the health
workers did anything to help them—they typically do not have
any information on aggregate outcomes, such as how many chil-
dren in their community did not survive beyond the age of 5 or
where citizens, on average, seek care, or what the community
can expect in terms of quality and quantity of service provision
(Khemani 2006). Partly as a response to this information problem,
and partly because monitoring a public facility is a public good
that may be subject to serious free-rider problems, few people ac-
tively participate in monitoring their service providers. Relaxing
261
these two constraints was therefore the main objective of the

intervention.
The key behavioral change induced by more extensive
community-based monitoring was expected to be increased effort
by the health unit staff to serve the community. In Uganda, as
in many other developing countries, health workers have few pe-
cuniary incentives to exert high effort. Public money does not
follow patients, and hiring, salaries, and promotions are largely
determined by seniority and educational qualifications—not by
how well the staff performs. An individual worker may of course
still put in high effort if shirking deviates from her ideal choice
(Akerlof and Kranton 2005). The effort choice may also be influ-
enced by social rewards from community members or social sanc-
tions against shirking workers. Social rewards and sanctions are
key instruments available to the community to boost the health
worker’s effort.
III.B. Experimental Design

The experiment involved fifty public dispensaries, and health
care users in the corresponding catchment areas, in nine districts
covering all four regions in Uganda. All project facilities were
located in rural areas. We define a facility’s catchment area, or
the community, as the five-kilometer radius around the facility.2
A community in our sample has, on average, 2,500 households
residing within the five-kilometer radius of the clinic, of which 350
live within a one-kilometer radius. For the experimental design,
the facilities were first stratified by location (districts) and then by
population size. From each group, half of the units were randomly
assigned to the treatment group and the remaining 25 units were
assigned to the control group.
III.C. Data
Data collection was governed by two objectives. First, data
were required to assess how the community at large views the
quality and efficacy of service delivery. We also wanted to contrast
the citizens’ view with that of the health workers. Second, data
were required to evaluate impact. To meet these objectives, two
surveys were implemented: a survey of the fifty providers and
2. Dispensaries are designed to serve households in a catchment area roughly

corresponding to the five-kilometer radius around the facility (Republic of Uganda
2000).
262
a survey of users. Both surveys were implemented prior to the

intervention (data from these surveys formed the basis for the
intervention) and one year after the project had been initiated.
A quantitative service delivery survey was used to collect
data from the providers. Because agents in the service delivery
system may have a strong incentive to misreport key data, the
data were obtained directly from the records kept by facilities
for their own need (i.e., daily patient registers, stock cards, etc.)
rather than from administrative records. The former, often avail-
able in a highly disaggregate format, were considered to suffer the
least from any incentive problems in record keeping. Data were
also collected through visual checks by enumerators.
The household survey collected data on both households’
health outcomes and health facility performance as experienced
by the household. A stratified random sample of households within
the catchment area of each facility was surveyed. In total, roughly
5,000 households were surveyed in each round.3 To the extent
that it was possible, patient records (i.e., patient exercise books
and immunization cards) supported the household’s response. The
postintervention household survey also included a shorter mod-
ule on health outcomes. Specifically, data on under-5 mortality
were collected and we measured the weight of all infants in the
surveyed households.
III.D. Intervention
A smaller subset of the findings from the preintervention sur-
veys, including utilization, quality of services, and comparisons
vis-à-vis other health facilities, were assembled in report cards.
Each treatment facility and its community had a unique report
card, translated into the main language spoken in the community,
summarizing the key findings from the surveys conducted in their
area.
The process of disseminating the report card information, and
encouraging participation, was initiated through a series of meet-
ings: a community meeting, a staff meeting, and an interface meet-
ing. Staff from various local NGOs (CBOs) acted as facilitators in
3. The sampling strategy for the baseline household survey was designed to
generate representative information on the core users’ variables in each commu-
nity (such as the proportion of patients being examined with equipment). In total,
88% of the households surveyed in the baseline survey were resurveyed in the
ex-post survey. The households that could not be surveyed were replaced.
263
End of 2004 Beginning of 2005 Beginning of 2006
Treatment areas
Collection of
Collection of Intervention (5 days): Community-based monitoring household and
baseline • report card dissemination; facility data for
household and • facilitate the agreement of program Time line
facility data a community contract. evaluation
Control areas
FIGURE I
Timing of the Project
these meetings.4 A time line of the intervention is depicted in

Figure I.
The community meeting was a two-afternoon event with ap-
proximately 100 invited participants from the community. To
avoid elite capture, the invited participants consisted of a se-
lection of representatives from different spectra of society (i.e.,
young, old, disabled, women, mothers, leaders). The facilitators
mobilized the village members by cooperating with village council
representatives in the catchment area. Invited participants were
asked to spread the word about the meeting and, in the end, a
large number of uninvited participants also attended the meet-
ing. More than 150 participants per day attended a typical village
meeting.
In the community meeting, the facilitators used a variety
of participatory methods to disseminate the information in the
report cards and encouraged community members to develop a
shared view on how to improve service delivery and monitor the
provider. Information on patients’ rights and entitlements was
also discussed. The participants were divided into focus groups
so that also more marginalized groups such as women and youth
could raise their voices and discuss issues specific to their group.
4. The eighteen participating CBOs had been active in 64% of the treatment
communities and half of the control communities prior to the intervention. A
handful of them covered more than one treatment community. The CBOs were
primarily focused on health, including issues of health education and HIV/AIDS
prevention, although other objectives such as agricultural development, women’s
empowerment, support of orphans and vulnerable children, and peace-building
initiatives, were also common. The CBO facilitators were trained for seven days
in data interpretation and dissemination, utilization of the participatory method-
ology, and conflict resolution and management. Various other CBOs also operate
in the project communities.
264
At the end of the meeting, the community’s suggestions for im-

provements, and how to reach them without additional resources,
were summarized in an action plan. The action plan contained
information on health issues/services that had been identified by
the community as the most important to address, how these is-
sues could be addressed, and how the community could monitor
improvements (or lack thereof). Although the issues raised in the
action plans differed across communities, a common set of con-
cerns included high rates of absenteeism, long waiting time, weak
attention of health staff, and differential treatment.
The health facility meeting was a one-afternoon event held at
the facility with all staff present. In the meeting, the facilitators
contrasted the information on service provision as reported by the
provider with the findings from the household survey.
An interface meeting with members from the community, cho-
sen in the community meeting, and health workers followed the
community and health facility meetings. During the interface
meeting, the community representatives and the health work-
ers discussed suggestions for improvements. The participants
discussed their rights and responsibilities as patients or medi-
cal staff. The outcome was a shared action plan, or a contract,
outlining the community’s and the service provider’s agreement
on what needs to be done, and how, when, and by whom. The
“community contract” also identified how the community could
monitor the agreements and a time plan. Because the problems
that were raised in the community meetings constituted the core
issues discussed during the interface meetings, the community
contract was in many respects similar to the community’s action
plan.
The three separate meetings aimed at kick-starting the pro-
cess of community monitoring. Thus, after the initial meetings
the communities were themselves in charge of establishing ways
of monitoring the provider. After a period of six months, the com-
munities and health facilities were revisited. The CBOs facili-
tated a one-afternoon community meeting and a one-afternoon
interface meeting with the aim of tracking the implementation
of the community contract. Health facility staff and community
members jointly discussed suggestions for sustaining or improv-
ing progress, or in the case of no improvements, why so.5
5. Details on the report cards and the participatory methods used, as well as
an example of an action plan, are provided in the Online Supplemental Appendix.
265
IV. EVALUATION DESIGN AND EXPECTED OUTCOMES

IV.A. Outcomes
The main outcome of interest is whether the intervention
increased the quantity and quality of health care provision and
hence resulted in improved health outcomes. We are also inter-
ested in evaluating changes in all steps in the accountability
chain: Did the treatment communities become more involved in
monitoring the health workers? Did the intervention change the
health workers’ behavior?
As a robustness test we also assess alternative explanations.
One concern is spillovers. Another concern is that the intervention
did not only (or primarily) increase the extent of community mon-
itoring, but had an impact on other agents in the service delivery
chain, such as the health subdistrict. The intervention could also
have affected the health workers’ behavior directly, or affected it
through the actions of the CBOs, rather than through more in-
tense community-based monitoring as we hypothesize. Although
this would not invalidate the causal effect of the intervention, it
would, of course, affect the interpretation. Therefore, these alter-
native hypotheses are also subject to a battery of tests.
IV.B. Statistical Framework

To assess the causal effect of the intervention we estimate
(1) yi jd = α + βT jd + X jdπ + θd + εi jd,
where yi jd is the outcome of household i (when applicable) in com-

munity/health facility j in district d, T jd is an indicator variable
for assignment to treatment, and εi jd is an error term. Equation
(1) also includes a vector, X, of preintervention facility-specific
covariates and district fixed effects (θd).6 Because of random as-
signment, T should be orthogonal to X, and the consistency of β
does not depend on the inclusion of X in the model. The regres-
sion adjustment is used to improve estimation precision and to
account for stratification and chance differences between groups
6. The baseline covariates included are number of villages in the catchment

area, number of days without electricity in the past month, indicator variable
for whether the facility has a separate maternity unit, distance to nearest public
health provider, number of staff with less than advanced A-level education, indi-
cator variable for whether the staff could safely drink from the water source, and
average monthly supply of quinine.
266
in the distribution of pre-random assignment (Kling, Liebman,

and Katz 2007).
We report the results of estimating equation (1) with X and
θ excluded in the Online Supplemental Appendix. For a subset of
variables we can also stack the pre- and postdata and explore the
difference-in-differences in outcomes; that is, we estimate7
(2) yi jt = γ POSTt + βDD (T j ∗ POSTt ) + µj + εijt ,
where POST is an indicator variable for the postintervention pe-

riod, µ j is a facility/community specific fixed effect, and βDD is the
difference-in-differences estimate (program impact).
For some outcomes we have several outcome measures. To
form judgment about the impact of the intervention on a family
of K related outcomes, we follow Kling et al. (2004) and estimate
a seemingly unrelated regression system,
(3) Y = [IK ⊗ (T X)]θ + υ,
where IK is a K by K identity matrix. We ! Kthen derive average stan-

dardized treatment effects, β̃ = 1/K k=1 β̂k/σ̂k, where β̂k and σ̂k
are the point estimate and standard error, respectively, for each
effect (see Duflo, Glennerster, and Kremer [2007]). The point esti-
mate, standard error, and p-value for β̃ are based on the parame-
ters, β̂k and σ̂k, jointly estimated as elements of θ in (3).
V. RESULTS
V.A. Preintervention Differences
The treatment and the control group were similar on most
characteristics prior to the intervention. Average standardized
pretreatment effects are estimated for each family of outcomes
(utilization, utilization pattern, quality, catchment area statis-
tics, health facility characteristics, citizen perceptions, supply of
resources, and user charges) using preintervention data. As shown
in Table I, we cannot reject the null hypotheses of no difference
between the treatment and the control group.8
7. It is a subset of variables because the postintervention surveys collected

information on more variables and outcomes.
8. We report the test of difference in means across control and treatment
groups for each individual variable in the Online Supplemental Appendix.
267
TABLE I
PRETREATMENT FACILITY AND CATCHMENT AREA CHARACTERISTICS AND AVERAGE
STANDARDIZED EFFECTS
Treatment Control
Variables group group Difference
Key characteristics
Outpatient care 593 675 −82
(75) (57) (94)
Delivery 10.3 7.5 2.8
(2.2) (1.4) (2.6)
No. of households in catchment area 2,140 2,224 −84.4
(185) (204) (276)
No. of households per village 93.9 95.3 −1.42
(5.27) (6.32) (8.23)
Drank safely today 0.40 0.32 0.08
(0.10) (0.10) (0.14)
No. of days without electricity in past month 18.3 20.4 −2.12
(2.95) (2.90) (4.14)
Average standardized pretreatment effects
Utilization 0.11
(0.77)
Utilization pattern −0.48
(0.33)
Quality measures −0.35
(0.84)
Catchment area statistics 0.11
(0.66)
Health facility characteristics 0.14
(0.31)
Citizen perceptions 0.37
(0.67)
Supply of drugs 0.73
(0.83)
User charges −0.65
(0.63)
Notes. Key characteristics are catchment area/health facility averages for treatment and control group
and difference in averages. Robust standard errors in parentheses. Description of variables: Outpatient care
is average number of patients visiting the facility per month for outpatient care. Delivery is average number
of deliveries at the facility per month. Number of households in catchment area and number of households per
village are based on census data and Uganda Bureau of Statistics maps. Drank safely today is an indicator
variable for whether the health facility staff at the time of the preintervention survey could safely drink from
the water source. Number of days without electricity in the month prior to preintervention survey is measured
out of 31 days. Average standardized pretreatment effects are derived by estimating equation (3) on each
family of outcomes. Utilization summarizes outpatients and deliveries. Utilization pattern summarizes the
seven measures in Supplemental Appendix Table A.I, reversing sign of traditional healer and self-treatment.
Quality measure summarizes the two measures in Table A.I, reversing sign of waiting time. Catchment area
statistics summarize the four measures in Table A.I. Health facility characteristics summarize the eight
measures in Table A.I and drank safely today and days without electricity, reversing sign of days without
electricity and distance to nearest local council. Citizen perceptions summarize the four measures in Table
A.I. Supply of drugs summarizes the five measures in Table A.I. User charges summarize the four measures
in Table A.I, reversing all signs. The χ 2 test-statistic on the joint hypothesis that all average standardized
effects are 0 is 4.70 with p-values = .79.
268
V.B. Processes
The initial phase of the project, that is, the three sepa-
rate meetings, followed a predesign structure. A parallel system
whereby a member of the survey team originating from the dis-
trict participated as part of the CBO team also confirmed that
the initial phase of the intervention was properly implemented.
After these initial meetings, it was up to the community to sus-
tain and lead the process. In this section we study whether the
treatment communities became more involved in monitoring the
providers.
To avoid influencing local initiatives, we did not have exter-
nal agents visiting the communities and could therefore not doc-
ument all actions taken by the communities in response to the
intervention. Still, we have some information on how processes in
the community have changed. Specifically, the CBOs submitted
reports on what type of changes they observed in the treatment
communities and we also surveyed the local councils in the treat-
ment communities. We use facility and household survey data to
corroborate these reports.
According to the CBO reports and the local council survey, the
community-based monitoring process that followed the first set of
meetings was a joint effort mainly managed by the local councils,
HUMC, and community members. A typical village in the treat-
ment group had, on average, six local council meetings in 2005. In
those meetings, 89% of the villages discussed issues concerning
the project health facility. The main subject of discussion in the
villages concerned the community contract or parts of it, such as
behavior of the staff.
The CBOs reported that concerns raised by the village mem-
bers were carried forward by the local council to the facility
or the HUMC. However, although the HUMC is an entity that
should play an important role in monitoring the provider, it was
in many cases viewed as being ineffective. As a result, misman-
aged HUMCs were dissolved and new members elected. These
claims are confirmed in the survey data: more than one-third of
the HUMCs in the treatment communities were dissolved and
new members were elected or received following the intervention,
whereas we observed no dissolved HUMCs in the control commu-
nities. Further, the CBOs report that the community, or individual
members, also monitored the health workers during visits to the
clinic, when they rewarded and questioned issues in the commu-
nity contract that had or had not been addressed, suggesting a
269
more systematic use of nonpecuniary rewards. Monitoring tools

such as suggestion boxes, numbered waiting cards, and duty ros-
ters were also reported to be put in place in several treatment
facilities.
In Table II, we formally look at the program impact on these
monitoring tools. We use data collected through visual checks by
enumerators during the postintervention facility survey. As shown
in columns (1) and (2), one year into the project, treatment facil-
ities are significantly more likely to have suggestion boxes (no
control facility had these, but 36% of the treatment facilities did)
and numbered waiting cards (only one control facility had one,
but 20% of the treatment facilities did). Columns (3) and (4) show
that a higher share of the treatment facilities also posted infor-
mation on free services and patients’ rights and obligations. The
enumerators could visually confirm that 70% of the treatment fa-
cilities had at least one of these monitoring tools, whereas only 4
of 25 control clinics had at least one of them. The difference is sta-
tistically significant (Online Supplemental Appendix, Table A.II).
Column (5) reports the average standardized effect of the moni-
toring tools. The estimate is significantly different from zero at
the 1% level.
The results based on household data mirror the findings
reported in columns (1)–(5). The performance of the staff is
more often discussed in local council meetings in the treatment
communities, shown in column (6), and community members in
the treatment group are, on average, better informed about the
HUMC’s roles and responsibilities, as reported in column (7).
Combining the evidence from the CBO reports and the household
survey data thus suggests that both the “quantity” of discussions
about the project facility and the subject, from general to specific
discussions about the community contract, changed in response
to the intervention.
V.C. Treatment Practices

The qualitative evidence from the CBOs and, to the extent
that we can measure them, the findings reported in Table II
suggest that the treatment communities became more involved
in monitoring the provider. Did the intervention also affect the
health workers’ behavior and performance? We turn to this
next.
We start by looking at examination procedures. The esti-
mate based on equation (2) with the dependent variable being
270
TABLE II
PROGRAM IMPACT ON MONITORING AND INFORMATION
Dependent Poster Poster on Average Discuss Received

variable Suggestion Numbered informing patients’ standardized facility in LC information
box waiting cards free services rights effect meetings about HUMC
Specification: (1) (2) (3) (4) (5) (6) (7)
Program impact 0.32∗∗∗ 0.16∗ 0.27∗∗∗ 0.14 2.55∗∗∗ 0.13∗∗∗ 0.04∗∗∗

(0.08) (0.09) (0.09) (0.10) (0.55) (0.03) (0.01)
Mean control group 0 0.04 0.12 0.12 — 0.33 0.08

Observations 50 50 50 50 50 3,119 4,996
Notes. Robust standard errors in parentheses. Disturbance terms are clustered by catchment areas in columns (6)–(7). Point estimates, standard errors, and average standardized
effect, columns (1)–(5), are derived from equation (3). Program impact measures the coefficient on the assignment to treatment indicator. Outcome measures in columns (1)–(4) are
based on data collected through visual checks by the enumerators during the postintervention facility survey. Outcome measures in columns (6) and (7) are from the postintervention
household survey. The estimated equations all include district fixed effects and the following baseline covariates: number of villages in catchment area, number of days without
electricity in the past month, indicator variable for whether the facility has a separate maternity unit, distance to nearest public health provider, number of staff with less than
advanced A-level education, indicator variable for whether the staff could safely drink from the water source, and average monthly supply of quinine. Specification: (1) indicator
variable for whether the health facility has a suggestion box for complaints and recommendations; (2) indicator variable for whether the facility has numbered waiting cards for its
patients; (3) indicator variable for whether the facility has a poster informing about free health services; (4) indicator variable for whether the facility has a poster on patients’ rights
and obligations; (5) average standardized effect of the estimates in columns (1)–(4); (6) indicator variable for whether the household discussed the functioning of the health facility at
a local council meeting during the past year; (7) indicator variable for whether the household has received information about the Health Unit Management Committee’s (HUMC’s)
roles and responsibilities.
COMMUNITY-BASED MONITORING IN UGANDA
*Significant at 10%.
**Significant at 5%.
***Significant at 1%.
749
271
TABLE III
PROGRAM IMPACT ON TREATMENT PRACTICES AND MANAGEMENT
Program Mean control

Spec. Dep. variable Model impact 2005 group 2005 Obs.
(1) Equipment used DD 0.08∗∗ −0.07∗∗∗ 0.41 5,280

(0.03) (0.02)
(2) Equipment used OLS 0.01 0.41 2,758
(0.02)
(3) Waiting time DD −12.3∗ −12.4∗∗ 131 6,602
(7.1) (5.2)
(4) Waiting time OLS −5.16 131 3,426
(5.51)
(5) Absence rate OLS −0.13∗∗ 0.47 46
(0.06)
(6) Management of clinic OLS 1.20∗∗∗ −0.49 50
(0.33)
(7) Health information OLS 0.07∗∗∗ 0.32 4,996
(0.02)
(8) Importance of family OLS 0.06∗∗∗ 0.31 4,996
planning (0.02)
(9) Stockouts OLS −0.15∗∗ 0.50 42
(0.07)
Notes. Each row is based on a separate regression. The DD model is from equation (2). The OLS model
is from equation (1) with district fixed effects and baseline covariates as listed in Table II. Robust standard
errors, clustered by catchment areas, are in columns (1)–(4) and (7)–(8), in parentheses. Program impact
measures the coefficient on the assignment to treatment indicator in the OLS models and the assignment to
treatment indicator interacted with an indicator variable for 2005 in the DD models. Specifications: (1) and
(2) indicator variable for whether the staff used any equipment during examination when the patient visited
the health facility; (3) and (4) difference between the time the citizen left the facility and the time the citizen
arrived at the facility, minus the examination time; (5) ratio of workers not physically present at the time
of the postintervention survey to the number of workers employed preintervention (see text for details); (6)
first component from a principal components analysis of the variables Condition of the floors of the health
clinic, Condition of the walls, Condition of furniture, and Smell of the facility, where each condition is ranked
from 1 (dirty) to 3 (clean) by the enumerators; (7) indicator variable for whether the household has received
information about the importance of visiting the health facility and the danger of self-treatment; (8) indicator
variable for whether the household has received information about family planning; (9) share of months in
2005 in which stock cards indicated no availability of drugs (see text for details).
*Significantly different from zero at 90% confidence level.
**Significantly different from zero at 95% confidence level.
***Significantly different from zero at 99% confidence level.
an indicator variable for whether any equipment, for instance, a

thermometer, was used during examination is shown in the first
row in Table III. Fifty percent (41) of the patients in the treatment
(control) community reported that equipment was used the last
time the respondent (or the respondent’s child) visited the project
clinic. The difference-in-differences estimate, a 20% increase, is
highly significant. The cross-sectional estimate in row (2), based
on equation (1), is less precisely estimated.
272
In row (3) we report the result with an alternative measure

of staff performance—the waiting time—defined as the difference
between the time the user left the facility and the time the user
arrived at the facility, subtracting the examination time. On aver-
age, the waiting time was 131 minutes in the control facilities and
119 in the treatment facilities. The estimate based on equation
(1), shown in column (4), is less precisely estimated.
The results on absenteeism are shown in row (3).9 The point
estimate suggests a substantial treatment effect. On average, the
absence rate, defined as the ratio of workers not physically present
at the time of the postintervention survey to the number of work-
ers on the list of employees as reported in the preintervention
survey, is 13 percentage points lower in the treatment facilities.
Thus, in response to the intervention, health workers are more
likely to be at work.
Enumerators also visually checked the condition of the health
clinics, that is, whether floors and walls were clean, the condition
of the furniture, and the smell of the facility. We combine these
variables through principal components analysis into a summary
score. Treatment clinics appear to have put more effort into keep-
ing the clinic in decent condition in response to the intervention.
The point estimate, reported in row (6), implies a 0.56 standard
deviation improvement in the summary score in the treatment
compared to the control facilities.
According to the government health sector strategic plan, pre-
ventive care is one of the core tasks for health providers at the
primary level. A significantly larger share of households in the
treatment communities have received information about the dan-
gers of self-treatment, reported in row (7), and the importance of
family planning, reported in row (8). The difference is 7 and 6
percentage points, respectively.
There is no systematic difference in the supply of drugs be-
tween the treatment and control groups (see Section V.F). How-
ever, as shown in row (9), stockouts of drugs are occurring at a
higher frequency in the control facilities even though, as reported
9. The postintervention survey was not announced in advance. At the start of

the survey, the enumerators physically verified the provider’s presence. A worker
was counted as absent if, at the time of the visit, he or she was not in the clinic.
Staff reported to be on outreach were omitted from the absence calculation. Four
observations were dropped because the total number of workers verified to be
present or reported to be on outreach exceeded the total number of workers on the
preintervention staff list. Assuming instead no absenteeism in these four facilities
yields a point estimate (standard error) of −0.20 (0.065).
273
below, the control facilities treat significantly fewer patients.

These findings suggest that more drugs leaked from health fa-
cilities in the control group.10
The findings on immunization of children under five are re-
ported in Table IV. We have information on how many times
(doses) in total each child has received polio, DPT, BCG, and
measles vaccines and vitamin A supplements. On the basis of
the recommended immunization plan, we create indicator vari-
ables taking the value of 1 if child i of cohort (age) j had received
the required dose(s) of measles, DPT, BCG, and polio vaccines,
respectively, and 0 otherwise.11 We then estimate (3), for each age
group, and calculate average standardized effects.
The average standardized effects are significantly positive
for the younger cohorts. Looking at individual effects (Online
Supplemental Appendix Table A.IV), there are significant posi-
tive differences between households in the treatment and control
community for all five vaccines, although not for all cohorts. For
example, twice as many newborns in the treatment group have re-
ceived vitamin A supplements, 46% more newborns have received
the first dose of BCG vaccine, and 42% more newborns have re-
ceived the first dose of polio vaccine as compared to the control
group.
V.D. Utilization
To the extent we can measure it, the evidence presented so far
suggests that treatment communities began to monitor the health
unit more extensively in response to the intervention and that the
health workers improved the provision of health services. We now
turn to the question of whether the intervention also resulted in
improved quantity and quality of care.
Cross-sectional estimates based on equation (3) are given
in Table V, Panel A. For outpatients and deliveries, we have
10. The dependent variable is the share of months in 2005 in which stock cards
indicated no availability of drugs, averaged over erythromycin, mebendazole, and
septrin. We find no significant difference between treatment and control clinics
for chloroquine—the least expensive of the drugs on which we have data. Not all
clinics had accurate stock cards and these clinics were therefore omitted.
11. According to the Uganda National Expanded Program on Immunization,
each child in Uganda is supposed to be immunized against measles (one dose at
nine months and two doses in case of an epidemic); DPT (three doses at six, ten,
and fourteen weeks); BCG (one dose at birth or during the first contact with a
health facility); and polio (three doses, or four if delivery takes place at the facility,
at six, ten, and fourteen weeks). Because measles vaccination should not be given
at birth, we exclude immunization against measles in the plan for infants under
twelve months.
274
TABLE IV
PROGRAM IMPACT ON IMMUNIZATION
Group Newborn Under 1 year 1 year old 2 years old 3 years old 4 years old
Specification: (1) (2) (3) (4) (5) (6)
Average standardized effect 1.30∗ 1.44∗∗ 1.24∗∗ 0.72 2.01∗∗∗ 0.86

(0.70) (0.72) (0.63) (0.58) (0.67) (0.80)
Observations 173 929 940 951 1,110 526
Notes. Average standardized effects are derived from equation (3) with the dependent variables being indicator variables for whether the child has received at least one dose of
measles, DPT, BCG, and polio vaccines and vitamin A supplement, respectively (see text for details), and with district fixed effects and baseline covariates listed in Table II included.
Robust standard errors clustered by catchment areas in parentheses. Groups: (1) Children under 3 months; (2) Children 0–12 months; (3) Children 13–24 months; (4) Children 25–36
months; (5) Children 37–48 months; (6) Children 49–60 months.
*Significant at 10% level.
**Significant at 5% level.
***Significant at 1% level.
753
275
276
TABLE V
PROGRAM IMPACT ON UTILIZATION/COVERAGE
754
Use of self-
Use of treatment/
Family Average project traditional Average
Dep. variable Outpatients Delivery Antenatal planning std effect facility healers std effect
(1) (2) (3) (4) (5) (6) (7) (8)

A: Cross-sectional data
Program impact 130.2∗∗ 5.3∗∗ 15.0 3.4 1.75∗∗∗ 0.026∗ −0.014 1.43∗
(60.8) (2.1) (11.2) (3.2) (0.63) (0.016) (0.011) (0.87)
Observations 50 50 50 50 50 50 50 50
(9) (10) (11) (12) (13) (14)

B: Panel data
Program impact 189.1∗∗∗ 3.48∗ 2.30∗∗∗ 0.031∗ −0.046∗∗ 1.96∗∗
(67.2) (1.96) (0.69) (0.017) (0.021) (0.89)
Observations 100 100 100 100 100 100
Mean control group 2005 661 9.2 78.9 15.2 – 0.24 0.36 –
Notes. Panel A reports program impact estimates from cross-sectional models with district fixed effects and baseline covariates as listed in Table II, with robust standard errors
in parentheses. Panel B reports program impact estimates from difference-in-differences models with robust standard errors clustered by facility in parentheses. Point estimates,
QUARTERLY JOURNAL OF ECONOMICS
standard errors, and average standardized effects in specifications (1)–(5), (6)–(8), (9)–(11), and (12)–(13) are derived from equation (3). Program impact measures the coefficient on
the assignment to treatment indicator in the OLS models and the assignment to treatment indicator interacted with an indicator variable for 2005 in the DD models. Specifications:
First column is average number of patients visiting the facility per month for outpatient care; second column is average number of deliveries at the facility per month; third column is
average number of antenatal visits at the facility per month; fourth column is average number of family planning visits at the facility per month; fifth column is average standardized
effect of estimates in specifications (1)–(4) and (9)–(10), respectively; sixth column is the share of visits to the project facility of all health visits, averaged over catchment area; seventh
column is the share of visits to traditional healers and self-treatment of all health visits, averaged over catchment area; eighth column is average standardized effect of estimates in
specifications (6)–(7) and (12)–(14), respectively, reversing the sign of use of self-treatment/traditional healers.
preintervention data and can also estimate difference-in-differ-

ences models, shown in Panel B, and value-added models, shown
in Table A.V in the Online Supplemental Appendix.12
One year into the program, utilization (for general outpatient
services) is 20% higher in the treatment facilities as shown in spec-
ification (1). For the difference-in-differences and the value-added
models (reported in specification (9) in Table V and specification
(ix) in Table A.V), the coefficients on the treatment indicator are
larger both in absolute magnitude and relative to their standard
errors. Thus, controlling for baseline outcomes, y jt−1 , improves
the precision of the treatment effect, which is to be expected
given the persistent nature of the outcome variable. The dif-
ference in the number of deliveries, shown in specification (2),
albeit starting from a low level, is 58% and is fairly precisely
estimated. There are also positive differences in the number
of patients seeking antenatal care (19% increase) and family
planning (22% increase), although these estimates are not in-
dividually significantly different from zero. The average stan-
dardized effect, reported in specification (5), however, is highly
significant.
The last three columns in Table V, Panels A and B, re-
port changes in utilization patterns based on household data.
We collected data on where each household member sought care
during 2005 in case of illness that required treatment and col-
lapsed this information by community. There is an 11%–13%
increase, specifications (6) and (12), in the use of the project fa-
cility in treatment as compared to the control group—a result
consistent with that reported in specification (1) using facility
records.
Households in the treatment community also reduced the
number of visits to traditional healers and the extent of self-
treatment, specifications (7) and (13), but there are no statisti-
cally significant differences across the two groups in the use of
other providers (not reported). Thus, as summarized in the av-
erage standardized treatment effects, specifications (8) and (9),
households in the treatment communities switched from tradi-
tional healers and self-treatment to the project facility in response
to the intervention.
12. The value-added specification is y jt = αVA + βVA T j + λy jt−1 + ε jt .
277
V.E. Health Outcomes

We collected data on births, pregnancies, and deaths of chil-
dren under five years in 2005. We also measured the weight of all
infants (i.e., under age 18 months) and children (between ages 18
and 36 months) in the surveyed households.
Health outcomes could have improved for several reasons. As
noted in the Introduction, access to a small set of proven, inexpen-
sive services could, worldwide, have prevented more than half of
all deaths of children under age 5. For a country with an epidemi-
ological profile as in Uganda, the estimate of preventable deaths
is 73 percent (Jones et al. 2003).13 In the community monitor-
ing project specifically, increased utilization and having patients
switch from self-treatment and traditional healers to seek care
at the treatment facility could have an effect. Holding utiliza-
tion constant, better service quality, increased immunization, and
more extensive use of preventive care could also have resulted in
improved health status.
As a reference point we review the set of health interven-
tions feasible for delivery at high coverage in low-income set-
tings with sufficient evidence of effect on reducing mortality from
the major causes of under-5 deaths (Jones et al. 2003). We fo-
cus on community-based, randomized, controlled field trials that
bear some resemblance (because they are community-based) to
our project. Several of these field trials document reductions in
under-5 mortality rates of 30%–50% one to two years into the
project.14 There is, however, a fundamental difference between the
13. This is likely to be a conservative number because only medical interven-

tions for which cause-specific evidence of effect was available were included in the
estimation. For example, increased birth spacing, which has been estimated to
reduce under-5 mortality by 19 percent in India, was not considered. Several peri-
natal and neonatal health interventions that could be implemented in low-income
countries were not included either (Darmstadt et al. 2005).
14. For example, a project in Tigray, Ethiopia, in which coordinators,
supported by a team of supervisors, were trained to teach mothers to recognize
symptoms of malaria in their children and provide antimalarials, reduced under-5
mortality by 40% (Kidane and Morrow 2000). Bang et al. (1990) document a 30%
reduction in under-5 mortality from an intervention that included mass education
about childhood pneumonia and case management of pneumonia by trained village
health workers—a result similar to the meta-analysis estimate by Sazawal and
Black (2003). Bang et al. (1999) evaluate a project in which trained village health
workers, assisted by birth attendants and supervisory visits, provided home-based
neonatal care, including treatment of sepsis. Two years into the project, they doc-
ument a reduction in infant mortality by nearly 50 percent. Rahmathullah et al.
(2003) assess the impact of a community-based project in two rural districts of
Tamil Nadu, India, where newborn infants in the treatment group were allocated
oral vitamin A after delivery. The intervention resulted in a 22% reduction in total
mortality at age 6 months. Manandhar et al. (2004) evaluate a project in which a
278
interventions discussed in footnote 14 and our work. The medi-

cal field trials study the impact of a biological agent or treatment
practice in a community setting when the community health work-
ers and medical personnel competently carry out their tasks. In
the experiment we consider, on the contrary, no new health in-
terventions were introduced and the supply of health inputs was
unchanged. Instead, we focused on incentivizing health workers
to carry out their tasks through strengthened local accountability.
Estimates for births and pregnancies are given in Table VI,
columns (1) and (2). To the extent that the intervention had an
effect on fertility, for example, through increased use of family
planning services, it would primarily affect the incidence of preg-
nancies in 2005, given the forty-week period between conception
to birth. The incidence of births is not significantly different across
treatment and control groups. However, the treatment groups had
10% fewer incidences of pregnancies in 2005.
Column (3) shows the treatment effect on under-5 mortality.15
The point estimate suggests a substantial treatment effect. The
average under-5 mortality rate in the control group is 144, close to
the official figure of 133 for 2005 (UNICEF 2006). In the treatment
group, the under-5 mortality rate is 97, which is a 33% reduction
in under-5 mortality. The difference is significant (and somewhat
larger in absolute magnitude) when controlling for district fixed
effects as reported in column (3). Although the effect is large, it is
worth emphasizing that the 90% confidence interval of our esti-
mate also includes much lower effects (90% CI: 8%–64% reduction
in under-5 mortality rate). With a total of approximately 55,000
households residing in the treatment communities, the treatment
effect corresponds to approximately 550 averted under-5 deaths
in the treatment group in 2005.
facilitator convened nine women’s group meetings every month in the Makwanpur
district in Nepal in which perinatal problems were identified and strategies to ad-
dress them formulated. Two years into the project they document a 30% reduction
in neonatal mortality. Rahman et al. (1982) evaluate the impact of immunization of
women with tetanus injections during pregnancy in rural Bangladesh. The inter-
vention reduced neonatal mortality by 45%. Mtango and Neuvians (1986) evaluate
a project in rural Tanzania in which trained village health workers visited families
at their homes every six to eight weeks, giving health education on recognition and
prevention of acute respiratory infections, treating children with pneumonia with
antibiotics or referring them to the next higher level of care. Within a two-year
period, they document a 27% reduction in under-5 mortality—a reduction slightly
lower than that found in a similar study in rural Bangladesh (Fauveau et al. 1992).
15. The under-5 mortality rate is the sum of the death rates for each cohort
(age groups 0–1, 1–2, 2–3, 3–4, and 4–5) per community in 2005, expressed per
thousand live births.
279
28 0
TABLE VI
PROGRAM IMPACT ON HEALTH OUTCOMES
Dependent variable Weight-for-age

758
Births Pregnancies U5MR Child death z-scores
Specification: (1) (2) (3) (4) (5) (6)
Program impact −0.016 −0.03∗∗ −49.9∗ 0.14∗∗ 0.14∗∗

(0.013) (0.014) (26.9) (0.07) (0.07)
Child age (log) −1.27∗∗∗
(0.07)
Female 0.27∗∗∗
(0.09)
Program impact × year −0.026∗∗
of birth 2005 (0.013)
Program impact × year −0.019∗∗
of birth 2004 (0.008)
Program impact × year 0.003
of birth 2003 (0.009)
of birth 2002 (0.006)
of birth 2001 (0.006)
Mean control group 2005 0.21 0.29 144 0.029 −0.71 −0.71
QUARTERLY JOURNAL OF ECONOMICS
Observations 4,996 4,996 50 5,094 1,135 1,135
Notes. Estimates from equation (1) with district fixed effects and baseline covariates as listed in Table II included. Specification (4) also includes a full set of year-of-birth
indicators. Robust standard errors in parentheses (3), clustered by catchment area (1)–(2), (4)–(6). Program impact measures the coefficient on the assignment to treatment indicator.
Specifications: (1) Number of births in the household in 2005; (2) indicator variable for whether any women in the household are or were pregnant in 2005; (3) U5MR is under-5
mortality rate in the community expressed per 1,000 live births (see text for details); (4) indicator variable for child death in 2005; (5)–(6) weight-for-age z-scores for children under
18 months excluding observations with recorded weight above the 90th percentile in the growth chart reported in Cortinovis et al. (1997).
Column (4) shows the age range of the mortality effects. We

have information on the birth year of all children (under age 5)
alive at in the beginning of 2005 and the birth year of all deceased
children in 2005. Using these data we estimate (1), replacing the
treatment indicator with a full set of year-of-birth indicators and
year-of-birth-by-treatment interactions. We can then address the
question: Conditional on having a child of age x at the end of
2004, or a child born in 2005, what is the probability that the
child died in 2005? As evident, children younger than two years
old drive the reduction in under-5 mortality. The point estimate
for the youngest cohort, for example, implies a 35% reduction in
the likelihood of death of a child born in 2005 in the treatment
compared to the control group.
The program impact on the weight of infants is reported in
columns (5) and (6). On the basis of weight-for-age z-scores, Ugan-
dan infants have values of weight far lower than the international
reference of the U.S. National Center for Health Statistics of the
Centers for Disease Control and Prevention (CDC) and the gap in-
creases for older infants, consistent with the findings in Cortinovis
et al. (1997).16 The difference in means of z scores of infants be-
tween the treatment and the control group is reported in column
(5): The estimated effect (difference) is 0.14 in weight-for-age. Fig-
ure II plots the distribution of z scores for the treatment and con-
trol groups. The difference in measured weight is most apparent
for underweight children. This is consistent with a positive treat-
ment effect arising from improved access and quality of health
care, rather than a general increase in nutritional status, because
underweight status causes a decrease in immune and nonimmune
host defenses and, as a consequence, underweight children are
at a higher risk of suffering from infectious diseases or severe
complications of infectious diseases, and therefore in higher
demand of health care. In column (6) of Table VI, we add controls
for age and gender. The results remain qualitatively unchanged.
The treatment effect is quantitatively important. For this
purpose, the baseline proportion of infants in each risk category
16. The z-score is a normally distributed measure of growth defined as the

difference between the weight of an individual and the median value of weight
for the reference population (2000 CDC Growth Reference in the United States)
for the same age, divided by the standard deviation of the reference population.
We exclude z-scores > |4.5| as implausible and omit observations with a recorded
weight above the 90th percentile in the growth chart reported in Cortinovis et al.
(1997). Because weight is measured by trained enumerators, the reporting error
is likely due to misreported age of the child. The coefficient estimate (standard
error) on the treatment indicator is 0.16 (0.09) when including these outliers.
281
0.3
Treatment group
Control group
0.2
Density
0.1
0
–4 –2 0 2 4
z -scores
FIGURE II
Distributions of Weight-for-Age z-Scores for Treatment and Control Groups
Weight-for-age z-scores for children under 18 months excluding observations
with recorded weight above the 90th percentile in the growth chart reported in
Cortinovis et al. (1997). Sample size is 1,135 children. Solid line depicts the dis-
tribution for the treatment group and dashed line the distribution for the control
group. Vertical solid line denotes mean in treatment group; dashed line denotes
mean in control group.
(severe, <−3 z-scores; moderate, −3 ≤ z-scores < −2; mild, −2 ≤

z-scores < −1) in the control group was calculated. Applying the
shift in the weight-for-age distribution (adding 0.14 z-score) with
the odds ratio for each category—children who are mildly (mod-
erately) [severely] underweight have about a twofold (fivefold)
[eightfold] higher risk of death from infectious disease (Jones et al.
2003)—the reduction in average risk of mortality is estimated to
be approximately 7 percent.17
V.F. Getting Inside the Box and Robustness Tests

The findings of large treatment effects on our proxies of
community-based monitoring and outcomes are consistent with
the community-based monitoring mechanism, but the findings do
17. To put this into perspective, a review of controlled trials designed to im-
prove the intake of complementary food for children ages six months to five years
showed a mean increase of 0.35 z-score (Jones et al. 2003). Jones and colleagues
argue that this is one of the most effective preventive interventions feasible for
delivery at high coverage in a low-income setting.
282
not rule out other explanations. In this section we assess a number

of these alternative hypotheses.
To examine the plausibility of community-based monitoring
as a key mechanism for the health utilization and health out-
comes treatment effects, we follow the methodology used by Kling,
Liebman, and Katz (2007). Specifically, we test whether the differ-
ences between treatment and control in outcomes across districts
are larger in districts with large treatment-control differences in
monitoring and information outcomes. This relationship is sum-
marized by the parameter δ, the coefficient on the summary index
of monitoring and information, in the outcome equation
(4) yj = δ Mj + X j π + ε j .
The summary index of monitoring M in (4) is the first compo-

nent from a principal components analysis of the six monitoring
and information variables in Table II. We examine two outcome
measures (y j ), under-5 mortality and number of outpatients.
Following Kling, Liebman, and Katz (2007), we estimate (4)
by two-stage least squares (2SLS), using a full set of district-
by-treatment interactions as the excluded instruments for the
monitoring index M, while controlling for district fixed effects.
The IV estimation of (4) will be consistent if M is the mediating
factor between treatment and outcomes.
The IV approach is depicted graphically in Figure III.18 There
is a consistent pattern across districts and groups that larger dif-
ferences in monitoring (relative to the district mean) are associ-
ated with larger differences in outcomes—a result in line with the
community-based monitoring mechanism.
Estimates based on equation (4) are given in Table VII. The
first two columns show 2SLS estimates of δ with district-by-
treatment interactions as excluded instruments for the the mon-
itoring index M. To increase precision, we control for baseline
outcomes y jt−1 , when data allow it (i.e., for number of outpatients
treated). The estimates are large in absolute terms and precisely
estimated.
18. If X contains only district indicators, the 2SLS estimate of δ using the
district-by-treatment interactions instruments is the slope of the line fit through a
scatterplot of the outcome and monitoring index means for the treatment and con-
trol groups in each of the nine districts, normalized so that each district has mean
0 (Kling, Liebman, and Katz 2007). We plot the average values by group (treat-
ment and control) for each district for y and M expressed in standard deviation
units relative to the control group overall standard deviation for each variable.
283
762 1 QUARTERLY JOURNAL OF ECONOMICS
T IG
Outpatients treated relative district overall mean
T SO
C MP
C WA
0.5
T AR
T MU
T AP T MA
T MB
0
C MB
C MA C AP T MP
C MU
–0.5
C AR
C SO T WA
–1
C IG
–1.5
–1 –0.5 0 0.5 1
Community monitoring score relative district overall mean
1
Infant mortality rate relative district over all mean
C IG C MB
0.5
C MU
C AR
T WA T MA
C AP
T SO
0
C MP T MP
C WA T AP
C MA
C SO T AR
–0.5
T IG
T MU
T MB
–1
–1 –0.5 0 0.5 1
Community monitoring score relative district overall mean
FIGURE III
Differences in Treatment-Control in Outcomes and Monitoring across Districts
Partial regression plots. The community monitoring index, outpatients, and
under-5 mortality rate in the community (all three variables are described in the
main text) are expressed in standard deviation units relative to the control group
overall standard deviation for each variable. The points are the average values by
group (treatment and control) for each district, normalized so that each district has
mean 0. The line passes through the origin with the slope from the 2SLS estimation
of equation (4) of the outcome on community monitoring and district indicators,
using district-by-treatment interactions as instrumental variables. T (C) denotes
treatment (control) group.
284
TABLE VII
MECHANISMS AND ROBUSTNESS
Dep. variable Out- Out- Out- Out-

patients U5MR patients U5MR patients U5MR patients U5MR
Specification: (1) (2) (3) (4) (5) (6) (7) (8)
Community monitoring index 0.77∗∗∗ −0.43∗ 0.86∗ −0.43 0.77∗∗ −0.54∗

(0.22) (0.25) (0.53) (0.82) (0.21) (0.30)
Staff ’s knowledge about −0.01 0.47
patients’ rights (0.28) (0.29)
Program impact −0.12 0.01 190.5∗∗ −41.3
(0.66) (0.88) (92.6) (45.8)
CBO presence −8.3 −21.0
(69.4) (37.9)
Program impact × CBO presence −127.9 −4.0
(126.1) (58.4)
F-test on program impact 6.17

(.05)
F-test on CBO presence 0.37
(.83)
F-test on program impact × CBO presence 1.03
(.60)
Notes. Columns (1)–(4) report 2SLS estimates from equation (4) with district-by-treatment interactions as the excluded instruments and district fixed effects and outpatientst−1
(in specifications (1) and (3)) as controls. The variables in columns (1)–(4) are expressed in standard deviation units relative to the control group overall standard deviation for
each variable. Robust standard errors are in parentheses. Program impact measures the coefficient on the assignment to treatment indicator. F-test statistics (with p-values in
parentheses) on the excluded instruments Community monitoring and Staff’s knowledge about patients’ rights are 15.9 (.00) and 7.23 (.00), respectively. Point estimates and standard
errors in columns (5)–(6) and columns (7)–(8), respectively, are jointly estimated from equation (3). Explanatory variables: Community monitoring is the first component from a
principal components analysis of the six monitoring and information proxies presented in Table II. Staff’s knowledge about patients’ rights is a measure of the in-charge’s knowledge
about patients’ rights and obligations (see text for details). CBO presence is an indicator variable for whether a participating CBO had been operating in the community before the
intervention. F-test on program impact (CBO presence) [Program impact × CBO presence] is the test statistic, with p-values in parenthesis, on the test that the coefficients on
program impact (CBO presence) [Program impact × CBO presence] are jointly 0 in columns (5)–(6) and (7)–(8), respectively.
763
285
A stricter test of whether the extent of the program impact

varies with the size of the community monitoring impact is to add
a treatment dummy (an overall treatment effect regardless of the
community monitoring impact) to the IV regressions in equation
(4). The community monitoring index is then identified by cross-
district variation in changes in community monitoring by treat-
ment from the district-by-treatment interactions as the excluded
instruments, with the main effect for treatment no longer ex-
cluded; the results are reported in columns (3) and (4) of Table VII.
Comparing the results without and with controls for treatment is
quite similar for both outpatients and under-5 mortality, while
the coefficients on the treatment indicator have the wrong sign
and are small relative to their standard errors, providing some
evidence that community monitoring had the primary effects on
outcomes as opposed to other effects induced by the intervention.
To examine the hypothesis that differences in monitoring are
driving the results as opposed to the supply-driven hypothesis
that health workers, once being informed that their effort de-
viates from what is expected (in the health facility staff meet-
ing), decided to exert greater effort in serving the community, we
augment specification (4) with a measure of the staff ’s knowl-
edge about patients’ rights and obligations.19 This model thus has
two endogenous variables. If large treatment effects on outcomes
across districts are associated with differences in staff knowl-
edge about patients’ rights rather than more intense community
monitoring, this would be evidence against the community-based
monitoring hypothesis. As reported in columns (5) and (6), the
coefficients on community monitoring remain largely unaffected,
and the coefficients on staff knowledge are insignificant and with
the wrong signs, providing additional evidence, albeit not conclu-
sive, that the demand-driven mechanism is more important than
the supply-driven mechanism.
The CBOs played an integral role in the intervention as fa-
cilitators of the meetings. However, it is possible that these CBOs
had a role (as educators or activists, for example) beyond the de-
scribed treatment itself. There is no definitive way to sort out the
19. The in-charge was asked to list patients’ rights and obligations according
to the Ministry of Health’s plan for basic health service delivery. Patients’ rights
were discussed in the interface meeting. Each correct answer (out of five) was
given a score of 0.2, and so this test score ranges from 0 to 1. We also examined
other measures of staff engagement, including number of staff meetings in 2005
and if the in-charge had initiated training of staff on proper conduct. The results
using these alternative proxies mirror those reported in Table VII.
286
role of community-based monitoring from the possible roles of the

CBOs, but because around 60 percent of the CBOs that took part
in the intervention had been operating in the communities before
the intervention, and several of them also had activities in the
control areas, we can investigate whether the outcomes are corre-
lated with preintervention CBO activity. This would be the case if
the CBOs that participated in the experiment, and that had been
present in the communities prior to intervention, had a direct im-
pact on health outcomes (through various preventive activities, for
example) or indirectly by being more involved in monitoring the
provider. The number of outpatients treated per month, shown in
column (7), and the under-5 mortality rate, shown in column (8),
are not significantly different in communities where the CBOs
had been active prior to the intervention. We have also exam-
ined whether the treatment effect varies conditional on observ-
able CBO characteristics or actions. For example, CBOs that are
located (have an office) in the community might, everything else
equal, be in a better position to monitor the health provider. More-
over, at ten of the treatment sites, the CBOs reported that they
regularly visited the clinic. If the CBOs, rather than the commu-
nity, were pushing the service providers into action, presumably
the effect would be more pronounced at sites where the CBO actu-
ally visited the clinic regularly. However, the treatment effects are
independent of whether the office of the CBO is located within a
five-kilometer radius of the health facility or if the CBO reported
that it regularly visited the clinic.20
Given that within each district there are both treatment and
control units, one concern with the evaluation design is the possi-
bility of spillovers from one catchment area to another. In practice,
there are reasons to believe spillovers will not be a serious con-
cern. The average (and median) distance between the treatment
and control facility is thirty kilometers, and in a rural setting,
it is unclear to what extent information about improvements in
20. Given the small sample size, we test whether the distribution of outcomes
in the subsample {T = 1 & CBO located in community = 1} is the same as in the
subsample {T = 1 & CBO located in community = 0}, and whether the distribution
of outcomes in the subsample {T = 1 & CBO regularly carries out monitoring
visits to the facility = 1} is the same as in the subsample {T = 1 & CBO regularly
carries out monitoring visits to the facility = 0}, using the Wilcoxon rank-sum test.
The test statistics (with p-values in parentheses) are 0.88 (.38) and −1.10 (.27) for
outpatients and 0.31 (.76) and −0.03 (.98) for under-5 mortality rate. We get similar
results if we enrich equation (1) with an interaction term T × CBO characteristic.
The estimates of the interaction term are not statistically different from 0 in any
of the specifications.
287
treatment facilities has spread to control communities. Still, the

possibility of spillovers is a concern. Following Miguel and Kremer
(2004), and taking advantage of the variation in distance to the
nearest treatment clinic induced by randomization, we estimate
spillovers from treatment to control groups by enriching X in equa-
tion (1) to include an indicator variable for whether the control
clinic is within ten kilometers of the nearest treatment clinic. The
results are presented in the Online Supplemental Appendix (for
utilization, delivery, and child death). We do not find evidence in
favor of the spillover hypothesis.
Another concern is if the district or subdistrict management
changed its behavior or support in response to the intervention.
For example, the health subdistrict or local government may have
provided additional funding or other support to the treatment
facilities. The results in Table A.VIII in the Online Supplemental
Appendix do not provide any evidence of this being the case. The
treatment facilities did not receive more drugs or funding from the
subdistrict or district as compared to the control facilities during
2005.
Upper-level authorities could also have increased their su-
pervision of treatment facilities in response to the intervention.
As shown in Online Supplemental Appendix Table A.IX, however,
supervision of providers by upper-level government authorities
remained low in both the treatment and the control group. As a
complement we also assessed sanctions. Only a handful of staff
were dismissed or transferred in 2005 and there is no system-
atic pattern that distinguishes treatment from control facilities.
There is also no difference between treatment and control facili-
ties in the number of staff that voluntarily left the facility during
2005 (Table A.IX).
VI. DISCUSSION
Based on a small but rigorous empirical literature on com-
munity participation and oversight, and extensive piloting in the
field, our conjecture was that lack of relevant information and fail-
ure to agree on, or coordinate expectations of, what is reasonable
to demand from the provider were holding back individual and
group action to pressure and monitor the provider. We designed
an intervention aimed at relaxing these constraints. Through two
rounds of community meetings, local NGOs initiated a process
288
aimed at energizing the community and agreeing on actions to

improve service provision.
We document large increases in utilization and improved
health outcomes that compare favorably to some of the more suc-
cessful community-based intervention trials reported in the med-
ical literature. However, whereas medical field trials address the
question of impact of a biological agent or treatment practice when
the health workers do what they are supposed to do, we focus on
a mechanism to ensure that health workers exert effort to serve
the community.
The project was implemented in nine districts in Uganda
with an estimated catchment population of approximately 55,000
households. In this dimension, therefore, the project has already
shown that it can be brought to scale. However, the literature on
how to enhance local accountability and participation is still in
its infancy. And although the results in the paper suggest that
community monitoring can play an important role in improving
service delivery when traditional top-down supervision is inef-
fective, there are still a number of outstanding questions. For
example, we know little about long-term effects and cross-sector
externalities. It may also be the case that combining bottom-up
monitoring with a reformed top-down approach could yield even
better results. Before scaling up, it is also important to subject
the project to a cost-benefit analysis. This would require putting
a value on the improvements we have documented. To provide
a flavor of such a cost-benefit analysis, consider the findings on
averting the death of a child under five. A back-of-the-envelope
calculation suggests that the intervention, including the cost for
collecting data for the report cards (the main cost item), at $3
per household in the catchment areas or $160,000 in total, only
judged on the cost per death averted, must be considered to be
fairly cost-effective. The estimated cost of averting the death of
a child under five is around $300, which should be compared to
the estimate that the average cost per child life saved through
the combined and integrated delivery of 23 interventions shown
to reduce mortality from the major causes of death in children
younger than 5 years is $887 (Bryce et al. 2003).
As argued in a recent Lancet article, a systematic program of
research to answer questions about how best to deliver health
(child survival) interventions is urgently needed (Bryce et al.
2003). In this paper we have focused on a mechanism that has been
highlighted, but not examined, in the literature—a mechanism of
289
accountability enabling (poor) people to scrutinize whether those

in authority have fulfilled their health responsibilities. Future
research should address long-term effects, identify which mech-
anisms or combination of mechanisms that are important, and
study the extent to which the results generalize to other social
sectors.
IGIER, BOCCONI UNIVERSITY, AND CEPR

INSTITUTE FOR INTERNATIONAL ECONOMIC STUDIES, STOCKHOLM UNIVERSITY, NHH,
AND CEPR
REFERENCES
Akerlof, George A., and Rachel E. Kranton, “Identity and the Economics of Orga-
nizations,” Journal of Economic Perspectives, 19 (2005), 9–32.
Banerjee, Abhijit, Angus Deaton, and Esther Duflo, “Wealth, Health, and Health
Service Delivery in Rural Rajasthan,” American Economic Review, 94 (2004),
326–330.
Bang, Abhay T., Rani A. Bang, Sanjay B. Baitule, M. Hanimi Reddy, and Mahesh
D. Deshmukh, “Effect of Home-Based Neonatal Care and Management of
Sepsis on Neonatal Mortality: Field Trial in Rural India,” Lancet, 354 (1999),
1955–1961.
Bang, Abhay T., Rani A. Bang, O. Tale, P. Sontakke, J. Solanki, R. Wargantiwar,
and P. Kelzarkar, “Reduction in Pneumonia Mortality and Total Childhood
Mortality by Means of Community-Based Intervention Trial in Gadchiroli,
India,” Lancet, 336 (1990), 201–206.
Besley, Timothy, and Robin Burgess, “The Political Economy of Government Re-
sponsiveness: Theory and Evidence from India,” Quarterly Journal of Eco-
nomics, 177 (2002), 1415–1451.
Black, Robert E., Saul S. Morris, and Jennifer Bryce, “Where and Why Are 10
Million Children Dying Every Year?” Lancet, 361 (2003), 2226–2234.
Bryce, Jennifer, Shams el Arifeen, George Pariyo, Claudio F. Lanata, Davidson
Gwatkin, and Jean-Pierre Habicht, the Multi-Country Evaluation of IMCI
Study Group, “Reducing Child Mortality: Can Public Health Deliver?” Lancet,
362 (2003), 159–164.
Chaudhury, Nazmul, Jeffrey Hammer, Michael Kremer, Karthik Muralidharan,
and F. Halsey Rogers, “Missing in Action: Teacher and Health Worker Absence
in Developing Countries,” Journal of Economic Perspectives, 20 (2006), 91–116.
Cortinovis, I., V. Vella, N. Ndiku, and S. Milani, “Weight, Height and Arm Circum-
ference of Children under 5 in the District of Mbarara, South-West Uganda,”
Annals of Human Biology, 24 (1997), 557–568.
Darmstadt, Gary L., Zulfiqar A. Bhutta, Simon Cousens, Taghreed Adam,
Neff Walker, and Luc de Bernis, Lancet Neonatal Survival Steering Team,
“Evidence-Based, Cost-Effective Interventions: How Many Newborn Babies
Can We Save?” Lancet, 365 (2005), 977–988.
Duflo, Esther, Rachel Glennerster, and Michael Kremer, “Using Randomization
in Development Economics Research: A Tool Kit,” CEPR Working Paper No.
6059, 2007.
Fauveau, Vincent, M. K. Stewart, J. Chakraborty, and S. A. Khan, “Impact on
Mortality of a Community-Based Programme to Control Acute Lower Respi-
ratory Tract Infections,” Bulletin of the World Health Organization, 70 (1992),
109–116.
Ferraz, Claudio, and Frederico Finan, “Exposing Corrupt Politicians: The Effects of
Brazil’s Publicly Released Audits on Electoral Outcomes,” Quarterly Journal
of Economics, 123 (2008), 703–745.
29 0
Jones, Gareth, Richard W. Steketee, Robert E. Black, Zulfiqar A. Bhutta, and Saul
S. Morris, Bellagio Child Survival Study Group, “How Many Child Deaths
Can We Prevent This Year?” Lancet, 362 (2003), 65–71.
Khemani, Stuti, “Can Information Campaigns Overcome Political Obstacles to
Serving the Poor,” World Bank Working Paper, 2006.
Kidane, Gebreyesus, and Richard H. Morrow, “ Teaching Mothers to Provide Home
Treatment of Malaria in Tigray, Ethiopia: A Randomised Trial,” Lancet, 356
(2000), 550–555.
Kling, Jeffrey R., Jeffrey B. Liebman, and Lawrence F. Katz, “Experimental Anal-
ysis of Neighborhood Effects,” Econometrica, 75 (2007), 83–119.
Kling, Jeffrey R., Jeffrey B. Liebman, Lawrence F. Katz, and Lisa Sanbonmatsu,
“Moving to Opportunity and Tranquility: Neighborhood Effects on Adult Eco-
nomic Self-Sufficiency and Health from a Randomized Housing Voucher Ex-
periment,” Princeton University Working Paper No. 5, 2004.
Manandhar, Dharma S., David Osrin, Bhim Prasad Shrestha, Natasha Mesko,
Joanna Morrison, Kirti Man Tumbahangphe, Suresh Tamang, Sushma Thapa,
Dej Shrestha, and Bidur Thapa, “Effect of a Participatory Intervention with
Women’s Groups on Birth Outcomes in Nepal: Cluster-Randomised Controlled
Trial,” Lancet, 364 (2004), 970–979.
McPake, Barbara, Delius Asiimwe, Francis Mwesigye, Mathias Ofumbi, Lisbeth
Ortenblad, Pieter Streefland, and Asaph Turinde, “The Economic Behavior
of Health Workers in Uganda: Implications for Quality and Accessibility of
Public Health Services,” Social Science and Medicine, 49 (1999), 849–865.
Miguel, Edward, and Michael Kremer, “Worms: Identifying Impacts on Education
and Health in the Presence of Treatment Externalities,” Econometrica, 72
(2004), 159–217.
Mtango, F. D., and D. Neuvians, “Acute Respiratory Infections in Children under
Five Years. Control Project in Bagamoyo District, Tanzania,” Transactions of
the Royal Society of Tropical Medicine and Hygiene, 80 (1986), 851–858.
Olken, Benjamin A., “Monitoring Corruption: Evidence from a Field Experiment
in Indonesia,” Journal of Political Economy, 115 (2007), 200–249.
Rahman, Makhlisur, Lincoln C. Chen, J. Chakraborty, Md. Yunus, A. I. Chowdhury,
A. M. Sarder, Shushum Bhatia, and George T. Curlin, “Use of Tetanus Toxoid
for the Prevention of Neonatal Tetanus. 1. Reduction of Neonatal Mortality by
Immunization of Non-pregnant and Pregnant Women in Rural Bangladesh,”
Bulletin of the World Health Organization, 60 (1982), 261–267.
Rahmathullah, Lakshmi, James M. Tielsch, R. D. Thulasiraj, Joanne Katz, Chris-
tian Coles, Sheela Devi, Rajeesh John, Karthik Prakash, A. V. Sadanand, N.
Edwin, and C. Kamaraj, “Impact of Supplementing Newborn Infants with Vi-
tamin A on Early Infant Mortality: Community-Based Randomised Trial in
Southern India,” British Medical Journal, 327 (2003), 254.
Reinikka, Ritva, and Jakob Svensson, “Local Capture: Evidence from a Central
Government Transfer Program in Uganda,” Quarterly Journal of Economics,
119 (2004), 679–705.
Republic of Uganda, National Health Policy and Health Sector Strategic Plan
2000/01-2004/05 (Kampala: Ministry of Health, 2000).
Sazawal, Sunil, and Robert E. Black, Pneumonia Case Management Trials Group,
“Effect of Pneumonia Case Management on Mortality in Neonates, Infants,
and Preschool Children: A Meta-analysis of Community-Based Trials,” Lancet
Infectious Diseases, 3 (2003), 547–556.
Strömberg, David, “Radio’s Impact on Public Spending,” Quarterly Journal of Eco-
nomics, 119 (2004), 189–221.
UNICEF (United Nations Children’s Fund), The State of the World’s Children
2007: Women and Children—The Double Dividend of Gender Equality (New
York: UNICEF, 2006).
World Bank, World Development Report 2004: Making Service Work for the Poor
People (Washington, DC: World Bank and Oxford University Press, 2003).
291
ARTICLE IN PRESS

Going to scale with community-based primary care:

An analysis of the family health program and infant mortality
in Brazil, 1999–2004
James Macinkoa,!, Maria de Fátima Marinho de Souzab,
Frederico C. Guanaisc, Celso Cardoso da Silva Simõesd
a
University of Pennsylvania, RWJF Health and Society Scholars program, 3641 Locust Walk, Philadelphia, PA 19104, USA
b
Secretariat of Health Surveillance, Ministry of Health, Brası´lia, Brazil
c
Ministry of Social Development and the Fight against Hunger, Brası´lia, Brazil
d
Institute of Geography and Statistics (IBGE), Rio de Janeiro, Brazil
Available online 8 August 2007
Abstract
This article assesses the effects of an integrated community-based primary care program (Brazil’s Family Health
Program, known as the PSF) on microregional variations in infant mortality (IMR), neonatal mortality, and post-neonatal
mortality rates from 1999 to 2004. The study utilized a pooled cross-sectional ecological analysis using panel data from
Brazilian microregions, and controlled for measures of physicians and hospital beds per 1000 population, Hepatitis B
coverage, the proportion of women without prenatal care and with no formal education, low birth weight births,
population size, and poverty rates. The data covered all the 557 Brazilian microregions over a 6-year period (1999–2004).
Results show that IMR declined about 13 percent from 1999 to 2004, while Family Health Program coverage increased
from an average of about 14 to nearly 60 percent. Controlling for other health determinants, a 10 percent increase in
Family Health Program coverage was associated with a 0.45 percent decrease in IMR, a 0.6 percent decline in post-
neonatal mortality, and a 1 percent decline in diarrhea mortality (po0.05). PSF program coverage was not associated with
neonatal mortality rates. Lessons learned from the Brazilian experience may be helpful as other countries consider
adopting community-based primary care approaches.
Keywords: Primary health care; Family health program; Brazil; Health services evaluation; Infant mortality
Introduction But there have been few peer-reviewed studies that

assess the effectiveness of national primary health
There is renewed worldwide interest in primary care strategies on improving population health in
health care and its potential for improving human the developing world. This study attempts to fill this
health (Pan American Health Organization, 2007). gap by analyzing the effects of the Brazilian Family
Health Program (Programa Saúde da Famı´lia or
!Corresponding author. Tel.: +1 2129985592. PSF in Portuguese) on child health. Child health
E-mail addresses: jmj5@nyu.edu (J. Macinko), outcomes are particularly important to examine
mfmsouza@usp.br (M. de Fátima Marinho de Souza). because of the higher-than-expected rates of infant
doi:10.1016/j.socscimed.2007.06.028
292
ARTICLE IN PRESS
J. Macinko et al. / Social Science & Medicine 65 (2007) 2070–2080 2071
and child mortality in Brazil as compared to other changes in health outcomes at the national level
countries of similar gross national income, the (Macinko, Guanais, & Marinho de Souza, 2006).
PSF’s emphasis on improving infant and child The present article expands and strengthens earlier
health, and the possibility of observing rapid work by employing local-level analyses, examining
changes in these outcomes over a relatively short several different outcomes, and by assessing the
period of time. effects of the rapid expansions in PSF coverage over
The PSF is the main approach to provide primary the past few years.
care services within Brazil’s national health system,
known as the Sistema Única de Saúde or SUS Methods
(Almeida & Pêgo, 2002). The PSF has its roots in
the community health agents program begun in the This study follows a quasi-experimental design
state of Ceará in the early 1990s (Cufino Svitone, since each municipality in Brazil adopted the PSF at
Garfield, Vasconcelos, & Araujo Craveiro, 2000). different times and coverage in each municipality
Since it was adopted as a national strategy in 1994, grew at different rates. To take advantage of this
the program had grown by 2007 to encompass heterogeneity, we use a pooled, cross-sectional, time
26,730 community-based teams responsible for series approach to assess the relationships between
providing care to about 85 million people, making dependent and independent variables over a 6-year
it one of the world’s largest systems of community- period. This technique pools together 6 years
based primary care (Brazilian Ministry of Health (1999–2004) of cross-sections (composed of all 557
Department of Primary Care, 2007). Brazilian microregions for each year) for a max-
The PSF is a decentralized approach to providing imum sample size of 3342 observations. The
core primary care functions, including first-contact approach provides an estimate of the health effects
access for each new health need, comprehensive and of program expansion by testing the association
person-focused care over the lifecourse, coordina- between differences in coverage in each microregion
tion of care between different providers and types of with differences in infant mortality outcomes, while
health services, and family and community-oriented controlling for potential confounders (Hsiao, 2003).
health promotion activities (Ministry of Health of In order to strengthen the study design, we
Brazil, 2003). These functions are achieved through analyze two types of outcomes. Based on previous
the program’s organization (municipalities manage literature, we hypothesize that the PSF will have a
the program with national supervision and each strong association with outcomes most sensitive to
PSF team is assigned to a geographical area with primary care: post-neonatal mortality (deaths of
responsibility for enrolling and monitoring the children from 30 days to 1 year per 1000 live births)
health status of about 3500 people), its financing and deaths from diarrheal diseases (deaths of
(services are delivered free of charge, are financed children under 1 year from diarrhea per 1000 live
on a capitation basis, and municipalities have births) (Caldeira, Franc- a, & Goulart, 2001; Cal-
incentives for increasing the number of neighbor- deira, Franc- a, Perpetuo, & Goulart, 2005). It
hoods with access to the program), and delivery should have a modest impact on IMR (all deaths
mechanisms (multidisciplinary teams are composed of children under 1 year per 1000 live births in the
of, at minimum, a physician and nurse who deliver same year) that will depend on the proportion of
clinic-based care and most teams include commu- IMR that is composed of post-neonatal mortality
nity health workers who make regular home visits (Moore, Castillo, Richardson, & Reid, 2003). We
and perform community-based health promotion hypothesize that there should be little or no relation
activities) (Ministry of Health of Brazil, 2003). between PSF coverage and neonatal mortality rates
Family Health Program teams in many areas also (deaths of children within their first month of life
include dental and social work professionals. per 1000 live births), since these outcomes are most
Despite the ambitious scope of this undertaking sensitive to care provided primarily by specialist and
there have been only a few evaluations of the hospital services outside the scope of the PSF
program (Conill, 2002; Escorel et al., 2002; Minis- (Lansky, Franc- a, & Leal Mdo, 2002).
tério da Saúde, 2004; Serra, 2005; Viana & The unit of analysis is the microregion. Each of
Pierantoni, 2002) although several more are under- the 557 microregions contains several of Brazil’s
way. To date, only one peer-reviewed article has 5564 municipalities that have been grouped together
assessed the relationship between PSF coverage and to be geographically contiguous and homogeneous
293
ARTICLE IN PRESS
2072 J. Macinko et al. / Social Science & Medicine 65 (2007) 2070–2080
in terms of demography, agriculture, and transpor- regressors and error terms. (Wooldridge, 2002) All
tation. Microregions represent smaller units of analyses were conducted using Stata 9 software and
analysis and thus capture greater variation than use robust standard errors to correct for hetero-
would analysis of the 27 Brazilian states. Micro- skedasticity (Statacorp, 2005).
regions also have a larger population size than Advantages of the fixed-effects model over cross-
individual municipalities, thus allowing for more sectional analyses include the fact that it is able
stable mortality estimates over time. establish temporal ordering between exposures and
Data on PSF coverage, health resources, and outcomes and it can control for unmeasured time-
outcomes are from the Brazilian Ministry of Health invariant characteristics of the microregion (such as
(Ministry of Health of Brazil, 2007). In this study, geography, historical disadvantages, urban/rural
we use official estimates of IMR that have been location, and local cultural practices) that might
adjusted for underreporting of child deaths (Rede influence health outcomes (Hsiao, 2003). One
Interagencial de informac- ões para a saúde (RIPSA), disadvantage of the fixed-effects approach is that
2002; Szwarcwald, Leal Medo, de Andrade, & the results obtained are conditional on the data used
Souza, 2002). All other outcomes (neonatal, post- to estimate them; that is, results cannot be general-
neonatal, and diarrhea mortality rates) are con- ized to other years or microregions not included in
structed directly from observed counts. the study (Hsiao, 2003).
Independent variables known to influence infant In order to compare how variables changed over
mortality include poverty (proportion of the popu- time, we calculate the mean values and standard
lation in the lowest income quintile), women’s deviations for 1999 and 2004 and the percent change
health and development (proportion of women over during this time. Differences in mean values
15 with no formal schooling, and proportion of between time periods were assessed using t-tests.
women with no prenatal care), child health (propor- Regression analyses are presented as a series of
tion of children with Hepatitis B immunizations, nested models. The F-test is used to assess whether
low birth weight defined as percent of births under the inclusion of an additional set of independent
2500 g), and health services (physicians and hospital variables improved regression models. In order to
beds per 1000) (Moore et al., 2003; Wang, 2003). compare the magnitude of the effects of the main
Data on these variables are based on population explanatory variables on the outcomes, we calcu-
surveys conducted by the Brazilian Institute of lated their marginal effects. This statistic represents
Geography and Statistics (IGBE) and developed for the percent change in the outcome given a one-
state-level representativity by the Institute of percent change in the independent variable, when all
Applied Economic Research (IPEA) (Brazilian other values are set at their mean (Greene, 2003).
Institute of Geography and Statistics, 2005; Insti- We also assessed several pathways by which the
tuto de Pesquisa Econômica Aplicada (IPEA), May PSF might influence IMR. Primary care access is
2005). associated with lower post-neonatal mortality and
Some independent variable data were missing for fewer deaths from diarrhea (UNICEF, 2002). In
some years. Missing data were imputed using non- order to test potential mechanisms of the health
linear interpolation methods that modeled within- effects of PSF expansion we developed a set of
municipal changes as a function of prior values at the dummy variables representing microregions in the
municipal level and contemporaneous values at the highest 75th percentile of under-five deaths from
state level (Allison, 2002; Guanais, 2006). All values both of these conditions (called ‘‘high diarrhea
were then summed up to the microregional level. deaths’’ and ‘‘high postneonatal deaths,’’ respec-
tively). We then created interaction terms between
Statistical analyses these binary variables and PSF coverage to test if
the PSF effect was higher in those microregions
The study uses a fixed-effects specification in where a greater share of infant and child mortality
order to correct for serial correlation of repeated was amenable to primary care. Other interactions of
measures and to control for time-invariant unob- the PSF term (with physicians per 1000 population
served or unobservable microregional characteris- and Hepatitis B coverage) were not significant and
tics. An alternative approach, the random effects therefore not included in the final models.
model, was rejected due to results of the Hausman Because there are great differences in health and
test (po.0001) that tested correlation between the economic development between the poorer north
294
ARTICLE IN PRESS
and northeastern regions of Brazil, as compared increased by 20 percent, and access to prenatal care
with the south, southeast, and central-west regions, increased by 50 percent. Hospital beds per 1000
we present analyses stratified by region. declined slightly. Average population size for
In order to test if there might be a threshold effect microregions increased by nearly a quarter and
for certain levels of PSF coverage, we transformed most of this increase occurred in large metropolitan
PSF coverage into quartiles of coverage and areas. The proportion of the population in the
included these in regression models. lowest income quintile increased slightly, while the
Finally, we performed a number of sensitivity proportion of mothers with no education declined
tests, including using statistical models to control by nearly one-third from 1999 levels.
for potential panel-level autocorrelation and using Table 2 presents the results of the fixed effects
Poisson regression to directly model count data for analyses. Model 1 shows the bivariate relationship
each outcome (Greene, 2003). None of these between PSF and IMR: the larger the proportion of
alternative specifications significantly affected the the state’s population served by the PSF, the lower
sign, significance, or main conclusions reached with the expected infant mortality rate. Model 2 adds
the fixed effects models, suggesting that the results health system covariates to model 1. PSF coverage
presented here are robust. remains significant and negatively associated with
IMR. In terms of covariates, physician supply and
Results Hepatitis B coverage were negatively associated
with IMR, while hospital beds were positively
Table 1 presents descriptive statistics. Between associated with it. The F-test is statistically sig-
1999 and 2004 some measures of infant mortality nificant, suggesting that addition of these covariates
declined: IMR was reduced by 13 percent, post- improves the explanatory power of model 2 over
neonatal mortality by 16 percent and diarrhea- model 1.
specific mortality by 44 percent. However, neonatal Model 3 adds a set of social and economic
mortality increased by 5 percent and the percentage variables. Population size was negatively associated
of births that were low birth weight increased 10 with IMR, suggesting that IMR is lower in
percent. By 2004, the PSF covered about 60 percent microregions with larger populations. Both the
of the population in the microregions, ranging from proportion of women with no formal education
a low of 6 percent to over 100 percent for the top and the proportion of the population in the lowest
90th percentile. Access to some forms of healthcare income quintile were positively associated with
appeared to increase: Physician availability in- IMR. The PSF coefficient remains significant and
creased by 87 percent, Hepatitis B coverage negative (although slightly reduced in magnitude),
Table 1
Descriptive statistics for Brazilian microregions 1999–2004 (n ¼ 557)
Indicator Statistic 1999 2004 Difference % Change

a
Adjusted infant mortality rate (per 1000 live births) Mean (s. dev) 30.155 (13.776) 26.096 (12.429) "4.059*** "13.46
Neonatal mortality rate (per 1000 live births) Mean (s. dev) 4.581 (0.108) 4.844 (0.087) 0.263* 5.74
Post-neonatal mortality rate (per 1000 live births) Mean (s. dev) 16.961 (8.286) 14.177 (4.670) "2.784*** "16.41
Neonatal mortality (as % of IMR) Mean (s. dev) 21.591 (9.532) 25.783 (9.480) 4.192*** 19.41
Proportionate mortality from diarrhea (as % of IMR) Mean (s. dev) 3.215 (0 .271) 1.782 (0.149) "1.432*** "44.54
Coverage of family health program (%) Mean (s. dev) 13.714 (15.862) 59.883 (31.137) 46.169*** 336.66
Physicians per 1000 populationb Mean (s. dev) 1.647 (1.141) 3.079 (1.983) 1.432*** 86.97
Hospital beds per 1000 populationb Mean (s. dev) 2.992 (1.867) 2.591 (1.662) "0.401*** "13.40
Children immunized against Hepatitis B (%) Mean (s. dev) 74.088 (33.192) 93.215 (12.098) 19.127*** 25.81
Population (1000s) Mean (s. dev) 294.953 (772.812) 318.211 (828.998) 23.259*** 7.88
Population in poorest income quintile (%)b Mean (s. dev) 20.703 (88.715) 23.756 (97.122) 3.053*** 14.74
Mothers with no formal education (%) Mean (s. dev) 5.722 (7.659) 4.015 (4.605) "1.707*** "29.83
Mothers with no prenatal care (%) Mean (s. dev) 6.247 (7.262) 2.855 (4.382) "3.392*** "54.30
Low birth weight (% of all births) Mean (s. dev) 6.765 (1.502) 7.423 (1.546) 0.658*** 9.73
Sources: IBGE, SIM/SINASC, IPEA, MAS.* po0.05; ** po0.01; ***po0.001; from paired t-test.
a
Adjusted infant mortality rate takes into account underreporting of infant deaths in some municipalities.
b
Missing values for 2001 and 2003–2004 calculated through interpolation.
295
ARTICLE IN PRESS
Table 2
Fixed effects regression models of infant mortality ratesa for the microregions of Brazil, 1999–2004
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Family health program (% !0.057** !0.042** !0.038** !0.030** !0.028** –

population covered) (0.001) (0.003) (0.002) (0.002) (0.002)
Family health program (Coverage – – – – – !0.715**
quartile 2) (0.066)
quartile 3) (0.105)
quartile 4) (0.147)
Physicians (per 1000 population) – !0.385** !0.333** !0.296** !0.299** !0.353**
(0.139) (0.128) (0.113) (0.112) (0.117)
Hospital beds (per 1000 population) – 0.478** 0.343** 0.350** 0.366** 0.365**
(0.063) (0.056) (0.055) (0.055) (0.056)
Hepatitis B vaccination (% children – !0.013** !0.014** !0.012** !0.011** !0.012**
covered) (0.001) (0.001) (0.001) (0.001) (0.001)
Population (1000s) – – !0.012** !0.009** !0.009** !0.009**
(0.002) (0.001) (0.001) (0.001)
Population in poverty (% in lowest – – 0.035** 0.027** 0.025** 0.025**
income quintile) (0.009) (0.007) (0.007) (0.007)
No formal education (% of all – – 0.158** 0.098** 0.097** 0.104**
mothers) (0.017) (0.015) (0.014) (0.015)
No prenatal care (% of all mothers) – – – 0.160** 0.149** 0.160**
(0.016) (0.015) (0.015)
LBW births (% of all births) – – – !0.214** !0.217** !0.213**
(0.028) (0.028) (0.028)
High diarrhea deaths (75th percentile) – – – – 0.452** –
(0.096)
Interaction: (PSF* high diarrhea – – – – !0.007** –
deaths) (0.002)
High post-neonatal mortality (75th – – – – 0.333** –
percentile) (0.092)
Interaction: PSF* high post-neonatal – – – – !0.003* –
mortality (0.001)
Constant 30.200** 30.340** 32.530** 32.531** 32.277** 32.059**
(0.059) (0.393) (0.574) (0.516) (0.512) (0.526)
Observations 3342 3337 3337 3337 3337 3337
Number of microregions 557 557 557 557 557 557
R-squared (within) 0.567 0.625 0.688 0.733 0.725 0.74
F-test (model 2 v 1) – 71.49** – – – –
F-test (model 3 v 2) – – 55.52** – – –
F-test (model 4 v 3) – – – 78.25** – –
Robust standard errors in parentheses. Microregion fixed effects not shown.

*po0.01; ** po0.001,
a
Infant mortality rate expressed as per 1000 live births and adjusted for underreporting of infant deaths in some municipalities.
and socioeconomic variables remain stable. Based direction and statistical significance. Results of the
on the results of the F-test, Model 3 is considered F-test indicate that Model 4 is superior to any
superior to the previous models. previous models. The R-squared value suggests that
Model 4 includes additional maternal and child the model explains up to 73 percent of the within-
health indicators. The proportion of women with no microregion variation in IMR from 1999 to 2004.
prenatal care is positively associated with IMR Model 5 further explores the relationship between
while the percentage of births that are low weight is PSF and IMR by including interaction terms
negatively associated with IMR. The PSF variable is between PSF coverage and microregions with high
lightly reduced in magnitude, but remains similar in proportionate mortality from diarrhea and high
296
ARTICLE IN PRESS
post-neonatal mortality. The coefficients for high Table 5 presents the marginal effects of the main
diarrhea mortality and for high post-neonatal explanatory variables included in the final model
mortality are positive and significant, suggesting (Model 4 in Table 2). Marginal effects have been
that IMR is higher in those microregions with very multiplied by 10 to give a measure of the percent
high levels of diarrhea-related and post-neonatal change in infant mortality associated with a 10
deaths. The interaction variable for PSF*diarrhea percent increase in the independent variable. Con-
and PSF*post-neonatal deaths is significant and trolling for all other covariates, a ten percent
negative, suggesting that increases in PSF coverage increase in PSF coverage was associated, on
have a particularly strong impact on lowering IMR average, with a 0.45 percent decrease in IMR, a
by reducing diarrhea and post-neonatal deaths in 0.6 percent decrease in post-neonatal mortality, and
areas where these rates are high. a 1 percent decrease in diarrhea-related mortality.
Model 6 tests a transformation of the PSF The largest contributor to reductions in all out-
variables to reflect quartiles of coverage. The results comes was the size of the microregion’s population,
show that as PSF coverage increases, the magnitude suggesting an important urban advantage. For
of the regression coefficient likewise increases, mortality from diarrhea, a ten percent increase in
suggesting a dose–response relationship. Hepatitis B coverage was associated with a 3.7
Table 3 presents results for neonatal, post- percent decline. Most other covariates had marginal
neonatal, and diarrhea mortality rates. All covari- effects near or less than that of PSF coverage.
ates are the same as in the full model (Model 4 from
Table 2). Family health program coverage was not
associated with neonatal mortality, although it was Discussion
negatively associated with both post-neonatal and
diarrhea mortality rates. The analyses presented here suggest that PSF
Table 4 presents analyses stratified by geographic coverage is independently associated with better
region. The main finding is that the PSF has a primary care-sensitive child health outcomes, in-
consistently significant negative association with cluding IMR, post-neonatal mortality, and deaths
IMR in each region. Covariates are generally from diarrhea. As hypothesized, PSF coverage was
similar to the full sample analysis, although in the not associated with neonatal mortality, which is
regional analyses poverty is significant only for the strongly influenced by the availability and quality of
north region, physicians are not significant in the care during and post-delivery, special care for low
southeast, and low birth weight is not significant for birth weight babies, and some aspects of prenatal
the south. care (Martines et al., 2005).
Table 3
Fixed effects regression of unadjusted infant mortality ratesa, Brazilian microregions 1999–2004
Variable Neonatal mortality rate Post-neonatal mortality rate Diarrhea mortality rate
Family health program (% covered) !0.004 (0.003) !0.022** (0.006) !0.012* (0.006)
Physicians (per 1000 population) !0.008 (0.069) !0.264 (0.14) !0.216 (0.182)
Hospital beds (per 1000 population) 0.084 (0.103) !0.297 (0.225) !0.176 (0.217)
Hepatitis B coverage (% of children covered) 0.016** (0.002) 0.006 (0.005) !0.020** (0.006)
Population (1000s) !0.003 (0.002) !0.015** (0.003) !0.007** (0.002)
Population in poorest income quintile (%) 0.011 (0.007) 0.035* (0.017) 0.017 (0.011)
Mothers with no formal education (%) !0.024 (0.024) 0.027 (0.082) 0.164** (0.056)
Mothers with no prenatal care (%) 0.046 (0.026) 0.279** (0.066) !0.066 (0.052)
LBW births (% of all births) 0.192** (0.063) 0.214 (0.156) !0.104 (0.124)
Constant 2.919** (0.77) 18.256** (1.775) 9.969** (1.479)
Observations 3336 3336 3228
Number of microregions 556 556 538
R-squared (within) 0.335 0.545 0.407
Robust standard errors in parentheses; microregion fixed effects not shown.

* po0.05; ** po0.01.
a
All rates expressed as per 1000 live births and are based observed counts that have not been adjusted for underreporting of infant
deaths in some municipalities.
297
ARTICLE IN PRESS
Table 4
Determinants of infant mortality rate by region, 1999–2004
Variable North Northeast Southeast South Central-west
Coverage of family health program (%) !0.037** (0.006) !0.023** (0.003) !0.038** (0.003) !0.014** (0.002) !0.013** (0.002)
Physicians (per 1000 population) !1.401** (0.232) !1.140** (0.111) !0.048 (0.055) !0.603** (0.065) !0.853** (0.091)
Hospital beds (per 1000 population) 0.338** (0.113) 0.222* (0.09) 0.371** (0.07) 0.433** (0.124) 0.255** (0.059)
Hepatitis B immunization (% of children !0.016** (0.003) !0.011** (0.002) !0.006* (0.002) 0.003* (0.001) !0.007* (0.003)
covered)
Population (1000s) !0.006* (0.003) !0.010** (0.003) !0.005** (0.001) !0.004 (0.004) !0.011** (0.001)
Population in poorest income quintile (%) 0.046* (0.022) !0.01 (0.036) 0.017 (0.009) !0.01 (0.017) !0.005 (0.005)
Mothers with no formal education (%) 0.060** (0.02) 0.147** (0.025) 0.062* (0.03) 0.071* (0.03) 0.088** (0.026)
Mothers with no prenatal care (%) 0.065** (0.023) 0.106** (0.021) 0.329** (0.04) 0.243** (0.048) 0.125** (0.043)
LBW births (% of all births) !0.106* (0.052) !0.194** (0.055) !0.212** (0.033) !0.066 (0.035) !0.107** (0.029)
Constant 31.888** (0.670) 47.913** (0.845) 22.874** (0.614) 19.754** (1.098) 26.871** (0.544)
Observations 378 1122 957 564 312
Number of microregions 63 187 160 94 52
R-squared (within) 0.76 0.84 0.74 0.76 0.90
Mean values for selected variables
(1999– 2004)
Infant mortality rate (IMR) 27.12 42.37 19.18 17.81 21.47
PSF coverage (%) 30.82 53.73 32.30 34.44 47.44
Physicians 1.22 1.83 3.34 2.67 2.16
Hospital beds 2.00 2.15 3.29 3.17 3.56
Hepatitis B coverage 80.69 85.49 96.98 97.02 92.41
Mothers with no prenatal care 7.48 10.24 1.97 1.77 2.34
Robust standard errors in parentheses; microregion fixed effects not shown.

* po0.05; ** po0.01.
Infant mortality rate expressed as per 1000 live births and adjusted for underreporting of infant deaths in some municipalities.
Table 5
Marginal effectsa by outcome, Brazilian microregions 1999–2004
Variable Infant mortality rate Post-neonatal mortality rateb Diarrhea mortalityb
Coverage of family health program !0.447** (!0.506, !0.387) !0.591** (!0.909, !0.273) !1.034* (!2.030, !0.037)
Physicians per 1000 population !0.251** (!0.439, !0.064) !0.401* (!0.816, 0.015) !1.088 (!2.880, 0.704)
Hospital beds per 1000 population 0.348** (0.242, 0.455) !0.526 (!1.310, 0.257) !1.038 (!3.536, 1.459)
Hepatitis B coverage (%) !0.376** (!0.452, !0.300) 0.359 (!0.252, 0.971) !3.770** (!5.951, !1.589)
Population (1000s) !1.048** (!1.340, !0.755) !2.873** (!4.154, !1.593) !4.715** (!7.439, !1.991)
Population in poorest income quintile (%) 0.213** (0.102, 0.323) 0.497* (0.035, 0.958) 0.801 (!0.229, 1.831)
Mothers with no formal education (%) 0.190** (0.134, 0.245) 0.093 (!0.459, 0.644) 1.879** (!0.628, 3.130)
Mothers with no prenatal care (%) 0.252** (0.204, 0.300) 0.784** (0.420, 1.147) !0.619 (!1.570, 0.332)
LBW births (% of all births) !0.546** (!0.686, !0.406) 0.973 (!0.414, 2.360) !1.572 (!5.245, 2.101)
Robust 95% confidence intervals in parentheses; microregion fixed effects not shown.
* po0.05; ** po0.01.
a
Marginal effects represent percent change in the outcome associated with a 10 percent change in the independent variable. All marginal
effects were calculated in terms of elasticities evaluated at the means of all other independent variables.
b
Rates expressed as per 1000 live births and are based on observed counts that have not been adjusted for underreporting of infant
deaths in some municipalities.
Our results are consistent with evidence of found to be associated with higher population rates
potential mechanisms through which the PSF might of breastfeeding, oral rehydration therapy, immu-
work to lower primary care-sensitive infant mortal- nizations, and treatment of respiratory and other
ity. For example, higher PSF coverage has been infections—interventions that address the leading
298
ARTICLE IN PRESS
causes of post-neonatal mortality (Emond, Pollock, 2006), they used measures of all immunization
Da Costa, Maranhão, & Macedo, 2002; Escorel schedules which are already over 90 percent in most
et al., 2002; Shi et al., 2004; Starfield, 1985). states. Hepatitis B vaccination is a more recent
The magnitude of the PSF effect was significant, initiative and coverage varies substantially between
albeit of lesser magnitude than observed in previous microregions, making it potentially a more sensitive
studies. This is likely to be due to the fact that IMR indicator of primary care access.
has experienced a dramatic decline throughout Availability of hospital beds was positively
Brazil as a function of a range of interventions, associated with outcomes—a result that was not
including PSF coverage, improved water and expected. One possible explanation is that in recent
sanitation, and better women’s health and develop- years hospitals may have experienced declines in
ment (Macinko et al., 2006). Moreover, as noted accessibility, quality, or both. This hypothesis is
above, as IMR declines a greater proportion of partially supported by the results in Table 3 which
infant deaths tend to happen within the first month show that hospital beds were not associated with
of life due to conditions that are less amenable to neonatal mortality, the outcome that should be
primary care. Neonatal mortality has been linked to most highly correlated with indicators of hospital
increased preterm and low birth weight births and care. Lansky, Franc- a, and Kawachi (2007) suggest
has become a more significant contributor to IMR that there is considerable variation in hospital
in Brazil as post-neonatal mortality declined quality and this variation is associated with elevated
(Barros et al., 2005; Caldeira et al., 2001). perinatal mortality from potentially avoidable con-
There were also important regional differences in ditions such as intrapartum asphyxia. Potentially
the effects of PSF coverage. In the region-stratified avoidable infant mortality was found to be espe-
analyses, the effect of the PSF program was reduced cially high for normal birth weight babies born in
for the more developed southern regions where government-contracted private hospitals in large
IMR has been lower relative to the north and urban areas, which were found to have lower quality
northeast. The apparent protective effect of popula- care (Lansky et al., 2007).
tion size may represent either an urban advantage or Finally, low birth weight births were found to be
the fact that since 1998, PSF expansion has focused negatively associated with IMR in this study. This
on municipalities with populations greater than ‘‘low birth weight paradox’’ has been observed
100,000 people. elsewhere and may be explained by the fact that low
Physician supply was also associated with lower birth weight infants from population groups in
infant mortality: a finding that is consistent with which LBW is most frequent often have a lower risk
other studies (Anand & Barnighausen, 2004). of death than low birth weight infants from the
Sensitivity tests using nurses per 1000 instead of general population (Hernandez-Dıaz, Schisterman,
physicians found similar results, although both & Hernan, 2006). Our ecological analysis might be
variables could not be included in the analyses due more prone to picking up this phenomenon than
to their high correlation (r ¼ 0.74; po0.001). This would an individual-level study. Removing LBW
results suggests that the PSF has made progress in from the analyses does not significantly change any
expanding primary care physician supply in under- of our conclusions.
served regions (such as the northeast) (Ministério da
Saúde, 2004). This argument is supported by the Limitations
observation that the physician supply effect was
significant in all regions except the southeast where This is an ecologic study, so it is not possible to
there has historically been less of a physician deficit test whether the reductions in IMR and other
than in other regions and where most physicians are outcomes occurred within families that actually
specialists (rather than family practitioners or other visited the Family Health Program. Ideally, we
primary care providers). would conduct a multi-level analysis but there are
Not surprisingly, measures such as poverty, currently no nationally representative data on
female illiteracy, lack of prenatal care, and low individual PSF users and non-users. Nevertheless,
levels of Hepatitis B immunization were all found to there is evidence that improving PSF coverage leads
be associated with higher mortality. Although to improvements in determinants of child health.
earlier studies found no relationship between For example, PSF clients regularly receive health
immunization rates and IMR (Macinko et al., education about breastfeeding, use of oral rehydration
299
ARTICLE IN PRESS
therapy, immunization, and infant growth monitor- that fact that the microregions excluded due to poor
ing (Emond et al., 2002; Escorel et al., 2002). In a quality data were also those with the highest rates of
study of several large urban centers, more than diarrheal deaths and underscores the importance of
three-quarters of PSF clients interviewed believed using adjusted rates when available.
that child health services were of good quality and
that the PSF was responsible for improvements in Conclusions
the health of the neighborhood and their family
(Escorel et al., 2002). There is also evidence to The study has shown that expanding coverage of
suggest that the PSF program decreases financial a community-based primary care program, hand-in-
barriers to access (Goldbaum, Gianini, Novaes, & hand with other socioeconomic developments, was
Cesar, 2005). Finally, other studies have confirmed consistently associated with reductions in primary
that in areas where the PSF or similar programs care-sensitive measures of infant mortality. Despite
have been implemented, infant mortality has the consistency of these findings, several issues need
actually declined (Cufino Svitone et al., 2000; to be addressed in order to assess the program’s
Macinko et al., 2006; Serra, 2005). overall effectiveness and potential relevance to other
Ecological analyses are vulnerable to omitted countries.
variable problems. That is, there could be some First, there is little data on the contribution of the
latent, unmeasured variable confounding the ap- PSF to health inequalities within Brazil. This study
parent relationship between PSF and IMR. In this provides some evidence that due to its expansion in
case, the existence of such a variable is unlikely the north and northeast regions of the country, the
given that we employed a comprehensive model of PSF may have contributed to reducing inter-
health determinants, included fixed effects to con- regional inequalities in primary care-sensitive infant
trol for time-invariant unobserved characteristics of mortality. But within regions, expanded PSF cover-
microregions, and tested several pathways and age has not always occurred in the most deprived
alternative explanations. The high R-squared values municipalities (Morsch, Chavannes, van den Akker,
of the main regression models suggest that they Sa, & Dinant, 2001). In order to maximize the
explain a large proportion of the variation in infant equity-enhancing potential of the program, national
mortality. efforts should be directed at encouraging adoption
Finally, conclusions about outcomes based on of the program in the poorest municipalities. Within
unadjusted rates (post-neonatal, neonatal, and municipalities, program expansion should be en-
diarrhea deaths) need to be interpreted with caution couraged within the most underserved neighbor-
since there is evidence of undercounting of child hoods. Such a strategy is likely to improve equity in
mortality in Brazil. Note that this undercounting outcomes since the greatest impact is likely to occur
has improved in recent years, so each year’s data where infant mortality is still the highest, especially
should be closer to real values. In this study, once outcomes have already improved for higher
adjusted IMR values corresponded with observed income groups (Victora, Vaughan, Barros, Silva, &
IMR rates 85 percent of the time with an average Tomasi, 2000).
difference of 4.7 deaths/1000 live births. Most of Second, financial incentives for municipalities to
this variation was in the Northeast region of the adopt the program are currently linked to increasing
country (60 percent agreement in the northeast, 84 population coverage, but there are few systematic
percent agreement in the north, 90 percent agree- monitoring and evaluation processes in place to
ment in the central-west, 99 percent agreement in assess municipal or service-level performance. Sur-
the south and southeast). In sensitivity tests that veys show that clients are generally satisfied with the
excluded the 982 (out of 3337) data points with quality of care delivered, but sustaining this level of
outcome data that was one or more standard satisfaction will be a critical challenge in maintain-
deviation above or below the adjusted IMR rates ing popular and political support for the program
for any year, there was no change in the main (Trad et al., 2002). New initiatives have been
conclusions of the relationship between PSF cover- proposed that would provide financial incentives
age and IMR, neonatal mortality, or post-neonatal for municipalities that reach or exceed certain
mortality. However, several covariates did become health targets as a means to enhance access and
non-significant as did the relationship between PSF quality of care. For these reasons, a major challenge
coverage and diarrhea mortality. This may be due to will be to develop and use systems to monitor and
30 0
ARTICLE IN PRESS
improve the quality of care delivered in order to programs, improving quality of care, and maximiz-
maximize the potential health gains of this innova- ing community-based health promotion. Taken
tive approach to integrated primary care delivery. together these actions may help to assure that the
Third, there is little data available on the cost- PSF becomes more than just another program, but
effectiveness of the PSF. In 2005, Federal govern- fulfills its promise as a central organizing feature of
ment transfers to municipalities totaled $5.7 billion a more accessible, effective, and equitable national
Brazilian Reais (approximately $US 2.6 billion), health system.
which represents about $US 14 per person covered.
This figure does not include the municipal contribu-
Acknowledgments
tion (which varies from zero to nearly 100 percent).
Thus we estimate that the true costs of the program
This study was partially supported by the
may be as much as $US 30 per capita. While this is
Brazilian Ministry of Health. Frederico Guanais
still a modest amount, there is, as yet, no national
was supported by the National Council for Re-
data to compare how well this program performs
search and Development (CNPq). The conclusions
vis-à-vis the status quo. Such information will
presented in this paper represent the opinion of the
become increasingly important to mobilize the
authors alone.
additional political and financial capital needed to
reach the rest of the Brazilian population not
currently covered and then to maintain adequate References
coverage in light of Brazil’s rapid epidemiologic and
demographic transition. Allison, P. (2002). Missing data. Thousand Oaks, CA: Sage
Fourth, Brazil has a large supply of health Publications.
workers, which might make it different from most Almeida, C., & Pêgo, R. A. (2002). Theory and practice in health
other developing countries. So far, the PSF strategy systems reform in Brazil and Mexico. Cad Saúde Pública,
18(4), 971–989.
has been successful in hiring more than 26,000 Anand, S., & Barnighausen, T. (2004). Human resources and
physicians and nurses and over 220,000 community health outcomes: Cross-country econometric study. Lancet,
health workers. In principle, any trained health 364(9445), 1603–1609.
professional may apply to the program and Barros, F. C., Victora, C. G., Barros, A. J., Santos, I. S.,
competitive salaries have made it an increasingly Albernaz, E., Matijasevich, A., et al. (2005). The challenge of
reducing neonatal mortality in middle-income countries:
attractive option. On-going training in primary care Findings from three Brazilian birth cohorts in 1982, 1993,
is an additional benefit of PSF affiliation, but there and 2004. Lancet, 365(9462), 847–854.
is not yet enough known on how well this in-service Brazilian Institute of Geography and Statistics. (2005). Banco
training prepares formerly specialty-trained physi- multidimensional de estatı́sticas. Available: /http://www.
ibge.gov.br/S.
cians to function as primary care providers. As the
Brazilian Ministry of Health Department of Primary Care (2007).
program continues to expand, health authorities Atenc- ão Básica e a Saúde da Famı́lia [Primary Care and
will need to develop longer-term plans for main- Family Health]. Available: /http://dtr2004.saude.gov.br/dab/
taining and expanding the health workforce, with abnumeros.phpS; Accessed May 24, 2007.
particular attention to improving the stability of Caldeira, A. P., Franc- a, E., & Goulart, E. M. (2001).
physician contracting mechanisms, and more con- Postneonatal infant mortality and quality of medical care: A
case-control study. Journal of Pediatr (Rio de Janeiro), 77(6),
certed efforts to enhance provider skills in commu- 461–468.
nity-based primary care. Caldeira, A. P., Franc- a, E., Perpetuo, I. H., & Goulart, E. M.
Finally, because it serves as part of the Brazilian (2005). Trends in avoidable causes of infant mortality in Belo
national health system, the PSF is vulnerable to Horizonte, Brazil, 1984 to 1998. Rev Saude Publica, 39(1),
health system level factors that could undermine its 67–74.
Chiesa, A. M., & Batista, K. B. C. (2004). Desafios da implantac-
potential impact, such as access to pharmaceuticals; ão do Programa Saúde da Famı́lia em uma grande cidade:
the quality and supply of needed specialty, diag- Reflexões acerca da experiência de São Paulo/challenges on
nostic, or hospital care; or the availability, training, the implementation of the family health program in a great
and salaries of health workers (Chiesa & Batista, city: Reflections on the experience São Paulo city. Mundo
saúde (1995), 28(1), 42–48.
2004; Franco, Bastos, & Alves, 2005). Key chal-
Conill, E. M. (2002). Primary care policies and health reforms:
lenges as the program moves forward include An evaluative approach based on an analysis of the family
ensuring coordination between different types of health program in Florianopolis, Santa Catarina, Brazil,
health services and vertically focused disease control 1994–2000. Cad Saude Publica, 18(Suppl), 191–202.
301
ARTICLE IN PRESS
Cufino Svitone, E., Garfield, R., Vasconcelos, M. I., & Araujo care model. Revista Brasileira de Saúde Materno-Infantil, 3(1),
Craveiro, V. (2000). Primary health care lessons from the 113–125.
northeast of Brazil: The Agentes de Saúde Program. Rev Ministry of Health of Brazil. (2007). Informac- ões de Saúde.
Panam Salud Publica, 7(5), 293–302. Available: /http://tabnet.datasus.gov.br/tabnet/tabnet.htmS.
Emond, A., Pollock, J., Da Costa, N., Maranhão, T., & Macedo, A. Moore, D., Castillo, E., Richardson, C., & Reid, R. J. (2003).
(2002). The effectiveness of community-based interventions to Determinants of health status and the influence of primary
improve maternal and infant health in the Northeast of health care services in Latin America, 1990–98. International
Brazil. Rev Panam Salud Publica, 12(2), 101–110. Journal of Health Planning and Management, 18(4), 279–292.
Escorel, S., Giovanella, L., Mendonc- a, M. H., Magalhães, R., de Morsch, E., Chavannes, N., van den Akker, M., Sa, H., &
Castro, M., & Senna, M. (2002). Avaliac- ão da Implementac- ão Dinant, G. J. (2001). The effects of the family health program
do Programa Saúde da Famı´lia em Dez Grande Centros in Ceará state, northeastern Brazil. Archives of Public Health,
Urbanos. Sıńtese dos Principais Resultados. Brası́lia: Ministér- 59, 151–165.
io da Saúde, Secretaria de Polı́ticas de Saúde. Pan American Health Organization. (2007). Renewing primary
Franco, A. L., Bastos, A. C., & Alves, V. S. (2005). The health care in the Americas: Declaration of montevideo. Washing-
physician–patient relationship under the family health pro- ton, DC: Pan American Health Organization Available http://
gram in three municipalities in Bahia State, Brazil. Cad Saúde www.paho.org/English/AD/THS/primaryHealthCare.pdf.
Pública, 21(2), 246–255. Rede Interagencial de informac- ões para a saúde (RIPSA). (2002).
Goldbaum, M., Gianini, R. J., Novaes, H. M., & Cesar, C. L. Indicadores básicos para a saúde no Brasil. Brası́lia, Brasil:
(2005). Health services utilization in areas covered by the Organizac- ão Panamericana da Saúde/OMS.
family health program (Qualis) in São Paulo City, Brazil. Rev Serra, R. (2005). Uma avaliac- ão empı́rica do impacto saúde da
Saude Publica, 39(1), 90–99. famı́lia sobre a saúde infantil no estado de São Paulo.
Greene, W. H. (2003). Econometric analysis. Upper Saddle River, Trabalho submetido ao Prêmio Economia da Saúde, IPEA.
NJ: Prentice-Hall. Brası́lia, DF.
Guanais, F. C. (2006). Evaluating the health impacts of primary Shi, L., Macinko, J., Starfield, B., Xu, J., Regan, J., Politzer, R.,
care decentralization in the context of a developing country. et al. (2004). Primary care, infant mortality, and low birth
Robert F. Wagner Graduate School of Public Service. New weight in the states of the USA. Journal of Epidemiology and
York, NY: Robert F. Wagner Graduate School of Public Community Health, 58(5), 374–380.
Service, New York University. Starfield, B. (1985). Postneonatal mortality. Annual Review of
Hernandez-Dıaz, S., Schisterman, E., & Hernan, M. (2006). The Public Health, 6, 21–40.
birth weight ‘‘paradox’’ uncovered? American Journal of Statacorp. (2005). Stata statistical software: Release 9.0. College
Epidemiology, 164(11), 1115–1120. Station, TX: Stata Corporation.
Hsiao, C. (2003). Analysis of panel data. Cambridge: Cambridge Szwarcwald, C. L., Leal Medo, C., de Andrade, C. L., &
University Press. Souza, P. R., Jr. (2002). Infant mortality estimation in Brazil:
Instituto de Pesquisa Econômica Aplicada (IPEA). (2005). Dados What do Ministry of Health data on deaths and live births
macroeconômicos e regionais. Available: /www.ipeadata. say? Cad Saúde Pública, 18(6), 1725–1736.
gov.brS; Accessed May 2005. Trad, L. A. B., Bastos, A. C. S., Santana, E. M., et al. (2002).
Lansky, S., Franc- a, E., & Kawachi, I. (2007). Social inequalities in Estudo etnográfico da satisfac- ão do usuário do Programa de
perinatal mortality in Belo Horizonte, Brazil: The role of Saúde da Famı́lia (PSF) na Bahia. Ciênc Saúde Coletiva, 7(3),
hospital care. American Journal of Public Health, 97(5), 867–873. 581–589.
Lansky, S., Franc- a, E., & Leal Mdo, C. (2002). Perinatal UNICEF. (2002). Facts for life. New York: United Nations
mortality and evitability: A review. Rev Saude Publica, 36(6), Children’s Fund.
759–772. Viana, A. L., & Pierantoni, C. R. (2002). Indicadores de
Macinko, J., Guanais, F. C., & Marinho de Souza, M. F. (2006). Monitoramento da Implementac- ão do PSF em Grandes Centros
Evaluation of the impact of the family health program on Urbanos. Brasilia: Brazilian Ministry of Health, Department
infant mortality in Brazil, 1990–2002. Journal of Epidemiology of Primary Care.
and Community Health, 60(1), 13–19. Victora, C. G., Vaughan, J. P., Barros, F. C., Silva, A. C., &
Martines, J., Paul, V. K., Bhutta, Z. A., Koblinsky, M., Soucat, A., Tomasi, E. (2000). Explaining trends in inequities: Evidence
Walker, N., et al. (2005). Neonatal survival: A call for action. from Brazilian child health studies. Lancet, 356(9235),
Lancet, 365(9465), 1189–1197. 1093–1098.
Ministério da Saúde. (2004). Avaliac- ão normativa do Programa Wang, L. (2003). Determinants of child mortality in LDCs.
Saúde da Famı´lia no Brasil. Brası́lia: Sectária de Assistência à Empirical findings from demographic and health surveys.
Saúde, Departamento de Atenc- ão Básica. Health Policy, 65, 277–299.
Ministry of Health of Brazil. (2003). The family health program: Wooldridge, J. (2002). Econometric analysis of cross section and
Broadening coverage to consolidate the change in the primary panel data. Cambridge, Massachusetts: The MIT Press.
302
Marchal et al. BMC Health Services Research 2010, 10:24
http://www.biomedcentral.com/1472-6963/10/24
RESEARCH ARTICLE Open Access
A realist evaluation of the management of a well-

performing regional hospital in Ghana
Bruno Marchal1*, McDamien Dedzo2, Guy Kegels3
Abstract
Background: Realist evaluation offers an interesting approach to evaluation of interventions in complex settings,
but has been little applied in health care. We report on a realist case study of a well performing hospital in Ghana
and show how such a realist evaluation design can help to overcome the limited external validity of a traditional
case study.
Methods: We developed a realist evaluation framework for hypothesis formulation, data collection, data analysis
and synthesis of the findings. Focusing on the role of human resource management in hospital performance, we
formulated our hypothesis around the high commitment management concept. Mixed methods were used in data
collection, including individual and group interviews, observations and document reviews.
Results: We found that the human resource management approach (the actual intervention) included induction of
new staff, training and personal development, good communication and information sharing, and decentralised
decision-making. We identified 3 additional practices: ensuring optimal physical working conditions, access to top
managers and managers’ involvement on the work floor. Teamwork, recognition and trust emerged as key
elements of the organisational climate. Interviewees reported high levels of organisational commitment. The
analysis unearthed perceived organisational support and reciprocity as underlying mechanisms that link the
management practices with commitment.
Methodologically, we found that realist evaluation can be fruitfully used to develop detailed case studies that ana-
lyse how management interventions work and in which conditions. Analysing the links between intervention,
mechanism and outcome increases the explaining power, while identification of essential context elements
improves the usefulness of the findings for decision-makers in other settings (external validity). We also identified a
number of practical difficulties and priorities for further methodological development.
Conclusion: This case suggests that a well-balanced HRM bundle can stimulate organisational commitment of
health workers. Such practices can be implemented even with narrow decision spaces. Realist evaluation provides
an appropriate approach to increase the usefulness of case studies to managers and policymakers.
Background the impact of human resource deficits on global health

In the wake of the numerous global health initiatives, initiatives and planning and training capacity issues.
the health workforce of low and middle-income coun- This relative neglect of studies of health workforce
tries is once again receiving a lot of attention [1-3]. management explains why the current evidence base on
While the key role of health workers in improving the effectiveness of HRM policies and strategies is rather
health care quality and implementing disease control weak. More specifically, there are a number of weak-
programmes is widely recognised [4,5] operational nesses that limit their potential to inform decisions of
aspects of health workforce management at service pro- policymakers or health service managers. First, the
vision level remain poorly studied. Indeed, the focus has determinants of health worker performance in poor
been mostly on macro-level aspects, such as brain drain, resource settings have not been studied well. Second,
the HRM policies and strategies, too, are under-
* Correspondence: bmarchal@itg.be researched [6]. Third, systematic reviews indicate that
1
Department of Public Health, Institute of Tropical Medicine-Antwerp,
Nationalestraat 155, B-2000 Antwerp, Belgium most of the studies are methodologically flawed. A
© 2010 Marchal et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
303
Marchal et al. BMC Health Services Research 2010, 10:24 Page 2 of 14
recent realist synthesis of the effect of human resource demonstrating the causal links between intervention and
management interventions on health worker perfor- outcome [29]. Much of this critique has its origins in
mance in LMIC found that very few studies provide quantitative criteria of validity, according to which case
adequate information on the assumptions, the context studies are based on too small numbers of cases and on
and the underlying mechanisms of these interventions non-randomised case selection, thus leading to problems
[7]. The same applies to a review of the effect of HRM of representation and inference [25]. It is exactly here
policies on supply, distribution, efficient use and perfor- that its adherents claim that theory-based methodologies
mance of health workers [8]. Rowe and collegues came can make a difference.
to similar conclusions [9]. Fourth, few of these studies During the 1980s, Chen and Rossi developed the the-
have been carried out in LMIC [8]. ory-based evaluation approach as an answer to policy
All this notwithstanding, policies and management and programme evaluation approaches that remained
strategies are still imported from other settings into limited to before-after and input-output designs or that
health services of LMIC without a blink of the eye. The focused narrowly on methodological issues (method-dri-
surge of performance based financing (PBF) provides a ven evaluation) [30,31]. The theories of change approach
good example. PBF is being introduced at different [32] and realist evaluation (RE) [33] are among the most
levels of the health system [10-13] and in a wide variety recent applications of theory-based evaluation. As we
of countries, including Nicaragua [13], Cambodia [14], will discuss in detail below, both approaches aim at
Rwanda [15,16], Zambia [17], Sri Lanka, Ghana, Zim- opening the black box between intervention and
babwe, Thailand and India [18]. The evidence base, outcome.
however, is very narrow [19]. Most PBF studies were For organisational research, realist evaluation seems to
found to lack controls and to neglect the analysis of offer a number of advantages. It promises, first, to
confounding factors [6], which reduces the validity of increase the external validity of case studies. Building
the attribution of the reported effects to the interven- upon existing knowledge, RE analyses why change
tion. Furthermore, very few studies offer indications of occurs, or why not, and in which conditions. It aims at
the conditions in which these approaches are working providing information that allows decision-makers to
(see [20] for an example of a study that does). judge whether the lessons learnt could be applied else-
In part, the methodological weakness of the health where [34]. Repeated case studies lead to more refined
workforce management research resides in insufficiently middle range theories that offer increasingly refined
rigorous studies. Some problems also stem from the information of context conditions, thereby increasing
widespread use of the case study. Indeed, although orga- generalisability of such case studies [21,27,29] and
nisational studies is a domain marked by a lack of con- improving our understanding of causal processes [35].
sensus on ontology and epistemology [21] and the Second, based on its generative perspective on causality,
consequent lack of consensus on methodology, the case it seeks to explain change by referring to the actors who
study is a common research design for a number of rea- change a situation under influence of particular external
sons. First, it allows exploring a “phenomenon within its events (such as an intervention) and under specific con-
real-life context, especially when the boundaries between ditions [33]. Accepting the role of actors in change
phenomenon and context are not clearly evident” [22], (agency), realist evaluation also considers structural and
and thus suits well the open systems-nature of human institutional features to exist independently of the actors
organisations. Second, it enables investigation of organi- and researchers. If human action is embedded within a
sational behaviour as it happens in its natural setting wider range of social processes and structures, then cau-
[23]. Case studies are also useful in dynamic and com- sal mechanisms reside in social relations and context as
plex situations where multiple, interacting variables may much as in individuals. As a consequence of this ontolo-
act upon intervention and outcome [24,25]. It is well gical perspective, evaluators need to unearth the social
suited to research on HRM [26]. Finally, Hartley argues layers in order to understand the root causes of the pro-
that case studies can help in probing and developing blem at hand [36] and to find the mechanism that
theory [27]. explains the outcomes of the intervention [33]. In short,
Since the publication of Experimental and quasi- Pawson & Tilley argue that realist evaluation indicates
experimental designs for research by Campbell & Stanley ‘what works in which conditions for whom’, rather than
[28], the major limitation held against the case study merely answering the question ‘does it work?’. Realist
design is its limited external validity, or the weak poten- evaluation is thus well suited to assessment of interven-
tial to generalise findings from one case to another. tions in complex situations, which most organisational
Other authors raise its limited attribution power: case research is all about.
studies are good at analysing the intervening processes While the merits of theory-driven and realist evalua-
or documenting evolution in time, but weak at tion have been amply discussed in journals on
30 4
evaluation (see for instance [36-41] and [42-44], there is MRT states how the intervention leads to which effect
little documented experience in the domain of health in which conditions. Lipsey & Pollard identify different
service organisation and public health, notable excep- mechanisms to develop this MRT [53]. It can be formu-
tions being [45] and [46]. This scarcity of realist studies lated on the basis of existing theory and past experience.
could be interpreted as a sign of the limited academic If the latter is not available, exploratory on-site research
credibility of theory-driven evaluation in general: ‘objec- can be done to unearth the models used implicitly by
tivist’ arguments overrule ‘subjectivist’ research [47]. the actors to make sense of the intervention - what
Other reasons may be practical in nature: carrying out a Pawson & Tilley call ‘folk theories’ [33]. Through indivi-
full-blown theory-driven evaluation is resource- and dual interviews or group discussions, the key elements
time intensive [48]. The need of assessing the underlying of the problem or intervention, the expected outcomes
theory in addition to the efficacy/outcome evaluation and potential moderating factors are to be identified
adds to the burden [41]. [50], p. 196). Additional information may be derived
In this paper, we examine whether and how a realist from programme or policy documents. Cause mapping
evaluation design can be applied in research of well per- or concept mapping can be used in this process [54].
forming hospitals. We present the case of Central Regio- Ideally, the resulting MRT is then compared with exist-
nal Hospital (CRH) in Cape Coast, Ghana and discuss ing knowledge. A literature review identifies studies
how we applied this method, from the stage of hypoth- reporting other causal chains, moderating factors or
esis formulation to the synthesis of the results. This case unintended outcomes, allowing a plausibility check of
study is part of a longitudinal study on the links the preliminary MRT. The result is then again discussed
between management and performance in well-perform- with the stakeholders and results in the middle range
ing hospitals. We describe the latter as hospitals that theory that will be tested. Byng constructed the middle
ensure equitable access to high quality care and that range theory on the basis of a literature review, a
provide such services in an efficient manner. We choose description of the intervention and discussions with
CRH both because it won the award for the best hospi- facilitators involved in the programmes in question [40].
tal of the Ghana Health Service in 2004 and on the Regarding designs and research methods, realist evalua-
basis of previous research. tion is neutral [33]: the hypothesis as expressed by the
The objective of the study was to analyse the manage- MRT is guiding the choice of data that should be collected
ment approach at CRH. We formulated the following and the methods and tools to do so. Most theory-driven
research questions: (1) What is the management team’s evaluations in healthcare used the case study design and
vision on its role?; (2) Which management practices are combine both quantitative and qualitative methods.
being carried out?; (3) What is the organisational cli- Pawson & Tilley call the working hypotheses that
mate? (defined by Takeuchi et al. as the perceptions of emerge during the analysis phase ‘Context-Mechanism-
employees regarding how the management approach is Outcome configurations’ (CMOC) [33]. Realist evalua-
practiced and implemented in their organisation [49]; tors describe not only the intervention and its outcome,
(4) What are the results?; (5) What are the underlying but also the context and the underlying mechanism.
mechanisms explaining the effect of the management They seek to establish patterns or regularities that
practices? explain outcomes of interventions. In practice, the data
from interview transcripts, document analysis and obser-
Methods vation are coded with codes drawn from the initial MRT
Principles of realist evaluation (See [40] for a practical example). Similar to other ana-
Drawing inspiration from [34,50,51], we structured our lysis methods, subsequent rounds of analysis lead to a
study in 4 steps: the formulation of the Middle Range refined set of themes, categories and codes. The emer-
Theory, the design of the study, the data analysis and ging findings are compiled as conjectural CMOCs,
synthesis, and presentation of the results. We briefly which indicate how the intervention led to particular
introduce these steps from a theoretical point of view, outcomes in which context and by which mechanism.
and then describe how we developed each step in Their fit with the data is checked to ensure internal
practice. validity. The retained CMOCs are then compared with
A realist evaluation research starts from a middle the MRT, which in turn is modified if necessary [55]. In
range theory (MRT), which is understood as “theor [y] some studies, the resulting ‘new’ MRT was discussed
that lie [s] between the minor but necessary working with key actors in order to validate it. A new study then
hypotheses (...) and the all-inclusive systematic efforts to further refines the MRT and this cyclical process leads
develop a unified theory that will explain all the to accumulation of better insights in how particular
observed uniformities of social behavior, social organiza- interventions work, in which conditions and how
tion and social change” [52] p. 39). In essence, this [33,34].
30 5
In order to be useful in decision-making, the synthesis A final element is the notion of ‘decision space’. This
should present the combinations of attributes required concept was developed by Bossert [64] to describe the
for an intervention to be effective, a presentation of the margins of freedom of health service managers at the
various alternative explanations, an indication of the operational level. His framework analyses how decentra-
potential of transferability by showing the links with lisation policies affect the management practice at
existing knowledge, and an indication of the preliminary operational level. We retained adequate decision spaces
nature of the findings [56]. as a potentially important context factor and a potential
Formulation of our MRT condition for HICOM to be possible.”
We formulated our preliminary MRT on the basis of an It should be noted that there is considerable debate
explorative study at CRH. During that study, intervie- about the outcomes of HRM, and even more about the
wees indicated the importance of trust between health methods to demonstrate these. In general, we would
workers and their management, and the high levels of describe the proximal outcomes of human resource
commitment of staff to the hospital. We also found management in terms of three categories: improved staff
arguments that pointed to the importance of a contin- availability, improved staff attitudes and affects (commit-
gency approach to management of health workers: effec- ment, job satisfaction) and better staff behaviour (in
tive managers implement management practices that terms of higher task performance and organisational
have a good fit with the nature of their workforce, the citizenship behaviour, and lower absenteeism). We
tasks of the organisation and its environment. selected organisational commitment and trust as proxi-
A second source of inspiration was our literature mal outcomes of human resource management, because
review of human resource management and hospital our literature review pointed out that these outputs are
performance, which led us to high commitment man- often found to explain the effect of HICOM.
agement (HICOM). We retained this concept because Combining all these elements with the findings of our
its comprehensive approach to management fitted well first exploration visit, we formulated the MRT as follows:
with our initial analysis. The central attribute of
HICOM is the combination of several complementary “Hospital managers of well-performing hospitals
practices (e.g. good selection of staff, providing training deploy organisational structures that allow decentra-
on a needs basis and individual mentoring) in what is lisation and self-managed teams and stimulate dele-
called ‘bundles’. Through their research in the indus- gation of decision-making, good flows of information
trial, commercial and service sectors, Pfeffer & Veiga and transparency. Their HRM bundles combine
identified a bundle of 7 elements, which they claim is employment security, adequate compensation and
universally valid [57]: providing employment security, training. This results in strong organisational com-
ensuring comparatively high compensation contingent mitment and trust. Conditions include competent
on organisational performance, instituting training and leaders with an explicit vision, relatively large deci-
development, putting in place selective hiring, institut- sion-making spaces and adequate resources.”
ing self-managed teams and decentralisation, reduction
of status differences and information sharing. Organi- Study design and data collection tools
sational commitment was identified as an outcome of As will be clear at this point, we used the case study
such HRM practices [58] and has been shown to con- design as the basis. We collected both qualitative and
tribute to higher organisational performance. Such quantitative data through document review of GHS and
balanced bundles of management practices lead to bet- hospital records and reports, focusing on hospital HRM
ter organisational performance [59-61]. We described policies, and staffing levels and skill mix data.
elsewhere the key elements of high commitment man- In-depth interviews with all 6 members of the hospital
agement in health care organisations [62]. Some of the management team (HMT) explored their management
mechanisms that link HICOM to better performance vision and practices. We based the HRM part of the
include positive psychological links between managers interview guide on the 7 elements set of Pfeffer & Veiga
and staff, organizational commitment and trust. [57](see some questions in Additional file 1). It must be
We drew another element from the work of Cameron noted that the interview guides only served as a guide to
& Quinn on organisational culture [63], which points to structure the interview when necessary, not as a ques-
the importance of the coherence between the vision of tionnaire list that must be applied similarly in all inter-
the managers on their role, the practices they choose to views. In line with the concern that most studies focus
implement, and the perception of their employees of on managers and ignore the perceptions of employees
these practices. Good fit between these would contribute [65], we also explored the perceptions of staff regarding
to better organisational performance. the management approach (the organisational climate).
30 6
In order to cover a wide range of views of different surveys, which in their opinion often lack assessment of
cadres, we made a purposive selection of staff. We iden- the intensity of application and coverage of the HRM
tified the main cadres and within these, we randomly practices. We selected coverage, intensity, internal fit
selected candidates for the interviews. This resulted in and external fit as dimensions. ‘Coverage’ is understood
individual in-depth interviews of 3 nurses, 1 midwife, 1 as the degree to which the elements of the HRM bundle
doctor, 1 radiographer, 1 physiotherapist, 2 laboratory are applied to all cadres. ‘Intensity’ looks at the intensity
technologists, 1 clerical officer and 1 ward assistant. We of application. ‘Internal fit’ examines the synergistic and/
also carried out 3 group discussions with heads of units, or counterbalancing effect of the different elements.
nurses, and paramedical staff (orderlies, clerical officers ‘External fit’ examines the appropriateness of the bundle
and account staff). Opportunistic non-participant obser- for the cadre and organisation in question.
vations were made of management meetings, ward pro- Reporting of findings
cedures and OPD clinics. The preliminary results were discussed with the man-
We also developed a data collection form that focused agement team of the hospital, and the final analysis sub-
on numbers of different cadres of staff (stocks) and on sequently refined. A research report was sent to the
movement of personnel in or out of the hospital (flows commissioner of the study, a policy brief posted on the
in terms of transfer in/out, deceased staff, dismissed web and the findings were presented at the 2008 Geneva
staff, absconded staff, retired staff). Health Forum.
During the preparation phase, a self-assessment of
ethical issues, based on the working paper “Notes Results
regarding ethical guidelines for health services research”, In this section, we present both primary findings and
of the Department of Public Health, Institute of Tropical results from the analysis of the qualitative data in terms
Medicine was done. This covered the following issues: of the management vision, the actual management prac-
Minimal risk to participants; Invitation, information and tices and the organisational climate. These sections cor-
informed consent; Feedback to interviewees and staff. respond with the research questions presented above
We sought and obtained a written informed consent and are drawn from a ‘thick’ description of the case, or
from all interviewees. Measures were taken to safeguard a detailed account of what the interviewees said, what
confidentiality and anonymity. All interviews were we observed and what we learned from our document
recorded and transcribed verbatim. review.
Data analysis The management vision
We used NVivo 2.0 software for data management and A first element we analysed was the views of the man-
analysis. The initial coding was based on a preliminary agement team members of their own role in the hospital
list of codes inspired by the MRT and on additional and on how they should manage the personnel. During
ideas that emerged during the fieldwork. the interviews, the management team members did not
In a second round of analysis, some themes and pat- use words like ‘bundle’ or ‘high commitment manage-
terns emerged (see below). In order to structure these ment’, but they nonetheless expressed a clear view of
as CMO configurations, we found it useful to borrow the hospital’s roles and of how the health workers
categories from theory-driven evaluation [66]. We should be managed accordingly. Key terms include striv-
described the intervention (in this case the HRM prac- ing for excellence, offering services to all, attention for
tices) in terms of content and application, and the their personnel and sound financial management. This
intended and actual outcomes. We drew on our inter- vision is transmitted through what they say during staff
views and observations to differentiate (proclaimed) meetings or write in the mission statement and the
vision (what the team wants), the discourse (what they annual reports.
say) and the actual practices (what they do). We This vision is well shared: not only do the director, the
described the organisational climate, defined as “the financial manager, the nursing manager and the non-
atmosphere that employees perceive is created in their medical administrator maintain the same discourse, also
organisation by practices, procedures and rewards” [67]. interviewees from the operational staff expressed this
In order to indicate how the intervention works, we vision clearly, from nurses to cleaners.
analysed both the context and the intervening mechan-
isms, and attempted to identify the essential conditions. “Their vision is that, they want this place to be a first
To assess the intensity of the implementation of the class hospital. Their aim is to save life, so that is
practices, we developed an analytical framework based their main focus. And whatever they want to do so
on the paper by Richardson & Thompson [59]. These that life is saved, to me is their agenda.” (Non-medi-
authors questioned the research tools used in HRM cal worker, group discussion Non-medical staff)
307
The actual human resource management - what the structure would foster active participation of staff in
management does decisions that affect the hospital.
Based on the analysis of our interviews, observations In this decentralised decision-making structure, we
and collected documents, we found that the actual set found that teamwork is understood as ’working all
of practices at CRH includes more and different ele- together, all engaged, all involved’. In the daily practice
ments than Pfeffer and Veiga [57] listed. These authors of curing and caring, teamwork was most visible at
list of seven elements includes: operational unit level. Deliberate efforts were made to
- putting in place selective hiring include cleaners, sweepers and auxiliary staff in deci-
- providing employment security sion-making.
- ensuring comparatively high compensation contin- The nursing cadre decided to introduce an all-white
gent on organisational performance uniform instead of the colour-coded uniforms. Inter-
- instituting training and development viewed nurses indicated this reduction of status differ-
- deploying self-managed teams and decentralisation ences as an important policy and perceived it as a sign
- reduction of status differences of respect by management. In contrast, reduction of sta-
- information sharing tus differences between the management team and the
We found that selective hiring took place at the start- operational staff seemed not a concern, neither for man-
up of the new hospital in 1998, when the medical and agement, nor for the staff.
para-medical staffs were almost handpicked from the Information sharing was one of the most striking fea-
pool of health workers in the region. At the time of the tures. Formal communication channels were in place at
study (2005), however, the Ghana Health Service (GHS) all levels, including regular unit and ward meetings,
regulations allowed only local recruitment of labourers heads of unit meetings and top management meetings.
and administrative staff. These were complemented by the committees men-
The employment security offered by the GHS to its tioned above. General quarterly meetings (staff “dur-
appointed staff was an often-mentioned reason why bars”), open to all staff, offered a voice even to the
interviewees prefer employment in the GHS rather the hierarchically lowest cadre. Observation showed that
private sector. durbars effectively contributed to low-threshold, two-
At the time of the study, setting compensation levels way communication.
was not within the decision space of the HMT. Only Additional practices
financial incentives for night duties and expatriate doc- We also found that the HMT developed HRM practices
tors could be given. Remuneration was not linked to not included in Pfeffer & Veiga’s set: they made substan-
actual performance. Just prior to the study, health sector tial efforts to ensure good physical working conditions,
strikes led to the Additional Duty Hours Allowance ensured good accessibility of the top managers and
(ADHA) policy, which significantly improved the pur- stressed hands-on involvement of managers and staff
chasing power of the health workers - the ADHA initi- socialisation.
ally constituted a mark-up of 100-250% to the salary of Major attention was given to creating optimal working
a doctor and of lesser proportions for other health conditions. The interviewees pointed to the good com-
workers. munication system in the hospital, the promptness of
Training and personal development was found to be repairs, the general cleanliness of building and com-
an important part of the HRM package. A full-time in- pound, the availability of air conditioning in virtually all
service training coordinator was appointed and a budget rooms and the good amenities for patients. Other ele-
allocated to organise continued medical education activ- ments of the physical environment that were appre-
ities, including clinical meetings, mortality meetings, ciated include the subsidised staff canteen, the internet
seminars and conferences. Staffs were actively stimu- café, the staff bus and the staff library. This points to
lated to follow external courses, even during working the leverage of improving the working conditions. In
hours and personnel from all cadres actually did. Ghana, this may be a management intervention that
We found decentralised decision-making to be a cen- increases not only the effectiveness of health workers,
tral feature. The different units enjoyed a moderate level but also their job satisfaction.
of autonomy in terms of decision-making and objective Top managers are accessible for all staff. As in most
setting. Considerable decision-making authority over a Ghanaian hospitals, we found a clear hierarchy, whereby
number of domains, including the highly sensitive distri- superiors should never be bypassed. Hierarchy was
bution of ADHA funds, was delegated to committees strong in the nursing and administrative cadres. How-
composed of different cadres of staff. The management ever, interviewees mentioned the possibility to see the
team members argued that such decision-making director or nursing manager in person when problems
30 8
could not be solved with their direct supervisor. Our nurses, midwives and doctors by providing adequate
observations showed that staff members of any cadre autonomy to the operational units regarding their daily
effectively made use of this open door policy. activities, while ensuring coordination between these
Management stays involved at the operational level. units. The management is also perceived to provide
Interviewees reported that the nursing managers were effective support, information and resources (see below).
regularly helping out staff in the wards during their As such, the bundle fits well to the task and mission of
twice-daily supervision rounds, while the director was the hospital and to the professional values.
still involved in clinical work. The interviews show that The organisational climate: the management practices as
this was a deliberate management strategy: the top man- perceived by the staff
agement aimed at boosting staff morale by actually Four themes emerged in the analysis of the perceptions
working with them and by leading by example. We also of the operational staff of the HMT’s actions: team
found that the heads of unit steered this process by work, strong perceptions of support by the management
inviting senior managers and heads of other depart- team, recognition and trust. As we will discuss below,
ments to their unit meetings in case of cross-border these themes point to mechanisms that help explain
problems. how the management strategies worked.
At the time of the study,socialisation of staff was a Teamwork stimulates staff from all cadres to be involved in
central element at CRH. Newcomers were given a for- care
mal induction course and rotated for a few weeks The interviews indicated a strongly shared feeling
through different units before being posted to their first among staff members that team work matters: they
station. Both close supervision and peer pressure con- maintain that quality of care can only improve if all
tributed to maintenance of the standards of work. Inter- types of staff are involved.
views show that unit heads would identify staff not
following the procedures and correct such behaviour “In some places, nobody gets close to the Nurse Man-
through tutoring. ager and it is like she only decides what she wants at
Intensity of implementation the place. (...) But here, everybody is important. We
We analysed the actual implementation of the HRM see everybody’s job as important aspect of the health
practices with the framework we presented under Sec- care delivery system, so we include everybody in the
tion ‘Study design and data collection tools’ and which care.” (IO 1, Unit head, Ind. interview)
was based on the paper by Richardson & Thompson
[60]. First, our observations and interviews show that Junior staff members pointed out the ‘free’ relations
the elements of the HRM bundle are applied to all with their superiors.
cadres (good coverage). The intensity of application was
variable. The management team, indeed, adapts its prac- “We are all free in our units. My head always comes
tices in response to emerging priorities. For example, round to see what is going on over here. If something
when confronted with problems of permanence of doc- is not in the right place, he will show you to do this
tors at the emergency department, a custom-made or that. So, always the heads are helping us, so we
incentive package was put in place. This unequal also feel free to work with them.” (Non-medical staff,
approach was not contested because all staff recognised GD Non-medical staff)
the role of doctors in the performance of the hospital.
Second, it seems the management team reached a ’Free relations’ strengthen the collaboration between
good internal fit of the bundle (good degree of synergy operational staff and their heads of units, but also with
between elements of the bundle). There were no prac- the top managers. Interviewees similarly mentioned the
tices that cancelled each other out, except perhaps for easy communication between the middle line staff and
the emphasis on training. This had the unintended effect the HMT.
of enabling staff to leave CRH for better posts. Most
other elements have mutually reinforcing effects: (1) “I would say there is good relationship both formally
information sharing, recognition and participative deci- and informally. We communicate by memos, but as
sion-making; and (2) bottom-up access to management soon as I came, I can just walk straight to Director
and managers getting involved in the wards. and tell him: ‘This is the problem’, and we just
Finally, the external fit of a HRM bundle is the fit of brainstorm to see how the problem can be solved.”
the management practices with the core activities of the (IM5, HMT member, Ind. interview)
hospital (caring and curing) and with the mission of the Perceptions of support by the management team
organisation (providing accessible quality care). The Interviewed staff members often mentioned that they
HRM practices stimulate good professional practice by feel supported by the HMT. First, interviewees
30 9
expressed the feeling that the HMT is effectively solving key event, not only because it helped set standards, but
problems. Unit meetings or ward conferences are a also because of its strong undertone of recognition.
good example of how formal meetings can prevent or
solve coordination problems. “To start with, I can surely say that, the standard
that was set right from the inception of the hospital
“The ward conference is very good. The accountant is has made such a mark. Because immediately when
there, the pharmacist is there, the lab man is there, this hospital was instituted, we were to come for an
everybody is there. The meetings or presentations are interview. So, a high standard was set (...) and they
not for fault finding. We pick issues from there and see if you have the call to work. On that note, in
we make our corrections or cover loopholes.” (Head of coming out to publish the names of those to come
unit 1, GD Unit heads) here, it is like Government releasing a white paper.
By that time, you feel as if you are in heaven. (...)
Informal and non-structured opportunities exist, too, With that alone, that standard was set and every-
and are used to good effect. Interviewees pointed out body was expected to give of his best.” (Nurse, GD
how open relationships and good access to top man- Nurses)
agers allows them to take a problem to the ‘next level’. Perceptions of trust
We explored the issue of trust, which we found to be an
“As a unit head, if I think that something is not going important element in the explorative study, by asking
on well, my demands are not being met, l can staff how they would rate the levels of trust at CRH and
approach the director and we sit down and talk how they believe trust is generated. The interviewees
about it. (...) You are free to enter his office anytime indicated fair levels of trust both amongst staff and
to discuss your problem, especially when you think between management and staff.
things are not going on well” (Head of unit 4, GD
Unit heads) In the whole hospital, there is some trust, but I don’t
think it is 100%. May be it is between 80% - 95%.”
Staff members appreciated not only the possibility to (Unit head 3, GD Unit heads)
discuss work-related problems with their superiors, but
also the attention given by the latter to their profes- Asked how management practices influence the levels
sional development. This also applies to members of the of trust, they pointed out the importance of meetings
hospital management team. during which information is exchanged, the willingness
of managers to discuss decisions and the resulting per-
“He [the director] made every opportunity for my ception of transparency.
career advancement. He is always looking out, listen-
ing and trying to help where he can, to see how he “At least, we have management meetings and after
can help people to progress. So, when you have some- that, management meets the unit heads and tells
one doing that for you, at least you also have to them what the institution wants to do, the pro-
return the same to him.” (IM5, hospital management grammes they have embarked on. They discuss with
team member, Ind. interview) the unit heads and if somebody does not understand
Strong perception of recognition something, management explains it. The unit heads
The interviewees expressed strong feelings of recogni- are supposed to go down and explain to their subor-
tion by the management team. They explained how a dinates. And when we have staff durbars, these things
range of practices, from a word of appreciation to tangi- are also brought up. So, transparency is there, we can
ble rewards expresses the appreciation of the HMT for understand things. Anything they want to do is
their work. explained to workers. (Unit head 3, GD Unit heads)
“At the end of the year, every staff here is given a These consultations and opportunities to discuss
token. Sometimes, something in the form of food, important issues contribute to perceived fairness of the
money, a get-together, occasionally words of motiva- decisions. Interviewees said that less rumour mongering
tion, a tap on your shoulder, meeting you and finding and suspicion arise when people are informed why cer-
out how is it, how is the work going on. This serves as tain measures are implemented and others not.
motivation.” (IO7, Head of unit, Ind. interview)
“At the end of the day, like we had our last year’s
Interestingly, several interviewees mentioned the initial meeting after we presented our reports, management
staff selection, when the hospital was started up, as a too presented their report, their financial report,
310
what they got and how they spent their expenditure management team’s vision. Indeed, in line with their
and those things. So, there will be no room to think vision, the management team motivates the staff
that somebody is cheating on you, or management is through different interventions: remuneration, effective
hiding certain things from us. So, we know what is support and recognition. The HRM practices are rein-
happening, you don’t need to or there is no room for forcing each other (good internal fit). The bundle is well
suspicion. (...) I think it is a fair deal between man- adapted to the different cadres of a healthcare organisa-
agement and staff.” (IO2, Midwife, Ind. interview) tion and its mission (good external fit). It is applied
similarly to all cadres (good coverage). The intensity is
Another source of trust is the effective support staffs variable, but this poses no problems for the staff.
receive from their superiors in case of problems. Realist evaluation improves external validity of a case
study by describing the implementation context. During
“Even the Director himself came here three days ago. the study, we found several potentially important ele-
So, what he said, he has done it. That is why I say I ments in the context of Central Regional Hospital. First,
trust him.” (Non-medical worker, GD Non-medical as testified by the brain drain, Ghana has a well-trained
staff) health workforce from which the GHS (and thus CRH)
“The trust comes from the urgent action taken when can draw personnel. Its medical and paramedical cadres
there is a problem. If there is any problem on disci- display a high degree of professionalism, and there is a
pline for example, an ad-hoc committee is set up and general culture of professionalism in the GHS. Second,
within days, the matter is settled.” (IM4, HMT mem- reasonably good resource availability in terms of hospital
ber, Ind. interview) funding and management capacity allows investing in
the workforce. Indeed, commitment eliciting manage-
Analysis ment practices are costly, especially in management
After categorising, and thus making sense, of the primary time and in terms of training costs.
data in the form of CMO Configurations, a realist evalua- We found that the outcomes of the HR management
tion seeks to examine the link between these findings and bundle at CRH included trust, commitment and strong
the middle range theory it set out to examine. In practice, perceptions of recognition and of support by manage-
we searched for potential causal pathways between the ment, which result in a positive organisational climate.
management practices and the apparent outcomes of com- CMO configurations
mitment and trust. To do so, we summarised the above During the later phases of the analysis, we found that
findings and then searched for CMO configurations. the management practices can be grouped according to
A summary of the intervention and its outcomes their key mechanism and this led to the description of
Our interviews and document review show that the two parallel CMO configurations, each with their own
Hospital Management Team identified good hospital outcome.
performance as the intended distant outcome of its man- The first CMO can be summarised as ‘keeping up
agement practices and a motivated and well-performing standards of excellence through organisational culture’.
workforce as the proximal outcome. As mentioned The hospital had a head start: staff members were
above, the scope of this study did not allow examining selected on professional and motivational grounds by
the association between management practice and hos- the management team. This lengthy selection procedure
pital performance, and we focus on the effect of these gave the staff a feeling of belonging to an elite corps of
practices on organisational commitment and trust, the health professionals and reinforced their professional
proximal outcomes we retained on the basis of our pre- identity. The management team used this opportunity
liminary theory-building. to initiate a culture of high standards of professional
The actual intervention can be summarised as a com- excellence. They set up an induction programme for
bination of HRM practices: socialisation of (new) staff, new staff, and much attention was paid to teamwork
training and personal development, good communica- and supervision. This reflects findings of Schein [68]:
tion and information sharing between different levels of such practices serve as strong embedding mechanisms
the organisation, and decentralised decision making to of the organisational culture. There was equally much
the level of ward and department teams. We also found attention for a clear role distribution and for task moni-
important additional management practices: the creation toring. In summary, both ‘hard’ and ‘soft’ management
of good working conditions, the good accessibility of top practices are balanced in the bundle. The former include
managers, and the active involvement of the manager on general rules and procedures, task distribution for clini-
the work floor. cal and administrative staff and monitoring of task per-
Regarding the process of implementation, we noted a formance; the latter include induction courses, peer
good coherence between the HRM practices and the pressure mechanisms and training/personal development
311
opportunities. All this reinforced the initial capital of Conditions for such management practices to work
professional excellence. Availability of a pool of profes- include competent leaders with an explicit vision, a
sional health workers is an important context element, minimum of resources and conducive institutional
and may be essential for such a bundle to work. arrangements, including effective decentralisation
The second CMO configuration can be summarised as and appropriate decision spaces (although the latter
follows: a hospital management team can attain higher can be narrow for HRM).”
organisational commitment if it strengthens positive
reciprocity relationships that are based on social
exchange, even in hospitals with limited HRM decision Discussion
spaces. Key practices in this set include creating open On the basis of this one study, we cannot yet draw firm
access to managers for all staff and grass-root involve- policy recommendations. Nevertheless, it offers interest-
ment of managers at operational level. This reinforces ing insights in health workforce management and in the
open relationships and contributes to solving opera- use of realist evaluation.
tional problems and conflict resolution. In turn, it sti- Lessons for policy and practice
mulates the feeling of perceived organisational support. First, we found a proof of concept for HICOM in
Eisenberger and colleagues describe this as the beliefs resource-poor health services. Second, our study found
and perceptions of employees regarding the support variant practices compared with the bundle described by
provided and the commitment demonstrated by the Pfeffer and Veiga, which supports the findings of
organisation in their staff [69]. Employees interpret deci- Richardson & Thompson [60] and Marchington & Gru-
sions and actions of their managers and their trust- gulis [71]. Third, this case reinforces the point that in
worthiness in terms of the commitment of managers to management of health workers, we need to apply coher-
their staff. At CRH, the leadership and management ent bundles of practices, and not focus on singular
style is indeed perceived to be effective (in meeting its interventions. In HRM, the quality of management prac-
promises and in ensuring adequate physical working tices counts more than the quantity. It is not the actual
conditions) and supportive, even on the personal level. number of practices, but rather the process by which
Ultimately, such practices stimulate reciprocity and as a these practices are put in place that is related with posi-
result, organisational commitment. This in turn contri- tive staff attitudes like commitment, job satisfaction and
butes to organisational performance [70]. procedural justice [65]. This is in line with conclusions
Availability of well-trained health workers and ade- of other studies in other sectors [26,72].
quate funding seem intuitively to be essential context Regarding the mechanisms, our findings relate to the
elements in both CMO configurations. analysis of Evans & Davis [73], who situate the underly-
The new MRT ing mechanisms of high commitment management at
Our analysis identified two CMOCs that indicate causal the level of the internal social structure of the organisa-
pathways between sets of HRM practices and HRM out- tion. Such practices improve knowledge, skills and abil-
comes, and we modified the MRT accordingly: ities, but they exert also major effects at the level of
relationships. Weak ties are strengthened [74], recipro-
“The management of a well-performing hospital city is established and maintained [75] and shared men-
deploys organisational structures that allow decen- tal models contribute to a strong organisational culture.
tralisation and self-managed teams and stimulates This in turn affects behaviour of staff and improves
delegation of decision-making, good flows of infor- organisational efficiency and flexibility, and ultimately,
mation and transparency. organisational performance. The evidence of the impact
In the management of health workers, they imple- of such reciprocity relations or of organisational com-
ment a balanced bundle of management practices mitment on organisational performance is not strong,
that includes both clear goal setting, role distribution and further research should investigate whether and
and task monitoring (hard HRM) and training, sup- how high commitment leads to better performance in
port and recognition (soft HRM). healthcare organisations.
Based on the mechanism of perceived organisational We found that the decision spaces managers require
support and reciprocity, such combinations lead to a to develop a responsive HRM approach may be smaller
positive organisational climate that includes recogni- than is often thought. At the time of study, the decision
tion, respect, commitment and trust. If these are spaces of regional hospital managers in Ghana were
taken up into the organisational culture and newco- quite limited concerning HRM. As important as the for-
mers are inducted into the OC, enduring effects of mal decision space is its actual utilisation. At CRH, the
such practices can be expected. team exploited its decision spaces well to create its own
312
way of management within the defined institutional What realist evaluation can do is to stimulate the
arrangements of a ‘regional hospital’ (e.g. by using com- researcher to describe a detailed picture of the causal
mittees and delegation of decision-making power). web that includes the multiple determinants and to
Finally, a balanced management approach is costly, categorise these as intervention, underlying mechanism
especially in management time. It requires reasonable or essential context factor. In our case, we have argu-
financial resources and a management capability to deal ments to say that both commitment-eliciting manage-
not only with administration but also with the less tan- ment and personnel administration are required, but we
gible issues of relationships, organisational culture and cannot (yet) indicate which among these two sets is the
motivation of staff. most important in which setting.
Future research should establish what other HRM The conclusion may be that one needs to accept that
approaches lead to high commitment, under which con- the kind of evidence provided by realist evaluation can
ditions HICOM works, and how it can be stimulated. never be put in the same categories of evidence pro-
This last question deserves attention. Health services in duced by controlled experimental methods, not only
many LMIC are both ill equipped and not sufficiently because of its perspective on causality, but also because
supported to implement a HRM approach that differs of the complexity of the subjects on which it will be
from a mere administrative approach. In the first place, applied.
the managers of health services are mostly medical doc- The MRT fallacy
tors. Human resource management is not an element of While any researcher adopts specific reference frame-
the medical education curriculum. Even if they received works during her research, realist evaluation asks
additional public health or management training, the researchers to make these frameworks explicit in the
curriculum mostly equates HRM to personnel adminis- form of a MRT. This implies a risk of developing a tun-
tration and this hardly prepares future health service nel vision: the researcher may remain blind for the
managers for responsive management. unexpected factors and alternative explanations. This
Methodological lessons risk can be reduced by the plausibility check during the
In this case study, we used a realist evaluation approach development of the initial MRT, triangulation of find-
because we consider health care organisations to be ings, analysis by multiple researchers and discussion
essentially social entities. Pawson argues that realist eva- with stakeholders and peers.
luation is well suited to investigate change in such social The MRT fallacy also operates at the stage of analysis
systems [34]. Its focus on the generative causality that and of dissemination. During analysis, we did several
underlies interventions, stimulates the analysis of how rounds of plausibility checks, because we kept finding
the intervention works and in which context conditions. alternative explanations in disciplines such as organisa-
This results in more detailed conclusions that indicate tional psychology, organisational theory and sociology.
how the intervention was carried out, which effect it The CMOCs and resulting MRTs are indeed most often
had and how it worked. It also offers insights in the just one way of explaining the findings. A middle-range
context elements. Such theory building helps to over- theory can indeed never cover all possible explanations
come the limits of traditional case studies, and specifi- of change [34]. In Pawson and Tilley’s view, a realist
cally their low external validity and low power to evaluator does not strive at nor pretend to provide the
explain change [42]. However, appealing as it is, realist ultimate evidence that the intervention works. Rather,
evaluation poses a number of challenges for the she aims at enlightening the decision-maker, a process
researcher. of utilisation of research that may be the most frequent
The attribution paradox in case of social science [78]. In such cases, a pragmatic
Perhaps the most critical issue is the attribution para- position should be taken, whereby one tries to refine the
dox. Because of its ontological and epistemological middle range theory as much as practically possible,
basis, realist evaluation is quite fit to assess complexity with the explicit aim of providing options for improve-
[76,77] and may contribute most in research of exactly ment rather than reaching a perfect understanding of
such topics. However, research of complex problems the intervention as such [56,79].
needs to confront multi-causality. In complex systems, The CMO dilemma
the behaviour of people and organisations alike is deter- As we mentioned, the CMO configuration is a powerful
mined by many interlinked factors. Health professionals model to go beyond the classic case study, as it forces
act under influence of their professional norms, social the researcher to go beyond description. However, a
pressure, management interventions, and not least, their true application of realist evaluation requires not only a
intrinsic motivation. Assessing the exact contribution of systematic description of the intervention in terms of
a set of management practices to overall organisational intervention, outcome, context and mechanisms. Also
performance may therefore be virtually impossible. the generative causal relationships between these
313
elements need to be assessed. In our analysis, this Conclusions

proved difficult at several levels. Realistic evaluation offers a comprehensive approach to
The first important hurdle is the differentiation of the assessment of interventions in complex situations that
effect of the context from that of the intervention. This can go beyond the simple efficacy question. We devel-
feeds the attribution paradox: is the outcome the result oped a realist case study that unravelled the manage-
of the intervention - and to which degree - or are there ment practices put in place by a hospital management
context elements that explain the change in outcome - team in Ghana. This study shows that it is possible to
and to which degree? Furthermore, some context ele- implement high commitment management practices in
ments can be expected to moderate the relation between LMIC and that these are perceived to be relevant by the
intervention and outcome, and in some cases, the out- health workers. We found that through a well-balanced
come of an intervention will influence its context (initi- bundle of HRM practices, management teams can sti-
ating feedback). Regarding our case, probably more mulate organisational commitment and an organisa-
attention needs to be given to the role of professional- tional culture of excellence. At CRH, the HRM bundle
ism. Professional values can steer providers’ behaviour included sound administrative management. Reciprocity
to an important degree and could partially explain the and perceived organisational support emerged as an
behaviour of certain staff, irrespectively of the manage- important underlying mechanism. In applying the realist
ment strategies. Most likely, we may find that the methodology, we also encountered a number of pitfalls
observed management strategies are in a close fit with and paradoxes. Only through further practical applica-
professional behaviour traditions. tions will we find out how these can be overcome.
Secondly, the realist researcher seeks to describe the
mechanism that is triggered by the intervention and that Additional file 1: Examples of interview questions from the interview
leads to the outcome. Confusion may result from the guide
Click here for file
finding that some context elements are essential for the [ http://www.biomedcentral.com/content/supplementary/1472-6963-10-
outcome: is this context element then part of the 24-S1.DOC ]
mechanism? We clarified this issue by considering con-
text elements as actors or factors that are external to
the intervention - that are present or occurring even if
Acknowledgements
the intervention does not lead to an outcome -, but We would like to thank all the staff of Central Regional Hospital (Ghana) for
which nevertheless may have an influence on the out- their warm welcome and collaboration during this study.
This work was funded through the framework agreement (Raamakkoord)
come. The mechanism is the causal pathway that
between the Belgian Directorate-General for Development Cooperation and
explains how the intervention leads to an observed out- the Institute of Tropical Medicine, Antwerp. The sponsors had no role in the
come in a particular context. In other words, the inter- study or in the writing of the paper.
vention leads to an outcome in specific contexts if it
Author details
triggers certain mechanisms. If the mechanism is found 1
Department of Public Health, Institute of Tropical Medicine-Antwerp,
to be context-dependent, which in health services may Nationalestraat 155, B-2000 Antwerp, Belgium. 2Volta Regional Health
Directorate, PO Box HP 72, Ho, Volta Region, Ghana. 3Department of Public
often be the case, essential context elements can be
Health, Institute of Tropical Medicine-Antwerp, Nationalestraat 155, B-2000
identified. In our case, the professionalism of the staff Antwerp, Belgium.
selected to work at CRH is a context element, the deci-
Authors’ contributions
sion to introduce an induction training was a manage-
All three authors contributed to the original design and analysis. BM and
ment decision, and the effect of building an MD carried out the data collection. BM and MD analysed the data. BM, MD
organisational culture was a mechanism. and GK contributed to the discussion section and to writing the manuscript.
BM edited the final draft. All authors read and approved the final
The efficiency question
manuscript.
By its very nature, RE may yield information that is par-
ticularly useful for policymakers. However, by its same Competing interests
The authors declare that they have no competing interests.
nature, a RE needs considerable expertise and ample
time and resources, because of its comprehensive scope. Received: 15 July 2009
Indeed, besides the efficacy/outcome evaluation, also the Accepted: 25 January 2010 Published: 25 January 2010
underlying theory and the context must be accounted
References
for [41,80]. In our case, work at CRH started in 2004
1. WHO: World Health Report 2006. Working together for health Geneva: World
with an exploratory visit, and much analysis went on Health Organisation 2006.
after the second visit in 2005 and the third visit in 2007. 2. Dal Poz MR, Quain EE, O’Neil M, McCaffery J, Elzinga G, Martineau T:
Addressing the health workforce crisis: towards a common approach.
Such timelines may still be acceptable in case of non-
Hum Resour Health 2006, 4:21.
urgent issues, but far less in case of high-interest policy 3. The Lancet: The crisis in human resources for health. Lancet 2006,
issues. 367(9517):1117.
314
4. Hongoro C, McPake B: How to bridge the gap in human resources for 32. Weiss C: What kind of evidence in Evidence-Based policy?. Third
health. Lancet 2004, 364(9443):1451-1456. International, Inter-disciplinary Evidence-Based Policies and Indicator Systems
5. JLI: Human Resources for Health. Overcoming the crisis. Joint Learning Conference, July 2001 Durham. CEM Centre, University of Durham 2001.
Initiative 2004, 1-207. 33. Pawson R, Tilley N: Realistic Evaluation London: Sage 1997.
6. Oxman A, Fretheim A: An overview of research on the effects of results- 34. Pawson R: Evidence-based policy. A realist perspective London: Sage 2006.
based financing. Report from Norwegian Knowledge Centre for the Health 35. Harrison S: Policy analysis. Studying the Organisation and Delivery of Health
Services Oslo: Norwegian Knowledge Centre for the Health Services 2008, Services: Research Methods London: RoutledgeFulop N, Allen P, Clarke A,
16. Black N 2001, 90-106.
7. Dieleman M, Gerretsen B, Wilt van der GJ: Human resource management 36. Julnes G, Mark M, Henry G: Promoting realism in evaluation. Evaluation
interventions to improve health workers’ performance in low and 1998, 4(4):483-504.
middle income countries: a realist review. Health Res Policy Syst 2009, 7:7. 37. Leone L: Realistic evaluation of an illicit drug deterrence programme.
8. Chopra M, Munro S, Lavis JN, Vist G, Bennett S: Effects of policy options Evaluation 2008, 14(1):19-28.
for human resources for health: an analysis of systematic reviews. Lancet 38. Koenig G: Realistic evaluation and case studies: stretching the potential.
2008, 371(9613):668-674. Evaluation 2009, 15(1):9-30.
9. Rowe AK, de Savigny D, Lanata CF, Victora CG: How can we achieve and 39. Davis P: The limits of realistic evaluation. Evaluation 2005, 11(3):275-295.
maintain high-quality performance of health workers in low-resource 40. Byng R: Using realistic evaluation to evaluate a practice-level
settings?. Lancet 2005, 366(9490):1026-1035. intervention to improve primary healthcare for patients with long-term
10. Wyss K, Prytherch H, Merkle R, Weiss S, Vogel T: Contributions to solving mental illness. Evaluation 2005, 11(1):69-93.
the human resource crisis in developing countries - with special 41. Blamey A, Mackenzie M: Theories of change and realistic evaluation. Peas
reference to German development cooperation. Swiss Centre for in a pod or apples and oranges?. Evaluation 2007, 13:439-455.
International Health. Basel: Swiss Tropical Institute 2006. 42. Costner H: The validity of conclusions in evaluation research: a further
11. Hammer J, Jack W: Designing incentives for rural health care providers in development of Chen and Rossi’s theory-driven approach. Evaluation and
developing countries. Washington: World Bank 2001. Program Planning 1989, 12:345-353.
12. Gilson L, Erasmus E: Supporting the retention of HRH: SADC policy 43. Chen H-T: The theory-driven perspective. Evaluation and Program Planning
context. Johannesburg: Equinet 2005. 1989, 12:297.
13. Jack W: Contracting for health services: an evaluation of recent reforms 44. Hallett M, Rogers R: The push for ‘Truth in sentencing’: evaluating
in Nicaragua. Health Policy and Planning 2003, 18(2):195-204. competing stakeholder constructions: the case for contextualism in
14. Soeters R, Griffiths F: Improving government health services through evaluation research. Evaluation and Program Planning 1994, 17:187-196.
contract management: a case from Cambodia. Health Policy Plan 2003, 45. Byng R: Using the ‘Realistic evaluation’ framework to make a
18(1):74-83. retrospective qualitative evaluation of a practice level intervention to
15. Meessen B, Kashala J, Musango L: Output-based payment to boost improve primary care for patients with long-term mental illness. The
productivity in public health centres: contracting in Kabutare district, 2002 EES Conference Three movements in Contemporary Evaluation: Learning,
Rwanda. Bulletin of the World Health Organisation 2006, 85(2):108-115. Theory and Evidence October 10-12: 2002 European Evaluation Society 2002.
16. Soeters R, Habineza C, Peerenboom PB: Performance-based financing and 46. Blaise P, Kegels G: A realistic approach to the evaluation of the quality
changing the district health system: experience from Rwanda. Bull World management movement in health care systems: a comparison between
Health Organ 2006, 84(11):884-889. European and African contexts based on Mintzberg’s organizational
17. Furth R: Zambia Pilot study of performance-based incentives. Quality models. Int J Health Plann Manage 2004, 19(4):337-364.
Assurance Project 2006. 47. Cole G: Advancing the development and application of theory-based
18. Mills A, Bennett S, Russell S: The challenge of health sector reform. What evaluation in the practice of public health. American Journal of Evaluation
must governments do? New York: Palgrave 2001. 1999, 20:453-470.
19. Eldridge C, Palmer N: Performance-based payment: some reflections on 48. Chen H-T, Rossi P: Issues in the theory-driven perspective. Evaluation and
the discourse, evidence and unanswered questions. Health Policy Plan Program Planning 1989, 12(4):299-306.
2009, 24(3):160-166. 49. Takeuchi R, Chen G, Lepak D: Through the looking glass of a social
20. Meessen B, Musango L, Kashala JP, Lemlin J: Reviewing institutions of rural system: cross-level effects of high-performance work systems on
health centres: the Performance Initiative in Butare, Rwanda. Trop Med employees’ attitudes. Personnel Psychology 2009, 62:1-29.
Int Health 2006, 11(8):1303-1317. 50. Donaldson S: Program theory-driven evaluation science. Strategies and
21. Murphy E: Micro-level qualitative research. Studying the Organisation and applications New York: Lawrence Erlbaum Associates 2007.
Delivery of Health Services: Research Methods London: RoutledgeFulop N, 51. Chen H-T: Practical program evaluation Thousand Oaks: SAGE Publications
Allen P, Clarke A, Black N 2001, 40-55. 2005.
22. Yin R: Case study research. Design and methods London: Sage Publications, 52. Merton RK: Social theory and social structure New York: The Free Press 1968.
Third 2003. 53. Lipsey M, Pollard J: Driving toward theory in program evaluation: more
23. Denscombe M: The good research guide for small-scale social research models to choose from. Evaluation and Program Planning 1989,
projects Maidenhead: Open University Press, 2 2003. 12:317-328.
24. Fitzgerald L: Case studies as a research tool. Qual Health Care 1999, 54. Rosas S: Concept Mapping as a Technique for Program Theory
8(2):75. Development: An Illustration Using Family Support Programs. American
25. Ragin CC: The distinctiveness of case-oriented research. Health Serv Res Journal of Evaluation 2005, 26:389-401.
1999, 34(5 Pt 2):1137-1151. 55. Barnes M, Matka E, Sullivan H: Evidence, understanding and complexity.
26. Truss C, Gratton L, Hope-Hailey V, McGovern P, Stiles P: Soft and hard Evaluation in non-linear systems. Evaluation 2003, 9(3):265-284.
models of human resource management: a reappraisal. Journal of 56. Pawson R, Tilley N: Realist Evaluation. DPRN Thematic Meeting 2006 Report
Management Studies 1997, 34(1):53-73. on Evaluation Development Policy Review Network 2006.
27. Hartley J: Case studies in organisational research. Qualitative methods in 57. Pfeffer J, Veiga J: Putting people first for organisational success. The
organizational research London: Sage PublicationsCassell C, Symon G 1994. Academy of Management Executive 1999, 13(2):37-48.
28. Campbell D, Stanley J: Experimental And Quasi-experimental Designs For 58. Whitener E: Do “high commitment” human resource practices affect
Research Skokie Illinois: Rand-McNally 1963. employee commitment? A cross-level analysis using hierarchical linear
29. Ferlie E: Organisational studies. Studying the Organisation and Delivery of modeling. Journal of Management 2001, 27(5):515.
Health Services: Research Methods London: RoutledgeFulop N, Allen P, Clarke 59. Richardson R, Thompson M: The impact of people management practices
A, Black N 2001, 25-39. on business performance: a literature review. London: Institute of
30. Chen H-T, Rossi P: The theory-driven approach to validity. Evaluation and Personnel and Development 1999.
Program Planning 1987, 10:95-103. 60. Porter L, Steers R, Mowday R, Boulina P: Organizational commitment, job
31. Anonymous: The origins of theory-based evaluation. Evaluation Practice satisfaction, and turn-over among psychiatric technicians. Journal of
1996, 17(2):169-171. Applied Psychology 1974, 59:603-609.
315
61. Steers RM: Antecedents and outcomes of organisational commitment.

Administrative Science Quarterly 1977, 22:46-56.
62. Marchal B, Kegels G: Focusing on the software of managing health
workers: What can we learn from high commitment management
practices?. International Journal of Health Planning and Management 2008,
23(4):299-311.
63. Cameron KS, Quinn R: Diagnosing and changing organizational culture San
Francisco: Jossey-Bass 2006.
64. Bossert T: Methodological guidelines for applied research on
decentralisation of health systems in Latin America. LAC HSR Health
sector reform initiative 2000.
65. Edgar F, Geare A: HRM practice and employee attitudes: different
measures - different results. Personnel Review 2005, 34(5):534-549.
66. Chen H-T: Theory-driven evaluations Newbury Park, California: Sage
Publications, 1 1990.
67. Schneider B, Gunnarson S, Niles-Jolly K: Creating the climate and culture
of success. Organisational Dynamics 2004, 23(1):17-29.
68. Schein E: Organizational culture. American Psychologist 1990, 45(2):109-119.
69. Eisenberger R, Fasolo P, Davis-laMastro V: Perceived organisational support
and employee diligence, commitment and innovation. Journal of Applied
Psychology 1990, 75:51-59.
70. Podsakoff P, MacKenzie S, Beth Paine J, Bachrach D: Organisational
citizenship behaviors: a critical review of the theoretical and empirical
literature and suggestions for future research. Journal of Management
2000, 26(3):513-563.
71. Marchington M, Grugulis I: “Best practice” human resource management:
perfect opportunity of dangerous illusion?. International Journal of Human
Resource Management 2000, 11(6):1104-1124.
72. Walton R: From control to commitment in the workplace. Harv Bus Rev
1985, (March-April), DOI: 10.1225/85219.
73. Evans W, Davis W: High-performance work systems and organisational
performance: the mediating role of internal social structure. Journal of
Management 2005, 31(5):758-775.
74. Granovettter M: The strength of weak ties: a network theory revisited.
Sociological Theory 1983, 1:201-233.
75. Eisenberger R, Huntington R, Hutchison S, Sowa D: Perceived
organisational support. Journal of Applied Psychology 1986, 71(3):500-507.
76. Sanderson I: Is it ‘What works’ that matters? Evaluation and evidence-
based policy making. Scott Policy Seminar - Economic Research Institute
of Northern Ireland, Undated.
77. Stame N: Theory-based evaluation and types of complexity. Evaluation
2004, 10(1):58-76.
78. Weiss C: The many meanings of research utilization. Public Administration
Review 1979, 39(5):426-431.
79. Knaap Van Der P: Theory-based evaluation and learning: possibilities and
challenges. Evaluation 2004, 10(1):16-34.
80. Pedersen L, Rieper O: Is realist evaluation a realistic approach for
complex reforms?. Evaluation 2008, 14(3):271-293.
Pre-publication history
The pre-publication history for this paper can be accessed here:http://www.
biomedcentral.com/1472-6963/10/24/prepub
doi:10.1186/1472-6963-10-24
Cite this article as: Marchal et al.: A realist evaluation of the
management of a well- performing regional hospital in Ghana. BMC
Health Services Research 2010 10:24.
Submit your next manuscript to BioMed Central

and take full advantage of:
• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at

www.biomedcentral.com/submit
316
5. Investigating policy India) clearly demonstrate how a range of different
decisions and interventions, taken at different times and
and system change sometimes with unexpected consequences, accumulate
over time over time to shape the current state and performance of
health systems. At a household level, meanwhile,
longitudinal work allows for the assessment of the
Lucy ilson impacts on livelihoods over time of, for example, health
seeking behaviour and the associated cost burdens.
reat ritain and orthern reland But how can change over time be tracked and investi-
A considerable body of HPSR work focuses on experience gated? The range of possible approaches include
at one point in time (see Part 4: Cross-sectional pers- prospective tracking of events, or phenomena, over time
pectives) and studies investigating (describing and and retrospective analysis of past events and experiences.
explaining) change over time are more rarely conducted. Historical research, for example, “is unusual in ... asking
Yet health policy change and health system development, big questions and in dealing with change” (Berridge,
around which many HPSR questions revolve, are pro- 2001:141) and these include “Why and how do we have
cesses that occur over time. Therefore the contextual our current health systems? How and why do they differ
influences over health policy and system experience are from the past?” or “How and for what reasons have
commonly recognized to include historical factors. Health different health professions established their areas of
systems never stop developing or evolving and past competence, and how have boundaries been establi-
experience influences current development – perhaps by shed?” (Berridge, 2001:141–2). Drawing on documen-
limiting or opening possibilities of future change. Indeed, tary, quantitative and oral sources of data, historical
‘path dependency’ is a notion widely applied in institu- work involves interpretive analysis of past experiences
tional analysis that suggests that what happened in the and seeks to open up debates rather than to draw direct
past directly influences, and limits, the possibilities of lessons. In contrast, fixed longitudinal study designs
institutional change today (North, 1998). Policy analysis involve repeated measures on the same variables for the
theory, meanwhile, recognizes that policy change is a same group, or groups, on an extended series of
dynamic process evolving over considerable periods of occasions and may support prospective analysis of trends
time. For example, punctuated equilibrium theory seeks over time (Robson, 2002).
to explain how and why policy processes are charac-
terized by largely incremental change for long periods of Rigour in studies of the
time, remaining fairly stable, but occasionally producing dynamics of policy change
large-scale departures from this pattern of change (True,
Jones & Baumgartner, 2007).
over time
Longitudinal perspectives are also particularly important The criteria for assessing the rigour and quality of studies
in understanding the complex causality embedded in examining the dynamics of policy and system change
processes of health policy and health systems change. At over time will vary with the disciplinary perspective or
a system level, for example, a recently published volume research strategy adopted and must be appropriate for
(Balabanova, McKee and Mills, 2011) demonstrates the the particular discipline and strategy (see also Part 2, and
value of taking a long-term perspective in examining the sections in Part 4 relating to the case-study approach
health system development. The country experiences and advances in impact evaluation).
presented (for example from Thailand, Tamil Nadu and

Balabanova D, McKee M and Mills A, eds (2011) ood The papers in this section were chosen to illustrate some
health at low cost years on. hat makes a successful of the different approaches that can be used to investi-
health system London, London School of Hygiene and gate change over time in health policy and systems
Tropical Medicine. experience.
Berridge V (2001). Historical research. In: Fulop N et al., Brown, Cueto & Fee (2006) address the changing
n
role of the World Health Organization over time.
eds. Studying the organisation and delivery of health
Using an historical approach based on documentary
services: research methods. London, Routledge:140–153. review, they argue that over time and in response to
larger political and historical processes, the World
North D (1998). Where have we been and where are we
Health Organization has sought to reconstruct itself
going? In: Ben-Ner A & Putterman L, eds. conomics, as the coordinator of global health initiatives, rather
values and organi ations. Cambridge, Cambridge than being the undisputed leader in international
University Press:491–508. health.
Crichton (2008) traces the experience over time of a
n
Robson C (2002). Real world research: a resource for
particular Kenyan health policy, using the theoretical
lens of policy analysis and what is in essence a
Oxford, Blackwell Publishing. process tracing approach.
Russell S (2005). Illuminating cases: understanding the Masanja et al. (2008) use statistical trends, based on
n
economic burden of illness through case study household epidemiological data, to support consideration of the
performance of the Tanzanian health system and how
research. Health Policy and Planning, 20(5):277–289.
and why its development has impacted positively on
True JL, Jones BD, Baumgartner FR (2007). Punctuated child survival.
equilibrium theory: explaining stability and change in Van Ginnekan, Lewin & Berridge (2010) use an
n
public policymaking. In: Sabatier P, ed. heories of the historical approach to examine the evolution over
time of the South African community health worker
policy process, 2nd ed. Cambridge MA, Westview
programme, drawing on data collected through oral
Press:155–188.
histories and witness seminars.
See also:
Russell & Gilson (2006) in the case-study approach
section which reports on prospective studies of Sri
Lankan case-study households in which change over
time in household livelihoods was tracked and analysed,
showing how these impacts were affected by the costs
associated with seeking health care.
Wang et al. (2009) in the advances in impact evaluation

section which reports a before and after, with control
group, evaluation of the impact on health status of a
community-based insurance scheme in China.
Brown TM, Cueto M, Fee E (2006). The World Health
Organization and the transition from ‘international’ to
‘global’ public health. merican ournal of Public Health.
January; 96(1):62–72.
http://dx.doi.org/10.2105/AJPH.2004.050831
Reproduced by permission of the American Journal
n
of Public Health. Copyright American Journal of
Public Health, 2006.
Crichton J (2008). Changing fortunes: analysis of

fluctuating policy space for family planning in Kenya.
http://dx.doi.org/10.1093/heapol/czn020
Reproduced by permission of the Oxford University
n
Masanja H et al. (2008).Child survival gains in Tanzania:

analysis of data from demographic and health surveys.
he Lancet. 371:1276–83.
http://dx.doi.org/10.1016/S0140-6736(08)60562-0
n
Elsevier, 2008.
Van Ginneken N, Lewin S, Berridge S. (2010). The

emergence of community health worker programmes in
the late apartheid era in South Africa: an historical
analysis. Social Science Medicine, 71:1110–1118.
n
Elsevier, 2010.

 PUBLIC HEALTH THEN AND NOW 
The World Health Organization and the Transition From

“International”to
“Global” Public Health
| Theodore M. Brown, PhD, Marcos Cueto, PhD, and Elizabeth Fee, PhD
The term “global health” is rapidly
replacing the older terminology of “in-
ternational health.” We describe the EVEN A QUICK GLANCE AT THE provide a critical analysis of the intergovernmental organizations
role of the World Health Organization titles of books and articles in re- meaning, emergence, and signifi- and agencies—for example, the
(WHO) in both international and global cent medical and public health cance of the term “global health” media, internationally influential
literature suggests that an impor- and to place its growing popular- foundations, nongovernmental
health and in the transition from one to
tant transition is under way. The ity in a broader historical context. organizations, and transnational
the other. We suggest that the term
terms “global,” “globalization,” In particular, we focus on the role corporations. Logically, the terms
“global health” emerged as part of
and their variants are every- of the World Health Organization “international,” “intergovernmen-
larger political and historical processes, where, and in the specific context (WHO) in both international and tal,” and “global” need not be mu-
in which WHO found its dominant role of international public health, global health and as an agent in tually exclusive and in fact can be
challenged and began to reposition “global” seems to be emerging as the transition from one concept understood as complementary.
itself within a shifting set of power the preferred authoritative term.1 to the other. Thus, we could say that WHO is
alliances. As one indicator, the number of Let us first define and differen- an intergovernmental agency that
Between 1948 and 1998, WHO entries in PubMed under the tiate some essential terms. “Inter- exercises international functions
moved from being the unquestioned rubrics “global health” and “international health” was already a with the goal of improving global
leader of international health to being national health” shows that term of considerable currency in health.
an organization in crisis, facing budget “global health” is rapidly on the the late 19th and early 20th cen- Given these definitions, it
rise, seemingly on track to over- tury, when it referred primarily to should come as no surprise that
shortfalls and diminished status,
take “international health” in the a focus on the control of epi- global health is not entirely an in-
especially given the growing influence
near future (Table 1). Although demics across the boundaries vention of the past few years. The
of new and powerful players. We argue
universities, government agencies, between nations (i.e., “interna- term “global” was sometimes used
that WHO began to refashion itself as and private philanthropies are all tional”). “Intergovernmental” well before the 1990s, as in the
the coordinator, strategic planner, and using the term in highly visible refers to the relationships be- “global malaria eradication pro-
leader of global health initiatives as a ways,2 the origin and meaning of tween the governments of sover- gram” launched by WHO in the
strategy of survival in response to this the term “global health” are still eign nations—in this case, with re- mid-1950s; a WHO Public Affairs
transformed international political unclear. gard to the policies and practices Committee pamphlet of 1958,
context. (Am J Public Health. 2006;96: We provide historical insight of public health. “Global health,” The World Health Organization: Its
62–72. doi:10.2105/AJPH.2004.050831) into the emergence of the termi- in general, implies consideration Global Battle Against Disease3; a
nology of global health. We be- of the health needs of the people 1971 report for the US House of
lieve that an examination of this of the whole planet above the Representatives entitled The Poli-
linguistic shift will yield important concerns of particular nations. tics of Global Health4; and many
fruit, and not just information The term “global” is also associ- studies of the “global population
about fashions and fads in lan- ated with the growing importance problem” in the 1970s.5 But the
guage use. Our task here is to of actors beyond governmental or term was generally limited and its
62 | Public Health Then and Now | Peer Reviewed | Brown et al. American Journal of Public Health | January 2006, Vol 96, No. 1
320
use in official statements and doc- dual aspect, one both promising hamstrung by financial constraints
uments sporadic at best. Now and threatening. and internal incompetencies, frus-
there is an increasing frequency In one respect, there was eas- trated by turf wars and cross-
of references to global health.6 Yet ier diffusion of useful technolo- national policies.”9 Given these
the questions remain: How many gies and of ideas and values such -realities, Yach and Bettcher’s pro-
have participated in this shift in as human rights. In another, there motion of “global public health”
terminology? Do they consider it were such risks as diminished so- while they were affiliated with
trendy, trivial, or trenchant? cial safety nets; the facilitated WHO was, to say the least, in-
Supinda Bunyavanich and marketing of tobacco, alcohol, triguing. Why were these spokes-
Ruth B. Walkup tried to answer and psychoactive drugs; the eas- men for the much-criticized and
these questions and published, ier worldwide spread of infec- apparently hobbled WHO so up-
under the provocative title “US tious diseases; and the rapid beat about “global” public health?
Public Health Leaders Shift To- degradation of the environment,
ward a New Paradigm of Global with dangerous public health THE WORLD HEALTH “War on the Malaria Mosquito!”
Health,” their report of conversa- consequences. But Yach and ORGANIZATION Poster produced by the Division of
tions conducted in 1999 with 29 Bettcher were convinced that Public Information, World Health
“international health leaders.”7 WHO could turn these risks into The Early Years Organization, Geneva, 1958.
Courtesy of the World Health
Their respondents fell into 2 opportunities. WHO, they ar- To better understand Yach and
Organization. Source: Prints and
groups. About half felt that there gued, could help create more effi- Bettcher’s role, and that of WHO Photographs Collection of the
was no need for a new terminol- cient information and surveil- National Library of Medicine.
ogy and that the label “global lance systems by strengthening its
health” was meaningless jargon. global monitoring and alert sys-
The other half thought that there tems, thus creating “global early
were profound differences be- warning systems.” They believed
tween international health and that even the most powerful na-
global health and that “global” tions would buy into this new
clearly meant something transna- globally interdependent world
tional. Although these respon- system once these nations real-
dents believed that a major shift ized that such involvement was in
had occurred within the previous their best interest.
few years, they seemed unable Despite the long list of prob-
clearly to articulate or define it. lems and threats, Yach and
In 1998, Derek Yach and Dou- Bettcher were largely uncritical as
glas Bettcher came closer to cap- they promoted the virtues of
turing both the essence and the global public health and the lead-
origin of the new global health in ership role of WHO. In an edito-
a 2-part article on “The Global- rial in the same issue of the Jour-
ization of Public Health” in the nal, George Silver noted that Yach
American Journal of Public and Bettcher worked for WHO
Health.8 They defined the “new and that their position was similar
paradigm” of globalization as “the to other optimistic stances taken
process of increasing economic, by WHO officials and advocates.
political, and social interdepend- But WHO, Silver pointed out, was
ence and integration as capital, actually in a bad way: “The
goods, persons, concepts, images, WHO’s leadership role has passed
ideas and values cross state to the far wealthier and more in-
boundaries.” The roots of global- fluential World Bank, and the
ization were long, they said, going WHO’s mission has been dis-
back at least to the 19th century, persed among other UN agen-
but the process was assuming a cies.” Wealthy donor countries
new magnitude in the late 20th were billions of dollars in arrears,
century. The globalization of pub- and this left the United Nations
lic health, they argued, had a and its agencies in “disarray,
January 2006, Vol 96, No. 1 | American Journal of Public Health Brown et al. | Peer Reviewed | Public Health Then and Now | 63
321
TABLE 1—Number of Articles Retrieved by PubMed, Using “International Health Organization, began its tionale d’Hygiène Publique, the
Health” and “Global Health” as Search Terms, by Decade: 1950 work in 1920.13 This organiza- League of Nations Health Organi-
Through July 2005 tion established its headquarters zation, and UNRRA merged into
International Global in Geneva, sponsored a series of the new agency. The Pan Ameri-
Decade Healtha Healtha international commissions on dis- can Sanitary Bureau—then
eases, and published epidemio- headed by Fred L. Soper, a for-
1950s 1 007 54
logical intelligence and technical mer Rockefeller Foundation offi-
1960s 3 303 155
reports. The League of Nations cial—was allowed to retain au-
1970s 8 369 1 137
Health Organization was poorly tonomous status as part of a
1980s 16 924 7 176
budgeted and faced covert oppo- regionalization scheme.14 WHO
1990s 49 158 27 794
sition from other national and informally divided the world into a
2000–July 2005 52 169b 39 759b
ternational organizations, includ- series of regions—the Americas,
a
Picks up variant term endings (e.g. “international” also picks up “internationalize” and “internationalization”; ing the US Public Health Service. Southeast Asia, Europe, Eastern
“global” also picks up “globalize” and “globalization”).
Despite these complications, Mediterranean, Western Pacific,
b
Number for 55 months only.
which limited the Health Organi- and Africa—but it did not fully
more generally, it will be helpful zation ’s effectiveness, both the implement this regionalization
to review the history of the or- Office Internationale d’Hygiène until the 1950s. Although an “in-
ganization from 1948 to 1998, Publique and the Health Organi- ternational” and “intergovern-
as it moved from being the un- zation survived through World mental” mindset prevailed in the
questioned leader of international War II and were present at the 1940s and 1950s, naming the
health to searching for its place in critical postwar moment when new organization the World
the contested world of global the future of international health Health Organization also raised
health. would be defined. sights to a worldwide, “global”
WHO formally began in 1948, An international conference in perspective.
when the first World Health As- 1945 approved the creation of The first director general of
sembly in Geneva, Switzerland, the United Nations and also voted WHO, Brock Chisholm, was a
ratified its constitution. The idea for the creation of a new special- Canadian psychiatrist loosely
of a permanent institution for in- ized health agency. Participants at identified with the British social
ternational health can be traced the meeting initially formed a medicine tradition. The United
to the organization in 1902 of commission of prominent individ- States, a main contributor to the
the International Sanitary Office uals, among whom were René WHO budget, played a contradic-
of the American Republics, Sand from Belgium, Andrija Stam- tory role: on the one hand, it sup-
which, some decades later, be- par from Yugoslavia, and Thomas ported the UN system with its
came the Pan American Sanitary Parran from the United States. broad worldwide goals, but on
Bureau and eventually the Pan Sand and Stampar were widely the other, it was jealous of its sov-
American Health Organization.10 recognized as champions of social ereignty and maintained the right
The Rockefeller Foundation, es- medicine. The commission held to intervene unilaterally in the
pecially its International Health meetings between 1946 and Americas in the name of national
Division, was also a very signifi- early 1948 to plan the new inter- security. Another problem for
cant player in international health national health organization. Rep- WHO was that its constitution
in the early 20th century.11 resentatives of the Pan American had to be ratified by nation states,
Two European-based interna- Sanitary Bureau, whose leaders a slow process: by 1949, only 14
tional health agencies were also resisted being absorbed by the countries had signed on.15
important. One was the Office In- new agency, were also involved, As an intergovernmental
ternationale d’Hygiène Publique, as were leaders of new institu- agency, WHO had to be respon-
which began functioning in Paris tions such as the United Nations sive to the larger political environ-
in 1907; it concentrated on sev- Relief and Rehabilitation Adminis- ment. The politics of the Cold
eral basic activities related to the tration (UNRRA). War had a particular salience,
administration of international Against this background, the with an unmistakable impact on
sanitary agreements and the first World Health Assembly met WHO policies and personnel.
rapid exchange of epidemiologi- in Geneva in June 1948 and for- Thus, when the Soviet Union and
cal information.12 The second mally created the World Health other communist countries
agency, the League of Nations Organization. The Office Interna- walked out of the UN system and
322
therefore out of WHO in 1949, modernization with limited so- He argued that it was now scien-
the United States and its allies cial reform.18 tifically feasible, socially desir-
were easily able to exert a domi- With the return of the Soviet able, and economically worth-
nating influence. In 1953, Union and other communist while to attempt to eradicate
Chisholm completed his term as countries in 1956, the political smallpox worldwide.19 The Soviet
director general and was replaced balance in the World Health As- Union wanted to make its mark
by the Brazilian Marcolino Can- sembly shifted and Candau ac- on global health, and Candau,
dau. Candau, who had worked commodated the changed bal- recognizing the shifting balance
under Soper on malaria control in ance of power. During the 1960s, of power, was willing to cooper-
Brazil, was associated first with malaria eradication was facing se- ate. The Soviet Union and Cuba
the “vertical” disease control pro- rious difficulties in the field; ulti- agreed to provide 25 million and
grams of the Rockefeller Founda- mately, it would suffer colossal 2 million doses of freeze-dried
tion and then with their adoption and embarrassing failures. In vaccine, respectively; in 1959,
by the Pan American Sanitary Bu- 1969, the World Health Assem- the World Health Assembly com-
reau when Soper moved to that bly, declaring that it was not feasi- mitted itself to a global smallpox
agency as director.16 Candau ble to eradicate malaria in many eradication program.
would be director general of parts of the world, began a slow In the 1960s, technical im-
WHO for over 20 years. From process of reversal, returning once provements—jet injectors and bi-
1949 until 1956, when the Soviet again to an older malaria control furcated needles—made the proc-
Union returned to the UN and agenda. This time, however, there ess of vaccination much cheaper,
WHO, WHO was closely allied was a new twist; the 1969 assem- easier, and more effective. The
with US interests. bly emphasized the need to de- United States’ interest in smallpox
In 1955, Candau was charged velop rural health systems and to eradication sharply increased; in
with overseeing WHO’s cam- integrate malaria control into gen- 1965, Lyndon Johnson instructed
paign of malaria eradication, ap- eral health services. the US delegation to the World
proved that year by the World When the Soviet Union re- Health Assembly to pledge
Smallpox Vaccination Program
Health Assembly. The ambitious turned to WHO, its representa- American support for an interna-
in Togo, 1967. Courtesy of the
goal of malaria eradication had tive at the assembly was the national program to eradicate small- Centers for Disease Control and
been conceived and promoted in tional deputy minister of health. pox from the earth.20 At that Prevention. Source: Public Health
the context of great enthusiasm Image Library, CDC.
and optimism about the ability
of widespread DDT spraying to
kill mosquitoes. As Randall
Packard has argued, the United
States and its allies believed that
global malaria eradication would
usher in economic growth and
create overseas markets for US
technology and manufactured
goods.17 It would build support
for local governments and their
US supporters and help win
“hearts and minds” in the battle
against Communism. Mirroring
then-current development theo-
ries, the campaign promoted
technologies brought in from
outside and made no attempt to
enlist the participation of local
populations in planning or imple-
mentation. This model of devel-
opment assistance fit neatly into
US Cold War efforts to promote
323
time, despite a decade of marked interests of international players, the 1950s, with the civil rights
progress, the disease was still en- the intellectual and ideological movement and other social
demic in more than 30 countries. commitments of key individuals, movements forcing changes in
In 1967, now with the support and the way that all of these national priorities.
of the world’s most powerful factors interact with the health This changing political envi-
players, WHO launched the policymaking process. ronment was reflected in corre-
Intensified Smallpox Eradication During the 1960s and 1970s, sponding shifts within WHO. In
Program. This program, an inter- changes in WHO were signifi- the 1960s, WHO acknowledged
national effort led by the Ameri- cantly influenced by a political that a strengthened health infra-
can Donald A. Henderson, context marked by the emer- structure was prerequisite to the
would ultimately be stunningly gence of decolonized African na- success of malaria control pro-
successful.21 tions, the spread of nationalist grams, especially in Africa. In
and socialist movements, and 1968, Candau called for a com-
The Promise and Perils of new theories of development prehensive and integrated plan
Primary Health Care, that emphasized long-term socio- for curative and preventive care
1973–1993 economic growth rather than services. A Soviet representative
Within WHO, there have al- short-term technological inter- called for an organizational study
ways been tensions between so- vention. Rallying within organiza- of methods for promoting the de-
cial and economic approaches to tions such as the Non-Aligned velopment of basic health serv-
population health and technology- Movement, developing countries ices.23 In January 1971, the Exec-
or disease-focused approaches. created the UN Conference utive Board of the World Health
These approaches are not neces- on Trade and Development Assembly agreed to undertake
sarily incompatible, although (UNCTAD), where they argued this study, and its results were
they have often been at odds. vigorously for fairer terms of presented to the assembly in
The emphasis on one or the trade and more generous financ- 1973.24 Socrates Litsios has dis-
Alma Ata Conference, 1978. other waxes and wanes over ing of development.22 In Wash- cussed many of the steps in the
Courtesy of the Pan American
time, depending on the larger ington, DC, more liberal politics transformation of WHO’s ap-
Health Organization. Source: Office
of Public Information, PAHO. balance of power, the changing succeeded the conservatism of proach from an older model of
health services to what would be-
come the “Primary Health Care”
approach.25 This new model
drew upon the thinking and ex-
periences of nongovernmental
organizations and medical mis-
sionaries working in Africa, Asia,
and Latin America at the grass-
roots level. It also gained saliency
from China’s reentry into the
UN in 1973 and the widespread
interest in Chinese “barefoot doc-
tors,” who were reported to be
transforming rural health condi-
tions. These experiences under-
scored the urgency of a “Primary
Health Care” perspective that in-
cluded the training of community
health workers and the resolution
of basic economic and environ-
mental problems.26
These new approaches were
spearheaded by Halfdan T.
Mahler, a Dane, who served as
director general of WHO from
324
“
1973 to 1988. Under pressure the World Bank, the vice presi-
from the Soviet delegate to the dent of the Ford Foundation, the The Declaration of Primary Health Care
executive board, Mahler agreed administrator of USAID, and the
and the goal of “Health for All in the Year
to hold a major conference on executive secretary of UNICEF.30
the organization of health serv- The Bellagio meeting focused 2000” advocated an “inter-sectoral” and
ices in Alma-Ata, in the Soviet on an alternative concept to that multidimensional approach to health
Union. Mahler was initially reluc- articulated at Alma-Ata—“Selec-
and socioeconomic development, emphasized
tant because he disagreed with tive Primary Health Care”—which
the Soviet Union’s highly central- was built on the notion of prag- the use of “appropriate technology,” and
ized and medicalized approach to matic, low-cost interventions that urged active community participation in health
”
the provision of health services.27 were limited in scope and easy to
care and health education at every level.
The Soviet Union succeeded in monitor and evaluate. Thanks
hosting the September 1978 con- primarily to UNICEF, Selective
ference, but the conference itself Primary Health Care was soon growth, thus providing a good ar-
reflected Mahler’s views much operationalized under the gument for social sector spend-
more closely than it did those acronym “GOBI” (Growth moni- ing. As the Bank began to make
of the Soviets. The Declaration of toring to fight malnutrition in direct loans for health services, it
Primary Health Care and the children, Oral rehydration tech- called for more efficient use of
goal of “Health for All in the niques to defeat diarrheal dis- available resources and discussed
Year 2000” advocated an “inter- eases, Breastfeeding to protect the roles of the private and public
sectoral” and multidimensional children, and Immunizations).31 sectors in financing health care.
approach to health and socioeco- In the 1980s, WHO had to The Bank favored free markets
nomic development, emphasized reckon with the growing influ- and a diminished role for na-
the use of “appropriate technol- ence of the World Bank. The tional governments.34 In the con-
ogy,” and urged active commu- bank had initially been formed in text of widespread indebtedness
nity participation in health care 1946 to assist in the reconstruc- by developing countries and in-
and health education at every tion of Europe and later ex- creasingly scarce resources for
level.28 panded its mandate to provide health expenditures, the World
David Tejada de Rivero has ar- loans, grants, and technical assis- Bank’s promotion of “structural
gued that “It is regrettable that af- tance to developing countries. At adjustment” measures at the very
terward the impatience of some first, it funded large investments time that the HIV/AIDS epi-
international agencies, both UN in physical capital and infrastruc- demic erupted drew angry criti-
and private, and their emphasis ture; in the 1970s, however, it cism but also underscored the
on achieving tangible results in- began to invest in population con- Bank’s new influence.
stead of promoting change . . . led trol, health, and education, with In contrast to the World
to major distortions of the original an emphasis on population con- Bank’s increasing authority, in
concept of primary health care.”29 trol.32 The World Bank approved the 1980s the prestige of WHO
A number of governments, agen- its first loan for family planning was beginning to diminish. One
cies, and individuals saw WHO’s in 1970. In 1979, the World sign of trouble was the 1982
idealistic view of Primary Health Bank created a Population, vote by the World Health
Care as “unrealistic” and unattain- Health, and Nutrition Depart- Assembly to freeze WHO’s
able. The process of reducing ment and adopted a policy of budget.35 This was followed
Alma-Ata’s idealism to a practical funding both stand-alone health by the 1985 decision by the
set of technical interventions that programs and health components United States to pay only 20%
could be implemented and mea- of other projects. of its assessed contribution to all
sured more easily began in 1979 In its 1980 World Development UN agencies and to withhold its
at a small conference—heavily in- Report, the Bank argued that both contribution to WHO’s regular
fluenced by US attendees and malnutrition and ill health could budget, in part as a protest
policies—held in Bellagio, Italy, be countered by direct govern- against WHO’s “Essential Drug
and sponsored by the Rockefeller ment action—with World Bank Program,” which was opposed
Foundation, with assistance from assistance.33 It also suggested that by leading US-based pharma-
the World Bank. Those in atten- improving health and nutrition ceutical companies.36 These
dance included the president of could accelerate economic events occurred amidst growing
325
tensions between WHO and reliance on WHO’s “regular assessment: “WHO is caught in a
UNICEF and other agencies budget”—drawn from member cycle of decline, with donors ex-
and the controversy over Selec- states’ contributions on the basis pressing their lack of faith in its
tive versus Comprehensive Pri- of population size and gross na- central management by placing
mary Health Care. As part of a tional product—to greatly in- funds outside the management’s
rancorous public debate con- creased dependence on extrabud- control. This has prevented
ducted in the pages of Social getary funding coming from WHO from [developing] . . . inte-
Science and Medicine in 1988, donations by multilateral agen- grated responses to countries’
Kenneth Newell, a highly placed cies or “donor” nations.39 By the long term needs.”41
WHO official and an architect period 1986–1987, extrabud- In the late 1980s and early
of Comprehensive Primary getary funds of $437 million had 1990s, the World Bank moved
Health Care, called Selective Pri- almost caught up with the regular confidently into the vacuum cre-
mary Health Care a “threat . . . budget of $543 million. By the ated by an increasingly ineffec-
[that] can be thought of as a beginning of the 1990s, extra- tive WHO. WHO officials were
counter-revolution.”37 budgetary funding had overtaken unable or unwilling to respond to
In 1988, Mahler’s 15-year the regular budget by $21 mil- the new international political
tenure as director general of lion, contributing 54% of WHO’s economy structured around ne-
WHO came to an end. Unexpect- overall budget. oliberal approaches to economics,
edly, Hiroshi Nakajima, a Japanese Enormous problems for the or- trade, and politics.42 The Bank
researcher who had been director ganization followed from this maintained that existing health
of the WHO Western Pacific Re- budgetary shift. Priorities and systems were often wasteful, inef-
gional Office in Manila, was policies were still ostensibly set ficient, and ineffective, and it ar-
elected new director general.38 by the World Health Assembly, gued in favor of greater reliance
which was made up of all mem- on private-sector health care pro-
Crisis at WHO, 1988–1998 ber nations. The assembly, how- vision and the reduction of public
The first citizen of Japan ever ever, now dominated numerically involvement in health services
elected to head a UN agency, by poor and developing coun- delivery.43
Nakajima rapidly became the tries, had authority only over the Controversies surrounded the
most controversial director gen- regular budget, frozen since the World Bank’s policies and prac-
eral in WHO’s history. His nomi- early 1980s. Wealthy donor na- tices, but there was no doubt that,
nation had not been supported tions and multilateral agencies by the early 1990s, it had be-
by the United States or by a like the World Bank could largely come a dominant force in interna-
number of European and Latin call the shots on the use of the tional health. The Bank’s greatest
American countries, and his per- extrabudgetary funds they con- “comparative advantage” lay in its
formance in office did little to as- tributed. Thus, they created, in ability to mobilize large financial
suage their doubts. Nakajima did effect, a series of “vertical” pro- resources. By 1990, the Bank’s
try to launch several important grams more or less independent loans for health surpassed WHO’s
initiatives—on tobacco, global of the rest of WHO’s programs total budget, and by the end of
disease surveillance, and and decisionmaking structure. 1996, the Bank’s cumulative
public–private partnerships— The dilemma for the organization lending portfolio in health, nutri-
but fierce criticism persisted that was that although the extrabud- tion, and population had reached
raised questions about his auto- getary funds added to the overall $13.5 billion. Yet the Bank recog-
cratic style and poor manage- budget, “they [increased] difficul- nized that, whereas it had great
ment, his inability to communities of coordination and continu- economic strengths and influence,
cate effectively, and, worst of all, ity, [caused] unpredictability in fi- WHO still had considerable tech-
cronyism and corruption. nance, and a great deal of nical expertise in matters of
Another symptom of WHO’s dependence on the satisfaction of health and medicine. This was
problems in the late 1980s was particular donors,”40 as Gill Walt clearly reflected in the Bank’s
the growth of “extrabudgetary” explained. widely influential World Develop-
funding. As Gill Walt of the Lon- Fiona Godlee published a se- ment Report, 1993: Investing in
don School of Hygiene and Tropi- ries of articles in 1994 and 1995 Health, in which credit is given to
cal Medicine noted, there was a that built on Walt’s critique.41 WHO, “a full partner with the
crucial shift from predominant She concluded with this dire World Bank at every step of the
326
preparation of the Report.”44 Cir-

cumstances suggested that it was
to the advantage of both parties
for the World Bank and WHO to
work together.
WHO EMBRACES
“GLOBAL HEALTH”
This is the context in which

WHO began to refashion itself
as a coordinator, strategic plan-
ner, and leader of “global health”
initiatives. In January 1992, the
31-member Executive Board of
the World Health Assembly de-
cided to appoint a “working
group” to recommend how
WHO could be most effective in
Current Director General Jong-wook
international health work in light for analyses of “global health Diseases: A Strategy for the 21st Lee with three former Directors-
of the “global change” rapidly interdependence.”47 In the same Century appeared, followed in General at the celebration to
overtaking the world. The execu- year, Milton and Ruth Roemer 2001 by the Institute of Medi- mark the 25th Anniversary of the
tive board may have been re- argued that further improvements cine’s Perspectives on the Depart- Alma Ata Declaration. From left:
G. H. Brundtland, H. Mahler,
sponding, in part, to the Chil- in “global health” would be de- ment of Defense Global Emerging
H. Nakajima, Lee JW. Courtesy of
dren’s Vaccine Initiative, pendent on the expansion of pub- Infections Surveillance and Re- the World Health Organization.
perceived within WHO as an at- lic rather than private health sponse System.52 Best-selling Source: Media Center, WHO.
tempted “coup” by UNICEF, the services.48 Another strong source books and news magazines were
World Bank, the UN Develop- for the term “global health” was full of stories about Ebola and
ment Program, the Rockefeller the environmental movement, es- West Nile virus, resurgent tuber-
Foundation, and several other pecially debates over world envi- culosis, and the threat of bioter-
players seeking to wrest control ronmental degradation, global rorism.53 The message was clear:
of vaccine development.45 The warming, and their potentially there was a palpable global dis-
working group’s final report of devastating effects on human ease threat.
May 1993 recommended that health.49 In 1998, the World Health
WHO—if it was to maintain lead- In the mid-1990s, a consider- Assembly reached outside the
ership of the health sector—must able body of literature was pro- ranks of WHO for a new leader
overhaul its fragmented manage- duced on global health threats. In who could restore credibility to
ment of global, regional, and the United States, a new Centers the organization and provide it
country programs, diminish the for Disease Control and Preven- with a new vision: Gro Harlem
competition between regular and tion (CDC) journal, Emerging In- Brundtland, former prime minis-
extrabudgetary programs, and, fectious Diseases, began publica- ter of Norway and a physician
above all, increase the emphasis tion, and former CDC director and public health professional.
within WHO on global health is- William Foege started using the Brundtland brought formidable
sues and WHO’s coordinating phrase “global infectious disease expertise to the task. In the
role in that domain.46 threats.”50 In 1997, the Institute 1980s, she had been chair of the
Until that time, the term of Medicine’s Board of Interna- UN World Commission on Envi-
“global health” had been used tional Health released a report, ronment and Development and
sporadically and, outside WHO, America’s Vital Interest in Global produced the “Brundtland Re-
usually by people on the political Health: Protecting Our People, port,” which led to the Earth
left with various “world” agendas. Enhancing Our Economy, and Summit of 1992. She was familiar
In 1990, G. A. Gellert of Interna- Advancing Our International with the global thinking of the en-
tional Physicians for the Preven- Interests.51 In 1998, the CDC’s vironmental movement and had a
tion of Nuclear War had called Preventing Emerging Infectious broad and clear understanding of
327
the links between health, environ- eases in the world’s poorest na- from a narrow emphasis on Requests for reprints should be sent to
ment, and development.54 tions, mainly through vaccines malaria eradication to a broader Elizabeth Fee, PhD, History of Medicine
Division, National Library of Medicine,
Brundtland was determined to and immunization programs.58 interest in the development of 8600 Rockville Pike, Bethesda, MD
position WHO as an important Within a few years, some 70 health services and the emerging 20974 (e-mail: elizabeth_fee@
player on the global stage, move “global health partnerships” had concentration on smallpox eradi- nlm.nih.gov).
This article was accepted January 11,
beyond ministries of health, and been created. cation. In the 1970s and 1980s, 2005.
gain a seat at the table where de- Brundtland’s tenure as director WHO developed the concept of
cisions were being made.55 She general was not without blemish Primary Health Care but then
Contributors
wanted to refashion WHO as a nor free from criticism. Some of turned from zealous advocacy to All authors contributed equally to the re-
“department of consequence”55 the initiatives credited to her ad- the pragmatic promotion of Se- search and writing.
able to monitor and influence ministration had actually been lective Primary Health Care as
other actors on the global scene. started under Nakajima (for ex- complex changes overtook intra- Acknowledgments
She established a Commission on ample, the WHO Framework and interorganizational dynam- The authors are grateful to the Joint
Learning Initiative of the Rockefeller
Macroeconomics and Health, Convention on Tobacco Control), ics and altered the international Foundation, which initially commis-
chaired by economist Jeffrey Sachs others may be looked upon today economic and political order. In sioned this article, and to the Global
of Harvard University and includ- with some skepticism (the Com- the 1990s, WHO attempted to Health Histories Initiative of the World
Health Organization, which has provided
ing former ministers of finance mission on Macroeconomics and use leadership of an emerging a supportive environment for continuing
and officers from the World Bank, Health, Roll Back Malaria), and concern with “global health” as our research.
the International Monetary Fund, still others arguably did not re- an organizational strategy that
the World Trade Organization, ceive enough attention from her promised survival and, indeed, References
and the UN Development Pro- administration (Primary Health renewal. 1. A small sampling of recent titles:
gram, as well as public health Care, HIV/AIDS, Health and But just as it did not invent David L. Heymann and G. R. Rodier,
“Global Surveillance of Communicable
leaders. The commission issued a Human Rights, and Child Health). the eradicationist or primary
Diseases,” Emerging Infectious Diseases
report in December 2001, which Nonetheless, few would dispute care agendas, WHO did not in- 4 (1998): 362–365; David Wood-
argued that improving health in the assertion that Brundtland suc- vent “global health”; other, ward, Nick Drager, Robert Beaglehole,
and Debra Lipson, “Globalization and
developing countries was essential ceeded in achieving her principal larger forces were responsible.
Health: A Framework for Analysis and
to their economic development.56 objective, which was to reposition WHO certainly did help pro- Action,” Bulletin of the World Health
The report identified a set of dis- WHO as a credible and highly mote interest in global health Organization 79 (2001): 875–881;
Gill Walt, “Globalisation of Interna-
ease priorities that would require visible contributor to the rapidly and contributed significantly to
tional Health,” The Lancet 351 (Febru-
focused intervention. changing field of global health. the dissemination of new con- ary 7, 1998): 434–437; Stephen J.
Brundtland also began to cepts and a new vocabulary. In Kunitz, “Globalization, States, and the
Health of Indigenous Peoples,” Ameri-
strengthen WHO’s financial posi- CONCLUSION that process, it was hoping to ac-
can Journal of Public Health 90 (2000):
tion, largely by organizing “global quire, as Yach and Bettcher sug- 1531–1539; Health Policy in a Global-
partnerships” and “global funds” We can now return briefly to gested in 1998, a restored coor- ising World, ed. Kelley Lee, Kent Buse,
and Suzanne Fustukian (Cambridge,
to bring together “stakeholders”— the questions implied at the be- dinating and leadership role.
England: Cambridge University Press,
private donors, governments, and ginning of this article: how does Whether WHO’s organizational 2002).
bilateral and multilateral agen- a historical perspective help us repositioning will serve to 2. For example, Yale has a Division of
cies—to concentrate on specific understand the emergence of the reestablish it as the unques- Global Health in its School of Public
targets (for example, Roll Back terminology of “global health” tioned steward of the health of Health, Harvard has a Center for Health
and the Global Environment, and the
Malaria in 1998, the Global Al- and what role did WHO play as the world’s population, and how London School of Hygiene and Tropical
liance for Vaccines and Immu- an agent in its development? this mission will be effected in Medicine has a Center on Global
nization in 1999, and Stop TB in The basic answers derive from practice, remains an open ques- Change and Health; the National Insti-
tutes of Health has a strategic plan on
2001). These were semiau- the fact that WHO at various tion at this time. ■ Emerging Infectious Diseases and Global
tonomous programs bringing in times in its history alternatively Health; Gro Harlem Brundtland ad-
substantial outside funding, often led, reflected, and tried to ac- dressed the 35th Anniversary Sympo-
About the Author sium of the John E. Fogarty Interna-
in the form of “public–private commodate broader changes Theodore M. Brown is with the Department tional Center on “Global Health: A
partnerships.”57 A very significant and challenges in the ever- of History and the Department of Commu- Challenge to Scientists” in May 2003;
player in these partnerships was shifting world of international nity and Preventive Medicine, University of the Centers of Disease Control and Pre-
Rochester, Rochester, NY. Marcos Cueto is vention has established an Office of
the Bill & Melinda Gates Founda- health. In the 1950s and 1960s, with the Facultad de Salud Pública, Uni- Global Health and has partnered with
tion, which committed more than when changes in biology, eco- versidad Peruana Cayetano Heredia, Lima, the World Health Organization (WHO),
$1.7 billion between 1998 and nomics, and great power politics Peru. Elizabeth Fee is with the History of the World Bank, UNICEF, the US
Medicine Division, National Library of Agency for International Development,
2000 to an international pro- transformed foreign relations Medicine, National Institutes of Health, and others in creating Global Health
gram to prevent or eliminate dis- and public health, WHO moved Bethesda, Md. Partnerships.
328
3. Albert Deutsch, The World Health Retrospective Memoir,” Perspectives in World Health Organization 76 (1998): 114–140. UNICEF was created in 1946
Organization: Its Global Battle Against Biology and Medicine 11 (1968): 233–235; Donald A. Henderson, “Erad- to assist needy children in Europe’s war
Disease (New York: Public Affairs Com- 273–285. ication: Lessons From the Past,” Bulletin ravaged areas. After the emergency
mittee, 1958). 13. Frank G. Bourdreau, “International of the World Health Organization 76 ended, it broadened its mission and con-
4. Randall M. Packard, “ ‘No Other Health,” American Journal of Public (Supplement 2) (1998): 17–21; Frank centrated resources on the needs of chil-
Logical Choice’: Global Malaria Eradica- Health and the Nation’s Health 19 Fenner, Donald A. Henderson, Issao dren in developing countries.
tion and the Politics of International (1929): 863–878; Bourdreau, “Interna- Arita, Zdenek Jevek, and Ivan Dali-
31. UNICEF, The State of the World’s
Health in the Post-War Era,” Parassitolo- tional Health Work,” in Pioneers in novich Ladnyi, Smallpox and its Eradica-
Children: 1982/1983 (New York: Oxford
gia 40 (1998): 217–229, and The Poli- World Order: An American Appraisal of tion (Geneva: WHO, 1988).
University Press, 1983). See also Cueto,
tics of Global Health, Prepared for the the League of Nations, ed. Harriet Eager 22. The New International Economic “Origins of Primary Health Care.”
Subcommittee on National Security Policy Favis (New York: Columbia University Order: The North South Debate, ed.
32. Jennifer Prah Ruger, “The Chang-
and Scientific Developments of the Com- Press, 1944), 193–207; Norman Jagdish N. Bhagwati (Cambridge, Mass:
ing Role of the World Bank in Global
mittee on Foreign Affairs, US House of Howard-Jones, International Public Health MIT Press, 1977); Robert L. Rothstein,
Health in Historical Perspective,” Ameri-
Representatives (Washington, DC: US Between the Two World Wars: The Orga- Global Bargaining: UNCTAD and the
can Journal of Public Health 95 (2005):
Government Printing Office, 1971). nizational Problems (Geneva: WHO, Quest for a New International Economic
60–70.
5. For example, T W. Wilson, World 1978); Martin David Dubin, “The Order (Princeton, NJ: Princeton Univer-
Population and a Global Emergency League of Nations Health Organisation,” sity Press, 1979). 33. World Development Report 1980
(Washington, DC: Aspen Institute for in International Health Organisations and 23. Socrates Litsios, “The Long and (Washington, DC: World Bank, 1980).
Humanistic Studies, Program in Environ- Movements, 1918–1939, ed. Paul Wein- Difficult Road to Alma-Ata: A Personal 34. Financing Health Services in Devel-
ment and Quality of Life, 1974). dling (Cambridge, England: Cambridge Reflection,” International Journal of oping Countries: An Agenda for Reform
University Press, 1995), 56–80. Health Services 32 (2002): 709–732.
6. James E. Banta, “From Interna- (Washington, DC: World Bank, 1987).
tional to Global Health,” Journal of Com- 14. Thomas Parran, “The First 12 24. Executive Board 49th Session, 35. Fiona Godlee, “WHO in Retreat; Is
munity Health 26 (2001): 73–76. Years of WHO,” Public Health Reports WHO document EB49/SR/14 Rev It Losing Its Influence?” British Medical
73 (1958): 879–883; Fred L. Soper, (Geneva: WHO, 1973), 218; Organiza-
7. Supinda Bunyavanich and Ruth B. Journal 309 (1994): 1491–1495.
Ventures in World Health: The Memoirs of tional Study of the Executive Board on
Walkup, “US Public Health Leaders Shift Fred Lowe Soper, ed. John Duffy (Wash- 36. Ibid, 1492.
Toward a New Paradigm of Global Methods of Promoting the Development of
ington, DC: Pan American Health Orga- Basic Health Services, WHO document
Health,” American Journal of Public 37. Kenneth Newell, “Selective Primary
nization, 1977); Javed Siddiqi, World EB49/WP/6 (Geneva: WHO, 1972),
Health 91 (2001): 1556–1558. Health Care: The Counter Revolution,”
Health and World Politics: The World 19–20. Social Science and Medicine 26 (1988):
8. Derek Yach and Douglas Bettcher, Health Organization and the UN System
25. Socrates Litsios, “The Christian 903–906 (quote on p. 906).
“The Globalization of Public Health, I: (London: Hurst and Co, 1995).
Threats and Opportunities,” American Medical Commission and the Develop- 38. Paul Lewis, “Divided World Health
15. “Seventh Meeting of the Executive ment of WHO’s Primary Health Care
Journal of Public Health 88 (1998): Organization Braces for Leadership
Committee of the Pan American Sani- Approach,” American Journal of Public
735–738, and “The Globalization of Change,” New York Times, May 1, 1988,
tary Organization,” Washington, DC, Health 94 (2004): 1884–1893; Litsios,
Public Health, II: The Convergence of p. 20.
May 23–30, 1949, Folder “Pan Ameri- “The Long and Difficult Road to Alma-
Self-Interest and Altruism,” American can Sanitary Bureau,” RG 90–41, Box 39. Gill Walt, “WHO Under Stress: Im-
Journal of Public Health 88 (1998): Ata.”
9, Series Graduate School of Public plications for Health Policy,” Health Pol-
738–741. Health, University of Pittsburgh 26. John H. Bryant, Health and the De- icy 24 (1993): 125–144.
9. George Silver, “International Archives. veloping World (Ithaca, NY: Cornell Uni-
versity Press, 1969); Doctors for the Vil- 40. Ibid, 129.
Health Services Need an Interorganiza- 16. WHO, “Information. Former Direc-
tional Policy,” American Journal of Public lages: Study of Rural Internships in Seven 41. Fiona Godlee, “WHO in Crisis,”
tors-General of the World Health Organi- Indian Medical Colleges, ed. Carl E. Tay-
Health 88 (1998): 727–729 (quote on British Medical Journal 309
zation. Dr Marcolino Gomes Candau,” lor (New York: Asia Publishing House,
p. 728). (1994):1424–1428; Godlee, “WHO in
available at http://www.who.int/archives/ 1976); Kenneth W. Newell, Health by the Retreat”; Fiona Godlee, “WHO’s Special
10. Pro Salute, Novi Mundi: Historia de wh050/en/directors.htm, accessed People (Geneva: WHO, 1975). See also Programmes: Undermining From
la Organización Panamericana de la July 24, 2004; “In memory of Dr M. G. Marcos Cueto, “The Origins of Primary Above,” British Medical Journal 310
Salud (Washington, DC: Organización Candau,” WHO Chronicle 37 (1983): Health Care and Selective Primary (1995):178–182 (quote on p. 182).
Panamericana de la Salud, 1992). 144–147. Health Care,” American Journal of Public
17. Randall M. Packard, “Malaria Health 94 (2004): 1864–1874; Litsios, 42. P. Brown, “The WHO Strikes Mid-
11. See John Farley, To Cast Out Dis-
Dreams: Postwar Visions of Health and “The Christian Medical Commission.” Life Crisis,” New Scientist 153 (1997):
ease: A History of the International Health
Development in the Third World,” Med- 12; “World Bank’s Cure for Donor Fa-
Division of the Rockefeller Foundation 27. See Litsios, “The Long and Difficult
ical Anthropology 17 (1997): 279–296; tigue [editorial],” The Lancet 342 (July
(1913–1951) (Oxford: Oxford University Road to Alma-Ata,” 716–719.
Packard, “No Other Logical Choice” 10, 1993): 63–64; Anthony Zwi, “Intro-
Press, 2003); Anne-Emmanuelle Birn,
Parassitologia 40 (1998): 217–229. 28. “Declaration of Alma-Ata, Interna- duction to Policy Forum: The World
“Eradication, Control or Neither? Hook-
tional Conference on Primary Health Bank and International Health,” Social
worm Versus Malaria Strategies and 18. Randall M. Packard and Peter J. Care, Alma-Ata, USSR, 6–12 September, Science and Medicine 50 (2000): 167.
Rockefeller Public Health in Mexico,” Brown, “Rethinking Health, Develop-
Parassitologia 40 (1996):137–147; Mis- 1978,” available at http://www.who.int/
ment and Malaria: Historicizing a Cul- 43. World Bank, Financing Health Ser-
sionaries of Science: Latin America and hpr/NPH/docs/declaration_almaata.pdf,
tural Model in International Health,” vices in Developing Countries.
the Rockefeller Foundation, ed. Marcos accessed April 10, 2004.
Medical Anthropology 17 (1997): 44. World Development Report, 1993: In-
Cueto (Bloomington: Indiana University 181–194. 29. David A. Tejada de Rivero, “Alma-
vesting in Health (Washington, DC: World
Press, 1994). Ata Revisited,” Perspectives in Health
19. Ian and Jennifer Glynn, The Life Bank, 1993), iii–iv (quote on pp. iii–iv).
12. Vingt-Cinq Ans d’Activité de l’Office Magazine: The Magazine of the Pan
and Death of Smallpox (New York: Cam- American Health Organization 8 (2003): 45. For a full account, see William
Internationale d’Hygiène Publique, bridge University Press, 2004), 1–6 (quote on p. 4). Muraskin, The Politics of International
1909–1933 (Paris: Office Internationale 194–196. Health: The Children’s Vaccine Initiative
d’Hygiène Publique, 1933); Paul F. 30. Maggie Black, Children First: The
20. Ibid, 198. and the Struggle to Develop Vaccines for
Basch, “A Historical Perspective on Inter- Story of UNICEF, Past and Present (Ox-
the Third World (Albany: State Univer-
national Health,” Infectious Disease Clin- 21. William H. Foege, “Commentary: ford: Oxford University Press; 1996), and
sity of New York Press, 1998).
ics of North America 5 (1991):183–196; Smallpox Eradication in West and Cen- The Children and the Nations: The Story of
W. R. Aykroyd, “International Health—A tral Africa Revisited,” Bulletin of the UNICEF (New York: UNICEF, 1986), 46. Bo Stenson and Göran Sterky,
329
“What Future WHO?” Health Policy 28

(1994): 235–256 (quote on p. 242).
54. Lawrence K. Altman, “US Moves to
Replace Japanese Head of WHO,” New
Fighting Global Blindness
47. G. A. Gellert, “Global Health Inter- York Times, December 20, 1992, p. 1. Improving World Vision
dependence and the International Physi- 55. Ilona Kickbusch, “The Develop- Through Cataract Elimination
cians’ Movement,” Journal of the Ameri- ment of International Health Priorities—
can Medical Association 264 (1990): Accountability Intact?” Social Science & By Sanduk Ruit, MD,
610–613 (quote on p. 610). Medicine 51 (2000): 979–989 (quote Charles C. Wykoff, MD, D.Phil., MD,
48. Milton Roemer and Ruth Roemer, on p. 985). Geoffrey C. Tabin, MD
“Global Health, National Development, 56. Commission on Macroeconomics
and the Role of Government,” American and Health, Macroeconomics and Health:
Journal of Public Health 80 (1990): Investing in Health for Economic Develop-
1188–1192. ment (Geneva: WHO, 2001); see also
49. See, for example, Andrew J. Howard Waitzkin, “Report of the WHO
Haines, “Global Warming and Health,” Commission on Macroeconomics and
British Medical Journal 302 (1991): Health: A Summary and Critique,” The
669–670; Andrew J. Haines, Paul R. Lancet 361 (February 8, 2003):
Epstein, and Anthony J. McMichael, 523–526.
“Global Health Watch: Monitoring Im- 57. Michael A. Reid and E. Jim Pearce,
pacts of Environmental Change,” The “Whither the World Health Organiza-
Lancet 342 (December 11, 1993): tion?” The Medical Journal of Australia
1464–1469; Anthony J. McMichael, 178 (2003): 9–12.
“Global Environmental Change and 58. Michael McCarthy, “A Conversation
Human Population Health: A Concep- With the Leaders of the Gates Founda-
tual and Scientific Challenge for Epide- tion’s Global Health Program: Gordon
miology,” International Journal of Epide- Perkin and William Foege,” The Lancet
miology 22 (1993): 1–8; John M. Last, 356 (July 8, 2000): 153–155.
“Global Change: Ozone Depletion,
Greenhouse Warming, and Public
Health,” Annual Review of Public Health
14 (1993): 115–136; A. J. McMichael,
U noperated cataract is the cause of millions
of cases of visual impairment and
blindness in poor populations throughout
Planetary Overload, Global Environmental
Change and the Health of the Human both the developing and the developed world.
Species (Cambridge, England: Cambridge This wonderfully written volume shares the
University Press, 1993); Anthony J. experiences of a team of surgeons who have
McMichael, Andrew J. Haines, R. Sloof, demonstrated how the surgical procedures can
and S. Kovats, Climate Change and be simplified and made more efficient,
Human Health (Geneva: WHO, 1996); accessible, and far less expensive.
Anthony J. McMichael and Andrew
ISBN 0-87553-067-2 • spiral bound • 2006
Haines, “Global Climate Change: The
$31.50 APHA Members • $45.00 Nonmembers
Potential Effects on Health,” British Med-
ical Journal 315 (1997): 805–809.
American Public Health
50. Stephen S. Morse, “Factors in the Association
Emergence of Infectious Diseases,” 800 I Street, NW,
Emerging Infectious Diseases 1 (1995): Washington, DC 20001
7–15 (quote on p. 7). www.apha.org
51. Institute of Medicine, America’s
Vital Interest in Global Health: Protecting TO ORDER: web www.aphabookstore.org
Our People, Enhancing Our Economy, and email apha@pbd.com
Advancing Our International Interests fax 888.361.APHA
(Washington, DC: National Academy phone 888.320.APHA M-F 8am-5pm EST
Press, 1997).
52. Emerging Infections: Biomedical Re-
search Reports, ed. Richard M. Krause
(San Diego: Academic Press, 1998); Pre-
venting Emerging Infectious Diseases: A
Strategy for the 21st Century (Atlanta:
Centers for Disease Control and Preven-
tion, 1998); Perspectives on the Depart-
ment of Defense Global Emerging Infec-
tions Surveillance and Response System,
ed. Philip S. Brachman, Heather C. O’-
Maonaigh, and Richard N. Miller (Wash-
ington, DC: National Academy Press,
2001).
53. For example, Laurie Garrett, The
Coming Plague: Newly Emerging Diseases
in a World Out of Balance (New York:
Farrar, Straus and Giroux, 1994).
330
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University
Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its
entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org
! The Author 2008; all rights reserved. Advance Access publication 24 July 2008 doi:10.1093/heapol/czn020
Changing fortunes: analysis of fluctuating

policy space for family planning in Kenya
Joanna Crichton*
Accepted 22 June 2008
Policies relating to contraceptive services (population, family planning and

reproductive health policies) often receive weak or fluctuating levels of
commitment from national policy elites in Southern countries, leading to slow
policy evolution and undermining implementation. This is true of Kenya, despite
the government’s early progress in committing to population and reproductive
health policies, and its success in implementing them during the 1980s. This key
informant study on family planning policy in Kenya found that policy space
contracted, and then began to expand, because of shifts in contextual factors,
and because of the actions of different actors. Policy space contracted during the
mid-1990s in the context of weakening prioritization of reproductive health in
national and international policy agendas, undermining access to contraceptive
services and contributing to the stalling of the country’s fertility rates. However,
during the mid-2000s, champions of family planning within the Kenyan
Government bureaucracy played an important role in expanding the policy space
through both public and hidden advocacy activities. The case study demonstrates
that policy space analysis can provide useful insights into the dynamics of
routine policy and programme evolution and the challenge of sustaining support
for issues even after they have reached the policy agenda.
Keywords Policy analysis, family planning, health policy, contraception
KEY MESSAGES
! Policy space for the issue of family planning in Kenya contracted during the late 1990s, and has since begun to expand,
due to changing contextual factors and the actions of different individuals.
! Proponents of family planning within two government ministries played an important role in expanding the policy
space through both public and intra-government advocacy activities.
! Policy space analysis can provide useful insights into the dynamics of routine policy and programme evolution and the
challenge of sustaining support for issues after they have made it onto the policy agenda.
Introduction policies during the 1960s and in contraceptive service provision

during the 1970s and 1980s, yet where resource allocations and
In many parts of the world policies relating to contraceptives
implementation subsequently declined (Chimbwete and Zulu
tend to receive weak or fluctuating levels of commitment from
2003). In Kenya, as elsewhere in sub-Saharan Africa, the past
national policy elites, leading to slow policy evolution and
decade has seen a weakening prioritization of contraceptive
undermining implementation. This is true of Kenya, where the
programmes in national and international policy agendas
government made early progress in committing to population
(Cleland et al. 2006), undermining access to services and
progress towards the Millennium Development Goals.
*African Population and Health Research Center, PO Box 10787, 00100-GPO,
Nairobi, Kenya. Tel: þ254 20 272 0400/1/2. Fax: þ254 20 272 0380. This key informant study examines factors affecting the
E-mail: jcrichton@aphrc.org fluctuating level of prioritization of contraceptive service
339
331
provision among Kenyan government policy-makers since the

mid-1990s. Contraceptive services are usually referred to as
‘family planning’ in national policy debates in Kenya and are
framed as cutting across reproductive health and population
concerns (Ministry of Health 2000, 2007; NCPD 2000, 2003,
2005, 2006a). Based on key informant interviews and a review
of academic and official publications and reports, the paper
focuses on the strategies and actions taken by a range of
actors to ‘reposition’ family planning in government policy and
to ensure the incorporation of contraceptive commodities in the
national government budget of 2005, for the first time in the
country’s history.
The problem of sustaining political and bureaucratic commit-
ment for the implementation and evolution of policies affects
a variety of policy issues (Grindle and Thomas 1991; Buse et al.
2005). Waning commitment can lead to stagnation in imple-
mentation, and can undermine the likelihood that political and
bureaucratic actors create new policies and strategies to adapt
to changing contexts, such as shifts in external funding trends. Figure 1 Factors affecting policy space
In Southern countries and elsewhere, reproductive health
policies are particularly vulnerable to weak political commit-
and have limited resources for dealing with them (Shiffman
ment, because they do not tend to have strong national support
2007). External actors and international structural trends
bases and have historically been controversial and perceived as
have a critical influence on national health policy processes,
driven by external actors (Jain 1998; Chimbwete and Zulu
with increasing diversity and fragmentation of international
2003). Thomas and Grindle (1994), in their review of popu-
actors and sources of funding (Walt and Buse 2000; Cerny
lation reforms in 16 countries, explain that sustained commit-
2002). These international factors often have contradictory
ment to the implementation of population policies tends to be
influences, particularly in contexts characterized by national
constrained by two main factors: the dispersed and long-term
government dependence on external funds, aid conditionalities,
nature of their impacts, and the lack of mobilized support from
shifting funding priorities, and persistence in vertical program-
users of contraceptive services. Reproductive health and
ming (Walt and Buse 2000; Cerny 2002; Mayhew et al. 2005).
population policies have therefore been vulnerable to depriori-
The background characteristics of policy elites are also
tization and neglect in many Southern countries, especially in
important pre-existing factors that shape policy space; for
the context of the shift in international attention and official example the values, level of expertise, experience, degree of
development assistance to HIV and AIDS programmes during influence and loyalties of elites influence both their receptive-
the 1990s (Cleland et al. 2006). ness to policy change, and their success in championing
In this paper, I contend that policy space analysis provides particular policies.
a useful framework for understanding why commitment to A second area affecting policy space is that of ‘policy
existing policies often fluctuates over time, and for mapping the circumstances’, or the ways in which policy makers’ perceptions
room for manoeuvre that advocates of particular policies have for about a policy issue shape the dynamics of decision making.
addressing policies that are being neglected. Policy elites can be The extent to which a policy issue is perceived by policy elites to
thought of as operating within a ‘policy space’, which influences be a matter of crisis or ‘politics-as-usual’ affects the level at
the degree of agency they have for reforming and driving policy which decisions are taken, the urgency with which decisions
implementation, but which can be expanded by the exercise of are made, and the extent of risk taking (Grindle and Thomas
that agency. These concepts are drawn from Grindle and Thomas 1991; Walt and Gilson 1994). Policy crises involve strong
(1991), who suggest that the scope of policy space is influenced by pressure on policy makers to act, as well as high political
the way in which policy elites manage the interactions between stakes, and can lead to radical shifts in the prioritization of
(1) national and international contextual factors, (2) the issues. When policies are not perceived as urgent, decision
circumstances surrounding the policy process, and (3) the making may be dominated by concerns about micropolitical
acceptability of the policy’s content. Figure 1 represents policy and bureaucratic costs and benefits. Policy circumstances differ
space as a balloon, which can be expanded, constrained or from contextual factors because of their dynamic element:
contracted by shifts in these factors and by peoples’ actions.
Firstly, contextual factors are the pre-existing circumstances How particular circumstances are perceived by policy elites
within which policy processes occur. They can act as opportu- [. . .] serves as a bridge between the ‘‘embedded orienta-
nities and constraints for policy elites’ prioritization of a policy tions’’ of individuals and societies and the kinds of changes
issue, and include historical, social, cultural, political, economic considered by decision makers confronted with specific
and demographic characteristics of a country and situational policy choices. (Thomas and Grindle 1994, p.53)
or focusing events, like epidemics, droughts or media coverage
of issues (Kingdon 1984; Grindle and Thomas 1991). Policy- Lastly, the policy’s characteristics are themselves influenced by
makers are confronted with a multitude of competing issues policy elites’ decisions, but also affect the scope policy makers
332
POLICY SPACE FOR FAMILY PLANNING IN KENYA 341
have for introducing a policy and prioritizing it. The acceptability in Kenya.1 Interviews were recorded in shorthand during the
of a policy is influenced by policy characteristics such as the interview and then typed up by the interviewer immediately
distribution of the costs and benefits associated with its afterwards. The notation I1, I2, IX is used in the results section
implementation across policy actors and society, which in turn as a code for the various key informants. I also reviewed official
affects the level of support or opposition to the policy from various and academic publications and grey material on family
stakeholders (Kingdon 1984). Characteristics of a policy that planning policy in Kenya, reports of relevant meetings, and
affect its acceptability include its implications for vested interests, the theoretical literature on budget and policy processes.
the level of public participation it involves, the resources required I investigated the factors affecting the policy space for reform
for implementation and the length of time needed for its impacts using the framework developed by Grindle and Thomas (1991).
to become visible (Grindle and Thomas 1991). I also carried out textual analysis (Ulin et al. 2005) of interview
In Grindle and Thomas’ model, the various factors interrelate transcripts to gain insights into the experiences of the different
in the following ways. Contextual factors shape the circum- individuals who played key roles in the policy process, and the
stances of decision making by policy elites concerning particular narratives they used to explain the importance of family
policies at particular times. These decisions in turn shape the planning as a policy issue.
characteristics of the policy, and public and bureaucratic While carrying out the analysis, I compared and triangulated
incentives to support or oppose it. These incentives in turn data from different key informant interview transcripts with
shape decisions by policy makers and policy managers about written resources to assess their validity and to mitigate the
resource allocation, and explain how prioritization and imple- impact of biased or partial testimony from key informants.
mentation may fluctuate over time. Though the framework was Where discrepancies and information gaps were found, I carried
initially developed for analysing processes of agenda setting, out further investigation through telephone interviews with key
decision-making circumstances directly affect policy makers’ informants and grey literature investigations, to resolve incon-
and managers’ decisions about subsequent implementation, for sistencies and address omissions.
example where shifts in perceptions of the issue among policy
elites affect decisions about resource allocation. Importantly, as
Figure 1 illustrates, policy makers can widen the policy space Results
they operate within by taking actions to influence the different This section begins with an overview of family planning policy
factors, for example by building consensus or by forming in Kenya. The remainder of the section examines each of the
coalitions in support of an issue. factors affecting the policy space for family planning, analysing
Indeed, analysis of agenda setting across different contexts the ways in which they helped to expand or contract policy space.
shows that individual politicians and bureaucrats often play a Box 1 summarizes Kenya’s long history of population
central role in championing issues and getting them onto the and reproductive health programmes. The first Population
policy agenda, in addition to non-government advocates Policy was introduced in 1967, however government involvement
(Grindle and Thomas 1991; Shiffman 2007). Such analyses in contraceptive service provision did not begin in earnest until
also show that the level of success of advocacy initiatives the 1980s (Chimbwete and Zulu 2003). During the 1980s
depends on a combination of factors including: clear indicators and early 1990s, the Kenyan government demonstrated consider-
to show the extent of the problem, the presence of political able commitment to family planning, through the development
entrepreneurs to champion the cause, and the organization of of national policies and guidelines, involvement of high-level
attention-generating focusing events; as well as the political politicians, the establishment of the National Council for
acceptability of policies (Shiffman 2007). Successful advocacy Population and Development (NCPD) in the Office of the Vice
may also require the ‘framing’ of contested or neglected issues President, and support for increased distribution of contra-
in a way that legitimizes them as an important issue for ceptives through governmental and non-governmental health
governments to address (Schön and Rein 1991; Joachim 2003), facilities, and extensive information, education and communica-
appealing to prevailing social norms (Shiffman 2007) and tion (IEC) campaigns (Ajayi and Kekovole 1999; Blacker 2006).
employing policy narratives, or stories, that simplify issues and Service provision expanded impressively during this period,
persuade others of their importance (Roe 1991; Keeley 2001). and the contraceptive prevalence rate in Kenya increased from
This case study has implications for government and non- 7 to 27% between 1980 and 1989 (Ajayi and Kekovole 1999).
governmental advocates aiming to sustain commitment to International factors played a leading role in this original
existing policies in shifting national and international contexts, expansion of policy space for family planning, with external
particularly policies relating to contraceptive services and other actors advocating for and supporting the implementation of the
neglected sexual and reproductive health issues. population policy. At this time, donors covered the costs of all
government and non-government contraceptives and IEC
campaigns. During the second half of the 1990s, however,
external funding for services and IEC declined, in the context
Methods of a shift in priorities to HIV and AIDS and donor fatigue
The material for this case study is based on 13 semi-structured (Aloo-Obunga 2003; NCPD 2003; I5; I13).
interviews and three unstructured discussions carried out The Kenyan government was slow to respond to the shifting
during 2006 and 2007 with high-level officials and programme international aid allocations. Combined with poor management
staff from government ministries and agencies, international of commodity procurement between the Ministry of Health and
non-governmental organizations (NGOs), national NGOs, a the Kenya Medical Supplies Agency (KEMSA)2 (I13; I4), the
bilateral donor and an academic with expertise in demography unreliable and dwindling international funds were a cause of a
333
Box 1 The history of family planning policy and programmes in Kenya

1962 Family Planning Association of Kenya (FPAK) established
1967 Government of Kenya’s first population policy, but contraceptive services and Information, Education and Communication
(IEC) mainly provided by the private sector
1975 The government launched a 5 year Family Planning Programme
1982 The National Council for Population and Development was established in the Office of the Vice President
1984 First National leader’s Population Conference in Nairobi
1994 United Nations International Conference on Population and Development (ICPD), Cairo
1996 NCPD published its National Population Advocacy and IEC Strategy for Sustainable Development 1996–2010
1997 National Reproductive Health Strategy published
2000 NCPD published the second Population Policy for Sustainable Development
2003 Kenya Demographic and Health Survey generates deteriorating indicators (published in 2004)
2004 NCPD became a semi-autonomous agency under the Ministry of Planning and Economic Development, the National
Coordinating Agency for Population and Development (NCAPD)
2005 The budget for 2005/6 presented to parliament and passed, allocating Kenyan government funds to family planning for the
first time
2007 National Reproductive Health Policy published
Sources: Ajayi and Kekovole (1998); Blacker et al. (2005); Aloo-Obunga (2000); NCPD (2000).
considerable weakening of government and voluntary sector as a result of these efforts, government funds were allocated to
contraceptive services (I2; I7; I4). In 1996, the NCPD launched contraceptives for the first time in Kenya’s history. The incor-
a National Population Advocacy and IEC strategy for poration of contraceptive programmes into the national budget
Sustainable Development 1996–2010, but this strategy floun- demonstrates national commitment (Shiffman 2006), and
dered when funding from UNFPA was withdrawn in 2000 (I5; enhances the potential for sustaining public programmes in
I6; The Global Gag Rule Project 2006). Some clinics suffered the face of potential fluctuations in external funding. The
from commodity stock outs and lack of method choice during government allocation for this line increased to 300 million
the early 2000s, while others closed altogether (I2; I4; I7). The Kenyan shillings, or US$4.17 million, in the 2006/7 budget.4
Kenya Service Provision Assessment Survey of 2004 found that However, it should be noted that this is still only around one-
in the 5 years preceding the survey, the proportion of health third of the cost of Kenya’s public sector provision of family
facilities offering any method of family planning declined from planning commodities according to 2000 projections (Ministry
88 to 75% (NCAPD et al. 2005). of Health 2003), and proponents of family planning continue to
The 2003 Kenya Demographic and Health Survey (KDHS) seek public funding from increased national allocations and
results revealed a stall in fertility decline at 4.8 in 1998–2003, from devolved government funds.
and the rate actually rose for women who had not completed
primary education (Blacker et al. 2005; CBS et al. 2005; Westoff Factors affecting policy space
and Cross 2006). The 2003 KDHS revealed increases in unmet
This section examines how policy elites interacted with each of
need for contraception and high contraception discontinuation
the three sets of factors in the policy space framework, to assess
rates (Blacker et al. 2005). These trends caused concern among
how each influenced the contraction and expansion of policy
national and international actors about the implications for the
space over time, ultimately leading to the inclusion of contra-
rate of population growth in Kenya.3 In 2004, UN predictions of
ceptive commodities in Kenya’s 2005 budget. Table 1 sum-
Kenya’s population by mid-2050 were revised from 48 to 70
marizes contextual factors, policy circumstances and policy
million, based on these new figures (Cleland et al. 2006).
characteristics, comparing their impact on policy space during
Various societal, economic and demographic factors may have
the second half of the 1990s with the years since 2000.
contributed to the worsening fertility and contraceptive use
trends, and there are differences of opinion among analysts
about the impact of declining donor resource allocations for (1) Contextual factors
contraceptives and weakening service delivery (Blacker et al. Changes over time in the political, bureaucratic, national and
2005; Bongaarts 2005; Westoff and Cross 2006). But in any international context had a major impact on the room for
case, the new data provided powerful evidence for reproductive manoeuvre open to proponents of family planning within the
health proponents, and catalysed a series of advocacy initiatives bureaucracy. Table 1 shows how, during the mid-2000s, there
with the aim of influencing the government to prioritize were shifts in all these areas that either increased opportunities
contraceptive services and allocate public funding to contra- for family planning to be prioritized within government, or
ceptive commodities. The advocacy initiatives included meetings reduced the contextual constraints against this occurring. The
with parliamentarians and informal advocacy in government role played by policy actors in working with these shifts and
budget meetings. A line item for contraceptive commodities was building on them is outlined in the text, below.
eventually included in the 2005 national budget, allocating 200
million Kenyan Shillings, or US$2.62 million.4 Influences on policy elites
The new budget line signifies a widening of policy space after Analysts of the national political environment for family planning
its contraction in the 1990s. Advocates had mobilized concern policy in Kenya contend that commitment to the issue by policy
among key decision-makers about the KDHS 2003 results and elites tended to be ambivalent during the 1960s and 1970s,
334
Table 1 Factors affecting policy space for family planning in Kenya
Mid to late 1990s, Policy space contracting Early 2000s, Policy space expanding
1. Contextual factors
Influences on policy elites # Lack of response to negative donor " Religious opposition becoming less vocal
funding trends by high-level politicians
# Religious opposition to contraceptives
" Government consensus building with
religious groups
Change of government in 2002 # Shortage of government resources " New government increasing
allocated to health sector resources to the health sector
" Passive support from high-level politicians
Bureaucratic # Conservative budget officials " Mandate and influence of NCAPD

# Intra- and inter-sectoral competition " Concern about weak service delivery
for resources within Ministry of Health
# Conservative budget officials
# Intra- and inter-sectoral competition
for resources
" Introduction of the MTEF
International # Vertical HIV and AIDS funding " Financial and technical support for
# Prioritization of HIV and AIDS family planning advocacy from international
# Reduced donor funding for NGOs and donors
contraceptive services and IEC
Availability of policy evidence " Availability of new evidence of a decline in

family planning
2. Policy circumstances # HIV and AIDS became a policy crisis, " HIV and AIDS policy is making a gradual transition
drawing attention and funding away from ‘crisis’ policy making to ‘politics-as-usual’
from family planning
3. Policy characteristics # Lack of mobilized support from users # Lack of mobilized support from users of
of contraceptive services contraceptive services
# Some religious sensitivity about " Decreasing religious sensitivity about contraceptive
contraceptive services services
# Vested interests undermining policy # Vested interests undermining policy implementation
implementation
#: Factors constraining or contracting policy space.

": Factors expanding policy space.
and that this was strongly influenced by contextual factors such international population funding than out of genuine conviction
as prevailing cultural and religious attitudes. During this period, (Chimbwete and Zulu 2003).
there was considerable popular opposition to contraceptives and During the 1980s, President Daniel Arap Moi appears to have
to population control in Kenyan society, especially outside the been less troubled than his predecessor by religious and cultural
narrow class of urban ‘modernising elites’ (Ajayi and Kekovole reservations about family planning, which enabled him to take
1999; Chimbwete and Zulu 2003). This included opposition to important measures to ensure effective implementation of the
the use of contraceptives from religious groups and from pro- population policy. Moi appears to have been more influenced
natalist attitudes associated with tribal politics. During this by neo-Malthusian arguments, using them in a number of public
period, some technocrats were convinced by arguments from the statements in support of the issue (Ajayi and Kekovole 1999;
international population control lobby about the beneficial Chimbwete and Zulu 2003; Blacker 2006). In addition, concerns
impacts of lowering fertility rates for economic development, about economic stagnation and heightened pressure from donors
but key policy elites expressed scepticism about family planning such as the World Bank also pushed Moi’s government into
on cultural, religious and pro-natalist grounds (Ajayi and prioritizing family planning (Ajayi and Kekovole 1999;
Kekovole 1999; Chimbwete and Zulu 2003). President Jomo Chimbwete and Zulu 2003). The government-led services and
Kenyatta is said to have never fully reconciled contraception IEC campaigns sparked a backlash from some religious orga-
with his cultural and religious attitudes, and believed that nizations and community leaders, who made public statements of
Kenyan society was too opposed to contraceptives for the opposition to the policy. However, the government and repro-
government to openly promote them or directly provide services. ductive health NGOs worked to create a supportive environment
Instead, he introduced the population policy more to impress and for family planning and population policies by sensitizing
build links with the international community and access religious organizations, the public and the media to the issue
335
(I5; I7; I14; I17). When multi-party elections were reintroduced Memorandum tabled by NCAPD in the same year, which called
in the early 1990s, all political parties included population for the government to make renewed efforts in family planning.
issues in their manifestos (Ajayi and Kekovole 1999), demon- In addition, one senior official in the Ministry of Health and
strating the success of these campaigns. one donor argued that the change of administration allowed
However, Moi’s commitment had significant limits, as family increasing government allocations to the health sector and
planning commodities remained totally funded by donors while made it more likely that politicians would take public health
he was in power, and his government failed to take action in issues such as reproductive health more seriously (I13; I17).
response to declining resource allocations from donors, allowing
implementation and policy evolution to stagnate (NCAPD
2003). This lends weight to the assertion by some key Bureaucratic culture, capacity and institutional arrangements
informants from donor agencies and NGOs (I14; I16; I17) Conservatism, lack of transparency and concentration of deci-
that policy elites in Kenya had never fully taken ownership of sion-making power in the budget process were factors constrain-
family planning policy, even during the 1980s. ing the policy space throughout the period examined. These
By the 2000s, pro-natalist attitudes appear to have much less were significant in preventing the government from allocating
influence on Kenyan politicians than in the past (I2; I6; I8; I14; resources to contraceptives until 2005. One key informant
I15; NCAPD 2006a). The influence of organized religious described budget officers in the ministries of health and finance
opposition to contraceptives has also considerably decreased as being opposed to any display of creativity or decisions that are
(I5; I6; I3; I4; I15). Efforts by the Kenyan government to build perceived as ‘radical’ (I6). Budget officials had to be convinced
consensus with religious groups during the 1990s appear to of the need to innovate by introducing government funding for
have helped to reduce the opposition. The 2000 Population an item that was already funded by donors:
Policy was a milestone in this process, with religious coalitions
being actively involved in the drafting of the policy before it Health indicators such as IMR [infant mortality rates] and MMR
was adopted in parliament (I5). [maternal mortality rates] are declining in Kenya. Our strategic
The increasing visibility of HIV and AIDS-related illness and plan 2005–2010 shows the need to reverse these trends. FP is
mortality over the past decade or so may also have helped to important for reducing MMR. One third of IMR is neonatal
make opposition less vocal. One key informant argued that HIV mortality. Economists understood this. But there was a feeling that
and AIDS have led religious groups to reconsider their partners were already supporting adequately. So why put money to
opposition to family planning, especially the use of condoms: this not drugs or infrastructure? (I4)
’. . . no one has not been affected by HIV/AIDS. Religious groups However, other bureaucratic factors helped to facilitate the new
have decided to lay low and remain silent’. (I5) budget line in 2005. One example is the existence of planning
units in each sectoral ministry, which supported the transfer
Although religious organizations continue to influence the of knowledge, information and skills between the Ministry of
government to exercise caution in their policy making in Planning and the Ministry of Health. The head of the Planning
persistently controversial areas such as abortion, emergency Unit, who was seconded from the Ministry of Planning, had been
contraception and sexuality education, key informants did not involved in the production of the 2003 KDHS, and therefore had a
consider general family planning policy to be affected by good understanding of population and contraceptive use trends,
religious opposition. In addition, high-level politicians in the and a personal stake in the issue (I12). This official was formally
2002–07 government appear to have strong personal convictions responsible for the initial drafting of the Ministry of Health
about family planning. President Mwai Kibaki is known to be budget. The introduction of the Medium Term Expenditure
convinced by economic arguments for limiting population Framework (MTEF) in 1999 (Ministry of Health 2005) may also
growth (Ajayi and Kekovole 1998; Chimbwete and Zulu have been a supporting factor. Since the MTEF allows for annual
2003), and the ministers of health and finance during that increases in resources for existing budget lines, allocations
period were considered to be sympathetic to reproductive health for family planning were much easier to pass in 2006 than in
issues (I4; I6; I17). 2005 (I1; I4; I7; I12; I13).
Since its creation as an agency in 2004, the existence of
Change of government in 2002 NCAPD has been an important factor expanding the policy
Moi’s government failed to address the declining implementa- space for family planning prioritization in Kenya. One key
tion of family planning policy during the 1990s, and it seems informant emphasized that the transformation of the National
that the change of administration in 2002 may have brought an Council for Population and Development into the agency
impetus of change that helped to mobilize action to address this NCAPD led to a considerable improvement in its effectiveness
issue. The new government may have helped to expand policy and policy influence. NCAPD is part of the Ministry of
space by bringing politicians who were more supportive of Planning, but is semi-autonomous, so has greater operational
family planning into key positions. The arrival of the new flexibility than its predecessor (I1; I7). Unlike the Division of
government certainly precipitated two actions that indicate Reproductive Health, NCAPD has a mandate to conduct high-
high-level sympathy for the issue. These were the creation of level advocacy (I2; I6, I14; NCAPD 2005). In 2003, shortly
the National Coordinating Agency for Population and before NCPD made its transition to an agency, a new Director
Development (NCAPD) through an act of parliament in 2004, was appointed, who was charismatic and influential within
with its new advocacy mandate, and the issuing of a Cabinet government and with donors, enabling him to take advantage
336
of this mandate to mobilize resources for family planning commodity stockouts. The ministry did not understand the donor’s
advocacy, and to sell the issue in high-level meetings (I9; I14). cycle. (I14)
The experience of poor implementation within the Ministry
of Health during the late 1990s and early 2000s was also A senior government official on the other hand, argued,
an important factor creating concern about the issue within
the ministry and triggering action to address it. In the Division Donors have no idea of our procurement schedule. You would find
of Reproductive Health and among NGO service providers, lorries arriving at KEMSA without any storage space. (I13)
the policy problem was identified because of stock outs of
family planning commodities from health facilities, leading to While external assistance for service delivery and IEC has
a concern that family planning policy implementation was dropped, international actors have increased their support to
ineffective and action needed to be taken to improve service ‘behind the scenes’ advocacy campaigns to reposition family
delivery. One official in the ministry stated that, planning. This includes the provision of financial and technical
assistance for advocacy on family planning from donors such as
The Ministry of Health had a general feeling that FP implementa- USAID, and of technical assistance from international NGOs
tion was not good enough. (I3) such as the Futures Group and the African Population and
Health Research Center (I2; I14). Since 2000, UNFPA has been
funding improvements in the division of responsibility and
International influences coordination between the Ministry of Health and NCAPD,
Population first made it onto the Kenyan government’s agenda which may have helped them to carry out joint advocacy for
because of the influence of external actors, and even at the family planning (I5). In the past few years, some donors have
height of prioritization of the issue during the 1980s and early been working with the Ministry to strengthen procurement
1990s, the government always relied on external resources to policy, though it is too early to assess the impacts of these
fund policy implementation (Ajayi and Kerkovole 1998; efforts (I6; I14). A key shift in international engagement
Chimbwete and Zulu 2003). As with the national government, between the 1980s and recent years is, therefore, that external
many international donors shifted their priorities to HIV and actors are now trying to create local ownership for family
AIDS during the 1990s, leading to declining foreign aid planning by supporting national advocates of the issue,
allocations for family planning (Aloo-Obunga 2003; NCPD particularly government officials and parliamentarians.
2003). The strong external pressure that had influenced
political elites to prioritize population and reproductive health Availability of policy evidence
issues during the 1980s and early 1990s declined. In addition, The availability of new data in 2003 demonstrating that a
some key informants described a situation of donor fatigue ‘policy problem’ existed was a catalyst for alerting policy
brought on by frustration with poor planning and lack of entrepreneurs to the need for family planning to be reprior-
ownership for family planning in the Ministry of Health. itized. Key informants from the NCAPD, Ministry of Health,
USAID and NGOs pointed to the importance of the 2003 KDHS
Donors got fed up with the lack of planning. DRH used to say, data in identifying and persuading others about the importance
‘‘we have a shortage of pills. UNFPA can give us an emergency of the issue.
drop’’. UNFPA would do this, but 6 months later they’d come back
and ask for another bail out. (I14) The plateau [of contraceptive use and fertility rates] was a critical
turning point. (I1)
Some key informants stated that donor agencies consider IEC The results showed clearly that unmet need for FP had not
to be expensive and lack conviction in its importance and changed for over 10 years. Contraceptive prevalence was the
effectiveness (I6; I2). There appears to have been complacency same. The TFR was beginning to show an increase.
among donors as well as national actors about fertility These figures rang a bell. So we did further analysis. Our finding
transition, and a belief that it would happen naturally without was that there was a shortage of commodities. [. . .] We needed a
the need for sustained interventions. broad program of high-level advocacy to lobby government, partners
and donors. (I2)
Implementation disappeared in the 1990s. There was an expecta-
tion that the transition would continue automatically. Resources Contrary to the previous quotation, those working on the
were moved away. (I1) issue in government had already expressed concern about
Donors no longer wanted to support community-based distribution, declining prioritization of family planning and decreasing donor
questioning its impact. (I2) funding before the KDHS funding before the KDHS results
were available (Ministry of Health 2000; NCAPD 2003). The
Government and donor key informants unsurprisingly dif- publication of this data provided an opportunity and a resource
fered as to where they put the blame for poor coordination and for champions of family planning to use in their advocacy.
commodity stockouts, with a USAID official stating that:
(2) Policy circumstances
[. . .] there was a major problem when the Germans picked up the Since the time of Kenya’s first population policy in the 1960s,
bulk of procurement, but there was a 6 month gap between projects family planning has consistently been regarded by policy elites
which the ministry had not picked up on, so there were almost as an issue of ‘business as usual’ rather than a crisis issue.
337
Government officials repeatedly stated that a difficulty for of supporters for the policy among users of contraceptive
securing prioritization of family planning in the Ministry of services, or the Kenyan public in general (I2; I6; I15; I16). The
Health is that it is not considered to be an emergency, unlike issue of family planning has therefore tended to involve low
other health issues such as epidemics (I6; I3; I4). During political stakes for the Kenyan Government, focusing the costs
the 1990s, the policy space for family planning narrowed and benefits of the policy in the bureaucratic domain.
further, when HIV and AIDS was perceived as a crisis issue In the bureaucracy, there seem to be no significant incentives
(Aloo-Obunga 2003; NCPD 2003). to oppose family planning programmes among government
officials, with the issue being treated as relatively uncontro-
FP has become routine. It has been overrun by other activities like versial (I2; I6). As with other health services, contraceptives
HIV/AIDS. (I4) have relatively intense administrative requirements because of
the need for continuous administrative resources to be allocated
This was exacerbated by a perception that family planning to procurement, storage and distribution of contraceptive
and HIV and AIDS are competing issues that can be traded off commodities, and the technical skills required for effective
against each other. This narrowed the policy space for family service delivery. The capacity of the government to distribute
planning by diverting resources away and undermining contraceptives beyond the district level to the facility level is
acknowledgement of the interdependence between the two weak (I17). As with other areas of the health sector, entrenched
services and the need for integrated policies and programmes. vested interests associated with procurement of family planning
One government official commented that: . commodities play an influential role in undermining the
implementation of family planning services (I14). These
There was the occasional minister who would prioritize HIV over interests continue to frustrate efforts to address inefficiencies
FP. (I2) in procurement and distribution by improving the effectiveness
of KEMSA.
During the 1990s, the deprioritization of family planning seems
to have been reinforced by complacency among government Procurement is worth billions [of Kenyan shillings]. KEMSA
officials and politicians about increasing contraceptive use rates became independent recently. But the Ministry of Health [still]
and declining fertility. There seems to have been a perception that wants it. How to let go of a cash cow? The previous minister
the fertility transition would continue without the need for selected a board chairman, but there is still no board. So there are
continuous government intervention, further undermining the many vested interests. It has become a donor issue. [Donors] keep
sense of importance of family planning as a policy issue. saying, ‘let KEMSA go!’. (I17)
People did not realise what was happening when the decline in FP
The role of advocacy strategies: expanding policy
funding started. For a long time, FP had been doing very well. It
space during the mid-2000s
was at the peak of its success when HIV/AIDS became a crisis issue.
[The decline in government prioritization of FP] was an The previous section has outlined how shifts in context, policy
involuntary decrease. (I5) circumstances and policy characteristics leading up to the mid-
2000s widened the policy space for family planning. This
As demonstrated in Table 1, changing perceptions of policy section focuses on the ways in which policy actors took
makers during the first half of the 2000s helped to create a more advantage of these shifts and widened policy space still further
supportive decision-making environment for family planning. through advocacy initiatives. It also examines strategies that
This involved both an increase in concern among policy makers were used effectively by these advocates in order to influence
about the issue, and an opening up of policy space because of key decision-makers.
changing attitudes to HIV and AIDS as a policy issue. By 2003, HIV From 2003 onwards, advocacy activities led by bureaucrats,
and AIDS was no longer seen as such an urgent crisis, opening the with support from political, international and civil society
policy space for policy makers to focus more on family planning. actors, led to increased recognition of the importance of
contraceptive services among key policy-makers and ultimately
resulted in the introduction of the new budget line for
(3) Policy characteristics contraceptive commodities in 2005. Certain advocacy strategies
As shown in Table 1, the policy content had an important appear to have been effective in encouraging increased
impact on the nature of the policy space for family planning, prioritization of the issue, including combining public and
but did not present a major change during the period examined intra-government advocacy, organizing focusing events, and
by this case study. The decreasing religious opposition to family using a variety of policy narratives to ‘reframe’ family planning.
planning during the 1990s may have helped to increase the The advocacy process involved a range of actors, loosely
acceptability of family planning policy among the electorate, coordinated through family planning and reproductive health
thus expanding policy space slightly. There appears to be committees chaired by the Ministry of Health, with member-
insufficient knowledge about how far family planning is ship including NCAPD, NGOs and donors. The aims were
accepted by individual Kenyans, but generally it is unlikely to multifaceted. They included ‘repositioning’ family planning by
meet strong opposition, although there are high levels of myth raising its profile as a government development priority, by
and suspicion about particular methods in some communities making it genuinely multi-sectoral, and enhancing integration
(Feldblum et al. 2001; I12; I15; 16). However, a defining feature with HIV and AIDS and other reproductive health issues such
of family planning policy is the lack of a mobilized constituency as maternal and child health (I1).
338
When preliminary results from the KDHS were circulated budget officials in the Ministries of Health and Finance to support
by the Central Bureau of Statistics (CBS, since renamed the public funding of contraceptive commodities (I1). NCAPD
Kenyan National Bureau of Statistics) in January 2004,5 the provided data and other support to the DRH in this process.
deteriorating trends were immediately noted, and the NCAPD A line of advocacy was necessary through government hierar-
carried out further analysis of the KDHS findings, with support chies, where the Head of the DRH took advantage of routine
from USAID, and held stakeholders’ meetings to discuss how to meetings to persuade Ministry of Health budget officials and
react (I12). A reproductive health working group, of govern- senior administrators such as the Director of Medical Services of
ment officials, NGOs and donors, chaired by the Ministry the importance of adding family planning to the budget (I8; I17).
of Health, identified a specific goal to address donor depen- In turn, these senior officials had to convey this message to
dency by ensuring the government allocated national resources the Ministry of Finance and during multi-sectoral planning
to family planning for the first time. meetings such as MTEF meetings.
Agenda setting to incorporate family planning in the 2005
budget process involved two advocacy processes. The first was [The Division of Reproductive Health (DRH)] needs to be able to
a public process to influence parliamentarians, senior bureau- push the DMS [Director of Medical Services] who oversees the
crats and the wider public, led by NCAPD. The second budget under the PS [Permanent Secretary] to make these
involved internal government advocacy to influence the decisions. There is a line of command from DRH to DMS to PS
budget process within the Ministry of Health and between to the Ministry of Finance. If Kibaru [Head of the DRH] is not
the Ministry of Health, the Ministry of Planning and the shouting enough to the DMS, the DMS will not be shouting to the
Ministry of Finance. PS, and so on. (I8)
The public efforts centred on the budget process. In April and
July 2005, two advocacy workshops were convened by NCAPD, The decision-making process to allocate government resources
with support from national and international NGOs and donors to contraceptive commodities began when bureaucrats in
(NCAPD 2005, 2006b). Presentations and speeches on the NCPAD, DRH and the Ministry of Health Planning Unit
importance of family planning and the deteriorating trends variously identified the need for the budget line (I4; I1; I2;
were delivered by NCAPD, the African Inter-Parliamentary I7). The process encompassed ministerial budget meetings and
Network on Reproductive Health and the Ministry of Health. the Medium Term Expenditure process and culminated in the
Advocacy materials and presentations (APHRC 2005; NCAPD acceptance of the budget by the Minister of Finance. The
2005) drew both from KDHS data and from evidence on the Planning Unit in the Ministry of Health started the process
correlation between higher contraceptive prevalence rates, lower officially, tabling arguments to the Ministerial Budget
fertility rates, and increased maternal and infant survival Committee charged with formulating the budget. Officials in
published by UNFPA (2003). These workshops targeted ministers, the Planning Unit presented key budget decision-makers in the
senior administrators and budget officials from the Ministries of Ministry of Health, including the Director of Medical Services
Finance, Planning and Health, and parliamentarians (I3; I4; I7). and the Permanent Secretary, with arguments about the need
The workshops were reported in the press, and key informants for the new budget line based on shortfalls in family planning
argue that this public profile of the event helped to persuade key funding from donors and the implications of declining KDHS
officials in the bureaucracy to accept and support the allocation indicators for health and development. In turn, the Ministry of
of national resources to family planning (I1; I2; I6; I7; I14). Health Budget Committee inserted the budget line into the
The exact role played by the parliamentarians is hard to ministerial budget and defended it to the cross-sector MTEF
pinpoint. Key informants involved in the advocacy argued that Secretariat in the Ministry of Finance (I12; I13).
the ultimate aim of targeting MPs was to make them become This intra-government advocacy can be seen as a strategy to
active in the budget process, advocating for resources to be create a sense of urgency about family planning as a policy
allocated to contraceptives (I6, I14). However, the parliamen- problem, in order to create more favourable decision-making
tarians’ direct impact on the budget is extremely small in dynamics. The KDHS data played an important role, and
Kenya, limited only to simply passing or rejecting the whole government economists were said to be receptive to arguments
budget (Mwenda and Gachocho 2003; Gomez et al. 2004; IPAR about the importance of access to contraceptives for improving
2004). Overall, targeting the parliamentarians may have a more maternal health and child health indicators (I2; I4; I12; I13; I17).
long-term effect through strengthening networks of support for The Public Expenditure Review, carried out by the Planning Unit,
reproductive health among politicians and paving the way for provided evidence of the fluctuating resources for family plan-
future work with parliamentarians (NCAPD 2006b), rather than ning, which was presented to the Minister and other senior
directly affecting the budget line. However, it is possible that policy-makers in the Ministry of Health to demonstrate that
the parliamentary workshops may have catalysed the budget donor funds were unreliable and inadequate without national
line decision from the Ministry of Health, by putting senior allocations (I13).
officials in the ministry under scrutiny about their response to In addition to the use of statistics, a wide range of policy
the deteriorating KDHS indicators. In this way, the workshops narratives were employed by different actors in their bid to
can be regarded as ‘focusing events’, which raised the profile of reframe family planning as an important issue for economic
the issue, strengthened networks of sympathetic individuals, growth, development and health, which should be prioritized in
and mobilized action. public policy-making. Arguments were made to counter a
In the parallel, hidden advocacy process, officials within the general perception among policy-makers that sustained fertility
Division of Reproductive Health (DRH) worked to influence transitions occur automatically due to socio-economic change,
339
without requiring government intervention (I2; I6). One key Discussion and conclusion
informant stated that ‘without continual family planning IEC,
This paper examines the challenge of sustaining commitment to
acceptance will decline’ (I6). Another key informant argued
existing policies in politics-as-usual circumstances, rather than
that argued that,
focusing on the agenda-setting phase of policy reform, as is
more common in the field of policy analysis. Policy space for
There is a tendency for poor communities to continually reduce
the issue of family planning in Kenya contracted during the late
their acceptance of FP [. . .]. FP is not readily accepted by the poor
1990s, and subsequently began to expand, due both to
except if they receive information and community-based distribu-
changing contextual factors and the ways in which advocates
tion. Hence the need for continuous IEC provision. (I2)
within and outside government worked with these factors.
The case study approach brings certain limitations to this
Particular individuals used various policy narratives, targeting
paper. In particular, it limits the potential for developing
arguments to particular audiences. Key informants explained
concrete assertions about causality in the policy process or for
how the Head of the Division of Reproductive Health used
generalizing about results. However, the paper does support
‘government language’ and internal advocacy within the
lessons on policy processes from other contexts, and also
Ministry of Health to make sure the issue did not seem radical
provides suggestions for how policy space analysis could be
or part of an external agenda (I7, I6). Advocates appealed to
utilized more widely in health policy analysis.
nationalism (I2; I3):
Firstly, the paper demonstrates the potential for the use of
policy space analysis to identify the challenges and opportunities
NCAPD’s argument to the government is: ‘‘don’t allow the life of
for sustaining or increasing commitment to existing policies
your citizens to hang on the whims of donors’’. We must have a
in politics-as-usual circumstances. This is particularly useful for
Plan B – of government money for family planning. (I2)
cases involving ‘unplanned drift’ of policies in response to trends
The slogan ‘Planning our families is Planning for our Nation’s such as political pressures or opportunities or shifts in funds
Development’ was used in advocacy materials distributed at the provided by global initiatives (Buse et al. 2005).
advocacy workshops (NCAPD 2005). In advocacy initiatives to Policy space analysis can be used both as an analytical
influence government officials and parliamentarians, propo- framework and as a tool that proponents of a policy issue can
nents of family planning focused on the importance of use to map the boundaries of policy space and identify the
family planning for economic and social development and actions that could be undertaken to expand it. Key advantages
poverty reduction, and specifically for achievement of the of the policy space analysis framework include its explicit focus
Millennium Development Goals (APHRC 2005; NCAPD 2005, on the dynamics of decision-making circumstances, the influ-
2006a,b). ence of vested interests in shaping policy outcomes, and the
There were also attempts to transform attitudes among policy agency of policy elites (Walt and Gilson 1994). In this way,
elites about the beneficiaries of contraception, highlighting the policy space is a powerful and under-utilized tool for analysis of
benefits for men, children, low-income families, and the nation the political economy of public health policies.
at large, countering popular assumptions that contraception is a The case study reveals the important role government officials
‘women’s issue’ (APHRC 2005). Some key informants for this can play in sensitizing colleagues within and between ministries
study described the importance of presenting family planning to neglected SRH issues. In Kenya this was dependent on the
as uncontroversial and in line with national Kenyan aspirations existence of highly motivated individuals in both the Ministry
and prevailing gender norms. of Planning and the Ministry of Health, and the existence of
With a couple of notable exceptions, reproductive health rights the NCAPD, which had the independence and mandate to carry
were very rarely used in advocacy materials (APHRC 2005; out advocacy on population-related issues.
NCAPD 2005), and remain controversial even among some This case study provides support for Thomas and Grindle’s
senior government officials (I17). However, population and observation that the ‘policy content’ of population policies,
sexual and reproductive health narratives were adeptly combined involving sustained bureaucratic demands, dispersed benefits
by some key informants, without explicitly referring to and low political stakes, is a likely reason why policies relating to
rights. One example was the argument that high quality contraceptive services tend to evolve slowly and are often poorly
contraceptive services based on a choice of methods are implemented (Thomas and Grindle 1994). In Kenya, the advocacy
essential for acceptance of contraception by the Kenyan public around family planning and the 2005 budget involved attempts to
and for lowering total fertility rates. Shortages of family planning counter this tendency by securing political commitment and
commodities in clinics and poor quality of service delivery government resources for the issue and addressing complacency
were blamed for causing discontinuation of contraceptive by feeding new evidence from the 2003 KDHS into policy.
use and decreasing acceptance of contraceptive methods (I1, The public advocacy events involving parliamentarians and the
I2, I8). media organized by NCPAD and other partners could be seen
as an attempt to move the issue from the purely bureaucratic
In the 1990s, there was unmet need for FP. Many women had arena into the public domain. The case study demonstrates that
unintended children. When they went to a facility, they did not find research examining policy processes would benefit from investi-
the contraceptive of their choice. They went away, meaning to come gating budget processes in more detail, because of their role
back another time, but did not [. . .] When there are shortfalls in in intra-government negotiation and advocacy for planning and
FP commodities, fertility goes up automatically. (I1) prioritizing policy issues.
34 0
3
In accordance with Walt and Buse (2000), Buse et al. (2005) The KDHS 2003 results were published in 2004 but were discussed in
meetings during late 2003 within the Ministry of Planning and
and Cerny (2002), UNFPA, USAID, other bilateral donors and
with other stakeholders.
international NGOs played a vital role in shaping the domestic 4
This figure is based on the conversion rate between Kenyan Shillings
policy process, first helping to contract, then to expand the policy and US Dollars in June 2005.
space for family planning through international support to 5
Although the specific agenda to use advocacy to ‘reposition family
local advocacy activities. However, while the original expansion planning’ began to appear in government documents during 2005,
of policy space during the 1980s was to a large degree led by the agenda appears to have its roots among actors in the then NCPD
and supporting US agencies from before the KDHS figures emerged.
international actors, national government officials and resources A 2003 document that does not feature KDHS results cites the
have played a greater role during the expansion since 2003, need for ‘renewed high-profile public commitment by high-level
providing some evidence of increased national ownership of leaders to reinvigorate FP in Kenya’ (NCPD 2003).
the issue.
The case study supports Shiffman’s assertion of the importance
of both the availability of reliable indicators to demonstrate Acknowledgements
the policy problem and the organization of focusing events The author expresses her appreciation for the financial support
(Shiffman 2007). As predicted by Thomas and Grindle (1994), (Grant HD4) to this study provided by the UK Department for
technical analyses of population problems played a central role International Development (DfID) for the Realising Rights
in persuading policy elites of the need for reform. Research Programme Consortium. The views expressed are not
The government officials and politicians who support family necessarily those of DfID.
planning appear to have been skilled at selecting from the range I am grateful to the key informants from the Ministry of
of policy narratives and tailoring their arguments for different Health, Kenya National Bureau of Statistics, National
audiences. Advocates’ use of arguments to reframe contraceptive Coordinating Agency for Population and Development, Futures
services as non-radical and in tune with national development Group, KAPAH, UNFPA, USAID, Marie Stopes International,
goals and prevailing gender norms can be seen as a useful strategy International Planned Parenthood Federation, and London
for increasing recognition of the importance of these services and School of Hygiene and Tropical Medicine for taking the time to
tackling sources of scepticism about them (Schön and Rein 1991; be interviewed and providing invaluable insights and information
Joachim 2003). Grindle and Thomas (1991) focus on the impli- for this study. I would like to thank Frederick Mugisha for his
cations of policy characteristics for the distribution of costs contributions to the study design and methodology, and Gill
and benefits to key stakeholders. However, where policy issues Walt, Lucy Gilson, Sally Theobald, and Eliya Zulu and other staff
that are highly influenced by social and cultural values are at APHRC for their helpful advice and comments on previous
concerned, including sexual and reproductive health policies, drafts of the paper.
the ways in which policies are framed to stakeholders may be
equally important.
Despite the significant expansion of policy space identified in
this case study, very few of the key informants interviewed were References
of the opinion that contraceptive service delivery and information
Ajayi A, Kekovole J. 1998. Kenya’s population policy: from apathy to
campaigns have returned to the levels of success experienced
effectiveness. In Jain A (ed.). Do population policies matter? Fertility
during the 1980s. Proponents of family planning in Kenya
and politics in Egypt, India, Kenya and Mexico. New York: Population
continue with their efforts to promote family planning as a Council, pp. 113–56.
priority in Kenya and to secure further resources for implementa- Aloo-Obunga C. 2003. Country analysis of family planning and HIV/AIDS:
tion. However, they may not be able to achieve major improve- Kenya. Washington, DC: the Policy Project.
ments in service delivery without successfully tackling the African Population and Health Research Center (APHRC), City Council
weaknesses in government procurement and distribution of of Nairobi, Ministry of Health, and National Coordinating Agency
contraceptive commodities. for Population and Development. 2005. Repositioning population
and reproductive health for the attainment of National and
Millennium Development Goals. Report of the Meeting of
Parliamentarians, Development Partners and Key Stakeholders in
Endnotes Population and Reproductive Health in Kenya, 19 April 2005,
Nairobi.
1
The key informants were from the Ministry of Health [one official in Blacker J, Opiyo C, Jasseh M, Sloggett A, Ssekamatte-Ssebuliba J. 2005.
the ministry’s Planning Unit and two officials in the Division of Fertility in Kenya and Uganda: a comparative study of trends and
Reproductive Health (I3; I4; I13)], the National Coordinating determinants. Population Studies 59: 355–73.
Agency for Population and Development (NCAPD) (I1; I2; I7),
Bongaarts J. 2005. The causes of stalling fertility transitions. New York:
the Kenyan National Bureau of Statistics (I12), the donors USAID
and UNFPA (I14; I17), and the NGOs Planned Parenthood Population Council. Working Paper No. 204.
Federation of America, Futures Group, Kenyan Association for Buse K, Mays N, Walt G. 2005. Making health policy. Maidenhead, UK:
the Promotion of Adolescent Health (KAPAH), and Marie Stopes Open University Press.
International (I5; I6; I8; I15; I16). Additional unstructured Campbell J. 1998. Institutional analysis and the role of ideas in political
discussions were carried out with an international adviser to the
economy. Theory and Society 27: 3.
Ministry of Health (I10), an NGO representative (I11), and a
demographer with expertise on family planning in Kenya (I9). Central Bureau of Statistics, Kenya, Ministry of Health, Kenya and ORC
2 Macro. 2004. Kenya Demographic and Health Survey 2003. Nairobi:
The public sector Medical Supplies Coordinating Unit (MSCU) was
transformed into a parastatal and renamed KEMSA in 2000. CBS, and Calverton, MD: ORC Macro.
341
Cerny P. 2002. Globalizing the policy process: from ‘iron triangles’ Ministry of Health. 2007. National Reproductive Health Policy: Enhancing
to ‘golden pentangles’? Paper presented at the annual convention reproductive health status for all Kenyans. Nairobi: Ministry of Health,
of the International Studies Association, New Orleans, 24–27 Republic of Kenya.
March 2002. Mwenda A, Gachocho M. 2003. Budget transparency: Kenyan perspec-
Chimbwete C, Zulu EM. 2003. The evolution of population policies in tive. Nairobi: Institute of Economic Affairs, Research Paper Series
Kenya and Malawi. Nairobi: African Population and Health No. 4. Online at: http://www.internationalbudget.org/resources/
Research Center, Working Paper No. 27. library/Kenyatransp.pdf.
Cleland J, Bernstein S, Ezeh A et al. 2006. Family planning: the National Council for Population and Development (NCPD). 2000. Family
unfinished agenda. The Lancet 368: 1810–27. Planning Financial Analysis and Projections for 1995 to 2020.
Druce N, Dickinson C, Attawell K, Campbell White A, Standing H. 2006. Nairobi: NCPD, Division of Primary Health Care, Ministry
Strengthening linkages for sexual and reproductive health, HIV and AIDS: of Health, Government of Kenya and the POLICY Project.
progress, barriers and opportunities for scaling up. London: DFID Health National Council for Population and Development (NCPD). 2003. Family
Resource Centre. Planning Achievements and Challenges. Nairobi: NCPD, Ministry
Feldblum P, Kuyoh MA, Bwayo JJ et al. 2001. Female condom of National Planning and Development, Division of Reproductive
introduction and sexually transmitted infection prevalence: results Health, Ministry of Health, Government of Kenya, Family Planning
of a community intervention trial in Kenya. AIDS 15: 1037–44. Association of Kenya and The POLICY Project.
The Global Gag Rule Project. 2006. Access denied: the impact of the NCAPD. 2005. ‘Calling the Nation to Action’. Fact sheet distributed
Global Gag Rule in Kenya. 2006 Updates. The Global Gag Rule at the parliamentary meeting on ‘Repositioning Population
Project. and Reproductive Health for the Attainment of National and
Gomez P, Friedman J, Shaprio I. 2004. Opening budgets to public Millennium Development Goals’, 19 April 2005, Nairobi.
understanding and debate: results from 36 countries. Washington, DC: NCAPD. 2006a. Proposed Workshop of Parliamentary Network on
International Budget Project. Population and Development to be held on 5–6 May 2006:
Government of Kenya. 2006. The Health Sector MTEF 2006/7–2008/9. A summary of initiatives to work with MPs to lobby for support
Sector Working Group Report (Final Draft), February 2006. to population and development issues. Nairobi: NCAPD.
Nairobi: Government of Kenya. Online at: http://www.treasury. NCAPD. 2006b. Launch and Agenda Setting Workshop, Summary of
go.ke/docs/sreports0506/HealthSectorReport.pdf. Proceedings. Nairobi: NCAPD.
Grindle MS, Thomas JW. 1991. Public choices and policy change: the political NCAPD, Ministry of Health (MoH), Central Bureau of Statistics (CBS),
economy of reform in developing countries. Baltimore and London: The ORC Macro. 2005. Kenya Service Provision Assessment Survey 2004.
John Hopkins University Press. Nairobi: NCAPD, MoH, CBS and ORC Macro.
Institute of Policy Analysis and Research (IPAR). 2004. Budgetary Schön DA, Rein M. 1994. Frame reflection: toward the resolution of
process in Kenya: enhancement of its public accountability. IPAR intractable policy controversies. New York: Basic Books.
Policy Brief 10(1). Nairobi: IPAR.
Shiffman J. 2007. Generating political priority for maternal mortality
Jain A. 1998. Do population policies matter? Fertility and politics in Egypt,
reduction in 5 developing countries. American Journal of Public
India, Kenya and Mexico. New York: Population Council.
Health 97: 796–803.
Joachim J. 2003. Framing issues and seizing opportunities: the
Thomas JW, Grindle MS. 1994. Political leadership and policy
UN, NGOs, and Women’s Rights. International Studies Quarterly 47:
characteristics in population policy reform. Population and
247–74.
Development Review 20 (Supplement: The new politics of population:
Keeley JE. 2001. Influencing policy processes for sustainable livelihoods: conflict and consensus in family planning): 51–70.
strategies for change. Lessons for Change in Policy and Organisations,
Ulin P, Robinson E, Tolley E. 2005. Qualitative methods in public health:
No. 2. Brighton: Institute of Development Studies.
a field guide in applied research. San Francisco, CA: Jossey-Bass.
Kingdon JW. 1984. Agendas, alternatives and public policies. New York:
Walt G, Buse K. 2000. Partnership and fragmentation in international
Harper Collins.
health: threat or opportunity? Tropical Medicine and International
Mayhew S, Walt G, Lush L, Cleland J. 2005. Donor agencies’ Health 5: 467–71.
involvement in reproductive health: saying one thing and doing
Walt G, Gilson L. 1994. Reforming the health sector in developing
another? International Journal of Health Services 35: 579–601.
countries: the central role of policy analysis. Health Policy and
Ministry of Health. 2000. Family planning and reproductive health commodities
Planning 9: 353–70.
in Kenya: background information for policymakers. Nairobi: Division of
Westoff CF, Cross A. 2006. The stall in the fertility transition in Kenya. DHS
Primary Health, Ministry of Health, Government of Kenya.
Analytical Studies No. 9. Calverton: ORC Macro.
Ministry of Health. 2003. Contraceptive Commodities Procurement Plan
2003–2006. Nairobi: Reproductive Health Advisory Board, Ministry
of Health, Government of Kenya.
342
Articles
Child survival gains in Tanzania: analysis of data from

demographic and health surveys
Honorati Masanja, Don de Savigny, Paul Smithson, Joanna Schellenberg, Theopista John, Conrad Mbuya, Gabriel Upunda, Ties Boerma,
Cesar Victora, Tom Smith, Hassan Mshinda
Summary
Lancet 2008; 371: 1276–83 Background A recent national survey in Tanzania reported that mortality in children younger than 5 years dropped
Ifakara Health Research and by 24% over the 5 years between 2000 and 2004. We aimed to investigate yearly changes to identify what might have
Development Centre, Ifakara, contributed to this reduction and to investigate the prospects for meeting the Millennium Development Goal for child
Tanzania (H Masanja PhD,
survival (MDG 4).
Prof D de Savigny PhD,
P Smithson MPH,
J Schellenberg PhD, Methods We analysed data from the four demographic and health surveys done in Tanzania since 1990 to generate
H Mshinda PhD); Swiss Tropical estimates of mortality in children younger than 5 years for every 1-year period before each survey back to 1990. We
Institute, Basel, Switzerland
estimated trends in mortality between 1990 and 2004 by fitting Lowess regression, and forecasted trends in mortality
(D de Savigny, Prof T Smith PhD);
London School of Hygiene and in 2005 to 2015. We aimed to investigate contextual factors, whether part of Tanzania’s health system or not, that
Tropical Medicine, London, UK could have affected child mortality.
(J Schellenberg); World Health
Organization, Dar es Salaam,
Tanzania (T John MSc); Ministry
Findings Disaggregated estimates of mortality showed a sharp acceleration in the reduction in mortality in children
of Health and Social Welfare, younger than 5 years in Tanzania between 2000 and 2004. In 1990, the point estimate of mortality was 141·5 (95% CI
Dar es Salaam, Tanzania 141·5–141·5) deaths per 1000 livebirths. This was reduced by 40%, to reach a point estimate of 83·2 (95% CI
(C Mbuya MPH, G Upunda MPH); 70·1–96·3) deaths per 1000 livebirths in 2004. The change in absolute risk was 58·4 (95% CI 32·7–83·8; p<0·0001).
World Health Organization,
Geneva, Switzerland
Between 1999 and 2004 we noted important improvements in Tanzania’s health system, including doubled public
(T Boerma PhD); and University expenditure on health; decentralisation and sector-wide basket funding; and increased coverage of key child-survival
of Pelotas, Pelotas, Brazil interventions, such as integrated management of childhood illness, insecticide-treated nets, vitamin A supplementation,
(Prof C Victora PhD) immunisation, and exclusive breastfeeding. Other determinants of child survival that are not related to the health
Correspondence to: system did not change between 1999 and 2004, except for a slow increase in the HIV/AIDS burden.
Honorati Masanja, Ifakara Health
Research and Development
Centre, Kiko Avenue, Plot N 463, Interpretation Tanzania could attain MDG 4 if this trend of improved child survival were to be sustained. Investment
Mikocheni, Dar es Salaam, in health systems and scaling up interventions can produce rapid gains in child survival.
Tanzania
hmasanja@ihrdc.or.tz
Funding Government of Norway.
Introduction birth-history surveys to obtain direct retrospective

The Millennium Development Goal (MDG 4) to reduce estimates of child mortality.11 Such national surveys are
mortality in children younger than 5 years by two-thirds done every 4–5 years and generally include measures of
between 1990 and 2015 has come into focus in recent coverage for priority child-health interventions.12 The
years as a galvanising force to align global and national surveys are standardised by national bureaux of statistics
efforts towards poverty reduction and better health.1–4 such as demographic and health surveys (DHS), which
Much of the current burden of mortality in children are sponsored by USAID, and multiple indicator cluster
younger than 5 years in low-income countries is surveys, which are sponsored by UNICEF. More than
preventable if effective coverage of available cost-effective 40 national mortality surveys from the 60 priority
interventions can be achieved.5 However, global countries will be available in 2005–07,12 one of the first of
assessments of the 60 priority countries where most which is from Tanzania.
children younger than 5 years die show that very few are In 1990, mortality in children younger than 5 years in
on track to reach MDG 4.1,6 Many of these countries are in Tanzania was 141 per 1000 livebirths; thus, Tanzania’s
sub-Saharan Africa, where little or no reduction in MDG 4 is to reduce this to 47 per 1000 by 2015. In
mortality in children younger than 5 years was evident Tanzania, demographic and health surveys were done
throughout the 1990s. Since 2000, global health initiatives in 1992, 1996, 1999, and 2005.13–16 The first three surveys
and resources for health have increased sharply,7 which showed that the rate of child mortality throughout
has increased coverage of life-saving child health the 1990s was high but static, oscillating between 141 and
interventions in several countries.8–10 We would therefore 147 deaths per 1000 children (table 1). The most recent
expect to see more evidence of progress towards MDG 4 survey, from late 2004 and early 2005, showed that the
in such settings in the mid-decade assessments. probability that a child would die before they reached
Since registration systems in sub-Saharan Africa have their fifth birthday fell by 24%, from 146·6 (95% CI
low coverage, most countries rely on periodic national 128·4–164·8) deaths per 1000 in 1999 to 112·0 (95% CI
1276 www.thelancet.com Vol 371 April 12, 2008
343
Articles
102·6–121·5) deaths per 1000 in 2004 (p<0·02).15 Similarly,

Midpoint of period Sample size Mortality (deaths per 1000)
the probability of dying before the first birthday (data not (households)
shown) fell by 31% from 99 to 68 deaths per 1000 over the
1987–9213 1990 8327 141·2 (95% CI 128·1–154·3)
same period. Reductions in mortality were concentrated
1991–9614 1993 7969 136·5 (95% CI 124·8–148·2)
in postneonatal infants (ie, those older than 28 days and
1994–99*16 1996 3615 146·6 (95% CI 128·4–164·8)
younger than 12 months) and were greater in rural areas.
2000–0515 2003 9735 112·0 (95% CI 102·6–121·5)
Neither neonatal nor maternal survival increased during
this period. The 24% drop in mortality in children Data from four demographic and health surveys in Tanzania since 1990.13–16 *The 1999 survey was an interim survey of
younger than 5 years, to 112 deaths per 1000, was reproductive and child health, that used a smaller sample size but identical methods for estimation of mortality.
calculated from the average mortality across the 5 years Table 1: Estimates of mortality in children younger than 5 years
before the survey.
Such a decline is unlikely to be due to one factor.17 But in 1991–92, and 22 000 in 2001–02. The sampling of the
what can account for it? What are the prospects now for survey was designed to allow estimates of household
Tanzania to reach MDG 4 over the ensuing 10 years? And variables for the 21 administrative regions of mainland
what can we learn that would help other countries to Tanzania. Household and individual indicators included
accelerate progress towards MDG 4? We aimed to measures of income poverty and performance of priority
calculate the annual rates to examine the pattern of the sectors as defined in a paper on the government’s
reduction in mortality and to see if the point estimate for poverty-reduction strategy.20,21 Data for trends in gross
the year 2004 differed from historical values or from the domestic product (GDP) per person were obtained from
period average. We also investigated Tanzania’s health- the Bank of Tanzania’s annual reports,22 the Penn World
system investments, including coverage of child-survival Tables,23 and the Tanzania public expenditure review.24
interventions between the late 1990s and 2000–04, and
examined other factors, not related to the health system, Statistical analysis
such as national economic growth, poverty reduction, We analysed the raw data from all four Tanzania DHS
food security, climate shock, fertility, maternal education, surveys (1992, 1996, 1999, and 2004)13–16 to generate several
and HIV/AIDS, that could plausibly have exerted large, estimates of mortality in children younger than 5 years
rapid effects on child survival. for every 1-year period before the respective survey back
to 1990, by use of direct methods based on complete birth
Methods histories. For every child recorded in these birth histories,
Data sources we computed survival for every month from birth until
To assess trends in mortality since 1990 we used all four either their fifth birthday or the date of the survey. We
Tanzanian demographic and health surveys, from 1992, grouped periods at risk and deaths for each calendar year,
1996, 1999, and 2004–05.13–16 These were nationally and constructed a separate life table for each year in the
representative cluster sample surveys that covered 8327, birth histories for which sufficient data were available to
7969, 3615, and 9735 households in 1992, 1996, 1999, and show, for a person at each age, the probability that they
2005, respectively. The surveys provided direct estimates would die before their next birthday. This generated
of child mortality through complete fertility (birth) 35 estimates of mortality over the 15-year period from 1990
histories of 32 877 women aged between 15 and 49 years. to 2004. We estimated trends in mortality from 1990
The surveys also provided detailed information about to 2004 by fitting Lowess regression25 of the natural log of
household demographics; asset ownership; dwelling mortality in children younger than 5 years [ln(5q0)] to
conditions; health and nutritional status of women and time with bandwidths ranging from 0·2 (representing
children; coverage of health-care services such as high sensitivity to recent data) to 2·0 (low sensitivity) and
immunisation, insecticide-treated nets, and maternal forecasted this trend for mortality from 2005 to 2015 with
and child health; and current knowledge and practices the same range of bandwidths. We calculated confidence
related to health. Survey data were obtained by trained intervals for probabilities with Greenwood’s formula.26
personnel, with the verbal informed consent of We obtained fiscal-year data on total health spending,
participants. To assess coverage of child-health both on-budget and off-budget, from the public-expenditure
interventions, we also used a 2003 survey on service reviews of the Tanzanian Ministry of Finance and Ministry
provision in Tanzania, which was a nationally of Health and Social Welfare.24 Spending data included all
representative facility-based survey of maternal and child domestic government health spending (including the
health and HIV/AIDS services.18 All the surveys provided government’s contribution to the national health insurance
cross-sectional data on intervention coverage in their fund) and all aid spending on health from official
respective years. documents. We did not include private out-of-pocket
We obtained data for poverty from Tanzanian household expenditure. We adjusted total government health
budget surveys in 1992 and 2002,19 which tracked the expenditure for each year with consumer price-index
progress of the government’s poverty-monitoring deflators on the 1998/99 base year to provide the total
strategy. These surveys sampled 4000 households government health expenditures per person per year. Thus,
www.thelancet.com Vol 371 April 12, 2008 1277
34 4
Articles
Factors not related to health systems included fertility,

200
GDP per person, and rates of poverty. We also examined
180 any major shocks, such as measles or meningitis
Mortality in children younger than 5 years
153·1
141·5
160 epidemics, famine, or increased food insecurity, that
(deaths per 1000 livebirths)
128·2
140 might have affected mortality differently in the 1990s and
108·6
120 after 2000.
100 83·2
Role of the funding source
80
The corresponding author had full access to all the data
60
in the study and had final responsibility for the decision
40 to submit for publication.
20
0 Results
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
Our results for disaggregated annual mortality (figure 1)
show that the rate of reduction accelerated between 2000
Figure 1: Annual mortality in children younger than 5 years from 1990 to 2005 and 2004. In 2004–05, the reduction in mortality
Data are from an analysis of the 2004–05 national demographic and health surveys in Tanzania.15 Dotted line between 1990 and 1999 was 1·4% per year whereas
shows Tanzania’s MDG-4 target of 47 deaths per 1000 livebirths by 2015. Vertical lines show 95% CIs for survival
probabilities.
for 2000 to 2005, this trend accelerated to 10·8% per year
(from regression trend analysis). The point estimate of
200
mortality in children younger than 5 years in 2004
was 83·2 (95% CI 70·1–96·3) per 1000, which was
180 40% lower than typical values seen in the 1990s
corresponding to a change of 58·3 per 1000 in absolute
160
risk (95% CI 32·7–83·9). This raises the question: is
Mortality in children younger than 5 years
140 MDG 4 more achievable than was previously appreciated?

(deaths per 1000 livebirths)
Figure 2 shows the family of smoothed regressions of the

120
combined disaggregated mortality data from all four
100 demographic and health surveys with extrapolation
to 2015 under different weightings for the recent past. All
80 these weighting projections suggest that MDG 4 is within
60 reach in Tanzania by 2015.
We compared the status of selected health-system
40 factors across the major functions of governance,
financing, resource allocation, and service delivery
20
for 1999 and 2005. Health systems improved substantially
0 on the basis of most of the indicators that we investigated.
1990 1995 2000 2005 2010 2015 With respect to governance, financing, and resources,
Year
Tanzania adopted a sector-wide approach (SWAp) for
Figure 2: Estimates of annual mortality in children younger than 5 years medium-term and long-term planning, in which a
Data are from reanalysis of four national demographic and health surveys in Tanzania, which included the birth coherent policy and expenditure programme, under
histories of 32 877 women aged 15–49 years in 1992, 1996, 1999, and 2004–05.13–16 The MDG-4 target in 2015 is government leadership, was jointly funded by pooled
shown by the horizontal line. The dotted line shows the rate of reduction needed to reach this target. Lowess
regression forecasts of possible future trends are shown by coloured lines, with red giving most weight and yellow
government and donor partners. A so-called basket
giving least weight to recent data trends. fund, jointly funded by partners, was created to provide
an additional US$0·50 per person to districts as
monetary amounts are represented in the international recurrent financing support. This approach was
dollar, a hypothetical unit of currency with the purchasing implemented in 2000–01 and constituted a major change
power that the US dollar had in the USA at a specific time. in the health system that decentralised substantial
We obtained total health expenditure including private financial resources for the first time. Moreover,
out-of-pocket spending from WHO statistics.27 between 1999 and 2004, we noted a 2·3-fold increase in
We searched for information on all contextual factors total government health expenditure, from US$4·70 to
that might have affected trends in child survival, $11·70 per person. Total health expenditure, including
differentiating factors related to health systems from private expenditure, increased from US$23 to $29 per
those that were not.28 Factors related to health systems person,27 indicating that most of the growth in health
included the comparative provision, use, or coverage of spending was due to increases in government
key child-survival interventions in 1999 and 2004; relevant expenditure.
policy changes regarding management, decentralisation, On the policy front, many health reforms planned
and services; and trends in public expenditure on health. during the 1990s started to be implemented during
345
Articles
the 2000s, including the sector-wide approach basket

1999
funding; new guidelines, methods, and informatics for 2005
district planning and management; and new policies
Antenatal care
(eg, substitution of more effective first-line anti-malarial
drugs). Under its poverty-reduction strategy, Tanzania’s
Measles immunisation
Ministry of Health and Social Welfare increased the
priority of cost-effective interventions which supported Fully immunised child at 12 months
national decisions and commitments to scale up and
strengthen several key child-survival interventions such Exclusive breastfeeding 0–2 months
as Integrated Management of Childhood Illness (IMCI),
vitamin A supplementation, immunisation, and Child given oral rehydration treatment
insecticide-treated nets. We did not record major gains
Intervention
in numbers of health professionals or physical Iron supplementation
infrastructure for health during this period.

Malaria intermittent preventive treatment
For service delivery, the coverage of interventions
relevant to child survival improved between the 1999
Exclusive breastfeeding 0–6 months
and 2004–05 surveys (figure 3). The most noticeable
changes were vitamin A supplementation (up from 14% Households with at least one mosquito net
in 1999 to 85% in 2005), IMCI (up from 19% to 73% of
districts), households with mosquito nets (up from 21% Districts with integrated management
of childhood illnesses
to 46%), children sleeping under insecticide-treated nets
(up from 10% to 29%), iron supplementation in Vitamin A supplementation
pregnancy (up from 44% to 61%), oral rehydration
Child sleeping under ever-treated
therapy for children (up from 57% to 70%), and exclusive insecticide-treated net
breastfeeding for those younger than 2 months of age 0 10 20 30 40 50 60 70 80 90 100
(up from 58% to 70%) and younger than 6 months (up Coverage (%)
from 32% to 41%). Coverage of other interventions did
not change significantly, since it was already high Figure 3: Comparison of access to key child survival interventions between 1999 and 2005
in 1999 (figure 3). Coverage of prevention of Data are from national demographic and health surveys in Tanzania.
mother-to-child transmission of HIV (PMTCT) and
antiretroviral therapy as of 2005 remained very low, and water was a protected source.29 The total fertility rate
therefore unlikely to have contributed to a reduction in did not change over this period, but the average age at
mortality in children younger than 5 years. As a risk first birth was 19·0 years in 2000, and 19·4 years in
factor for child mortality, rates of underweight and 2002; the rate of adolescent childbearing diminished
stunted children improved from 29% and 44% in 1999, (from 26·1% to 24·6%); and median birth intervals did
to 22% and 38% in 2005, respectively. not change (33·3 and 33·4 months, respectively).
Of all the factors not related to Tanzania’s health Tanzania had a low rate of food-energy deficiency (43·9%)
system that could possibly have affected child survival in 2000.30
(table 2), the only change was a worsening of the The estimated prevalence of HIV in adults aged
manifestations of the HIV epidemic. Over the 5 years 15–49 years was 8%, according to the demographic and
of our study, Tanzania’s national wealth (in GDP per health survey in 1999,16 whereas the first national
person) increased by 93 international dollars: from $819 community-based survey of HIV prevalence in 2003–04
to $912 per person between 1999 and 200423 (or US$256 established the rate to be 7% in adults aged 15–49 years.18
to $303). The proportion of households living below the Urban areas had higher rates of HIV than did rural areas.
poverty line was 36% in 2001–02 and 39% in 1991–92 In 2004, only 3–9% of health facilities ran programmes
(p=0·29). Poverty in urban areas, excluding Dar es for prevention of mother-to-child-transmission of HIV;
Salaam, decreased from 29% to 26% during this period most of these were district and faith-based hospitals and
(p=0·60), whereas that in rural areas dropped from 41% a few health centres. By 2006, only 13% of health facilities
to 39% (p=0·52).29 The educational attainment of adults offered at least one of the four components of the PMTCT
improved only marginally between 1999 and 2004, with programme.31
greater gains for women than for men (table 3).14,15 When we analysed the differentials between coverage
Similarly, literacy rates did not change; about two-thirds of health interventions between 1999 and 2004 using a
of the women were reported to be literate throughout modelling system,32 we extrapolated a 33% reduction of
this period. Population-based statistics on access to safe mortality in children younger than 5 years, from 129 to
water in Tanzania were sparse; those that were available 86 deaths per 1000 livebirths. These effects would mainly
indicated no change between 2000 and 2002 in the be attributable to reduction of postneonatal mortality in
proportion of households for which the main supply of children younger than 5 years.
346
Articles
suggest that Tanzania is on the trajectory necessary to

1999 2004–05
achieve MDG 4 by 2015, for a range of different weightings
Gross domestic product per person US$256 US$303 of past performance in the distant or near past, back
Number living in poverty (food) 21·6% 18·7% to 1990. Our results differ from those of an analysis of all
Number living in poverty (basic 38·6% 35·7% available data from direct and indirect estimates of
needs)
mortality disaggregated into 2-year intervals, including
Food security No change
data before 1990, which concluded that Tanzania would
Climate shock No change
not be able to achieve this goal.6 However, the data from
Total fertility rate (children for 5·6 (5·0–6·1) 5·7 (5·4–5·9)
every mother)
before 1990 can have little bearing on the ability to achieve
a goal for which the starting point is 1990, especially
Low birthweight 133/3206 (3·8%) 344/8551 (3·7%)
since the purpose of the MDG was to elicit changes in
Age at first birth (years) 19·2 (17·4–21·6) 19·4 (17·7–21·8)
trends. To assume that the trend before 1990 continued
Birth spacing (months) 33·3 (28·0–41·0) 33·4 (28·0–42·4)
would imply that setting the goal was futile. In this
Meningitis epidemics None None
specific instance, performance was poor before 1990, and
Measles epidemics None None
inclusion of earlier data biases the conclusion towards a
HIV/AIDS epidemics Increasing
mortality slower improvement. Furthermore, all extrapolations
must, of necessity, assume a degree of continuity in the
Data are number (%), rate (95% CI), or median (IQR), unless otherwise specified. underlying processes, and so tend to over-smooth if a
Table 2: Comparison of non-health system changes between 1999 trend accelerates, as it seems to have done in Tanzania in
and 2005 that might be expected to affect child survival about 2000. Since aggregation of the data into longer
time-units tends to increase the degree of this smoothing,
we disaggregated the data into shorter time units.
Completed primary education Completed secondary Years of schooling The large reduction in mortality evident since 2000
education
immediately raises questions about the quality of surveys
1999 2004–05 1999 2004–05 1999 2004–05 and data and about comparability over time. Additional
Women 46% 49% 5% 9% 6·1 6·2 quality control was provided for the 2004–05 demographic
Men 51% 52% 7% 11% 6·2 6·3 and health survey and its data precisely because fewer
child deaths were recorded than were expected.15
Data are from demographic and health surveys in Tanzania in 1999 and 2005.15,16
Under-reporting bias could also have occurred, for
Table 3: Educational attainment of men and women aged 15–49 years example if maternal mortality increased because of
HIV/AIDS or other factors. The cross-sectional
Discussion demographic and health surveys did not gather
In Tanzania, the most recent demographic and health information about children whose mothers died.
survey in 2005 showed a 24% improvement in child However, the demographic surveillance systems in
survival, with mortality rates in children younger than Tanzania, which track entire populations longitudinally,
5 years down from 147 deaths per 1000 for 1994–99 to also reported reductions in mortality in children younger
112 deaths per 1000 for 2000–04 (p<0·02).15,33 In national than 5 years, which substantiates the data from the
birth-history surveys, these 5-year averages conceal the demographic and health surveys.35 Furthermore,
pattern and degree of change in yearly rates. Since this demographic and health surveys in 1999 and 2005 did
5-year change substantially exceeded 15%, the minimum not detect any major increases in maternal mortality
regarded by Korenromp and colleagues34 as indicative of between these two periods, although such changes would
a true reduction, we decided to calculate the yearly rates be difficult to detect in sample sizes used in the
to examine the pattern of the reduction and to see if the demographic and health surveys. With respect to deaths
point estimate for the year 2004 differed from historical of mothers due to HIV/AIDS, reduced mortality in
values or from period average. children younger than 5 years is probably not an artifact
Our analysis of the annual rates shows a pattern of caused by the under-reported deaths, since the estimated
continuous reduction in mortality reaching 83·2 (95% CI magnitude of this effect in a rural Tanzanian population
70·1–96·3) deaths per 1000 in 2004. Within the with an HIV prevalence of 4·3% would underestimate
2004–05 survey data, five of the six lowest values over the deaths in children younger than 5 years by only 2·3%.36
15 years were recorded in the last 5 years,15 indicating that If we assume that our finding of a reduction in mortality
mortality in this group fell by 40% between 1990 and 2004. for children younger than 5 years is real, what can explain
Based on Tanzania’s 2002 population of 34·4 million, this apparent acceleration of survival in Tanzania after a
this finding suggests that 280 000 children’s lives were decade of high but static mortality rates in the 1990s?
saved between 1999 and 2005 that would otherwise have And can this improvement be sustained? We examined
been lost had the prevailing rate of the 1990s continued. differences in the health system in Tanzania between 1999
Our analyses of data from all four demographic and and 2004 and in external factors that could reasonably be
health surveys, analysed by year of reference, thus expected to have contributed to large survival gains over
347
Articles
this short period. Between 1999 and 2004, Tanzania more interventions, such as antenatal care an immunisation,
than doubled its public expenditure on health; such coverage was already high, and did not change.
increased expenditure has been strongly correlated with Modelling showed that a 33% reduction of mortality in
increased survival in children younger than 5 years in children younger than 5 years could be expected
developing countries, especially in poor people.37 between 1999 and 2004, from 129 to 86 deaths per
Increased public expenditure on health could also be 1000 livebirths. These effects would mainly be in
especially powerful in decentralised health systems when reduction of postneonatal mortality in children younger
such resources are targeted towards essential cost-effective than 5 years. The predicted failure to affect neonatal (and
interventions.38 Tanzania implemented such governance maternal) mortality draws attention to problems with the
shifts towards greater decentralisation in 2000, by continuum of care necessary to achieve MDGs. The
introducing sector-wide capitation grants that gave general scarcity of data and analyses continues to limit
districts substantial financial resources. This was perhaps programme efforts and monitoring of progress.
one of the most important distinctions in Tanzania’s Among factors not related to the health system, gains
health system between the 1990s and the 2000s, since it in wealth would be expected to exert a major effect on
opened opportunities for local problem solving and survival in children younger than 5 years. Tanzania has
provided resources for districts to selectively increase enjoyed many decades of political stability and, in recent
resources for key interventions, as has been shown in years, steady economic growth. Nevertheless, GDP per
pilot studies since 1996.38 person has increased by only 93 international dollars
Decentralisation allowed the introduction and scale-up (US$47) over the 5 years between 1999 and 2004. An
of new interventions such as the integrated management increase of this size corresponds to an expected decrease
of childhood illness, which facilitated adoption of new in mortality in children younger than 5 years of 2·2%, on
treatment policies for malaria that replaced failing the basis of a regression of GDP (in international dollars)
first-line treatments with more effective case management per person and mortality in children younger than 5 years
for the largest single cause of death for children. The for 45 sub-Saharan countries (data reanalysed from WHO
IMCI programme also assisted promotion of the use of statistics).27 Although important, this growth in national
insecticide-treated nets for malaria prevention. Sentinel wealth would be unlikely to account for much of our
districts had piloted the introduction of IMCI from 1997, finding of a 40% reduction in mortality, especially since
with full provision, increased use, and effective coverage the proportion of the population living below the absolute
by 1999–2000.39 Impact studies showed that, after a 2-year poverty line and food poverty line in the 1990s had
follow-up, IMCI was associated with 13% lower child improved only slightly in 2002. Although gains have been
mortality in pilot districts that had health-system made in the education of Tanzania’s current cohort of
strengthening than in other districts.40 Other pilot studies schoolchildren, child-health outcomes are affected by the
in Tanzania showed the high local effectiveness of educational status of parents, which had improved only
insecticide-treated nets for reduction of mortality in marginally by 2004. Early child-bearing and short
children of this age.41 birth-spacing both raise the risk of child mortality, and
Tanzania started nationwide scale-up of the total fertility rate, average age at first birth, adolescent
insecticide-treated nets in 1999 and of IMCI in 2000, and childbearing, and median birth intervals remained
changed its drug policy for malaria in 2001. Since malaria similar in the two periods. Hence changes in fertility
mortality in Tanzania is concentrated in postneonatal probably did not contribute to our findings of a large
infants younger than 5 years,42 the survival gains recorded improvement in child survival.
in the 2004–05 demographic and health survey were We did not find evidence of any major epidemics (for
highest for postneonatal infants, suggesting that example, of measles or meningitis) that might have
malaria-specific mortality reduction has made progress. occurred in the late 1990s but not in the early 2000s.
Moreover, several sentinel sites in Tanzania, which Conversely, adult and child mortality due to HIV/AIDS
monitor cause-specific mortality by use of continuous continued to increase slowly,43 and therefore differentials
longitudinal demographic surveillance systems, also in HIV/AIDS interventions might have affected overall
reported reductions in mortality in children younger mortality, since 25% of children who are born to
than 5 years before the findings of the 2004–05 HIV-positive mothers are infected. The PMTCT
demographic and health surveys, and detected declines programme is a proven cost-effective combination of
in malaria and acute febrile illness deaths in children strategies and interventions that can be tailored to specific
younger than 5 years.34,43 These findings add plausibility local conditions. These interventions and strategies,
to the hypothesis that the collective effect of a multifaceted including voluntary and confidential counselling and
approach to malaria contributed to child-survival gains testing, provision of antiretroviral drugs to HIV-positive
during this period.44 Coverage of other child-survival pregnant women, planning of safe delivery procedures,
interventions, such as vitamin A supplementation,45 and counselling about appropriate infant-feeding options,
exclusive breastfeeding, oral rehydration therapy and can reduce mother-to-child transmission by 50%.
iron supplementation for children, increased. For other However, in Tanzania access to HIV/AIDS interventions
348
Articles
such as voluntary counselling and testing, PMTCT, and We were unable to estimate the relative contributions
antiretrovirals was not yet sufficient as of 2004 to have of different factors in the health system to reduction of
affected child survival on a national scale. Epidemic child mortality since 2000. However, the collective weight
patterns, including HIV/AIDS and its response, can of so many positive changes in the health system, in the
therefore be excluded as an explanation for the reduction absence of other explanations, is compelling. Rather, we
in child mortality, and could even have worked against could ask why we would not expect to see gains in
this trend. survival.5 Broad, multifaceted progress in stewardship,
Nutrition can be determined by health systems (eg, public expenditure on health, decentralised financing,
micronutrient supplementation and other health sector resource allocation, and better coverage of essential
interventions) and by other factors (eg, food insecurity, child-survival services can work synergistically to effect
poverty, climate shocks, and natural disasters). We did important progress towards MDG 4 in low-income
not identify evidence of major events outside the health countries such as Tanzania. Increased health resources
system that could have contributed to changes in combined with strengthening of decentralised health
nutritional status in Tanzania during the study period. systems to ensure that life-saving interventions reach
However, the nutritional status of children did improve those in need is a key child-survival strategy.
slightly, possibly because of better access to various Contributors
general health interventions (eg, IMCI, insecticide-treated HM and DDS led the conceptualisation of the paper with contributions
nets, and vitamin A supplementation), and slight gains from all authors and wrote the first draft. PS compiled statistical data,
and HM, TS, and DDS did statistical analyses. JS, TJ, CM, GU, TB, and
in wealth. Improved nutritional status is likely to have CV contributed to the interpretation and writing of this manuscript. All
contributed to the reduced risk of mortality in children authors have seen and approved the final version.
younger than 5 years. Conflict of interest statement
If we assume that the trend is real, and is due to a We declare that we have no conflict of interest.
strengthening health system and increased access to key Acknowledgments
child-survival interventions, can this trend be continued? We thank the Government of Norway for encouragement and financial
It should be noted that the most recent demographic and assistance.
health survey, in 2004–05, preceded the potential effect of References
increased funding to Tanzania from the Global Fund to 1 Horton R. A new global commitment to child survival. Lancet 2006;
368: 1041–42.
Fight HIV/AIDS, Tuberculosis and Malaria. Although 2 Martines J, Paul VK, Bhutta ZA, et al. Neonatal survival: a call for
the first grants were announced in late 2002, the actual action. Lancet 2005; 365: 1189–97.
programmes that they supported did not begin until 3 Lawn JE, Cousens S, Bhutta ZA, et al. Why are 4 million newborn
late 2004, and the benefits would not have been detectable babies dying each year? Lancet 2004; 364: 399–401.
4 Bryce J, el Arifeen S, Bhutta ZA, et al. Getting it right for children:
in the last demographic and health survey but can be a review of UNICEF joint health and nutrition strategy for 2006–15.
expected to assist the downward trend into the future. Lancet 2006; 368: 817–19.
For children, these funds will boost access to 5 Black RE, Morris S, Bryce J. Where and why are 10 million children
dying every year? Lancet 2003; 361: 2234.
insecticide-treated nets through a national voucher
6 Murray CJ, Laakso T, Shibuya K, Hill K, Lopez AD. Can we achieve
scheme, which is designed to provide the nets to all Millennium Development Goal 4? New analysis of country trends
pregnant women and their newborn babies, which and forecasts of under-5 mortality to 2015. Lancet 2007; 370: 1040–54.
started in late 2004. Scaling up the PMTCT programme 7 Horton R. The coming decade for global action on child health.
Lancet 2006; 367: 3–5.
and antiretroviral therapy started in 2005; programmes 8 Bryce J, Terreri N, Victora CG, et al. Countdown to 2015: tracking
for zinc supplementation and oral rehydration therapy intervention coverage for child survival. Lancet 2006; 368: 1067–76.
started in 2007; and access to improved antimalarial 9 Victora CG, Fenn B, Bryce J, Kirkwood BR. Co-coverage of
preventive interventions and implications for child-survival
treatment through artemisinin combination therapy strategies: evidence from national surveys. Lancet 2005; 366: 1460–6.
in 2007. 10 UNICEF and Roll Back Malaria. Malaria & children: progress in
Since neonatal mortality remains constant and forms intervention coverage. New York, USA: UNICEF, 2007.
an increasing share of the mortality in children younger 11 Morris SS, Black RE, Tomaskovic L. Predicting the distribution of
under-five deaths by cause in countries without adequate vital
than 5 years, it could emerge as a barrier to continued registration systems. Int J Epidemiol 2003; 32: 1041–51.
reductions in mortality and attainment of MDG 4. 12 UNICEF. Countdown to 2015 child survival: tracking progress in
Renewed efforts are being planned to address neonatal child survival. New York, USA: UNICEF, 2005.
and maternal mortality in Tanzania. These efforts will 13 National Bureau of Statistics, Tanzania, and Macro International.
Tanzania demographic and health survey 1991–02. Dar es Salaam,
coincide with a doubling in the sector-wide district basket Tanzania: National Bureau of Statistics and Calverton, MD, USA:
fund for the Tanzanian health system to US$1·00 per Macro International, 1993.
person per year. Such continued efforts at scaling up will 14 National Bureau of Statistics, Tanzania, and Macro International.
Tanzania demographic and health survey 1996. Dar es Salaam,
need concomitant investments in strengthening of health Tanzania: National Bureau of Statistics and Calverton, MD, USA:
systems, including management of human resources, Macro International, 1997.
procurement and supply chain management, health 15 National Bureau of Statistics, Tanzania, and ORC Macro. Tanzania
demographic and health survey 2004–05. Dar es Salaam, Tanzania:
information management, and constant attention to National Bureau of Statistics and Calverton, MD, USA: ORC
enhancing quality of care. Macro, 2005.
349
Articles
16 National Bureau of Statistics, Tanzania, and Measure Evaluation. 32 The Futures Group International. Spectrum policy modeling
Tanzania reproductive and child health survey 1999. Dar es Salaam, system: projecting future requirements for improving reproductive
Tanzania: National Bureau of Statistics and Chapel Hill, NC, USA: health. Washington, DC, USA: The Futures Group
Measure Evaluation, 2000. International, 2008.
17 Victora CG, Schellenberg JA, Huicho L, et al. Context matters: 33 Janson A. Shed some light on darkness: will Tanzania reach the
interpreting impact findings in child survival evaluations. millennium development goals? Acta Paediatr 2007; 96: 781–86.
Health Policy Plan 2005; 20: 18–31. 34 Korenromp EL, Arnold F, Williams BG, Nahlen BL, Snow RW.
18 National Bureau of Statistics, Tanzania, ORC Macro International, Monitoring trends in under-5 mortality rates through national birth
and Tanzania Commission for AIDS. Tanzania HIV/AIDS indicator history surveys. Int J Epidemiol 2004; 33: 1293–301.
survey 2003–04. Dar es Salaam, Tanzania: National Bureau of 35 Adult Morbidity and Mortality Project and Ministry of Health,
Statistics, 2005. Tanzania. The policy implications of Tanzania’s mortality burden:
19 National Bureau of Statistics, Tanzania. Tanzania household budget A ten-year community based perspective. Dar es Salaam, Tanzania:
survey 2000–01. Dar es Salaam, Tanzania: Government of Ministry of Health, 2004.
Tanzania, 2003. 36 Ng’weshemi J, Urassa M, Isingo R, et al. HIV impact on mother
20 Government of Tanzania. Poverty reduction strategy paper. Dar es and child mortality in rural Tanzania. J Acquir Immune Defic Syndr
Salaam, Tanzania: Government of Tanzania, 2001. 2003; 33: 393–404.
21 Government of the United Republic of Tanzania and Vice 37 Houweling TA, Kunst AE, Looman CW, Mackenbach JP.
President’s Office. National strategy for growth and reduction of Determinants of under-5 mortality among the poor and the rich:
poverty (NSGRP). Dar es Salaam, Tanzania: Government of a cross-national analysis of 43 developing countries. Int J Epidemiol
Tanzania, 2005. 2005; 34: 1257–65.
22 Bank of Tanzania. Tanzania: Economic and Financial Indicators. 38 de Savigny D, Kasale H, Mbuya C, Reid G. Fixing health systems.
Dar es Salaam, Tanzania: Bank of Tanzania. http://www.bot-tz.org/ Ottawa, Canada: International Development Research Centre, 2004.
Publications/EconomicIndicators/EconomicAndFinancialIndicator 39 Armstrong Schellenberg JRM, Bryce J, de Savigny D, et al. The
s.htm (accessed on Jan 23, 2007). effect of Integrated Management of Childhood Illness on observed
23 Heston A, Summers R, Aten B. Penn world table version 6.2. quality of care of under-fives in rural Tanzania. Health Policy Plan
Philadelphia, PA, USA: University of Pennsylvania, 2006. 2004; 19: 1–10.
http://pwt.econ.upenn.edu/ (accessed Nov 25, 2007). 40 Armstrong Schellenberg JRM, Adam T, Mshinda H, et al.
24 Ministry of Health, Tanzania. Public expenditure review (PER) Effectiveness and cost of facility-based Integrated Management of
2000–2006. Dar es Salaam, Tanzania: Ministry of Health, United Childhood Illness (IMCI) in Tanzania. Lancet 2004; 364: 1583–94.
Republic of Tanzania, 2006. 41 Armstrong Schellenberg JRM, Abdulla S, Nathan R, et al. Effect of
25 Cleveland WS, Devlin SJ. Locally weighted regression: an approach large-scale social marketing of insecticide-treated nets on child
to regression analysis by local fitting. J Am Stat Assoc 1988; survival in rural Tanzania. Lancet 2001; 357: 1241–47.
83: 596–610. 42 Abdulla S, Adazu K, Masanja H, et al. Patterns of age-specific
26 Collet D. Modelling survival data in medical research. London, UK: malaria mortality in children in endemic areas of sub-Saharan
Chapman & Hall, 1994. Africa. Am J Trop Med Hyg 2007; 77: 99–105.
27 WHO. World health statistics 2007. Geneva, Switzerland: World 43 Ministry of Health and Social Welfare, Tanzania. Annual health
Health Organization, 2007. statistical abstract. Dar es Salaam, Tanzania: Ministry of Health and
28 Habicht JP, Victora CG, Vaughan JP. Evaluation designs for Social Welfare, 2006.
adequacy, plausibility and probability of public health programme 44 Rowe AK, Steketee RW, Arnold F, et al. Viewpoint: evaluating the
performance and impact. Int J Epidemiol 1999; 28: 10–18. impact of malaria control efforts on mortality in sub-Saharan
29 United Republic of Tanzania Research and Analysis Working Africa. Trop Med Int Health 2007; 12: 1524–39.
Group. Poverty and human development report 2005. Dar es 45 Masanja H, Schellenberg JA, Mshinda HM, et al. Vitamin A
Salaam, Tanzania: Mkuni na Nyota Publishers, 2005. supplementation in Tanzania: the impact of a change in
30 Smith LC, Alderman H, Aduayom D. Food insecurity in programmatic delivery strategy on coverage. BMC Health Serv Res
sub-Saharan Africa: new estimates from household expenditure 2006; 6: 142.
surveys. Washington, DC, USA: International Food Policy Research
Institute, 2006.
31 National Bureau of Statistics, Tanzania, and Macro International.
Tanzania service provision assessment survey 2006. Dar es Salaam,
Tanzania: National Bureau of Statistics and Calverton, MD, USA:
Macro International, 2007.
35 0
Social Science & Medicine 71 (2010) 1110e1118

The emergence of community health worker programmes in the late apartheid

era in South Africa: An historical analysis
Nadja van Ginneken a, *, Simon Lewin a, b, c, Virginia Berridge d
a
Faculty of Public Health and Policy, London School of Hygiene and Tropical Medicine, Keppel St, London WC1E 7HT, UK
b
Health Systems Research Unit, Medical Research Council of South Africa
c
Norwegian Knowledge Centre for the Health Services
d
Centre for History in Public Health, London School of Hygiene and Tropical Medicine, United Kingdom
Article history: There is re-emerging interest in community health workers (CHWs) as part of wider policies regarding
Available online 23 June 2010 task-shifting within human resources for health. This paper examines the history of CHW programmes
established in South Africa in the later apartheid years (1970se1994) e a time of innovative initiatives.
Keywords: After 1994, the new democratic government embraced primary healthcare (PHC), however CHW
Community health workers initiatives were not included in their health plan and most of these programmes subsequently collapsed.
Community health worker (CHW) policy
Since then a wide array of disease-focused CHW projects have emerged, particularly within HIV care.
South Africa
Thirteen oral history interviews and eight witness seminars were conducted in South Africa in April
Oral history
Apartheid
2008 with founders and CHWs from these earlier programmes. These data were triangulated with
Task-shifting written primary sources and analysed using thematic content analysis. The study suggests that
Community participation 1970se1990s CHW programmes were seen as innovative, responsive, comprehensive and empowering
for staff and communities, a focus which respondents felt was lost within current programmes. The
growth of these earlier projects was underpinned by the struggle against apartheid. Respondents felt
that the more technical focus of current CHW programmes under-utilise a valuable human resource
which previously had a much wider social and health impact. These prior experiences and lessons
learned could usefully inform policy-making frameworks for CHWs in South Africa today.
Introduction interest waned in the late 1980s and 1990s for several reasons:
structural adjustment programmes; government failure in coun-
Community health workers (CHWs) are increasingly advocated tries where large programmes were operational; and changes in
as a potential solution to overcoming current shortfalls in human ideology (Frankel, 1992; Walt & Gilson, 1990; WHO, 1986).
resources for health in different settings (Chopra, Munro, Lavis, South Africa has a rich history of CHW projects that burgeoned
Vist, & Bennett, 2008; Lewin et al., 2010; WHO, 2008). CHW is an during the repressive regime of apartheid (Table 1 juxtaposes key
umbrella term used for a heterogenous group of lay health workers. historical and project events). Under this racially and politically
Their remit can range from implementing biomedical interventions divided regime, healthcare was intentionally inequitably distrib-
to acting as community agents of social change (Lewin et al., 2010; uted (WHO, 1983). Among the first CHWs were malaria assistants
Werner & Bower, 1982). This paper defines CHWs as people chosen trained in the late 1920s by G.A. Park Ross, a senior health officer in
within a community to perform functions related to healthcare Natal and Zululand (MacKinnon, 2001). In the 1940s, despite an
delivery, who have no formal professional training or degree. CHWs early Smuts government advocating racial segregation, supporters
initially gained global support at the 1978 Alma Ata conference on of social medicine initiated the ‘health centre’ movement. Chief
primary healthcare. They were seen as a key element of the strategy among the politicians involved was Henry Gluckman, the then
to achieve WHO’s goal, set in 1975, of ‘Health for All by the year Minister of Health, who had been influenced by the United King-
2000’. Many CHW programmes were established in the 1970s in dom’s Beveridge Report (1942). The ambitious 1942 National
low- and middle-income countries (Walt & Gilson, 1990). However, Health Service Commission and 1945 Gluckman Report set out to
provide “unified healthcare to all sections of the people of South
Africa”. They addressed both the social and biomedical causes of
* Corresponding author.
disease, responding in part to concerns regarding the effects of poor
E-mail addresses: nadja.vanginneken@lshtm.ac.uk, nvanginneken@yahoo.co.uk
(N. van Ginneken).
health on black migrant labourers’ and miners’ productivity
0277-9536/$ e see front matter ! 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2010.06.009
351
N. van Ginneken et al. / Social Science & Medicine 71 (2010) 1110e1118 1111
Table 1 to the ‘homeland’ governments (van Rensburg & Harrison, 1995).

Timeline of CHW projects and historical landmarks in South Africa. Most of these governments were under-resourced and corrupt and
Date Event thus neglected health service funding.
1913 Natives Land Act (7.3% of South African land Another phase of CHW projects began in the 1970s and
dedicated for Africans’ habitation) continued into the 1990s, established mostly by individuals or
1930s Park Ross: Malaria assistants small civic or religious organisations (Tollman & Pick, 2002). There
1940s Revival of African nationalism. African National
was a growing conviction from the late 1980s that apartheid would
Congress (ANC) rebuilt
1940 Pholela Health Unit founded by the Karks soon end, particularly after African nationalist organisations were
(Kwazulu) unbanned and their leaders released (Baldwin-Ragavan, Gruchy, &
1942e1944 National Health Services Commission London, 1999). This encouraged progressive thinkers and
1945 Gluckman Report: intersectoral academics to develop community initiatives and formulate advice
recommendations for comprehensive health
service. Only health centres and IFCH
which they hoped would inform a new government’s health poli-
established. cies (Price, 1993).
1948 Nationalist Party comes to power In 1994, South Africa welcomed its first democratic government.
1951 Bantu Authorities Act: forcible relocation of Though the government adopted the district health system (DHS)
blacks to ‘homelands’
as the cornerstone of their national health plan, CHWs were not
1951 Valley Trust established
1958 Karks’ exodus included. Subsequently, many CHW projects collapsed as interna-
1960 Sharpville massacre; ANC and Pan African tional donors withdrew their earlier support, or redirected their
Congress (PAC) banned. funds through government departments. In recent years, an
1970 Chalumna/Newlands project started uncoordinated array of CHW programmes has re-emerged, mainly
1976 Soweto uprising: sparks nationwide uprisings
1976 Elim Care Groups started
within healthcare for people living with HIV/AIDS (Friedman,
1977 Steve Biko (leader of Black Consciousness 2005).
Movement) tortured/dies in detention The early history of the 1930s and 1940s CHW projects has been
1977 Public Health Act: dual/segregated healthcare analysed elsewhere (Jeeves, 2000, 2005; Marks, 1997; Tollman,
1979 Health Care TrusteVHW project started
1994). However the late apartheid period (post-1970s) lacks
1980 Valley Trust establishes CHW programme;
SACLA clinic historical analysis. This study aims to explore the factors affecting
1982 National Medical and Dental Association these late apartheid projects’ evolution. This historical analysis
founded to address human rights intends to contribute to current debates on the appropriateness,
1985e1986 3 states of emergency: detentions/repression/ effectiveness and sustainability of CHW initiatives within South
assassinations
1986 SACLA clinic taken over by army
Africa and to similar global debates.
1986 Mamre Community Health Project and Rural
Foundation Health Project started
1987 National Progressive Primary Health Care Methods
Network founded
1989 Relaxing of emergency laws
1990 ANC/PAC unbanned; progressive release of Our approach used oral history interviews (in-depth, open-
imprisoned leaders (including Mandela) ended interviews seeking people’s reconstruction and interpreta-
1991 Mamre/Zibonele projects start CHW tion of events) with founders, coordinators and health workers of
programmes CHW initiatives active during the study period. This approach was
1992e1994 Preliminary ANC health plan; national CHW
workshops/conferences
chosen as people working within CHW projects at this time were
1992 HCTe Brown’s Farm and Agincourt started busy ‘doing’ rather than ‘documenting’. This technique recognises
(CHWs until mid-1990s) individual experience, often missing from standard social histories,
1994 First democratic elections (Mandela elected); and gathers unrecorded information (Perks, 1992).
final health plan
To select interviewees, four contacts known to one of the
1994e2003 Social/health policies promoting development
and intersectoral collaboration (e.g. 1994: authors (SL) helped identify further participants through snow-
Reconstruction and Development Programme); balling. Participants were chosen from urban and rural projects
Many late-apartheid CHW programmes close where documentation was available. Of 54 potential interviewees,
2004 Community Health Policy Framework 39 were selected on the basis of representativeness of professional
background, involvement in CHW programmes, and availability.
Tragically, one key participant, Ivan Toms, died unexpectedly before
(Phillips, 1993). However, the government only adopted the his interview. A total of 38 participants were therefore interviewed
recommendation to establish health centres (Jeeves, 2000, 2005; from 10 projects (Table 2). One author (NvG) visited five provinces
Marks, 1997). Modelled on the rural health centre in Pholela (Western Cape, Eastern Cape, Kwazulu-Natal, Mpumalanga and
(near Durban) initiated by Sidney and Emily Kark e a progressive Limpopo) in South Africa in April 2008 to conduct 10 oral history
and politically well connected medical couple e these centres were interviews. An additional two interviews were conducted by
staffed by community nurses and assistants who treated and phone, and one in London. Eight witness seminars (focus groups
surveyed health problems (Kark, 1951; Tollman, 1994). Only 40 of with two to seven participants from the same project) were also
the suggested 400 centres were eventually built to serve black conducted, partly out of convenience but also because groups may
communities. This service became racialised and gained a reputa- encourage recollection of events. This research’s principal limita-
tion for being a “second class service” (Marks, 1997). tions are that snowballing may not have reached saturation and the
As the Afrikaner National Party strengthened from 1948, the chosen CHW initiatives may not have included the full range of
government withdrew its support from these centres. Many centres contemporaneous programmes.
closed and a number of their founders, including the Karks, went Most interviews were conducted in English; four required an
into exile (Marks, 1997). From the 1960s to 1980s, when the interpreter (Zulu and Shangaan) and professional translation.
bantustans (‘homelands’) became ‘independent’, the responsibility Interviews explored how respondents became involved in CHW
for the forcibly relocated black population’s health care was given programmes, their experiences and their views on CHW
352
1112
Table 2
Details of projects.
Project Comments CHW characteristics Current Status Interviewees
CHW C D/N F
(PERI-)URBAN INITIATIVES
South African Christian Leadership CHW programme in several peri-urban Paid; generic and specialist (rehabilitation) Running 1 3 2 1
Assembly Health Project townships in Cape Town
Mamre Community Health Project Coloured community. 3 components: Paid; specialist (youth, chronic illnesses, Closed 1 1 1
research, CHW project, student teaching hypertension)
(academic)
Health Care Trust e Brown’s Farm Peri-urban township in Cape Town Paid; generic Closed 1
Project
Zibonele Peri-urban township in Cape Town Paid; specialist (women, children) Closed 1 1
(academic)
RURAL INITIATIVES
The Elim Care Group Project Originally focussed on trachoma, Most volunteers; generic: (care-group volunteers, Running 8 1 1
then expanded to general health motivators, CHWs)
(Northern-Province)
Chalumna and NewlandsVillage In the former Ciskei (Eastern Cape) Stipend; generic Closed 1
Health Worker Project Focus on nutrition, immunisation, TB
Health Care Trust e Village Health Xalanga district in the former Transkei Volunteers; generic Closed 1
Worker Project (Eastern Cape)
The Valley Trust Community Care CHW project: part of Valley Trust e a large Volunteers then paid; generic Running 1
Project influential organisation, Kwazulu-Natal.
Focus on health and ecology.
The Rural Foundation Health Nationwide CHW programme for farm Paid by farmers; generic CHW programme 2 4 1
Programme workers (started in Transvaal) closed
N. van Ginneken et al. / Social Science & Medicine 71 (2010) 1110e1118
National Progressive Primary Umbrella organisation for progressive Closed 1 1

Health Care Network community health projects. Set up a CHW
National Training Centre.
Agincourt Community health project with a focus on Paid research assistants Running 1
surveillance in Gazankulu (academic)
C, coordinator; CHW, community health worker; D, doctor; N, nurse; F, founder.
353
programmes’ development. The interview guide was adapted to The motive for helping the black population was not always
incorporate emerging themes. altruistic. There was also fear of a spill-over of ‘black diseases’ to the
Primary and secondary historical sources were obtained from white community. This provided an incentive for a study on health
libraries (UK and South Africa), government databases, the South and urbanisation to assess the impact of black urban migration on
African National Archives in Cape Town and from bibliographies. white city dwellers. Prevention strategies to ‘sanitise’ the most
Grey literature held by interviewees (conference papers, reports, disadvantaged are globally recognised in history across public
minutes, theses, photographs) was reviewed. The interviews were health reform (Pelling, Berridge, Harrison, & Weindling, 2001). This
manually transcribed, coded and analysed by one author (NvG) study ultimately led to the creation of the Centre for Epidemio-
using thematic content analysis. The other authors read selected logical Research in Southern Africa (CERSA), which included
transcripts and commented on emerging themes. progressive thinkers concerned with documenting and addressing
The analysis involved an inductive process to identify emerging the ill health of the underprivileged (F5).
themes. Constant comparison ensured that the themes reflected The Karks’ Community Oriented Primary Care (COPC) model,
the original data. Oral sources were cross-examined with written developed in South Africa, contributed to shaping the 1977 Alma-
material. This methodological triangulation allowed the identifi- Ata declaration and subsequent global community health move-
cation of critical perspectives and emerging themes (Green & ments. It also influenced later projects in South Africa. The Karks’
Thorogood, 2004). visits to Johannesburg and Durban in the 1980s and 1990s
Because some respondents requested anonymity, participants contributed to academics reviving surveillance/research-based
have been kept anonymous. Their quotes are coded according to projects based on the COPC approach. Mamre (in the Western Cape)
participants’ professional background (C: coordinator; CHW: and the Agincourt site (in Gazankulu, now Limpopo Province) of
community health worker; D: doctor; N: nurse; F: founder). This the University of Witwatersrand Health Systems Development
research was approved by the ethics committees of the London Unit, developed and utilised participatory research approaches to
School of Hygiene and Tropical Medicine and the University of create an important body of evidence on community health needs
Cape Town. (Katzenellenbogen, Hoffman, & Miller, 1990; Tollman, 1999).
Leaders of non-academic civic projects drew less influence from
the Karks’ model. Though South Africa did not attend the Alma Ata
Results
conference due to international sanctions, project founders
embraced these principles as they reflected and justified their
How CHW programmes started
efforts. One founder explained why:
The driving force for non-governmental organisations’ (NGO) or “Immediately.I was taken up with the idea. In fact Alma-Ata was
rural health initiatives was often the desire of individuals to address in 1978, so ideas about primary healthcare were floating around at
the health of the under-served black majority. Most leaders of these that time and were starting to get formalised. What was clear to me
initiatives were white doctors or nurses as oppression and poverty was that [our project] had been practising PHC for nearly two
made it difficult for blacks to establish such infrastructures. Ithuseng decades before that. Because if you looked at what the principles of
health centre project set up by Mamphela Ramphele was one excep- Alma-Ata were, things like community involvement, community
tion to this (Ramalepe, 1992). Involvement was sometimes fuelled by development, appropriate health technologies, using a basic
religious conviction (CHW12, C3) or by guilt about their privileged approach, even. basic equity. I mean there were things of course
position compared to racially-oppressed black counterparts (C9). that weren’t being done, but some of those principles were being
Founders of many programmes explained that these projects implemented and I felt very much at home. And for the next decade
arose during a time of growing discontent with apartheid, we really tried to make it a living example of primary healthcare in
expressed through uprisings and demonstrations. The promise that action.” (F2)
the 1977 Public Health Act would expand healthcare for the black
Some respondents, particularly from repressive regime areas
population remained unfulfilled (De Beer, 1984; Digby, 2006). The
like the Ciskei, felt that their projects started in isolation and had
health and social problems experienced by the black majority
few external influences as political sanctions hindered communi-
worsened, as documented in the Second Carnegie Inquiry into
cation and access to information from outside of South Africa:
Poverty and Development in South Africa (1984), to which some
project founders contributed (Wilson & Ramphele, 1989). “I was the only one. Mine was the only community health
During this period, CHW projects often started as single inter- department. There weren’t any others in this province. There was
ventions to address what was seen as the greatest need (Table 2). no such thing as community health work.you know. I was just the
The Elim Care Groups, spear-headed by the Swiss ophthalmologist clinic doctor and then the sense of a community health service
Erika Sutter, responded to trachoma (an eye infection causing grew.” (F8)
blindness). The Newlands and Chalumna projects, led by Trudy
In the late 1980s, conditions became more favourable to infor-
Thomas, a paediatrician, set up a nutrition scheme to respond to
mation exchange. Health activism grew alongside anti-apartheid
kwashiorkor (protein-energy malnutrition). The success of these
activism. A network of local community health organisations
single interventions led them later to address wider health issues in
formed the Progressive Primary Health Care Network (PPHCN).
their communities.
Supported by the National Medical and Dental Association and the
Health projects, such as the Empilisweni SACLA (South African
(Kaiser-Foundation, 1988), it strengthened project cross-fertilisa-
Christian Leadership Association) in an informal settlement outside
tion and collaboration to formulate a future primary healthcare
Cape Town, and Health Care Trust’s (HCT) rural health initiative in
strategy (NPPHCN, 1986).
Cala in the Transkei (now Eastern Cape) were motivated partly by
community requests:
The political nature of CHW projects
“We weren’t looking for long term projects. We were approached to
do these so I think it was something that we had as part of our Most respondents felt that healthcare provision was inseparable
values. We weren’t just ‘go and plonk ourselves’ in communities. It from democracy, reflecting De Beer’s (1984) description of apart-
had to be something that we were approached by.” (F7) heid as the most important ‘disease’ affecting South Africa. Some
354
1114 N. van Ginneken et al. / Social Science & Medicine 71 (2010) 1110e1118
respondents’ conviction that politics and health are connected overload and demotivation of staff. Some projects, however,
explained their involvement in political activism. This put them at successfully involved communities. Brown’s Farm health-clinic lay
significant risk of detention without trial (D1, F3, F6), receiving managerial committees, and Elim and Rural Foundation coordina-
threats (C11) or being harassed (CHW1), but did not hinder their tors were good mediators for enhancing community participation
commitment to work. and dissipating personal and political disputes. However, commu-
Other respondents were not politically active or found it too nity participation was never comprehensive e rather, it was
dangerous. They masked their desire for political change under the restricted to certain tasks within projects. With the exceptions of
banner of healthcare provision while simultaneously challenging SACLA and Elim, projects were established and run exclusively by
the status quo by empowering CHWs to become agents of social people from outside the communities served.
change. The Valley Trust, HCT and SACLA, for example, successfully Experienced programme clinicians and leaders developed and
introduced democratic community structures and elections within undertook hands-on supervision and ongoing training of CHWs
their projects. As one of the founders said: and coordinators. Most CHWs described their supervision as
informative and non-threatening:
“I felt we needed to bring in the social aspects, where we needed to
bring in elements of community involvement. Dangerous stuff at “On the farm [the supervisor] walk with me and the house-visit.
that time, because working with black communities was on the And she look at us. And when something not right she don’t say: ‘He
fringe of social revolution, but luckily primary healthcare permitted he he, no’. She go with us in the clinic. In the container, we sit down,
that ideology.” (F2) and she say: ‘Do you remember, what did you learn?’” (CHW3)
Most of the respondents who were active politically worked in
areas where major political and social injustices had been carried Appropriateness and adaptability
out. The government’s systematic attempt in the 1980s to crack
down on ‘illegal’ squatter areas through encouraging community Adapting the project’s goals to community needs was impor-
riots led to a local SACLA clinic closing in 1986. Individuals working tant. Selina Maphorogo, the first CHW motivator (and later
within projects that had some approval from their ‘homeland’ director) of the Elim Care Groups, re-shaped the project by
governments, such as Agincourt/Manguzi in Gazankulu and Valley adopting culturally-sensitive methods for delivering community
Trust in former Kwazulu, were less likely to be heavily involved in health messages. These methods were reported as effective and
political activism. A project coordinator felt that their work was sustained through the project’s history (Maphorogo, Sutter, &
part of the struggle for democracy. Jenkins, 2003).
Projects in their early stages, or which were geographically
“During apartheid our main struggle was for freedom. Once that
and ideologically isolated, were innovative in their use of
was achieved our remit was over.” (C10)
appropriate technology and training approaches. Many suc-
This statement, which was reflected by many respondents, rai- ceeded despite sanctions in accessing international literature and
ses the question whether the same level of commitment of health low technology resources. They adapted key CHW training guides
workers to communities can be reproduced in a more democratic including the Chinese Barefoot doctors manual (Hunan-Zhong,
political climate in which human rights are less threatened. 1977), Werner’s books Where there is no doctor (1977) and
Helping health workers learn (1982) as well as WHO guidelines
Innovative and experimental leadership, supervision and training (1992). The Rural Foundation and Elim also used UNICEF tools
such as Road-to-Health charts. In the late 1980s, networks
Respondents saw the presence of a charismatic idealistic leader, wanted to create a feasible training model for the post-apartheid
who had a firm development approach, as key to six projects’ era. Emerging training centres (such as the PPHCN learning
success (Valley Trust, SACLA, HCT, Elim, Chalumna/Newlands and centre) were modelled, in part, on the Institute of Family and
Rural Foundation). Ivan Toms, who helped establish the Empi- Community Health (IFCH) (1945e1961) which trained the 1940s’
lisweni SACLA clinic, was seen as an example of such a leader and as health centres.
crucial to the project’s success. In addition to actively defending the In the late 1980s, these projects also adapted to a changing
clinic and community during the mid 1980s’ riots, respondents disease burden in South Africa, moving away from child survival
described how he enlisted and trained lay people to work as CHWs towards chronic diseases and HIV (Bradshaw et al., 1999). SACLA,
or management staff, and empowered community members and Mamre and HSDU trained CHWs who specialised in rehabilitation,
staff to later adopt full managerial and clinical responsibility (C2, chronic disease and HIV. This also coincided with the move to
CHW1). a more selective PHC approach, influenced by international criti-
Respondents from all projects admitted to being experimental. cisms that the comprehensive PHC approach provided few concrete
Supervision, training and management of staff and CHWs were recommendations (Cueto, 2004).
often done on an ad-hoc basis, as outlined by a SACLA doctor: There was an interesting paradox, which several key infor-
mants recognised. These CHW projects, they suggested, experi-
“Those first CHWs were a huge experiment. We were just flying by
enced a ‘golden’ era under the constraints of apartheid and lack of
the seat of our pants, we didn’t know what we were doing. We
political freedoms. Projects were free to respond innovatively to
equipped them with basic medications and dressings and so on.
needs. Funders e whether international donors (for most projects)
And they were fantastic, so they were with the project for many,
or ‘homeland’ governments (as for Valley Trust) e had minimal
many years.” (C3)
requirements. Project leaders felt their impact was greatest on
Project leaders were health professionals or academics with community health and development during apartheid. In contrast,
little experience in management e they were “trying things and they criticised current funding for being constrictive and condi-
seeing if they worked” (C10). Management difficulties sometimes tional, and thus hindering creativity and local adaptation.
developed, such as when SACLA and Rural Foundation became However these divergent views may result from a tendency to
larger and more complex (C3, F10). One Elim report (Annual report, romanticise the achievements possible in times of struggle and to
1980) outlined difficulties of project expansion such as staff resist, as many did globally, the emerging funding bureaucracy of
shortages and inadequate delegation. These caused management the 1990s.
355
Links with communities Many dedicated nurses (including five coordinators and three
founders interviewed) played a significant role in supporting and
In the late apartheid era, some local authorities felt threatened by training CHWs within NGO projects (Clarke, 1991; Mamre, 1992).
the growing influence of projects (Toms, 1987). Also communities Some CHW literature advocates nurses as ideally placed to support
sometimes found it difficult to accept CHW programmes. With CHWs (Buch, Evian, Maswanganyi, Maluleke, & Waugh, 1984;
individuals expressing jealousy regarding CHWs’ status (C6, F10, Roscher, 1990). However some CHW respondents were dispar-
CHW-workshop, 1982). In addition, local expectations were hard to aging of clinical supervision by hospital nurses, one commenting
meet. For example, within the HCT-Cala project, villagers “did not that it was only by “the will of God” that nothing disastrous
get involved unless remuneration for services or products was happened during childbirth (CHW7). One founder suggested why
guaranteed” (Alperstein & Bunyonyo, 1998). Participation fluctuated nurses’ supervision was poor:
and depended on social and power relationships, and satisfactory
“[The nurse facilitators] wouldn’t have that kind of vision or
incentivising, as described in the wider literature (Frankel, 1992).
experience of working with communities in a democratic way, so
Despite these challenges, rural and peri-urban projects reported
they would tend to be bureaucratic and play things by the book.
some success in retaining CHWs in voluntary or partially paid work
They could supervise but it was much more mechanical. And that I
and in community ownership of projects:
found that was not helpful in developing the analytic skills of the
“So we started January 1987. and we had patients that followed CHWs.”(F2)
us from Old Crossroads. Because we moved to New Crossroads, that
CHWs were not necessarily welcomed by formal health system
community welcomed us. So we had patients who followed us, the
staff. Although the South African Nursing Council e the main
chronic patients saying that ‘we can’t do without you.’” (C2)
regulatory body for the profession e supported CHWs in principle
With the shift to employment within the public health system (Marks, 1994), in practice nurses on the ground were reportedly
following the democratic elections, many CHWs felt that their intimidating and rude to them (C8). One doctor interviewed
accountability to the community had changed. They were no longer explains:
flexible community-based workers, but located in health clinics
“The attitude of the nurses is very, very problematic, they’re also so
full-time. A few missed community work intensely (CHW3, CHW6).
hierarchical now. My thesis is that nurses are fighting a feminist
However many CHWs now resented unpaid requests from fellow
battle in their work place, black nurses in particular, because they
villagers:
have been so oppressed and the present health system oppresses
“They used to come to my house asking for help after even after them too, but there’s a bit of a perverse feminism acting there.” (F8)
working hours. I used to help them but now I am unable, I tell that I
This quote illustrates that nurses have, in both the late apartheid
am tired as I have started working at 06h00 in the morning until
and post-apartheid eras, enforced hierarchies within healthcare
16h00 in the afternoon.” (CHW7)
with CHWs in health centres often becoming nursing subordinates
Coordinators (C11, C6) and founders (F2, F8) also commented on (Schneider et al., 2008). These limitations of nursing care in South
changes in CHWs’ attitudes since 1994: Africa have been discussed extensively elsewhere (Marks, 1994;
Stein, Lewin, & Fairall, 2007; Wood & Jewkes, 2006), with Marks
“There’s a very serious materialist dependency. I hate it but I have
(1994) noting that nurses may feel their role is threatened within
to face up that it has happened. It’s not that I am saying it was
primary healthcare, particularly in light of potentially pro-
idealistic, the community health workers were at least as enthu-
fessionalising CHWs.
siastic as I. .I can’t talk about the CHWs now in those glowing
terms. The government now has this huge thing, they’ve got this
small business programme e the pay roll. And the village health The end of an era: the closure of CHW programmes in the 1990s
worker., if the pay doesn’t come out, they ‘toyetoy’, they don’t go
to work” (F8) Respondents noted that the early 1990s were a time of transi-
tion out of the bleak system of apartheid (F9) towards a more
There is likely to be an element of romanticising the past in
idealistic vision of the future (F8). Progressive community health
describing volunteers as only committed during the apartheid era.
leaders were active in formulating health policy which informed
However, introducing a stipend would expectedly reduce a volun-
the ANC’s forthcoming national health plan (NPPHCN, 1992a,
teer’s willingness to work unpaid. The CHWs interviewed, who had
1992b). Sidney Kark also held many meetings with health officials
worked in both the old and new systems, rejected volunteering.
and academics to promote the Community Oriented Primary Care
This is supported by contemporaneous literature (Binedell, 1990;
decentralised approach with strong community involvement (Kark,
HCT, 1982, pp. 45e50) and by recent findings in the Free State
1981).
province (Schneider, Hlophe, & van Rensburg, 2008).
Respondents were surprised that the new 1994 government had
dropped support for CHW projects (A national health plan for South
CHWs within the health system
Africa, 1994), as the 1992 draft of the ANC health plan had favoured
CHW coverage (Slabber, 1992). Despite decentralisation being a key
Whether CHWs can adequately bridge the gap between the
element of the new health plan (van Rensburg, 2004) the incoming
formal health service and the community has long been debated
minister of health, Nkosasana Zuma, believed that CHWs would
internationally (Walt & Gilson, 1990). CHWs in Mamre and Elim
provide second rate care (F2). Rather, a professionalised team of
reported that knowledge of their communities allowed them to
doctors and nurses was visualised. In addition, health service
successfully bridge the gap between researchers and communi-
leadership was preoccupied with structural transformation of the
ties. For example, they explained the purpose of community
health sector (Tollman & Pick, 2002). Respondents felt the 1994
surveys in culturally-sensitive ways. But because some CHWs, for
health plan was poorly informed with regards to CHWs:
example in Elim, officially reported to the nurse-in-charge at the
hospital, they did not bridge the gap with health services but “[It was] highly ambitious, had little connection to and had not
were instead viewed as the lowest level of the health service consulted contemporary projects on what they were actually doing
hierarchy. and achieving.” (F9)
356
Interviewees understood why international funders had redir- This statement equates to what Lund (1993) called a policy
ected their support to the new democratic government. However paradox in describing the 1992 national CHW policy draft. The
they were frustrated that established community-based initiatives policy, she argues, attempts “to give the category of CHW a place,
folded because of the government’s idealistic vision of a profes- but whilst so doing, has failed to recognise diversity, needs and
sional-driven DHS which, in their view, failed to incorporate flexibility, thus invalidating its initial aim.”
adequately existing South African experiences. Some projects Others suggested that implementation of the CHW Framework
survived with meagre charitable contributions (Elim, SACLA). remains rigid and gives insufficient focus to rural areas (Friedman,
A few, such as the Valley Trust, continued to receive government 2005). The Framework has also been criticised for its vague and
support and helped to develop a provincial health plan. conflicting statements about remuneration and responsibility for
The government’s decision in the mid-1990s to provide free CHWs (Schneider et al., 2008). Respondents felt that community
primary healthcare was a further blow to organisations such as the health leaders’ recommendations for appropriate incentives (C10)
Rural Foundation which had relied partly on community contri- were distorted into meagre stipends. Keeping CHWs as volunteers
butions. This decision changed community expectations of projects. may have serious implications for their motivation, retention and
The HIV epidemic also negatively impacted several small the quality of care they provide (F2).
vertical projects. Respondents suggested providing care to people The 2004 CHW framework suggests that the “government
living with AIDS in the pre-antiretroviral drugs era had diverted would provide grants to NGOs who would employ CHWs” (DoH,
funds away from CHW projects. In the late 1990s, when PPHCN 2004). Three respondents identified partnership with civic orga-
struggled for funds, a budget ten times its global budget was nisations as positive but that the framework abdicates government
allocated to it to run the national AIDS programme (F2). The AIDS financial responsibility and diminishes NGOs to stipend distribu-
epidemic also shifted CHWs towards being single-purpose tors (Schneider et al., 2008). In addition, the policy was seen to put
workers e a phenomenon also noted in Britain (Berridge, 1996). insufficient emphasis on supervision, with directors deploring the
A comparison of past and present CHW programmes is shown in fact that so few current budgets included adequate funds for
Table 3. Respondents saw these changes as having shattered the supervision (F2). Quality of supervision, they suggested, is poor and
ideal of community-oriented and comprehensive primary care is performed by inexperienced and overburdened staff.
(Oppenheimer & Bayer, 2007).
Discussion
Has more recent community health policy been informed by the
This research contributes to filling a gap in the history of South
past?
African CHW programmes in the late apartheid period. These
projects had similarities to the earlier community health initiatives
The Community Oriented Primary Care approach informed the
of the 1940s. For example, as a consequence of the socio-political
development of the DHS in the late 1990s. However, it has been
context, primarily outsiders, many of whom were white, middle-
noted that its community focus and epidemiological surveillance
class professionals, started these projects. However, early and late
have not been implemented widely, mainly because of a poor
apartheid projects evolved in different contexts. Due to a more
management capacity and accountability to communities (Moosa,
liberal political leadership, the 1940s projects received government
2006; Tollman & Pick, 2002).
support as potential models for universal health care. In contrast,
More recently, South Africa established a CHW Framework
the late-apartheid projects studied were initiated within an era of
(2004) to guide the development of a national CHW programme.
heightened repression and segregation; were intertwined with the
This aimed to establish cohesion between old and new CHW
contemporaneous wider social aims of democracy and social
community-based organisations and to address the growing crisis
justice, and did not generally receive backing from the state.
of health worker shortage. Interviewees felt “this policy [had] come
Unfortunately, many late-apartheid programmes were poor at
too late” (C10, F3) as this Framework had only drawn upon the
documenting the process and impact of their work, in part because
newer single-purpose fragmented projects not upon apartheid era
of the repressive conditions. Projects of both apartheid periods,
comprehensive community health initiatives. One interviewee, one
which were influenced by Community Oriented Primary Care, were
of the few to be consulted prior to the finalisation of this policy, was
much better at this, given the epidemiological focus of this model.
disappointed:
Our respondents argued that the strong socio-political motiva-
“The policy had ignored the recommendations we had made based tions of the late apartheid period projects were mostly not carried
on the spirit and the experience of the CHW projects [in the through into the post-apartheid period. The current struggle to
1970e1990s]” (C10) redress the economic, health and racial inequalities is not, it could
Table 3
Comparison of past and current CHW programmes in South Africa.
Characteristics 1970e1990s programmesb Current programmes

Supervision and training Experimental but applied; done by experienced and Supervisors are of lower grades and less motivated/committed.
inspiring people Variable training qualitya
Funding International donors. More flexibility of allocation given Government-channelled funding, or charitable fundsa
to project by funders Rigid spending allowance, often determined by fundersb
Remuneration Variable, some CHWs well paid, many volunteers Discontent with voluntary contributionsb, low government
stipenda
Scope Started with a vertical issue, then extended to integrate larger Most are single-disease focused (e.g. HIV, TB) community-based
health issues. PPHCN network established to coordinate projects. organisationsa
Relationship with community Project more dependent on community and linked to activism. Different expectations since democracy (less dependence,
Community more participatory. more individualistic attitude)b
Relationship with health sector Filling a large gap that health service was not providing. Poor coordination among projects/health sector, leading to
Mixed acceptance by health sector. competition and duplication of worka
a
Based on (Friedman, 2005).
b
Based on interviewees responses/contemporaneous literature.
357
be argued, fuelled by the same fervour for action, partly because the Acknowledgements
country is now a stable democracy (Friedman, 2002). In addition,
the changing burden of disease (HIV, TB, chronic diseases) means This research was funded as part of a studentship awarded by
that a national CHW programme would now have to incorporate the Wellcome Trust to the Centre for History in Public Health as
significantly different needs (Oppenheimer & Bayer, 2007). part of its Enhancement Award. We wish to thank the academic
However, a number of contemporary social movements, such as staff at the University of Cape Town who helped facilitate this
that to improve access to treatment for people living with HIV/ research, in particular Leslie London and Margaret Hoffman. Our
AIDS, may have benefited from the leadership and experience of deepest gratitude goes to all participants whom we interviewed
activists from the earlier projects (Ballard, Habib, Valodia, & Zuern, and consulted for this study.
2005).
Though these small-scale programmes were a product of their
References
times, they have important lessons for current CHW programmes
and policy within South Africa and potentially globally. It is sug- Alperstein, M., & Bunyonyo, B. (1998). Collective healthcare. A review of the Health
gested that the now predominant, single-purpose CHWs’ focus on Care Trust’s village health worker project, Cala. In C. Barberton, M. Blake, &
clinical conditions fails to address the social determinants of health H. Kotze (Eds.), Creating action space: The challenge of poverty and democracy in
South Africa (pp. 229e243). Cape Town: Idasa and David Philip Publishers.
(Friedman, 2005). Reinvigorating the political nature of community Annual report. (1980). Elim: Elim careeGroup project.
healthcare by addressing social, economic and environmental Baldwin-Ragavan, L., Gruchy, J.d, & London, L. (1999). An ambulance of the wrong
issues, may have a greater impact in tackling ill health of the most colour: Health professionals, human rights and ethics in South Africa. Cape Town:
University of Cape Town Press.
disadvantaged. Ballard, R., Habib, A., Valodia, I., & Zuern, E. (2005). Globalization, marginalization
Although interviewees claimed that dissatisfaction among and contemporary social movements in South Africa. African Affairs, 107(417),
CHWs with volunteering was minimal during apartheid, this may, 614e634.
Berridge, V. (1996). AIDS in the UK: The making of policy, 1981e1994. Oxford: Oxford
in part, be a romanticisation of the past. In reality, many late University Press.
apartheid programmes had very limited funding and repressive Beveridge, W. (1942). Social insurance and allied services. London: HMSO.
conditions often allowed few opportunities for local people other Binedell, J. (1990). Community health workers talk about their experiences. Honours
Project. Department of Psychology: University of Cape Town.
than voluntary involvement. Within their current work, CHW Bradshaw, D., Mbweu, A. D., Brink, P. A., Walker, A. R. P., Van Der Merwe, P. L., &
respondents indicated dissatisfaction with volunteering. Social Mokhobo, K. P. (1999). Epidemiology of cardiovascular disease in South Africa e
changes in South Africa have created better local opportunities for part I. South African Medical Journal, 89, C38eC44.
Buch, E., Evian, C., Maswanganyi, S., Maluleke, T., & Waugh, R. (1984). Do the primary
people to contribute to health and seeking employment. Poorly- or
healthcare nurses in Gazankulu provide second class cheap care to the poor?.
un-remunerated involvement is perhaps no longer possible, and Conference-paper no.197 South Africa: Carnegie Inquiry into Poverty and
might be seen as exploitative (Lehmann, Friedman, & Sanders, Development in Southern Africa.
2004). Carnegie-papers. (1984). Second Carnegie inquiry into poverty and development in
Southern Africa. Post Conference Series, CapeTown.
The current debate about the potential professionalisation of Chopra, M., Munro, S., Lavis, J. N., Vist, G., & Bennett, S. (2008). Effects of policy
CHWs makes the ideal of ‘bridging a gap’ between the community options for human resources for health: an analysis of systematic reviews.
and the health system more remote. For many younger workers, Lancet, 371(9613), 668e674.
Clarke, M. (1991). The lay health worker as a member of the primary health team.
a CHW position is a stepping stone to a nursing career, not a long- Faculty of Education: Magister Curationis in Community Health Nursing. Rand
term commitment to this cadre (Schneider et al., 2008). CHWs Afrikaans University.
interviewed reported the community contact and trust experi- Cueto, M. (2004). The origins of primary healthcare and selective primary health
care. American Journal of Public Health, 94(11), 1864e1874.
enced during the 1970s and 1980s as crucial to their work, although De Beer, C. (1984). The South African disease. Apartheid health and health services.
the degree of community participation was perhaps not as exten- Johannesburg: Southern African Research Service.
sive and empowering as respondents claimed. Within the current Digby, A. (2006). Diversity and division in medicine: Healthcare in South Africa from
the 1800s. Oxford: Lang.
model, CHWs risk becoming over-medicalised and no longer
DoH. (2004). Community health workers policy framework. Pretoria, South Africa:
embedded in communities. Professionalisation may also blur the D. o. Health.
differences between nurses and CHWs, thereby contributing to Frankel, S. (1992). Community health workers: Effective programmes for developing
countries. Oxford University Press.
power struggles between these two cadres. Furthermore, funding
Friedman, I. (2002). Community-based health workers. Durban: Health Systems
bodies’ centralised control on the remit and extent of these projects Trust.
may curtail projects’ adequate community responsiveness. Friedman, I. (2005). CHWs and community caregivers. Towards a unified model of
Un-surprisingly, interviewees highlighted that good leadership practice. South African Health Review 176e188.
Green, J., & Thorogood, N. (2004). Qualitative methods for health research. London:
and supervision, even though not always achieved, were essential SAGE.
to the success of programmes. Furthermore, community involve- HCT. (1982). Report on a workshop for village health workers. Village health workers in
ment and adequate financial capacity were seen as crucial for South Africa. South Africa: Cecilia Makewane Hospital/Health Care Trust.
Hunan-Zhong, H. (1977). A barefoot doctor’ s manual. London: Routledge & Kegan
sustainability. Indeed, the ethos and funding flexibility of earlier Paul.
programmes were seen to have led to community acceptance, CHW Jeeves, A. (2000). Health, surveillance and community: South Africa’s experiment
job satisfaction and health gain. Many problematic managerial with medical reform in the 1940s and 1950s. South African Historical Journal, 43,
244e266.
issues then are also now further complicated by changes in the Jeeves, A. (2005). Delivering primary care in impoverished urban and rural
burden of disease and in user and provider expectations of health communities: the institute of family and community health in the 1940s.
services. Many ongoing CHW programmes would probably benefit In A. Jeeves, & S. Dubow (Eds.), South Africa’ s 1940s: Worlds of possibilities
(pp. 87e107). Cape Town: Double Storey.
from stronger professional management and better integration Kaiser-Foundation. (28th January 1988). Letter to Dr. Noddy Jinabhai responding to
within the district primary health system. funding proposal. Office for Health and Development in South Africa, Kaiser
The issues identified in this historical analysis of CHW pro- Family Foundation.
Kark, S. L. (1951). Health centre service. In E. H. Cluver (Ed.), Social medicine
grammes are still recognised as important today but often remain
(pp. 20e40). Johannesburg: Central News Agency.
poorly addressed, particularly in larger scale initiatives (Lehmann Kark, S. L. (1981). The practice of community-oriented primary health care. New York:
et al., 2004; Walt & Gilson, 1990). Given the renewed growth of Appleton-Century-Crofts.
CHW programmes within South Africa and globally, lessons learned Katzenellenbogen, J. M., Hoffman, M., & Miller, T. (6 September 1990).
Research in progress: Mamre community health project. Community
from past programmes should play a stronger role in informing health worker programme. Urbanisation and Health Newsletter, MNR/MRC,
current policies. CERSA, 6, 3e6.
358
Lehmann, U., Friedman, I., & Sanders, D. (2004). Review of the utilisation and Ramalepe, M. (1992). Ithuseng village health program. Newsletter of the National
effectiveness of community-based health workers in South Africa. Joint learning Progressive Primary Health Care Network. Johannesburg: Networker.
initiative on human resources for health and development. Working Paper 4-1. Roscher, C. I. (15th May 1990). Panel discussion. Viewpoints of different profes-
Cape Town. sionals. Nursing. Pretoria: Forum: Community Health Workers.
Lewin, S., Munabi-Babigumira, S., Glenton, C., Daniels, K., Bosch-Capblanch, X., van Schneider, H., Hlophe, H., & van Rensburg, D. (2008). Community health workers
Wyk, B. E., Odgaard-Jensen, J., Johansen, M., Aja, G. N., Zwarenstein, M., and the response to HIV/AIDS in South Africa: tensions and prospects. Health
Scheel, I. B., et al. (2010). Lay health workers in primary and community health Policy & Planning, 23(3), 179e187.
care for maternal and child health and the management of infectious diseases. Slabber, C. (1992). A new South Africa e a new health care strategy. South African
Cochrane Database of Systematic Reviews. doi:10.1002/14651858. Medical Journal, 82(6), 388e391.
CD004015.pub3. Issue 3. Art. No.: CD004015. Stein, J., Lewin, S., & Fairall, L. (2007). Hope is the pillar of the universe: healthcare
Lund, F. J. (1993). Paradox and policy: some lessons from CHW projects. Critical providers’ experiences of delivering anti-retroviral therapy in primary health-
Health, 42(April), 60e67. care clinics in the Free State province of South Africa. Social Science & Medicine,
MacKinnon, A. S. (2001). Of Oxford bags and twirling canes: the state, popular 64(4), 954e964.
responses, and Zulu antimalaria assistants in the early-twentieth-century Tollman, S. M. (1994). The Pholela Health Centreethe origins of community-
Zululand malaria campaigns. Radical History Review, 80, 76e100. oriented primary healthcare (COPC). South African Medical Journal, 84(10),
Mamre (1992). Minutes of CHW training group. 24/01/1992. Mamre Community 653e658, An appreciation of the work of Sidney and Emily Kark.
Health Project. Tollman, S. M. (1999). The Agincourt field siteeevolution and current status. South
Maphorogo, S., Sutter, E., & Jenkins, J. (2003). The community is my university: A voice African Medical Journal, 89(8), 853e858.
from the grass roots on rural health and development. Amsterdam: Royal Tropical Tollman, S. M., & Pick, W. M. (2002). Roots, shoots, but too little fruit: assessing the
Institute. contribution of COPC in South Africa. American Journal of Public Health, 92(11),
Marks, S. (1994). Divided sisterhood: Race, class and gender in the South African 1725e1728.
nursing profession. London/New York: Macmillan; St Martin’s Press. Toms, I. (1987). Empilisweni SACLA clinic. Cape Town: National Medical and Dental
Marks, S. (1997). South Africa’s early experiment in social medicine: its pioneers Association.
and politics. American Journal of Public Health, 87(3), 452e459. van Rensburg, H. C. J. (2004). Health and health care in South Africa. Pretoria: Van
Moosa, S. A. H. (1994). A national health plan for South Africa. Johannesburg: South Schaik Publishers.
Africa. van Rensburg, H. C. J., & Harrison, D. (1995). History of health policy. South Africa
Moosa, S. A. H. (2006). Community-oriented primary care in district health services Health Review 95e118.
of Gauteng, South Africa. South African Journal of Family Practice, 48(4), 6e8. Walt, G., & Gilson, L. (1990). Community health workers in national programmes: Just
NPPHCN. (1986). Proposal for the development of a progressive primary health care another pair of hands? Milton Keynes: Open University Press.
network in South Africa. Durban: National Progressive Primary Health Care Werner, D. (1977). Where there is no doctor: A village healthcare handbook. Hesperian
Network. Foundation.
NPPHCN (1992a). ‘The way forward and resolutions’. CHW Regional Conference Werner, D., & Bower, B. (1982). Helping health workers learn. Hesperian Foundation.
proceedings. South African Medical Journal. Uluntu Centre, Gugulethu, Cape Town. WHO. (1983). Apartheid and health. Report of an international conference held in
NPPHCN. (1992b). National community-based health worker conference proceedings. Brazzaville, People’s Republic of the Congo. Geneva: World Health Organisation.
Durban: National Progressive Primary Health Care Network. WHO. (1986). Community health workers. Report on an African Regional Seminar.
Oppenheimer, G. M., & Bayer, R. (2007). Shattered dreams? An oral history of the Brazzaville: WHO Regional Office for Africa.
South African AIDS epidemic. Oxford: Oxford University Press. WHO. (1992). Guidelines for the selection of community health activities Brazzaville.
Pelling, M., Berridge, V., Harrison, M., & Weindling, P. (2001). The era of public WHO Regional office for Africa.
health, 1848 to 1918. In C. Webster (Ed.), Caring for health: History and diversity WHO. (2008). Primary health care e Now more than ever. Geneva: World Health
(pp. 97e132). Buckingham, Philadelphia: Open University Press. Organisation.
Perks, R. (1992). Oral history. Talking about the past. London: Historical Association Wilson, F., & Ramphele, M. (1989). Uprooting poverty: The South African challenge.
and the Oral History Society. Report for the Second Carnegie Inquiry into Poverty and Development in
Phillips, H. T. (1993). The 1945 Gluckman report and the establishment of South Southern Africa. Cape Town: David Phillip.
Africa’s health centers. American Journal of Public Health, 83(7), 1037e1039. Wood, K., & Jewkes, R. (2006). Blood blockages and scolding nurses: barriers to
Price, M. (1993). The NPPHCN/SAHSSO policy conference 1992: an overview. Critical adolescent contraceptive use in South Africa. Reproductive Health Matters, 14
Health, Policy in Transition, 42, 7e14. (27), 109e118.
359
6. Cross-national and Part 4: ‘The case study approach’). However, given
the scale, complexity and cost of conducting any form of
analysis cross-national HPSR work, there remain relatively few
such studies. The criteria for assessing study quality and
Lucy ilson rigour must clearly be appropriate to the particular
niversity of Cape own, South frica and London School overarching research approach adopted (fixed, flexible or
of Hygiene and ropical Medicine, nited ingdom of mixed-method: see Part 2: Step 3).
reat ritain and orthern reland
A different role for cross-national analysis is in the
Heath policy and system developments are often country- assessment of various dimensions of health system
wide in scope, as in a national policy change or nation- performance drawing on standardized data and
wide implementation of a new health system interven- classification systems. Stimulated by the publication of
tion. Therefore, analysing these experiences to under- the World Health Organization’s World Health Report of
stand the impacts of particular changes or interventions 2000 on health systems’ performance, the work using
and the pathways of change (i.e. how these impacts are National Health Accounts is one example of such
achieved) must be undertaken at country level. However, analysis. Cross-national health and health systems
the transferability of health policy and systems lessons analysis is now also the subject of wider debate and
from one country to another is commonly questioned development, although the development of appropriate
because the long and complex causal pathways databases and rigorous analytic tools remains in its
underlying their effects allow contextual features to infancy.
influence their effects in many ways (Mills, 2012). As a
result, various analysts have called for studies that References
identify plausible rather than causal links between health
policy and systems interventions and their impacts, and Mills A (2012). Health policy and systems research:
for direct examination of the contextual factors under defining the terrain; identifying the methods. Health
which particular interventions achieve their impacts Policy and Planning 27(1):1-7.
(Janovsky & Cassels, 1995; McPake & Mills, 2000; Janovsky K, Cassels A (1995). Health policy and systems
Victora, Habicht & Bryce, 2004). research: issues, methods, priorities. In: Janovsky K, ed.
Cross-national analysis may, therefore, be helpful in not Health policy and system development: an agenda for
only understanding the forces driving health policy research. Geneva, World Health Organization:11–24.
and systems interventions but also influencing their
McPake B, Mills A (2000). What can we learn from
impacts. Such comparative analysis should allow critical
international comparisons of health systems and health
contextual features to be identified and their influence
system reform? ulletin of the orld Health rgani ation,
over interventions and subsequent impacts to be
78(6):811–820.
considered. Recent advances in impact evaluation and,
particularly, ideas around theory-based evaluation offer Victora CG, Habicht JP, Bryce J (2004). Evidence-based
valuable approaches for use in such analyses (see Part 4: public health: moving beyond randomized control trials.
‘Advances in impact evaluation’). At the same time, cross- merican ournal of Public Health, 94(3):400–405.
national studies can be seen as, in effect, country-level
case studies, with comparative analysis then allowing
general conclusions about particular interventions and
influences over their effects to be teased out through the
approach of analytic generalization (see Part 1: Section 7,

Overview of selected papers References for selected papers
The papers in this section illustrate the types of questions Bryce J et al. (2005). Programmatic pathways to child
and approaches that can be analysed in cross-national survival: results of a multi-country evaluation of Inte-
HPSR. grated Management of Childhood Illness. Health Policy
and Planning, 20 (supplement 1):i5–i17.
Bryce et al. (2005) report a seminal intervention
n
evaluation that drew on a plausibility approach to http://dx.doi.org/10.1093/heapol/czi055
assessing impact and examined the implementation Reproduced by permission of Oxford University
n
of one health policy and system intervention (the Press. Copyright Oxford University Press, 2005.
integrated management of child illness programme)
in different national contexts. The aim was to under- Gilson et al. (2001). Strategies for promoting equity:
stand what contextual factors were of most influence experience with community financing in three African
over the intervention’s impacts. Countries were selec- countries. Health Policy, 58:37–67.
ted for inclusion because they had implemented the
http://dx.doi.org/10.1016/S0168-8510(01)00153-1
Integrated Management of Childhood Illness (IMCI)
strategy. Reproduced by permission of Elsevier. Copyright
n
Elsevier, 2001.
Gilson et al. (2001) report a study that, using policy
n
analysis theory to guide it, adopted a comparative Lee K et al. (1998). Family planning policies and program-
case study analytical approach to gain insight into
mes in eight low income countries: A comparative policy
how to support implementation of a financing policy
(the Bamako Initiative) in any setting. Countries were analysis. Social Science Medicine, 47(7):949–959.
selected for inclusion because they had implemented http://dx.doi.org/10.1016/S0277-9536(98)00168-3
some form of the Bamako Initiative in Africa. Reproduced by permission of Elsevier. Copyright
n
Lee et al. (1998) report an eight-country study that,
n Elsevier, 1998.
using policy analysis theory, adopted a comparative
case study analytical approach to draw out general O’Donnell O et al. (2007). The incidence of public spend-
conclusions about how to strengthen the imple- ing on healthcare: comparative evidence from Asia. orld
mentation of family planning programmes. Countries ank conomic Review, 21:93–123.
were selected on the basis of available data and to http://dx.doi.org/10.1093/wber/lhl009
allow comparison and contrast of experience between
n
strong and weak national family planning program-
mes in four pairs of contrasting national socio- Press. Copyright Oxford University Press, 2007.
economic contexts.
O’Donnell et al. (2007) report a study that uses
n
comparable, quantitative data from household
surveys to conduct statistical analyses of the inci-
dence of public health expenditure in 11 Asian
countries and provinces. They concluded that pro-
poor health care requires limiting the use of user
fees, or protecting the poor from them, and building
a wide network of health facilities.
! The Author 2005. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. All rights reserved.
doi:10.1093/heapol/czi055
Programmatic pathways to child survival: results

of a multi-country evaluation of Integrated Management
of Childhood Illness
JENNIFER BRYCE,1 CESAR G VICTORA,2 JEAN-PIERRE HABICHT,3 ROBERT E BLACK4 AND
ROBERT W SCHERPBIER ON BEHALF OF THE MCE-IMCI TECHNICAL ADVISORS5
1
WHO Consultant, 2081 Danby Road, Ithaca, NY, USA, 2Universidade Federal de Pelotas, Pelotas, Brazil,
3
Division of Nutritional Sciences, Cornell University, Ithaca, NY, USA, 4The Johns Hopkins Bloomberg School
of Public Health, Baltimore, MD, USA and 5Department of Child and Adolescent Health and Development,
Family and Child Health Cluster, World Health Organization, Geneva, Switzerland
Objective: To summarize the expectations held by World Health Organization programme personnel
about how the introduction of the Integrated Management of Childhood Illness (IMCI) strategy would
lead to improvements in child health and nutrition, to compare these expectations with what was
learned from the Multi-Country Evaluation of IMCI Effectiveness, Cost and Impact (MCE-IMCI), and
to discuss the implications of these findings for child survival policies and programmes.
Design: The MCE-IMCI study designs were based on an impact model developed in 1999–2000 to
define how IMCI would be implemented at country level and below, and the outcomes and impact it
would have on child health and survival. MCE-IMCI studies included: feasibility assessments
documenting IMCI implementation in 12 countries (1999–2001); in-depth studies using compatible
designs in Bangladesh, Brazil, Peru, Tanzania and Uganda; and cross-site analyses addressing the
effectiveness of specific subsets of IMCI activities.
Results: The IMCI strategy was successfully introduced in the great majority of countries with
moderate to high levels of child mortality in the period from 1996 to 2001. Seven years of country-
based evaluation, however, indicates that some of the basic expectations underlying the development
of IMCI were not met. Four of the five countries (the exception is Tanzania) had difficulties in
expanding the strategy at national level while maintaining adequate intervention quality. Technical
guidelines on delivering interventions at family and community levels were slow to appear, and in
their absence countries stalled in their efforts to increase population coverage with essential
interventions related to careseeking, nutrition, and correct care of the sick child at home. The full
weight of health system limitations on IMCI implementation was not appreciated at the outset, and
only now is it clear that solutions to larger problems in political commitment, human resources,
financing, integrated or at least coordinated programme management, and effective decentralization
are essential underpinnings of successful efforts to reduce child mortality.
Conclusions: This analysis highlights the need for a shift if child survival efforts are to be successful.
Delivery systems that rely solely on government health facilities must be expanded to include the
full range of potential channels in a setting and strong community-based approaches. The focus
on process within child health programmes must change to include greater accountability for
intervention coverage at population level. Global strategies that expect countries to make massive
adaptations must be complemented by country-level implementation guidelines that begin with local
epidemiology and rely on tools developed for specific epidemiological profiles.
Key words: child survival, IMCI, public health programme evaluation, child health
Introduction
Health Organization (WHO) and their technical partners
The Integrated Management of Childhood Illness developed the strategy in a stepwise fashion, seeking to
(IMCI) strategy address limitations identified through experience with
disease-specific child health programmes, and especially
Integrated Management of Childhood Illness (IMCI) is those addressing diarrhoeal disease and acute respiratory
a strategy for reducing mortality among children under infections (Claeson and Waldman 2000). Elements
the age of 5 years (Tulloch 1999). UNICEF, the World of the strategy were developed in a rough sequence
363
i6 Jennifer Bryce et al.
Improve family &

Improve health Improve health
community
worker skills systems
practices
! Case management ! District planning and ! Appropriate

standards & management careseeking
guidelines
! Availability of IMCI drugs ! Nutrition
! Training of
facility-based public ! Quality improvement and ! Home case
health care providers supervision at health management &
facilities adherence to
! IMCI roles for private
! Referral pathways and recommended
providers
services treatment
! Maintenance of ! Community
competence among ! Health information involvement in health
trained health system services planning &
workers monitoring
Source: WHO (1999a).
Figure 1. The three components of IMCI as presented by WHO in 1998
from: (1) evidence-based guidelines for health workers as a treatment for measles (Gove 1997). Undernutrition
serving high-mortality populations that defined clinical was addressed by having health workers counsel care-
case management actions to respond to common infec- takers about appropriate feeding, including breastfeeding.
tious diseases in childhood and the delivery of key The guidelines were adapted in each country (WHO
prevention services including immunization and nutrition 1998b), resulting in a set of tasks to be performed by the
interventions; (2) health worker training in the guidelines health worker(s) including a full assessment and classifi-
based on paedological principles of supervised practice in cation of the child’s condition leading to a determination
clinical settings and follow-up of trainees to assist with the of treatment, and counselling of the caretaker on
establishment of new practices; (3) attention to needed administration of medicines, appropriate home care, and
health system supports for child health and development, the conditions under which the child should be brought
based on the recognition that health workers are not back to the facility. The guidelines also recommend
isolated, but work in systems that, if not strengthened, the use of the illness episode as an opportunity for the
would limit their abilities to perform good work; and delivery of preventive interventions, including vaccines
(4) strengthening of family practices needed to prevent and nutritional counselling.
disease, to stimulate appropriate utilization of health
services, and to improve home care for sick children. The generic IMCI training course was developed based
Figure 1 presents the components of the IMCI strategy on these guidelines, and emphasized supervised clinical
and the interventions WHO and UNICEF initially practice (Gove 1997). In addition, the IMCI training
proposed for inclusion within each component (WHO approach recommends that each participant receive a
1999a). follow-up visit from their trainer within 4 to 6 weeks
after the initial training in order to help them implement
The IMCI case management guidelines for the integrated their new skills (WHO 1999a).
management of sick children in a first-level health
facility were designed to address the major causes of IMCI programme developers incorporated the need
child mortality in countries with infant mortality rates of for specific health system supports into the strategy
40 per 1000 live births or greater (Gove 1997; WHO itself (see Figure 1), an important step forward from
1998a). Undernutrition, an underlying cause contributing the disease-oriented programmes of the past. The
to over 50% of deaths in children between the ages of expectation was that introducing IMCI would contribute
1 month and 5 years (Pelletier et al. 1995; Caulfield et al. to these needed health systems changes, strengthening
2004), was also a major target. Interventions in the generic existing systems for supervision, drug supply and health
IMCI guidelines therefore included the provision of information.
antibiotics for pneumonia and dysentery, antimalarials
for fever in settings where malaria was endemic, oral The vision for the strategy also included the need to
rehydration therapy for the prevention and treatment of deliver interventions at the community level aimed
dehydration due to diarrhoea, and the use of Vitamin A at improving family practices – such as appropriate
364
Pathways to child survival i7
careseeking and home management of illnesses – that governments, and the results therefore have relevance to
would act synergistically with improving health worker efforts to improve the delivery and utilization of a broad
skills at the facility level. WHO and UNICEF defined range of public health interventions (Bryce et al. 2003;
a set of 12 key family and community practices and Victora et al. 2004).
commissioned a synthesis of evidence supporting
their importance relative to child health and survival
The IMCI impact model
(Hill et al. 2004).
The MCE-IMCI Technical Advisory Group was created
Implementation of the IMCI strategy in 1998, and included experienced researchers and
evaluators in the fields of child survival, economics and
IMCI was first introduced at country level in 1996 by health policy. Advisors worked closely with IMCI
Tanzania and Uganda. In the 9 years since then, over 100 developers from WHO and UNICEF to develop an
additional countries across all geographic regions have impact model for IMCI. This model was needed as a basis
adopted the strategy and gained significant experience in for defining the specific types and magnitude of changes
its implementation (WHO 2005a). expected from the introduction of IMCI, for choosing
indicators and for calculating sample sizes. Parts of this
The global planning guidelines for use by countries in model were then computerized using an approach that
implementing IMCI recommended three stages (WHO was similar to that of Becker and Black (1996) and used to
1999b). In the introductory phase, countries conducted estimate the magnitude of mortality reduction that could
orientation meetings, trained key decision makers in be expected from introducing IMCI in different settings.
IMCI, defined a management structure for preparing for
IMCI, planning and early implementation, and built Figure 2 presents a greatly simplified version of the
government commitment to move forward with the IMCI model; the full model is available for review at [http://
strategy. In the early implementation phase, countries www.who.int/imci-mce/]. Each of the arrows in Figure 2
gained experience while implementing IMCI in limited reflects an expectation among WHO programme staff
geographic areas. They developed their national strategy in the late 1990s about the pathways through which
and plan, adapted the IMCI guidelines to their national the introduction of IMCI at country level would lead to
context, developed management and training capacity in improvements in child survival and nutrition. Important
a limited number of districts, and started implementing exceptions are the boxes on coverage, which were added
and monitoring IMCI. The end of this phase was marked only in 2004 based on the MCE-IMCI findings.
by a review meeting with the objective of synthesizing
early implementation experience and planning for The temporal dimension of the model moves from level 1
expansion. In the expansion phase, countries increased to level 4. The first level defines the planning steps and
both the range of IMCI interventions and IMCI coverage. inputs needed to initiate IMCI-related activities. The
An important challenge emphasized in planning for the second level outlines how these activities were expected
expansion phase was maintaining quality while expanding to lead to implementation of the IMCI interventions.
coverage. The third and fourth levels specify the pathways through
which these IMCI interventions were expected to lead
The Multi-Country Evaluation of IMCI Effectiveness, to intermediate behavioural outcomes and to impact on
health status, respectively.
Cost and Impact (MCE-IMCI)
The MCE-IMCI includes studies of the effectiveness, The objective of this paper is to compare the findings of
cost, and impact of the IMCI strategy in Bangladesh, the MCE-IMCI relative to the programme expectations
Brazil, Peru, Tanzania and Uganda (Bryce et al. 2004). reflected in the IMCI impact model. We review five of the
In-depth studies assessing the feasibility of conducting a most important programme expectations from the impact
large-scale impact evaluation like the MCE-IMCI were model and describe the extent to which each was realized
conducted in seven additional countries. Planning for the in IMCI implementation among countries participating
MCE-IMCI began in 1997, just as the first countries were in the MCE-IMCI. These expectations are: (1) The generic
adapting the IMCI strategy and moving into the early IMCI guidelines could and should be adapted and
implementation phase. The evaluation objectives were to implemented in developing countries with an infant
assess the behavioural, nutritional and mortality impact mortality of more than 40/1000 live births (WHO
of IMCI, as well as to document the effect of IMCI 1998a); (2) IMCI case management training would lead
interventions on health worker performance, health to improved quality of care at first-level health facilities;
systems and family behaviour. The MCE-IMCI was (3) The introduction and implementation of IMCI would
planned as one part of a larger research agenda that contribute to strengthening health system supports;
included efficacy evaluations of the individual interven- (4) Families would respond to improved quality of care
tions within IMCI, as well as qualitative and operations in government health facilities, leading to increases in
research. Details about the development, design and utilization and reductions in child mortality; and (5) All
implementation of the MCE-IMCI are available elsewhere three components of the IMCI strategy could be imple-
(Bryce et al. 2004). A key focus of the MCE-IMCI was mented in a coordinated fashion at country level within
the implementation of the IMCI strategy in the hands of a time frame of 3 to 5 years. In our conclusions we
365
Model
Level
Introduction of
IMCI
(1)
Health system Family and
improvements Training of community
health workers interventions
Increased training coverage*
Improved
(2) quality of care in
health facilities
(3) Improved Improved Improved

household careseeking & preventive
compliance/care utilization practices
Increased coverage for curative & preventive interventions*
Improved
(4) health / nutrition
Reduced mortality
*added later as a result of MCE-IMCI findings.
Figure 2. A simplified version of the IMCI impact model developed in 1999–2000
summarize what has been learned from the MCE-IMCI Data sources
about effective child survival programmes and highlight
implications for other public health initiatives. The MCE-IMCI includes three different types of studies,
each of which provides important findings relative to the
impact model:
Methods (1) The 12-country assessment of IMCI implementation.

The country selection process for the MCE-IMCI
Design included visits by teams of MCE-IMCI Advisors and
WHO staff to countries selected by WHO as
The MCE-IMCI consisted of a series of independent representing the best examples of IMCI implementa-
studies with compatible designs, each tailored to the tion at that time. All countries in each of the six
stage and characteristics of IMCI implementation in the WHO regions were evaluated against a set of criteria
participating country (Bryce et al. 2004). The set of that included the probability that the government
site-specific studies included prospective, retrospective would be successful in implementing all three
and mixed designs. They reflected a continuum from components of the IMCI strategy over the subse-
efficacy to effectiveness, with variable degrees of influence quent 5 years. Further information on eligibility
from the evaluation team on programme implementation. criteria are presented elsewhere (Bryce et al. 2004).
Each study addressed the need to document the plausi- Based on this review, in each region one or two
bility of an effect of IMCI on intermediate steps defined countries judged most likely to meet the criteria were
in the impact model. All studies measured an identical selected for assessment visits. The assessment proto-
set of indicators and, with few exceptions, used identical col included in-depth reviews of country-level plans
data collection tools. Observation-based surveys were and progress in child health activities, including but
used to assess the quality of child health care provided not limited to IMCI. More than one assessment visit
in health facilities. Cost data were collected at the was made to several countries in which small studies
household, health facility, district and national levels. were commissioned to evaluate the potential for
Household surveys assessed preventive practices and successful IMCI implementation. Although the
family responses to illness. All tools were adapted to countries visited had been implementing IMCI for
respond to local characteristics and questions, and in varying periods of time, the search was restricted
some sites the variables necessary to assess equity were to those likely to implement IMCI fully, in large
added. geographical areas, within the 2 years after the
366
Pathways to child survival i9
assessment visit, allowing an impact evaluation
areas of 10 study districts

Comparison of 10 districts
2001 & 2002 in catchment

period of 2 to 3 years within the time frame of the
with different levels of

IMCI implementation
Included in survey tools

Ongoing rolling sample
MCE-IMCI. Bangladesh was included as a site even
though IMCI implementation had not yet begun, to
serve as an efficacy study in which the investigators
2000 (baseline)
could collaborate with the Government in imple-
Plausibility
menting the strategy under relatively ideal condi-
Uganda
Surveys
tions. Findings from the 12-country assessment
62
20 554
141
330
None
provide important information on the validity of
Yes
those parts of the IMCI impact model related to
planning and implementing activities across the

Comparison of 25 departments Pre–post comparison of
three model components (WHO 1999b); some of
2 IMCI and 2 non-

their implications have been reviewed and discussed
elsewhere (Victora et al., in press).
IMCI districts
1999 (baseline)
2000 (midway)
(2) In-depth studies in five sites. Based on the findings
surveillance
Demographic
of the 12 country assessments described above,
Plausibility
Tanzania
Bangladesh, Brazil, Peru, Tanzania and Uganda
69
32 102
160–180
210
were selected as in-depth study sites. In Peru, IMCI
None
2004
Yes
had already been taken to scale and implemented
nationwide, so the evaluation used a fully retro-
with different levels of IMCI

spective design and relied heavily on routine data
sources. IMCI implementation was in the expansion
phase in Brazil, Tanzania and Uganda, and each
selected departments)
design represented a mixture of retrospective and
1999 (pilot study in

implementation
prospective elements. In Bangladesh, as explained
above, a fully prospective design was possible
Vital statistics
Not included
because IMCI implementation had not yet begun
Plausibility
at national level. In both Bangladesh and Tanzania,
Variable
24 797
2610
58
88
MCE-IMCI investigators are participating actively
None
None
Peru
Table 1. Characteristics of in-depth study sites in the Multi-Country Evaluation of IMCI, 2000
in the Government’s plans for IMCI implementa-

tion. Table 1 presents a summary of characteristics
and MCE-IMCI data collection activities in the five
Comparison of 32 IMCI

in-depth sites. Full descriptions of the methods and
and 64 non-IMCI
results for each study site are available at [http://

www.who.int/imci-mce/].
municipalities
(3) Cross-site analyses. The use of standard indicators 2002 (midway)
Plausibility
and analysis plans permitted comparisons across
Surveys
165 851
4790
70
83
the five MCE-IMCI study sites. Topics addressed to

Brazil
None
None
None
date include the effect of IMCI in improving care
quality in first-level health facilities (Gouws et al.
2004), health system barriers to scaling-up (Victora
health facilities with IMCI
et al. 2004), and the importance of context-specific

and 10 without IMCI
Randomized trial of 10
delivery mechanisms (Bryce et al. 2003), as well as

surveillance þ survey
methodological issues (Bryce et al. 2004; Bryce and

*Source of demographic estimates: UNICEF (2000).
Victora 2005; Gouws et al. 2005).

Household coverage surveys 2000 (baseline);
2007 (planned)
2000 (baseline)
Demographic
Bangladesh
In addition, other documentation and research efforts

Probability
Yes (2001)
related to IMCI were reviewed carefully and the findings

124 774
96
38
360
None
were taken into account in interpreting MCE-IMCI

results.
GNP per capita (2000 US$)
Baseline under-5 mortality
Analytic approaches
Health facility assessments
Total population (1000)
Total adult literacy rate
Mortality assessment
The analytic approach used in the MCE-IMCI varied

among countries. As shown in Table 1, all evaluations
Type of inference
Cost assessments
entailed a comparison, either between areas with and

Randomization
Demographic*
without IMCI (Bangladesh, Brazil and Tanzania) or

MCE-IMCI
among areas with variable degrees of implementation

Design
Malaria
(Peru and Uganda). Details of the analytical approaches

are available in the country-specific publications from
Bangladesh (Arifeen et al. 2005), Brazil (Amaral et al.
367
2004), Peru (Huicho et al. 2005), Tanzania (Armstrong for children under 5 years of age, and the incorporation
Schellenberg et al. 2004a) and Uganda (Pariyo et al. 2005). of interventions designed to reduce deaths from causes
in the neonatal period.
Results Another part of the expectation was that IMCI could

and should be implemented fully regardless of the strength
Expectation 1: The generic IMCI guidelines could and of the health service system. Again this was an implicit
should be adapted and implemented in developing countries expectation, but was supported by the fact that virtually
every developing country was approached by WHO
where infant mortality is higher than 40 per 1000 live births
to introduce IMCI. IMCI implementation guidelines
Although the original target for the IMCI case manage- suggested that countries with weak health systems
ment guidelines was countries with infant mortality rates should begin slowly with IMCI implementation, and
of at least 40 per 1000 (WHO 1998a), other countries build toward stronger health system strength and inte-
or specific geographic areas within countries found the grated programmes simultaneously and synergistically
concept of integration attractive and moved to adapt (Lambrechts et al. 1999; WHO 1999b).
and adopt them as well. For example, the Pan American
Health Organization ‘. . .urges all countries to incorporate Expectation 2: IMCI case management training would lead
IMCI as a basic standard for child care’ (PAHO, to improved quality of care at first-level health facilities
undated). There was an expectation in the early years of
IMCI introduction that the generic guidelines could and One part of this assumption, that IMCI case management
would be adapted by any country or area to reflect their training would improve health worker performance and
specific epidemiological profile and health system char- thus contribute to improved care quality, has been
acteristics. WHO therefore worked in the late 1990s to repeatedly borne out through MCE-IMCI findings
develop guidelines for the country adaptation process, (Amaral et al. 2004; Armstrong Schellenberg et al. 2004b;
including evidence for intervention choices, models for El Arifeen et al. 2004; Gouws et al. 2004). In all settings
how to incorporate additional diseases and conditions where case management training was implemented at
into the training materials, and how to conduct local minimum standards of quality, and where sufficient
studies to identify terminology and local foods (WHO coverage of trained workers was able to be maintained
1998b). Cadres of ‘IMCI adaptation consultants’ were at health facility level, the quality of care improved.
trained at regional and global levels. Ill children managed by health workers trained in IMCI
receive a more thorough assessment than children cared
The resource-intensive efforts at country level required to for by workers without IMCI training, and are more likely
adapt the generic IMCI guidelines were necessary because to receive correct treatment. Caretakers are more likely
the specific pneumonia-diarrhoea-malaria profile under- to receive key messages about how to continue care at
lying the generic IMCI algorithm represents countries home and when to return to the facility.
that accounted for only about 35% of under-five deaths
in 2000 (Black et al. 2003). The remaining 65% of deaths Expectation 3: The introduction and implementation
occurred in epidemiological contexts without endemic of IMCI would contribute to strengthening health
malaria, dominated by neonatal disorders or in a few
system supports
countries with generalized epidemics of HIV/AIDS.
The widespread uptake of the IMCI concept resulted in Early experiences with IMCI implementation suggested
overextension of the guidelines to settings with disease that the inter-programme working groups at national
profiles that varied widely from those for which they were level that were recommended as a mechanism to plan for
developed. IMCI, and specific planning steps such as the review and
updating of child health policies and essential drug lists,
The IMCI strategy as defined in 1996 applied only to would lead to activities designed to improve health system
children from the ages of 1 week to 5 years (Gove 1997), supports for child health activities (WHO 1999b). In most
and did not include interventions addressing deaths in the countries this assumption, at this level, was borne out.
early neonatal period. The cause structure of infant deaths The introduction of IMCI led to the rationalization of
was not well understood at that time, and few interven- child health policies and the updating of essential drug
tions had been fully developed and evaluated for efficacy. lists in most countries in Africa, for example (Lambrechts
et al. 1999; WHO 2000).
The eventual expectation that a set of generic algorithms
based on the global distribution of causes of death, In three of the 12 countries assessed, IMCI benefited
combined with support for adaptation at country level, from activities designed to strengthen the health system.
would be an efficient way to improve case management In Tanzania, the Tanzania Essential Health Intervention
in all countries proved over-ambitious. With benefit of Project (TEHIP) introduced basic management tools at
hindsight, greater technical efficiency might have been district level (De Savigny et al. 2004) which permitted an
gained if lower mortality countries had been encouraged effective use of decentralized health resources and resulted
to develop, or wait for, epidemiologically driven algo- in the adoption of IMCI. Other districts with the same
rithms more consistent with their cause-of-death profiles resources but without the TEHIP tools did not adopt
368
Health Policy 58 (2001) 37 – 67
www.elsevier.com/locate/healthpol
Strategies for promoting equity: experience with

community financing in three African
countries!
Lucy Gilson a,*, Denny Kalyalya b, Felix Kuchler c,
Sally Lake a, Hezron Oranga d, Marius Ouendo e
a
Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine,
London, UK
b
Department of Economics, Uni!ersity of Zambia, Lusaka, Zambia
c
Swiss Tropical Institute, Basel, Switzerland
d
African Medical and Research Foundation, Nairobi, Kenya
e
Institut Regionale de Santé Publique, Cotonou, Benin
Received 9 June 2000; accepted 28 March 2001
Abstract
Although the need for a pro-poor health reform agenda in low and middle income
countries is increasingly clear, implementing such policy change is always difficult. This
paper seeks to contribute to thinking about how to take forward such an agenda by
reflection on the community financing activities of the UNICEF/WHO Bamako Initiative. It
presents findings from a three-country study, undertaken in Benin, Kenya and Zambia in
1994/95, which was initiated in order to better understand the nature of the equity impact of
community financing activities as well as the factors underlying this impact. The sustained
relative affordability gains achieved in Benin emphasise the importance of ensuring that
financing change is used as a policy lever for strengthening health service management in
support of quality of care improvements. All countries, however, failed in protecting the
most poor from the burden of payment, benefiting this group preferentially and ensuring that
their views were heard in decision-making. Tackling these problems requires, amongst other
things, an appropriate balance between central and local-level decision-making as well as the
creation of local decision-making structures which have representation from civil society
!
This paper is in memory of one of the co-authors, Hezron Oranga.
* Corresponding author. Present address: Centre for Health Policy, PO Box 1038, Johannesburg 2000,
South Africa. Fax: + 27-11-4899900.
E-mail address: lucyg@mail.saimr.wits.ac.za (L. Gilson).
0168-8510/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved.
PII: S0168-8510(01)00153-1
369
38 L. Gilson et al. / Health Policy 58 (2001) 37–67
groups that can voice the needs of the most poor. Leadership, strategy and tactics are also
always important in securing any kind of equity gain—such as establishing equity goals to
drive implementation. In the experiences examined, the dominance of the goal of financial
sustainability contributed to their equity failures. Further research is required to understand
what equity goals communities themselves would prefer to guide financing policy. © 2001
Elsevier Science Ireland Ltd. All rights reserved.
Keywords: User fees; Community financing; Affordability; Participation; Equity; Policy analysis; Imple-
mentation; Evaluation; Benin; Kenya; Zambia; Africa
1. Introduction
The need for a pro-poor health reform agenda in low and middle income
countries has become increasingly evident in the face of inequities in access and
payment for care [1] and disrespectful treatment of patients [2]. The World Health
Organisation (WHO) has, therefore, combined fairness in financial contributions
and responsiveness to the legitimate expectations of the population with improving
the level and distribution of health as the criteria it is promoting for assessing
health system performance [3]. Implementing pro-poor health reform is, however,
always difficult as it has to confront the challenges associated with any politically
controversial policy change [4–6].
This paper seeks to contribute to thinking about pro-poor reform by reflection
on an earlier phase of health policy change, the UNICEF/WHO Bamako Initiative
(BI). Building on previous experiences of community financing, local-level co-oper-
ative action associated with material or financial support for health care activities
[7], the BI sought to accelerate and strengthen the implementation of primary
health care, with the goal of achieving universal accessibility to these services. Its
three main strategies were: decentralised decision-making including the involvement
of community members in managing primary health care activities; user-financing
of health services under community control; and the provision of essential drugs
within the framework of a national drugs policy [8,9]. From its inception the BI was
caught up in a wider debate about the potential equity impact of any form of user
financing, given its potential to undermine the access to health care of lower
socio-economic groups [10–14].
The investigation reported here was initiated in the mid-1990s to add to the
available empirical evidence about the equity impacts of community financing and
BI activities. They investigated both the perceived and demonstrated impacts on
equity of such activities, as well as the mechanisms and processes through which
these impacts were obtained. The studies’ focus on understanding how and why any
perceived and demonstrated changes in equity came about, or what obstacles there
were to securing such impact, is unusual in the health care financing literature.
However, such investigation is important both in better understanding the nature of
the equity impacts and in generating policy-relevant findings. Experience suggests
that understanding the factors influencing the pattern and nature of public policy
change is essential in determining how to better achieve policy goals in the future
370
L. Gilson et al. / Health Policy 58 (2001) 37–67 39
[15– 20]. Policy-makers and managers seeking to learn lessons from existing experi-
ences are, therefore ‘…demanding information on what is being done elsewhere,
what works, what does not work, why, whether it can be imported, adapted, and
how’ ([21] p. 18).
The study was undertaken in Benin, Zambia and Kenya in 1995 –96 [22 –25]. The
Benin BI programme adopted the ‘classical’ BI approach [26] as its main health
reform strategy, seeking to improve the quality of care available at existing primary
care facilities, staffed by trained primary care health workers, and to develop the
financial sustainability of services offered within them. The package of interventions
included the introduction of charges to fund improved drug supplies and support
the provision of immunisation services, the formation of local committees,
combining community representatives and health staff, to participate in decision-
making about drug control, revenue collection and revenue use and clinical training
and enhanced supervision. The Kenyan BI programme, in contrast, was imple-
mented in parallel to other changes within the health system, and sought rather to
extend primary care coverage beyond the existing facility network by establishing
new community pharmacies in areas otherwise not served by government health
facilities. The pharmacies were staffed by community members who received a short
period of basic training to allow them to offer simple curative and preventive care.
They were also associated with a wider network of community health workers
(CHWs) based in the villages served by the pharmacy who had health education
and preventive care roles. The pharmacies stocked and sold both a limited range of
drugs and bed nets, for use in protecting against malaria transmission, and were
managed by community committees established with the support of the local
leaders. Finally, the pharmacies were also intended to be the focus for the wider
community development action, particularly income generating activities, needed to
combat ill health and poverty. The very different experience of the third country,
Zambia, involved, in 1994/5, an almost exclusive focus on decentralisation to
district management teams and boards as the main reform strategy for improving
the efficiency and equity of the health care system. The introduction of user fees in
the early 1990s was, therefore, only of secondary importance in its overall reform
programme. Zambia was, nonetheless, included in this study because its different
experiences were expected to provide interesting comparisons with the other coun-
tries’ activities.
The full evaluation of the equity impacts of these community financing activities
is presented in a sister paper [27]. It was rooted in consideration of three equity
principles: payment on the basis of ability to pay; equal opportunity of use for
equal need; and effective representation of all community interests in decision-mak-
ing [22,28]. Whilst the first two principles are commonly associated with distributive
justice concerns, that is the distribution of the outcomes of decision-making, the
third reflects a concern for procedural justice— the respectful treatment of all
groups in decision-making [29– 31]. The equity successes of the Benin and Kenyan
BI programmes resulted from the relative affordability gains associated with
reducing the cost of accessing care by, respectively, improving existing services and
bringing new services closer to people’s homes. In Kenya, however, these gains were
371
undermined by two factors. First, the limited range of services provided through the
BI programme meant that people still had to access more distant services for many
health problems. And, second, the provision of even this basic set of services was
not sustained over time (as evidenced by the drug supply problems experienced in
pharmacies towards the end of the study period). Yet in these relative affordability
gains went hand in hand with absolute affordability problems, as the most poor
received little protection from, and struggled to cope with, the burden of fee
payment. Absolute affordability problems were, moreover, evident in both coun-
tries as neither established effective exemption mechanisms and so the poorest
groups were unfairly burdened with paying for care. These problems were seen
most clearly in Zambia where the introduction of user fees without concurrent
quality improvements or effective exemption practices led to declining utilisation
levels, as large proportions of the population experienced reduced access to health
care (although these levels may have stabilised over time [32]). Finally, the voice
and needs of the poorest within communities were largely ignored within decision-
making practices in each country, a failing in terms of the third equity principle
used in the study.
This paper seeks specifically to identify the factors that explain this pattern of
equity impacts within and across countries, and to draw policy-relevant conclusions
from this analysis. Section 2, first, describes the framework used in the analysis.
Attention is then given to the three key sets of factors identified as shaping the
country experiences: the leadership given to policy development and implementa-
tion (Section 3); the contribution of policy design in sustaining relative affordability
gains (Section 4); and the interacting problems of policy design and process that
failed the poorest within communities (Section 5). Finally, policy relevant conclu-
sions are outlined (Section 6).
2. The analytical framework
Cross-country analysis of experience in developing and implementing health

policies is recognised as important in informing broad questions of policy direction
as well as implementation strategies [21,33,34]. The analytical framework used both
within each country study and in reflecting on the three different experiences is
summarised in Fig. 1.
In stage 1 the impact of the community financing schemes on equity was assessed
against the study’s three guiding principles of equity using available utilisation data,
investigations of the experiences of different population groups, especially the
poorest, in accessing care and in decision-making, and assessment of the design of
the schemes of focus (details presented in [27]).
In stage 2 (the focus of this paper) the factors influencing the equity impacts of
the community financing activities in each country, and across countries, was
investigated by combining a grounded approach to data analysis with the applica-
tion of a broad lens through which to filter experience. This lens built on the policy
analysis approach of Walt and Gilson [20] and highlighted four broad groups of
factors as having potential influence over impacts:
372
1. contextual factors: the socio-economic context of implementation, the previous

condition and financing patterns of the health system, socio-cultural traditions
and practices of decision-making;
2. the design of each scheme: its objectives, the nature and level of fees, practices
regarding the retention and use of revenue, the existence and nature of an
exemption scheme, the structures and practices of community involvement in
decision-making;
Fig. 1. Analytical framework of the study.
373
3. the particular processes used in initiating and implementing the schemes: the
speed and manner of implementation, and the relative inputs of technicians,
service providers and community members in design and implementation;
4. the actors affecting decision-making at all levels of the system (groups within
communities, community leaders, service providers, health managers and exter-
nal donors): their interests, concerns and roles in the activities.
The methods used to gather the data used in this analysis are outlined in Table
1 (see also [22,27]). Document reviews and semi-structured interviews with key
informants (policy-makers, programme managers, donor agency representatives) in
each country allowed initial analysis of the policy environment and aspects of the
process of policy development and implementation. More detailed data on imple-
mentation practices were drawn from the two rounds of community inquiry
conducted within study sites, that is the commune, within which the primary care
facility is located, in Benin; villages served by a BI pharmacy/CHW network in
Kenya and districts in Zambia. The first round of these inquiries involved a rapid
appraisal of purposively selected sites, in which information about the history and
performance of the site was gathered by record review and semi-structured inter-
views with health workers/managers and a small number of community representa-
tives. In the second round of site visits a wider range of structured interview and
qualitative data collection approaches were used in a purposively selected sub-set of
the initial sample of sites (see Table 1). Community respondents’ (including the
poorest in Benin and Kenya) views about their experiences of the services and
decision-making processes were identified.
As only a limited number of sites were investigated in each country it is clearly
important to be careful in generalising from the study findings. However, investigat-
ing the complexity of implementation experience is at least equally as important in
informing future policy development as identifying common patterns across a large
number of sites. An understanding of how and why equity has been promoted or
undermined can, moreover, be better generated by small-scale, intensive case study
evaluations than by large-scale, extensive assessments [35]. Qualitative methods are
particularly relevant within such an approach: ‘Quantitative methods can identify
‘how’ individuals behave in certain circumstances, while qualitative methods… are
better equipped to answer the diagnostic question of ‘why’’ ([36], p. 445).
3. The importance of leadership in effective policy design and implementation
The overall success of the Benin BI activities, evident in the restoration of

services in previously ineffective rural facilities, contrasted with the poorly sustained
BI pharmacies in Kenya and the equity losses consequent on reduced utilisation in
Zambia. What explains these different experiences?
The first explanation lies in the three countries’ differing processes of policy
development and implementation. Although actors played critical roles in each
case, in Benin they demonstrated an ability to shape and mould the interactions
between themselves and the other three sets of factors influencing policy change
374
Table 1
Summary of methods used by phase of study and country
Benin Kenya Zambia
Policy review Document reviews and key informant interviews

(in Zambia, key members of the research team
had also been involved in financing debates)
Rapid appraisal One commune randomly selected from each of 18 Two districts purposively selected Eight districts purposively
purposively selected sous-prefecturesa, to provide because among first to develop BI selected, each from a
a sample representative of each of the country’s schemes (so longer experience) and different province (of which
six departments and to cover all of the ‘partner’ areas of most poor health status; 12 there are nine); districts
institutions involved in supporting the primary sites purposively selectedc: six from included fairly even balance
care network (international bilateral & multilateral each district located in different of rural and urban areas,
organisations, and NGOsb); semi-structured agro-ecological potential zones areas of different
interviews with four purposively sampled health (reflective of socio-economic status), socio-economic status; in
professionals and six randomly selected members ten government-sponsored sites of one district services run by
of the community per site commune visit lasted different ages (five from each district) mission; in six districts,
2–3 days and two NGO sites (one from each visits included collection of
district); conducted three focus group available data on utilisation
discussions with village/pharmacy patterns, and semi-structured
committee, CHWs and TBAs interviews with district
(traditional birth attendants), using managers, local government
interview guide; semi-structured managers and health care
interviews with chairman, treasurer providers; for remaining two
and pharmacist in each site; collected districts, data on utilisation
available health service statistics and and staff perceptions
revenue data; site visit lasted 1–2 days collected from parallel
study; in each district
L. Gilson et al. / Health Policy 58 (2001) 37–67
looked specifically at
experiences of hospital
located in it and sample of
two to four health centres
or clinics; 34 facilities of
focus: ten hospitals (all
levels; three church run);
nine urban clinics (eight
council run); 14 rural health
centres (two mission run);
43
district visits lasted 2 days
375
376
44
Table 1 (Continued)
Benin Kenya Zambia
Detailed case studies Seven sites purposively selected from initial 18, Seven sites purposively selected from Four sites purposively
five ‘typical’ (i.e. said they were implementing initial 12, including sites from both selected from initial eight
national BI principles) and two atypical (i.e. said districts and the two NGO sites, on districts; semi-structured
they were not implementing national BI grounds of level of function and ease interviews and focus group
principles), on grounds of ease of access to of access to information; two discussions with health
information, quality of information collected, household surveys across all sites: (a) service users and other
focus on needs of poor; self-administered random sample of 30 households per community members
questionnaires completed by three purposively site (210 in total) (b) 87 ‘poorest’
selected health workers; conducted interviews with households; Participatory rapid
ten poor households, 20 randomly selected service appraisal techniques applied including
users, and undertook four focus group discussions wealth ranking, social mapping,
(members of the commune committeed, women, transects in community where
young people and village notables) pharmacy located in four sites; first
round of focus group discussions with
community representatives in all sites
and second round in government sites
only with village health committeese;
collection of additional health service
statistics and other data
a
From a total of 67 rural sous-prefectures (districts).
b
NGO, non-governmental organisation.
c
From a total of 52 in the two districts of focus; there were 237 sites across the country in 1994.
d
The Comite de Gestion de Commune (COGEC).
e
This round of focus group discussions was specifically undertaken to review site experiences following the withdrawal of UNICEF support for the BI
programme.
Fig. 2. The mutually reinforcing interactions of BI programme implementation in Benin.
and, in particular, to create mutually reinforcing interactions in support of, and

through, implementation (Fig. 2). In Kenya and Zambia, however, leading actors
failed to build such interactions and so either did not or could not take action to
offset obstacles and opposition.
3.1. Leadership and !ision in Benin
The design of the Benin BI programme was rooted in a context characterised by

poor economic performance and a deteriorating health system. The government
budget allocation for health fell by more than half between 1987 and 1990, leaving
total government health expenditure per capita at just under US$2 [37]. Rural
health centres frequently lacked drugs and other supplies and health staff were
poorly motivated. Patients using their services had to purchase drugs from distant
private pharmacies and few, if any, preventive or other services were provided to
the surrounding population.
The Benin BI programme sought, therefore, to build demonstrable improvements
in the quality of curative primary care and in the coverage of immunisation
services. It built both on the country’s diverse range of community financing
experiences and on a government decision to allow a district management board,
composed of representatives of all sectors, to generate funds locally and decide on
their use, rather than returning them to the central government. The WHO/
UNICEF Bamako declaration of 1987 then acted as a catalyst for the development
of a coherent framework within which to extend a similar financing approach to all
government health facilities. The first steps were to establish the legal framework
for the activities and to strengthen drug procurement and supply.
Equally important was the early government action to forge ‘alliances’ [38] with
international agencies and non-governmental organisations (NGOs) in support of
BI activities. The Benin UNICEF country office (BCO) was, for example, a key
external partner for the Ministry of Health and supported the first steps in
management training for community committee members through its expanded
programme of immunisation (EPI). Subsequent support for BI activities was
provided through the World Bank’s project for the development of health services
377
(1990), whilst bilateral donors and NGOs supporting financing activities in different
parts of the country funded drugs, equipment, renovation, training, supervision,
and the development of tools such as clinical pathways for diagnosis. Although
government sought to promote some degree of coherence between these external
partners’ activities, it also provided an environment in which they were encouraged
to experiment and to feed back new design and management ideas into the BI
programme. The relatively gradual growth in the numbers of BI-supported health
centres (increasing from 44 in 1988 to 250 in 1992 [39]) also enabled lessons from
experience to be fed back into the programme.
In addition, both the programme’s design and the manner of its implementation
generated wider support for it. The commitment and enthusiasm of local-level
health workers was, for example, partly promoted through the provision of direct
benefits (such as a financial incentive for each fully immunised child) as well as
through overall service improvements. These improvements in turn promoted
community support of the programme, as did their direct involvement in decision-
making; and with local-level ownership and enthusiasm came the continued support
of government and external donors.
Overall, therefore, a virtuous cycle of policy change was founded on an alliance
between a range of actors. They either shared the common vision underlying the
scheme design or were persuaded of its relevance through successful implementa-
tion. At a technical level, Knippenberg et al. [40] identify three strategies as
particularly important to the development of the Benin BI activities: analysis of best
practices, applying lessons learnt from earlier national and international experi-
ences; translation of best practices into a coherent set of operational strategies and
management systems through experimentation; adaptation of the strategies through
a bottom-up approach involving community participation, peer support, network-
ing and regular monitoring. But, finally, the leadership of the Ministry of Health
was critical in sustaining the implementation process over time as ‘sustainability
depends on the internal capacity to manage the process of change’ ([38] p. 24).
3.2. Actor failure in Kenya
The development of the Kenyan BI programme was, like that of Benin, rooted in
the earlier community financing experiments of NGOs whilst the harambee tradi-
tion, a form of community financing for local development activities, provided
evidence on the potential role of community-based charges [41]. Again as in Benin,
Kenya initially extended its BI activities through a fairly gradual increase in
numbers of BI-supported pharmacies, to try and ensure that the increase in sites
could be adequately supported. Policy guidelines were also developed to support
this expansion, and were allowed to evolve as new lessons and approaches were
developed. The initial successes of the programme only bred further support for the
programme, as parliamentarians saw advantages for their own constituents at an
early stage and began pressing for the faster development and spread of the
approach. The number of BI pharmacies, thus, rose from one in 1989, to three in
1990, to 84 in 1992 and to 237 (including NGO-supported sites) in 1994.
378
However, the Kenyan BI programme, unlike its Benin counterpart, was not
adequately rooted in the context of its development. The programme sought
specifically to extend primary health care coverage to previously under-served areas
on the flawed understanding that the most critical factor undermining the effective-
ness of the Kenyan primary care network in the late 1980s was poor coverage [42].
Yet by the late 1980s this network suffered as much from quality weaknesses as
from poor coverage [41,43], due to the biased allocation of health system resources
towards urban areas and growing balance of payment problems [44]. Weaknesses in
the drug supply and distribution system thus bedevilled the existing primary care
network and, ultimately, the BI pharmacies. At the same time, the programme
failed to build on wider international experience with CHW programmes [45] and
so suffered similar problems—such as communities’ poor perceptions of the low
level of care offered by CHWs, CHW attrition and a failure to provide support to
CHWs through the broader health system.
The Kenyan Ministry of Health, like its counterpart in Benin, played an
important role in the programme’s initiation. Its delegation attended the 1987
WHO/UNICEF Bamako conference and officials working with the Ministry of
Health’s national primary health care unit were subsequently involved in shaping
BI activities, including developing training programmes and supervision manuals.
However, the Kenyan UNICEF Country Office (KCO), to which a key member of
the Ministry of Health Bamako delegation moved shortly after 1987, remained the
stronger partner. Together with a few bilateral agencies, the KCO funded all the
costs associated with pharmacy-based activities, even including the non-salary costs
of the officials working within the national primary health care unit, as well as
being the sole distributor of drugs and bed nets to pharmacies. The significant
dependence of BI activities on UNICEF support explains why they were severely
disrupted by the suspension of this support in 1995/96 during a period of reorgan-
isation within the UNICEF.
It also suggests that, in practice, the UNICEF KCO drove the development of
the BI programme. Thus, it was the KCO officials who were primarily responsible
for the frequent introduction of new ideas, such as changes to the service package,
into the BI programme. It was also the KCO that refused to consider basing drug
procurement systems on the existing national Essential Drugs Programme (EDP)
and instead sought to establish an alternative distribution approach using NGOs.
However, as these innovations were generally based on ‘what might be good to do’
rather than resulting from reflection on experience or the changing context, they
were often flawed. The decision to ignore the EDP, for example, partly reflected the
economic and management difficulties faced by this programme but supporting
NGO distributors was equally problematic and did not survive the withdrawal of
UNICEF’s financial support. This failure to establish sustainable drug supplies was
a critical weakness of the BI programme.
At the same time, Minister of Health policy-makers were responsible for isolating
the BI programme from the wider developments that could have supported it by
following the common pattern of establishing parallel management structures based
on donor funding directed at specific purposes [46]. Run from the central primary
379
health care unit as a vertical programme and only weakly tied to the existing health
facility network, there were few links between BI pharmacies and nearby primary
care facilities. These facilities simply had no funds for, and no interest in, the
activity. At a national level the programme was never given government recurrent
budget support and was kept separate from the management of the broader
cost-sharing programme that developed over the 1990s. As the first level primary
care facility remained free, the failure to link up the two systems of charging not
only created the potential for perverse incentives over utilisation patterns [47] but
also prevented BI activities from being strengthened through the cost-sharing
programme.
Ultimately, therefore, its two central actors, the UNICEF KCO and the Ministry
of Health undermined the Kenyan BI programme. The design of the programme,
its evolution over time and the support it received were simply not adequate to
allow effective implementation. The imaginative approaches developed within it
remained experiments that were not sustained in the face of changing
circumstances.
3.3. The contradictions of implementation strategies in Zambia
The Zambian experience was clearly very different from that of the other two
countries because financing reforms took second place to decentralisation, and so
were both given less consideration by policy-makers and also subjected to other
policy changes. Initiated after the election of the first democratic government in
1991, the decentralisation programme was intended to address the critical weak-
nesses of the health system by strengthening management and quality.
By 1995, the time of this study, the reforms had primarily focused on the
appointment of district health management teams (DHMTs), as well as training
and systems development to strengthen their capacity to manage the budgets
allocated to them. Despite the importance of community participation in decision-
making, less consideration had been given to the appointment and support of
district health boards (to be a governance structure working with management
teams), area boards (to act as a link between the population and district boards) or
neighbourhood health committees (to act as a forum for community-based deci-
sion-making, with representation on health facility management committees). Few
of these bodies were functioning in the districts visited in this study. Following the
guidance of the 1992 National Health Policy and Strategies document [48] fees had
been introduced in some facilities, but the extent and level of fees varied consider-
ably between districts as did revenue retention and use practices.
A major review of the nature and consequences of the Zambian reform imple-
mentation strategy undertaken in 1996 identified the strong leadership and pragma-
tism of the reformers as being fundamental to the achievements in district
development that had by then been secured [49]. Yet at the same time, it suggested
that the incremental nature of the strategy and delays in tackling ‘difficult-to-win’
problems, such as the development of a national drug policy, resulted in a
piecemeal package of reforms and generated uncertainty that undermined imple-
38 0
mentation (Section 4). In particular it suggested that there had been an ‘apparent
ambivalence… to the whole issue of financing, which contrasts sharply with the
clarity and sureness of touch which has characterised many other aspects of the
reform process’ ([49], pp. 23–24). Comparison of the Benin and Zambia experiences
emphasises this point. Whilst an incremental process was adopted in both countries,
in Benin this was rooted in a clearly specified policy design as well as implementa-
tion and monitoring procedures that allowed experience to be reviewed and fed
back into policy development. In contrast, the purpose and design of financing
reform in Zambia was unclear and the general lack of monitoring precluded lessons
being learnt from the process of reform [49,50]. Tackling such problems requires
stronger leadership and vision in the development of financing policy change.
4. Strengthening management through fee introduction: the contribution of policy

design to equity gains
The second explanation for the differing equity impacts of the three countries’ BI
activities lies in seven key differences in the design of the financing activities
investigated in the three countries (Table 2).
1. The Benin BI programme was rooted in an enabling legal and policy frame-
work. New legislation permitted the sale of drugs within health facilities, the
retention of revenue by the collecting facilities and decision-making on revenue use
by community management committees. The overall policy framework comple-
mented legal change and guided the coherent development of BI activities in
different areas of the country-for example, specifying practice concerning fee levels
and revenue use (point 2) and the tasks and functions of community decision-mak-
ing committees (point 6).
Although similar actions were taken in Kenya and Zambia, they did not provide
such clear guidance for implementation in either country. A policy framework [51]
was only established in Kenya after 5 years of experience, whilst its adaptation over
time simply generated uncertainty around key aspects of practice. Not surprisingly
there was considerable variation across Kenyan sites in fee-setting practices and
levels (point 2), the implementation of income generating activities and the extent
of community consultation (Section 5). Zambian fee-related practices also varied
between districts (point 2), largely because, as the health managers and providers
interviewed in this study indicated, the various circulars and verbal official an-
nouncements supposed to guide implementation were commonly perceived as
confusing.
2. In Benin fee levels for curative care (in the form of a drug rather than a
consultation fee), antenatal care and deliveries were established by national man-
agers and community committees were not allowed to adjust them. The prices were
based on the cost of drugs used for complete treatment with a mark-up, varying by
20– 300% between treatment types. This mark-up generated sufficient revenue to
cross-subsidise immunisation outreach activities (which were free of charge) and
curative care for children, and to cover the costs of drug supplies and staff
381
382
50
Table 2
Key elements of policy design
Design element Benin Kenya Zambia
1. Legal and policy framework A clear framework promoted No legal framework; guidelines Inadequate legal framework and
coherent development across country developed late and remained flexible, confusing guidance
generating uncertainty
2. Fee design and fee setting Nationally set fee levels ensured Weak national guidelines adapted by No national guidance and so
practices adequate revenue generated to allow VHCsa on basis of broad assessment DHMTsb made own decisions on
expected cross-subsidisation of other of local circumstances unclear grounds
activities
3. Funding sources supporting Government and donor support Total reliance on donor funding Significant reliance on donor funds
service provision provided to primary care facilities to despite local revenue generation within health system as a whole,
complement local revenue generation and so at district level
4. Strengthening drug availability Deliberate parallel action taken to No action to improve drug supply; No action to improve drug supply
improve drug availability few drugs available in basic package
of care offered in pharmacies
5. Strengthening clinical skills In-service training and supervision Little action Little action
deliberately strengthened
6. Supporting local management Community committees given clear VHC guidelines applied flexibly in DHMTs trained but given weak
structures guidelines, specific training and practice and key roles undermined guidance on roles
regular supervision
7. Strengthening information Clinic information system Steps to develop community-based Focus only on district financial
systems strengthened and used in monitoring information system weak and not information system
activities sustained
a
VHC, village health committee.
b
DHMT, district health management team.
incentives. In practice, the revenue generated through community contributions

covered, on average, nearly 30% of total recurrent costs of the primary care facility
[52].
Fee design and fee-setting practice were quite different in Kenya and Zambia.
The rule of thumb established by central managers to guide drug and bed net fee
levels in Kenya (two to three times the purchase price) was not based on careful
analysis of revenue needs, and village health committees (VHCs) were anyway
allowed to adapt national guidelines on the basis of local circumstances. Price levels
varied, on average, by 400% across drug items between the BI sites assessed. The
lack of a guiding policy framework in Zambia was similarly reflected in the
considerable variability in health centre price levels between districts: rising from
200– 300% for outpatient fees between rural health centres to 400–500% in outpa-
tient fees between urban health centres. The revenues generated were barely
adequate to re-supply drugs and nets in Kenya and made negligible contributions
to total operating costs in Zambia.
3. Within the Benin BI programme’s funding package government and donor
support complemented the revenue generated by fees [52]. Government contribu-
tions, representing about one half of total facility recurrent costs, fully covered the
costs of salaries and partially covered the costs of supervision, whilst donor
contributions, representing less than one-quarter of total facility recurrent costs,
supported the cost of transport, training, cold chain requirements and building
renovation and maintenance. At the same time, some steps had been taken by the
mid-1990s to reduce reliance on donor funding. For example, external funding for
fuel for immunisation outreach services was being gradually withdrawn as health
facilities began to cover these costs fully from their own revenue surpluses [40].
In contrast, the Kenyan BI programme’s almost total reliance on external
funding led to the severe disruption of its activities when the UNICEF KCO
withdrew its support in 1995/96. Whilst the Zambian health system’s reliance on
donor funding [53] made it similarly vulnerable to changing donor priorities,
donors were broadly in support of the reform programme at the time of this study.
4. The Benin BI programme’s efforts to tackle low quality within primary care
facilities was supported by improving drug availability through parallel action to
promote essential drug lists and use international tendering procedures.
In direct contrast, no steps had been taken by the time of this study to develop
an effective drug procurement and supply system in support of BI pharmacies in
Kenya or the wider health system in Zambia. The weaknesses of the Zambian
system meant that fees were introduced without any concomitant improvement in
drug availability at health facilities. As a drug, rather than consultation, fee was
levied, patients complained that they effectively had to pay twice, once for
consultation in the public facility and a second time in purchasing drugs from other
sources [54]. Perhaps not surprisingly, the initial evidence suggested that utilisation
levels fell considerably after fees were introduced—for example, by 40–100% in
selected clinics in Lusaka Urban District [22,55]. In Kenya, focus group discussions
with VHCs held after UNICEF had stopped supplying drugs and nets to pharma-
cies indicated that pharmacies were then experiencing major problems in drug
383
availability and had turned to local, private sources, despite concerns about the
quality of their supplies. In practice, therefore, the access gains achieved by locating
pharmacies in previously under-served areas were undermined by the failure to
develop a secure, local, drug procurement system. In addition, as the benefit
package offered through the programme included only first aid care, and little
access to referral services, community members still had to use other sources of care
for some, particularly more serious, conditions, with the consequent cost implica-
tions. In the household surveys undertaken within this study, the limited range of
drugs was the most frequently identified community criticism of the BI activities.
5. Within the Benin BI programme, various actions were taken to strengthen the
clinical skills of primary care staff. They were given in-service training to promote
rational prescribing of drugs, the use of clinical pathways (such as flowcharts) in
diagnosis and risk screening in the provision of care to pregnant women. Efforts
were made to strengthen supervision practices, including the development of a tool
to help facility staff and supervisors monitor coverage and identify and address the
obstacles to improved coverage. Fee revenue was also partly channelled into
supporting regular supervision, through a flat-rate levy of 2500CFA on all health
centres paid to the local Direction Departmentale de la Santé (i.e. regional health
office). As a result, nearly all (99%) of the health staff interviewed in this study
indicated that their health centre received financial, material and technical support
from higher levels (although another study identified weaknesses in supervision
practices [40]).
In contrast, clinical skills’ development was weak in both Kenya and Zambia.
Indeed, at the time of this study, the Zambian health reforms explicitly focussed on
the development of management rather than clinical skills. In Kenya, data collected
from household surveys in case study sites indicated that the limited skills of CHWs
was the second most frequently identified community criticism of the BI pharma-
cies. Subsequent in-service training rarely followed the short-period of initial
training given to CHWs, and little supervision was provided. Pharmacy staff at
only one out of the 12 sites visited in this study indicated that they had received
support from the neighbouring health facility whilst national supervision was,
again, ultimately undermined by the lack of secure funding for BI activities.
6. Local management structures were developed in Benin by clearly defining the
tasks and functions of community committees, and providing relevant training for
their members. The Comite de Gestion de Commune (COGEC) was given respon-
sibility for managing drugs (receiving drugs, stock control, being informed on drug
orders made by staff), managing funds (banking money and keeping one of the two
keys to the facility safe), employing and paying local workers such as drug
dispensers, and deciding on how to use money. Clear guidelines, training and
supervision also promoted common practices across communes: thus, 74% of the
health workers interviewed in this study indicated that revenue use in their facility
followed policy guidance.
Although guidelines were established to guide the establishment and functioning
of VHCs in Kenya [51], the establishment, size, composition and activities of the
committees varied considerably between sites. Their revenue management function
384
was anyway undermined by UNICEF’s continued provision of financial support

and drug and bed net supply. Rather than being used to support BI activities the
revenue generated by fees largely remained in bank accounts, earning interest but
losing value, and sometimes being misused. In Zambia although community
committees had not been established at the time of this study, district management
teams had been strengthened using an on-the-job training approach, rooted in plan
development and performance monitoring. However, as already noted, the guid-
ance district managers received on fee-related issues was often confusing. At the
time of this study no attempts had been made to develop the management skills of
health facility staff or community committees.
7. The management information system was strengthened in Benin by linking it
to local decision-making concerning health care provision, resource management,
supervision of quality of care and monitoring coverage, drug use and cost recovery
[38]. Steps were taken to involve both health staff and community members in
simplifying the system, so increasing their understanding of the information avail-
able. This also promoted transparency at a local level.
Similar efforts to strengthen the Zambian district financial information system,
through a process involving district management staff, were not, however, extended
to other relevant management information or to the health facility and community
level. A local-level information system developed to record basic community
statistics (such as births, deaths, pit latrines constructed etc.) within the Kenyan BI
sites, the ‘chalk and board’ system, was simply not sustained after the withdrawal
of UNICEF support.
Overall, this cross-country comparison of design issues emphasises that the Benin
BI’s promotion of relative affordability gains was not simply a function of levying
fees. Rather, as Knippenberg et al. ([40], p. 42) comment, ‘while the cost sharing
mechanism initially seemed revolutionary at the national and international levels,
the linkage with strengthened clinic management, staff quality and morale, drug
supply and relations with the community as a whole were visibly more important
factors in revitalising’ the health centres. The management change associated with
fee introduction was, ultimately, the key to improving the service quality and
coverage of primary care facilities in Benin, whilst management weaknesses under-
mined the Kenyan and Zambian financing activities.
5. Failing the poorest: the interacting problems of policy design and process
Despite its other successes, the Benin BI programme shared a common equity
problem with the financing activities examined in Kenya and Zambia: all three
failed to protect and benefit preferentially the poorest within communities.
A critical factor underlying this equity problem was the failure to establish the
protection of the poorest as a clear goal of the activities. The Benin BI programme
sought, rather, to improve quality of care, and the Kenyan programme, to support
both improved access to drugs at community level and health-promoting commu-
nity development actions. Whilst the Zambian reforms sought broadly to improve
385
equitable access to cost-effective health care, fees were introduced with the specific
goals of creating community ownership of the health system and raising revenue.
Given these goals, the subsequent design and implementation of the relevant
financing activities in all countries simply failed to recognise and tackle the specific
needs of the poorest. For example, neither Benin nor Kenya took action to offset
differences in revenue generating capacities between communities of 200% annually
(Benin) and 900% monthly (Kenya). Although not fully investigated, there were
signs that more wealthy communities generated higher levels of revenue, and
benefited from greater service improvements, than less wealthy communities [27,38].
Zambian action to implement a resource re-allocation mechanism between districts
may, however, provide lessons for other countries on this issue [55].
The three design problems promoting intra-community inequities are highlighted
in Table 3, together with the key factors explaining them. However, for each issue
and in each country there were important features of context that influenced
practice concerning the poorest, and that cannot easily be off-set by actions within
the health sector alone. The health needs of the poorest and their ability to
contribute to local decision-making clearly require much broader action if the
socio-economic and socio-political roots of these problems are to be effectively
addressed.
5.1. Ignoring financial barriers
A critical gap in the design of all the schemes of focus was the lack of an effective
means to protect the poorest from the burden of payment. Zambia was the only
country in which guidance on who to exempt was established by the central
Ministry of Health. In Benin and Kenya the decision of whether or not to protect
the poorest groups from payment, and how, was left to the local-level management
committee on the grounds that it could best make case-by-case exemption judge-
ments. Yet in all countries the weak guidance on who to exempt and how to
provide for the poorest groups’ needs was commonly identified by interviewees in
these studies as a reason why exemptions or reduced prices were usually not
offered.
Exemption practice in all countries was, however, primarily undermined by the
conflict between financial sustainability and protection of the poor. Even in
Zambia, where revenue generation was not an explicit goal of the fee system,
providers interviewed in this study complained that if the exemptions of policy were
applied fully it would prevent revenue generation. In Benin the need to recover
costs in order to maintain the quality of services was the most important reason
given by service users for why protection was not offered to the poorest, and was
also one of the reasons given by health staff. ‘More and more, social assistance and
the desire to help the sick who are targeted by the health services is undermined by
profit’ (focus group discussion, young people). The pre-eminence of financial
sustainability was almost inevitable given the programme’s insistence on generating
revenue to promote service improvements. The training and supervision offered to
primary care workers and community members stressed their responsibility to raise
386
Table 3
Explaining intra-community equity losses across countries
Equity losses Explanatory factors
Context/actors Design/actors Process/actors
No protection for Benefiting majority poor is the Primary goal of financial sustainability; Top-down and inconsistent implementation
poorest established and accepted equity goal; unclear or no guidance on who to process undermines authority of local
communities may not wish to implement exempt; vague or weak exemption actors to offer protection; limited training
protection for poorest (prices affordable, mechanism; no other mechanism to and supervision to develop relevant
danger of leakage to non-poor); weak tackle financial barriers; poor have no management capacity
management capacity voice (see below)
Limited benefit Community demand/preference for Limited health promotion benefit Top-down and inconsistent implementation
strategies curative care; low cash incomes limits packages; curative care dominance; process undermines authority of local
revenue generation possible; weak limited curative care package (Kenya); actors to widen benefit package; training
management capacity target group primarily defined in and supervision to develop relevant
disease terms (at risk); limited management capacity; limited consultation
inter-sectoral collaboration; poor have within community
no voice (see below)
Not listening to the Characteristics of poorest; socio-cultural ‘Community participation’ seen as Socio-cultural realities dominate practice of
poorest realities of local communities strategy of implementation not implementation; implementation through
objective in its own right; formal local structures promotes exclusion of
guidance that promoted exclusion of poorest; top-down implementation
poorest; no mechanisms to promote undermines local ownership and
inclusion of poorest decision-making by local structures
55
387
revenue and so both groups expressed a concern about the need to avoid making
a loss and to ‘balance the books’. ‘We do not see any sense of equity in the
decisions taken about the health centre. Perhaps the health workers and the
COGEC members have not been sensitised to this issue’ (focus group discussion,
village leaders). Similar practices in Kenya may also have influenced decision-mak-
ing despite external support for the provision of drug and bed net supplies.
Given the dominance of financial sustainability, it is perhaps not surprising that
little consideration was generally given to other possible strategies for addressing
the financial barriers faced by the poorest. Yet in three Kenyan sites, VHCs had
been encouraged by the DHMT to consider imaginative ways of addressing the
issue, such as an approved list of those entitled to exemptions and a special bank
account to cover the costs of care for the poorest. Some NGOs in Kenya and Benin
had also developed broader protection strategies. In one Kenyan site, a community
solidarity fund (which was set up and funded separately from the health care fee
system) was used to pay for the health care provided to the indigent. And in Benin,
the poor were protected through mechanisms such as reduced prices and a
pharmacy providing free drugs to the poor.
The failure to develop such innovative protection strategies in most communities
may itself reflect the limited authority given to local decision-makers within the BI
programmes. In both Benin and Kenya the composition and tasks of community
committees were determined within fairly limited parameters by higher levels. In
Benin, for example, they were neither allowed to determine price levels nor given
much freedom in terms of revenue use. There was, in effect, limited management
flexibility to respond to the financial needs of the poor, as highlighted in discussions
with COGEC members.
‘‘…the COGEC has regulations to respect, which considerably limit its field of
action. Drugs must be disbursed at a small cost, we have no authority to
distribute them freely and the stocks must be replaced.’’
‘‘…the COGEC is ruled by regulations which deprive it of its autonomy.’’
Although Zambian providers were given authority to offer specified exemptions,

the guidelines were implemented differentially between districts because of a failure
effectively to communicate them either to health staff or the community at large.
Indeed, guidance on exemptions was only provided after fees had been introduced
and been negatively received by the population. Thus, staff at one rural health
centre indicated that no official communication had been received about exempting
under fives or the elderly and so ‘being just a rumour [they] did not exempt the two
from paying’. Many others complained that policy was changed often and that the
changes only came as verbal pronouncements. In 1994 the Deputy Minister of
Health had even announced that nothing should be considered official until written
notification had been received from the permanent secretary, given the number of
verbal pronouncements being made from the central level.
388
Ultimately, the voice and views of the poorest were often simply not heard or
considered in decision-making on price structures and levels. In Benin, for example,
price levels were largely thought to be acceptable by the general population. Yet
whilst only 1% of the community-level key informants felt that prices should be
related to socio-economic status, 62% of those interviewed from the poorest group
said they would like to obtain exemptions and 87% said current price structures
deterred some people from accessing services.
5.2. Inadequate de!elopment of pro-poor benefit strategies
The importance of benefit strategies to equity gains is shown, for example, in the
contrast between the relative affordability gains of fees with quality improvements
in Benin and in the decline of utilisation rates that appeared to be associated with
the introduction of fees without quality improvements in Zambia. The contrasting
experiences of Benin and Kenya also suggest that the nature of benefit strategies
influence the extent to which the poorest preferentially benefit from health care. The
broader health promotion and development strategies pursued in Kenya had the
potential to generate equity gains by cross-subsidising the spread of benefits within
communities beyond the group of health care users. Although the cross-subsidisa-
tion of immunisation services in Benin did generate some similar gains for the
health vulnerable groups of mothers and children, the dominant focus on curative
care channelled most benefits only to those using these services. Yet financial
barriers continued to constrain access to these benefits by at least some of the
poorest [38,56].
The potential benefits of the broader Kenyan benefit strategy were, moreover
undermined by the limited development of such activities. In practice, only four
sites initiated income-generating activities (IGA) and of these, only one site
supported activities through an IGA that spread benefits widely within the commu-
nity (the construction of a road and a school). In other sites the IGAs generated
benefits for only a limited group, sometimes as incentives to CHWs. Even relative
affordability gains were constrained in Kenya by the limited package of care
provided, as it required continued use of more expensive and more distant health
providers especially for more serious, and potentially expensive, conditions. These
weaknesses of the Kenyan BI programme reflected four main factors:
1. The programme adopted a curative care ‘entry point’ in initiating its activities,
with the intention of building broader primary health care activities over time.
However, the pharmacies came to be seen by the community almost solely as
places that sold drugs and bed nets, perhaps reinforcing a general preference for
curative services and undermining the intended role of the BI programme in
health promotion.
2. It is always difficult to raise revenue at primary care level: price setting has to
balance the potential impact on demand with the generation of funds [12]. In
practice, the revenue generated within BI sites was barely adequate to re-supply
drugs and bed nets and no site visited in this study had generated enough
revenue to give CHWs incentives for providing preventive services, or broader
development activities.
389
3. To offer broader benefit strategies it is necessary that local management

committees are trained in a wider range of skills and better supported, than in
more narrowly focused approaches. Yet the skills and training needed to support
the diverse range of IGAs initiated were simply not available within the BI
programme, and would have required inter-sectoral collaboration.
4. Community members expressed strong concern that decision-making around
IGAs, in particular, was in the hands of the VHC and/or CHWs rather than the
whole community in four out of the ten government-supported BI sites visited
in this study.
‘‘We find ourselves at a crossroads now because there is nothing we can ask the
VHC about this project because we were not part and parcel of its inception.’’
‘‘We cannot comment on the IGAs because even at present none of us knows the
number of bags of maize which were brought to be sold.’’
IGAs may, therefore, have become simply a way of generating benefits for a small
elite rather than promoting health and development activities of benefit to the wider
community.
5.3. Not listening to the !oice of the poorest
The failure to hear the voice of the poorest reflected a broader problem: there were
signs in all three countries that the community at large, let alone the poorest, did
not feel involved in decision-making. In some Kenyan BI sites, activities were initiated
by a specific group or person (such as pre-selected CHWs or the chief) and this
influenced the wider community’s perception of who ‘owned’ the BI pharmacy. Even
when elections were undertaken without the overt influence of the local administra-
tion, the chief’s real influence would be understood by the community and he, or an
assistant chief, might be present at the baraza (chief’s assembly). The very fact that
the baraza was seen as a key instrument in initiating BI activities underlined the
potential for the chiefs to manipulate the activities to their own ends. In one case,
a chief took control of the dairy cattle owned by the BI for ‘safe keeping’ and then
declared the animal his, in spite of opposition from community members. In contrast,
there were other instances when the district BI co-ordinator (a DHMT member)
directly involved himself in local decision-making concerning the appointment of
office bearers and price levels. Whilst perhaps undertaken to promote ‘good practice’,
this may also have undermined local ownership. Not surprisingly, community
members often thought that ‘the project’ belonged to the VHC, the BI co-ordinator,
the Minister of Health or UNICEF. Similarly in Zambia, although cost sharing was
introduced ostensibly with the aim of promoting partnership, few community
members felt they could participate in decision-making or influence practice. One
analysis of the Zambian experience expressed concern that decision-making had been
taken over by some health staff and so had discouraged the community [49].
39 0
As noted, in all three countries key aspects of implementation remained effec-

tively controlled by higher levels and so precluded opportunities to listen to the
poorest. All community-based key informants interviewed in Benin, therefore,
stated that the health authorities set prices. In Kenya, the revenue generated largely
remained stored in bank accounts whilst community bodies awaited instructions on
when and how to use them. Only when UNICEF deliveries of bed nets and drugs
failed to arrive did communities begin to think they could use the revenue they had
collected. In Zambia, although decisions concerning price levels, exemptions and
revenue generation were being taken at district level, there were also signs that the
district sometimes blocked decision-making by lower levels on these issues. Thus, in
six out of eight districts visited in this study, facilities were required to bank fee
revenue at the district level— leading to some confusion about how the revenues
could be used and who could decide on their use. The lack of clear guidance only
exacerbated the issue: guidelines requiring that a proportion of revenue be retained
for use by individual facilities were drafted and verbally communicated to district
managers, but never signed and given official status.
The policy guidance implemented through these top-down practices sometimes
directly excluded the poorest. Thus, in Kenya, VHC members and CHWs had to be
literate and the selection of both groups had to occur through the baraza. Yet 78%
of the poorest households surveyed in case study sites had not attended a baraza in
the previous year, compared to 43% of those surveyed in the initial survey.
‘‘The vulnerable members do not get an opportunity to be a CHW or join a VHC

because… that selection is one only for the fittest members in society.’’
‘‘The poor do not take part in the decisions regarding exemptions because they
do not take part in meetings.’’
Clearly, however, the diverse range of personal and material factors that charac-
terised the poorest in all countries [27] are likely themselves to have had a
marginalising effect on their role in the community. The extreme poverty from
which the poorest suffer inevitably places an enormous burden of survival on them
and may simply prevent them from engaging in any voluntary activity. Women may
be most excluded from decision-making because of deep-rooted beliefs about the
traditional roles of men and women and so, despite policy guidance, the VHC
chairperson was a man in all Kenyan BI sites visited.
Perhaps the tendency towards top-down implementation approaches was in-
evitable in all countries. The problems were defined as technical in nature, the
technicians played a dominant role in generating solutions, the traditional decision-
making practices of most communities and public sectors were hierarchical and
external, international agencies played a strong role in supporting these activities.
Certainly, despite stated intentions, an appropriate balance between central level
control and local decision-making seems never to have been achieved. Some
decisions, such as who to exempt, were left to the community in apparent reflection
391
of the international view that this was the most effective way of identifying and
addressing some community needs. However, this approach ignores the clear
pressures to focus on other priorities at the expense of the poorest, as well as the
socio-cultural and political realities of communities. In addition, the practice of
implementing change in all three countries gave only limited roles to these local
decision-making structures and consistently excluded both direct and indirect
consideration of the voice of the poorest. Only in sites supported by NGOs, where
special mechanisms had been established to address the needs of the poorest and
the parent organisation had taken responsibility for providing funding, were these
mechanisms implemented effectively. Overall, therefore, the community decision-
making bodies created to strengthen accountability by giving a ‘voice’ to the
community often did not appear to serve the interests of the poorest.
6. Conclusions and recommendations: meeting the needs of the poor and the
poorest
6.1. Strategy is always important
The three-country studies all illustrate the critical importance of leadership and
strategy to the effective implementation of policy change. Managing such change
requires both political skills, to develop and mobilise support, and technical skills,
to inform and guide the reform process [57 –59]. The careful design of reforms can
aid implementation by reducing the potential for confusion or conflict by stating
clear goals, outlining simple technical features and establishing clear implementa-
tion steps. Within a clear guiding framework, incremental approaches then allow
capacity for implementation to be developed, give implementors the flexibility to
learn from experience and enable support for change to be developed.
The continual adaptation of reforms in pursuit of goals is also only possible if
there are sound procedures for monitoring and evaluating experience [57,58,60].
For pro-poor policies it is particularly important to monitor the impact of policy on
the poorest. Dis-aggregated data are essential for this task. For example, it must be
possible to identify and compare the utilisation of different population groups as
well as to track changes in utilisation over time. This study has also highlighted the
usefulness of looking at various aspects of equity, and the interaction between
them, as well as the need to understand why and how change is brought about-not
only what change is achieved.
6.2. Sustaining the potential equity gains of community financing schemes
The country experiences reviewed here also suggest that the key factor in
sustaining the potential relative affordability gains of community financing activi-
ties is to use the introduction of fees as a policy lever for strengthening manage-
ment. The key, interacting steps required to ensure these gains include:
392
! establishing a clear design that includes local retention of most revenue and
cross-subsidisation of a limited range of preventive services;
! developing a legal and policy framework enabling implementation;
! ensuring that parallel action is taken to support implementation-in particular,
reforms to improve drug availability and to support decentralised decision-
making;
! providing clear and detailed guidance on pricing practice and revenue use;
! providing management and clinical training and supervision for health facility
staff, possibly supported by a financial contribution from each facility;
! encouraging health facility staff to monitor local health facility performance;
! involving local community structures in decision-making with appropriate guid-
ance and support;
! generating in-country support for change through incentives and sustained
improvements;
! maintaining government financial support for at least the salaries of staff and
using donor funds as flexibly as possible to support the overall approach;
! adopting a gradual but progressive implementation process.
6.3. Seeking to meet the needs of the poorest
However, the experience of all three countries highlights the difficulty of estab-
lishing effective exemption mechanisms to protect the poorest from payment,
especially within systems seeking to promote financial sustainability.
An alternative approach, proposed by respondents in both Benin and Kenya, is
to establish a separate ‘community solidarity fund’ which can fund the use of care
by the poorest, alleviating the tension between financial sustainability and concern
for their needs:
‘‘To better care for the impoverished and vulnerable, the political authorities
must count the indigent. The state must, moreover, give the health centre a
special drug supply to care for the impoverished and vulnerable who don’t have
support.’’ (Benin focus group discussion, young people)
A first step would be to develop mechanisms for determining who should be

given support. Drawing on the 1992 Zambian experience of drought relief proce-
dures, Booth et al. ([54]; see also [61]) suggest that local, democratically elected
committees could be strengthened by NGOs in assessing each household within the
catchment area of health centres and determining which should be exempted. The
approach has some similarities to that of the Thai low income card scheme which
brings local leaders and health workers together to determine on the basis of a
nationally-determined income threshold who within a community should be allo-
cated a card entitling them to free care. Over 15 years of implementation experience
has shown that such an approach can be implemented relatively effectively [62]. de
Kadt and Tasca [63], similarly, propose a geographic targeting approach based on
393
identifying vulnerable groups by living conditions, rather than income, through a

process that uses both available technical information, such as health statistics, and
the knowledge of the local population. They suggest that health interventions, and
inter-sectoral action, should then be directed to these target populations in response
to their worse access to care or experience of a particular health problem.
Although there are no easy options in meeting the needs of the poorest, the
experiences examined here suggest that the following actions are always important
to consider:
! maintaining government and donor support within an overall financial plan for
the health sector, so that the full burden of financing, and especially the burden
of financing necessary support, is not left to communities;
! the creation of local decision-making structures which try to take into consider-
ation the needs of the poorest by specifically seeking representation from civil
society groups such as churches and NGOs, women and others, and by proce-
dures which allow broader views to be heard (e.g. community-wide meetings,
specific attempts to hear the needs of the poorest);
! developing broad approaches to targeting which involve local people working
within central guidelines, and which are managed and funded separately from
the local revenue generation mechanism;
! developing benefit packages broader than curative care to ensure the wide
dispersion of benefits within the community (recognising the particular impor-
tance of strong local-level administrative capacity);
! a package of training and supervision which strengthens local management
practices and emphasises the importance of addressing the needs of the poorest;
! a monitoring approach, perhaps building on a targeting mechanism, which
allows changes in the situation of the poorest to be identified and fed back into
health service planning and local decision-making;
! an appropriate balance between local and central decision-making.
The last issue is possibly one of the most critical and is also emphasised by wider
decentralisation experience [64]. Rather than simply leaving protection of the
poorest to communities, governments need to provide financial assistance, guidance
and appropriate support to communities in this task and in promoting inter-com-
munity equity. By themselves community financing schemes can do little for the
poorest, instead much broader action, backed by political support, is required.
6.4. The continuing debate
Ultimately, however, the nature of the equity goal established to guide any health
programme’s development will influence the equity gains it actually promotes.
Whose views and values should underlie the selection of this goal? Some argue that
the concern for the poorest groups is imposed on African cultures by external
agents [65]. Carrin, thus, ([66], p. 186) suggests that:
‘‘…equity does not normally seem to be perceived as a priority at the outset of

a community financing scheme. One of the reasons is that feelings of interfamily
394
solidarity may be rather weak so that the population may resist the implementa-
tion of certain equity rules in a financing scheme… Greater equity should be kept
as a long-run goal. Schemes are invited to monitor equity and to move gradually
towards this goal.’
However, during focus group discussions undertaken in these studies community

members expressed concern for the poorest as well as the broad community, and
recognised the difficulties faced by the poorest in accessing fee-paying care:
‘‘Equity requires equality of rights for all at the health centre with, nonetheless,
some priority for the worst sufferers and the children.’’ (Benin)
‘‘The poor should be chosen as leaders of the project as well so that they can
speak on behalf of other poor colleagues about their requirements.’’ (Kenya)
‘‘Equity in health care means that everyone, whether they are rich or poor and
whatever ethnic group they come from should have access to health care when
they need it… Equity is not possible because every intervention has its own price
and those who have no money dare not even come to the health centre.’’ (Benin).
Although an inadequate analysis, these community voices may be suggesting that

strategies to promote equity must achieve gains for the majority poor and the
minority poorest. Further research on understanding how communities perceive
equity and how to achieve it would be an important foundation for future policy
development.
Acknowledgements
This three-country study was undertaken by an international research team

involving: the Institut Regionale de Santé Publique and the Ministere de la Santé in
Benin; the African Medical and Research Foundation and the Ministry of Health
in Kenya; the University of Zambia and the Ministry of Health in Zambia; the
Swiss Tropical Institute; and the London School of Hygiene and Tropical
Medicine, UK. The authors are particularly grateful to the community members
and health sector officials who were interviewed during the course of the studies
and to the following members of the country research teams: Benin: Professors
M.P. Diallo, T. Zohoun, Dr A. Alassane, C. Assegnisso, A. Attalou, Dr L. Kessou,
M. Parais; Kenya: W. Liambila, I.O. Okulla, S. Ong’ayo; Zambia: Drs J. Milimo;
S. Bennett. The study was financially supported by UNICEF’s central Bamako
Initiative Operations Research Unit, and received operational support from
UNICEF country offices. Additional financial support for the Kenya study was
provided by the Tropical Diseases Research programme of the World Health
395
Organisation. L.G. and S.L. worked on this study as members of the Health
Economics and Financing Programme of the London School of Hygiene and
Tropical Medicine, which receives financial support from the UK’s Department for
International Development.
References
[1] Mills A. Reforming health sectors: fashions, passions and common sense. In: Mills A, editor.
Reforming Health Sectors. London: Kegan Paul, 2000.
[2] Segall M. From cooperation to competition in national health systems — and back?: impact on
professional ethics and quality of care. International Journal of Health Planning and Management
2000;15:61– 79.
[3] World Health Organisation. The World Health Report 2000. Health Systems: Improving Perfor-
mance. Geneva: World Health Organisation, 2000.
[4] Gilson L, Russell S, Buse K. The political economy of user fees with targeting: developing equitable
health financing policy. Journal for International Development 1995;7(3):369 –402.
[5] Reich M. The politics of health reform in developing countries. In: Berman P, editor. Health Sector
Reform in Developing Countries. Harvard: Harvard University Press, 1996.
[6] Nelson J. The politics of pro-poor adjustment. In: Nelson J and contributors Fragile Coalitions,
editors. The Politics of Economic Adjustment. Washington, DC: Overseas Development Council,
1889.
[7] Abel-Smith B, Dua A. Community financing in developing countries: the potential for the health
sector. Health Policy and Planning 1998;3(2):95 – 109.
[8] WHO. Guidelines for implementing the Bamako Initiative. Regional Committee for Africa, 38th
session, Brazzaville, 7 – 14 September, 1988 AFR/RC38/18 Rev.1.
[9] UNICEF. The Bamako Initiative Planning Guide. New York: Bamako Initiative Management
Unit, UNICEF, 1990.
[10] Gilson L. Charging for Government Health Care: Is Equity Being Abandoned? EPC publication
no. 15. London: London School of Hygiene and Tropical Medicine, 1988.
[11] Gilson L. Financing Systems within PHC. In: Streefland P, Chabot J, editors. Implementing
Primary Health Care. Amsterdam: Royal Tropical Institute, 1990.
[12] Gilson L. The lessons of user fee experience in Africa. Africa Health Policy and Planning
1997;12(4):273– 85.
[13] McPake B. User charges for health services in developing countries: a review of the literature. Social
Science and Medicine 1993;36(11):1397 – 405.
[14] Reddy S, Vandemoortele J. User Financing of Basic Social Services: A Review of Theoretical
Arguments and Empirical Evidence. Office of Evaluation, Policy and Planning, United Nation’s
Children’s Fund: New York, 1996.
[15] Brinkerhoff DW. Process perspectives on policy change: highlighting implementation. World
Development 1996;24:1395 – 401.
[16] Gilson L, Mills A. Health sector reform in sub-Saharan Africa: lessons of the last 10 years. Health
Policy 1995;32:215 – 43.
[17] Glassman A, Reich M, Laserson K, Rojas F. Political analysis of health reform in the Dominican
Republic. Health Policy and Planning 1999;14(2):115 –26.
[18] Grindle MS, Thomas JW. Policy Choices and Policy Change: the Political Economy of Reform in
Developing Countries. Baltimore, MD: The Johns Hopkins University Press, 1991.
[19] Nelson JM. Poverty, equity and the politics of adjustment. In: Haggard S, Kaufman RR, editors.
The Politics of Adjustment. Princeton: Princeton University Press, 1992.
[20] Walt G, Gilson L. Reforming the health sector in developing countries: the central role of policy
analysis. Health Policy and Planning 1994;9(4):353 – 70.
396
[21] Janovsky K, Cassels A. Health policy and systems research: issues, methods and priorities. In:
Janovsky K, editor. Health Policy and Systems Development: an Agenda for Research. Geneva:
World Health Organisation, 1996.
[22] Gilson L, Kalyalya D, Kuchler F, Lake S, et al. Promoting Equity Within Community Financing
Schemes: Experiences from Three African Countries. PHP Departmental publication no. 31.
London: London School of Hygiene and Tropical Medicine, 1999.
[23] Kalyalya D, Lake S, Milimo J. Promoting Equity in Cost-Sharing Schemes. Report of the Zambia
Case Study. Lusaka/London: University of Zambia/London School of Hygiene and Tropical, 1997.
[24] Oranga HM, Gilson L, Ong’ayo S, Liambila W, Okulla O. Promoting Equity Within Bamako
Initiative Schemes: the Kenyan Experience. Summary Report. Nairobi/London: African Medical
and Research Foundation/London School of Hygiene and Tropical Medicine, 1997 (mimeo).
[25] Ouendo EM, Diallo P, Zohoun T, Kuchler F. Operational Research on Equity in the Implementa-
tion of the Bamako Initiative. Cotonou/Basle: Institut Regional de Sante PubliqueP/Swiss Tropical
Institute, 1996 (mimeo).
[26] Jarrett S, Ofosu-Amaah S. Strengthening health services for MCH in Africa: the first four years of
the Bamako Initiative. Health Policy and Planning 1992;7(2):164 – 76.
[27] Gilson L, Kalyalya D, Kuchler F, Lake S, et al. The equity impacts of community financing
activities in three African countries. International Journal of Health Planning and Management,
2000;15:291 –317.
[28] Gilson L. What is the future for equity in health policy? Health Policy and Planning 1989;4(4):323–
7.
[29] Donaldson C, Gerard K. Economics of Health Care Financing: the Visible Hand. Basingstoke:
Macmillan, 1993.
[30] Gilson L. Re-addressing equity: the importance of ethical processes. In: Mills A, editor. Reforming
Health Sectors. London: Kegan Paul, 2000.
[31] Mooney G. And now for vertical equity? Some concerns arising from Aboriginal health in
Australia. Health Economics 1996;5:99 – 103.
[32] Daura M, Mabandhla M, Mwanza K, Bennett S. District Cost-Sharing. Bethesda MD and Lusaka:
Partnerships for Health Reform, Abt Associates Inc and Department of Economics, University of
Zambia, 1998.
[33] McPake B. Financing. In: Janovsky K, editor. Health Policy and Systems Research: Issues,
Methods and Priorities. Health Policy and Systems Development: an agenda for research. Geneva:
World Health Organisation, 1996.
[34] Bennett S. The public/private mix in health care systems. In: Janovsky K, editor. Health Policy and
Systems Research: Issues, Methods and Priorities. Health Policy and Systems Development: an
agenda for research. Geneva: World Health Organisation, 1996.
[35] Yin RK. Case Study Research: Design and Methods. Thousand Oaks: Sage, 1994.
[36] Folch-Lyon E, Trost JE. Conducting focus group sessions. Studies in Family Planning
1981;12:443 – 9.
[37] Inoussa S, Padanou P, Faton M, et al. Primary health care under difficult conditions: experiences
in Benin. In: Chabot J, Harnmeijer JW, Streefland PH, editors. African Primary Health Care in
Times of Economic Turbulence. Amsterdam: Royal Tropical Institute, 1995.
[38] Knippenberg R, Soucat A, Oyegbite K, et al. Overview: sustainability of primary health care
including immunisations in Bamako Initiative Programmes in West Africa: an assessment of 5
years’ field experience in Benin and Guinea. International Journal of Health Planning and
Management 1997;12(Supp. 1):S9 – S28.
[39] Levy-Bruhl D, Soucat A, Osseni R, et al. The Bamako Initiative in Benin and Guinea: improving
the effectiveness of primary health care. International Journal of Health Planning and Management
1997;12(Supp. 1):S49 – 80.
[40] Knippenberg R, Alihonou E, Soucat A, et al. Implementation of the Bamako Initiative: strategies
in Benin and Guinea. International Journal of Health Planning and Management 1997;12(Supp.
1):S29 –48.
[41] Bennett FJ, Maneno J. National Guidelines for the Implementation of Primary Health Care in
Kenya. Nairobi: Ministry of Health/World Health Organisation/United Nations Children’s Fund,
1986.
397
[42] UNICEF (United Nations Children’s Fund). Community Financing of Primary Health Care: the
Bamako Initiative in Kenya. UNICEF Kenya Country Office, 1991 (mimeo).
[43] UNICEF (United Nations Children’s Fund). Project: Strengthening Primary Health Care Through
Community-Based Action. UNICEF Kenya Country Office, 1992 (mimeo).
[44] Ministry of Health. Investing in Health: Kenya’s Health Policy Framework. Nairobi: Government
Printers, 1994.
[45] Walt G, editor. Community Health Workers in National Programmes: Just Another Pair of Hands?
Milton Keynes: Open University Press, 1988.
[46] Cohen S, Mwanzia J, Omeri I, Ong’ayo S. Decentralisation and Health Systems Change in Kenya.
Geneva: World Health Organisation, 1995 (mimeo).
[47] McPake B, Ajuong F, Forsberg B, et al. SIDA/ODA/LSHTM Evaluation of the Bamako Initiative
in Kenya. London: London School of Hygiene and Tropical Medicine, 1991.
[48] Ministry of Health. National Health Policies and Strategies: Health Reforms, Lusaka: Ministry of
Health, 1992.
[49] Ministry of Health, World Health Organisation, United Nations Children’s Fund, World Bank.
Independent Review of Zambian Health Reforms. Lusaka: Ministry of Health, 1997.
[50] Choongo D, Milimo J. Decentralisation and Health Systems Change: Zambia Case Study. Geneva:
World Health Organisation, 1995 (mimeo).
[51] Liambila W, Maneno J, Hill J, editors. A Practical Guide to the Planning and Implementation of
the Bamako Initiative in Kenya. Nairobi: Ministry of Health/United Nations Children’s Fund,
1994.
[52] Soucat A, Levy-Bruhl D, De Behune X, et al. Local cost sharing in Bamako Initiative Systems in
Benin and Guinea: assuring the financing viability of primary health care. International Journal of
Health Planning and Management 1997;12(Supp. 1):S109 – 36.
[53] Lake S, Musumali C. Zambia: the role of aid management in sustaining visionary reform. Health
Policy and Planning 1999;14(3):254 – 63.
[54] Booth D, Milimo J, Bond G, Chimuka S, et al. Coping with Cost Recovery Report to Swedish
International Development Agency (SIDA), commissioned through the Development Studies Unit,
Department of Social Anthropology, Stockholm University, 1995.
[55] Lake S, Daura M, Mabandhla M, Masiye F, Mulenga S, Antezana I, Mwikisa C, Bennett S.
Analyzing the Process of Health Financing Reform in South Africa and Zambia: Zambia Country
Report. Major Applied Research 1 Technical Paper No. 2, Bethesda, MD: Partnerships for Health
Reform Project, Abt. Associates Inc., 2000.
[56] Soucat A, Gandaho T, Levy-Bruhl D, et al. Health seeking behaviour and household heath
expenditures in Benin and Guinea: the equity implications of the Bamako Initiative. International
Journal of Health Planning and Management 1997;12(Supp. 1):S137 – 64.
[57] Leighton C. Strategies for achieving health financing reform in Africa. World Development
1996;24:1511– 25.
[58] Crosby BL. Policy implementation: the organisational challenge. World Development
1996;24:1403– 15.
[59] Walt G. Implementing health care reforms: a framework for discussion. Paper prepared for the
Workshop on Analysis of Health Systems Reforms, 10 – 11 November, Copenhagen, World Health
Organisation, 1995.
[60] Kohlemainen-Aitken R-L, Newbrander W. Decentralizing the Management of Health and Family
Planning Programmes. Boston, MA: Management Sciences for Health, 1996.
[61] Sauerborn R, Adams A, Hien M. Household strategies to cope with the economic costs of illness.
Social Science and Medicine 1996;43(3):291 – 301.
[62] Gilson L, Russell S, Rauyajin O, et al. Exempting the Poor: a Review and Evaluation of the Low
Income Card Scheme in Thailand. Public Health and Policy Departmental publication no. 30.
London: London School of Hygiene and Tropical Medicine, 1998.
[63] de Kadt E, Tasca R. Promoting Equity: a New Approach from the Health Sector. Washington,
DC: PAHO, 1993.
[64] Collins C, Green A. Decentralisation and primary health care: some negative implications for
developing countries. International Journal of Health Services 1994;24:459 – 75.
398
[65] Herbst J. The Politics of Reform in Ghana, – 1991. Berkeley, CA: University of California Press,
1982:1993.
[66] Carrin G. Strategies for Health Care Finance in Developing Countries: with a Focus on Commu-
nity Financing in sub-Saharan Africa. London: Macmillan, 1992.
399
-%., -./, 0"', 6>AL bK9 (>L K9 HHL ^bî^h^9 J^^_
! J^^_ ,AQFUEFG -?EF;?F %I<L "AA GED@IQ GFQFGUF<
9,,: 7;<==>?@ABC?DE;;FBD>A 'GE;IF< E; )GF:I WGEI:E;
gdKK3^hjfX^_ kJ^Lgg l gLgg
!"#$%& '%"(($() '*%$+$,- "(. '/*)/"##,- $(

,$)01 %*23$(+*#, +*4(1/$,-5 " +*#'"/"1$6,
'*%$+& "("%&-$-
7,%%,& %,,89 %*4$-$"(" %4-09 )$%% 2"%1 :;< =*0( +%,%"(.
%>;<>; -?@>>A >B 0CDEF;F :;< 1G>HE?:A #F<E?E;F9 0F:AI@ '>AE?C 4;EI9 7FHHFA -IGFFI9
%>;<>; 2+J, K019 4L7L
!"#$%&'$M1@F FNIF;I I> O@E?@ B:PEAC HA:;;E;D HG>DG:PPFQ :GF QR??FQQBRA :I GF<R?E;D BFGIEAEIC GFP:E;Q
: P:S>G <FT:IF :P>;D H>HRA:IE>; Q?@>A:GQL " ?>PH:G:IEUF H>AE?C :;:ACQEQ >B B>RG H:EGQ >B A>O3E;?>PF
?>R;IGEFQ VW:;DA:<FQ@X':YEQI:;9 1@:EA:;<X'@EAEHHE;FQ9 1R;EQE:X"ADFGE: :;< ZEPT:TOFXZ:PTE:[ O:Q ?:G3
GEF< >RI I> R;<FGQI:;< O@C Q>PF ?>R;IGEFQ <FUFA>H :HHG>HGE:IF :;< F\F?IEUF HG>DG:PPFQ9 O@EAF >I@FG
?>R;IGEFQ <> ;>IL 1@F QIR<C B>R;< I@:I I@F B>GP:IE>; >B ?>:AEIE>;Q :P>;D H>AE?C FAEIFQ9 QHGF:< >B H>AE?C
GEQY9 :;< E;QIEIRIE>;:A :;< ];:;?E:A QI:TEAEIC OFGF B:?I>GQ O@E?@ QRHH>GIF< >G E;@ETEIF< I@F :<>HIE>; >B
QIG>;D H>HRA:IE>; H>AE?EFQ :;< B:PEAC HA:;;E;D HG>DG:PPFQL ! J^^_ ,AQFUEFG -?EF;?F %I<L "AA GED@IQ
GFQFGUF<
!"# $%&'(MB:PEAC HA:;;E;D9 H>HRA:IE>; H>AE?C9 @F:AI@ H>AE?C
*;F >B I@F DGF:I <FT:IFQ :P>;D H>HRA:IE>; Q?@>A:GQ H>HRA:IE>; :;< @F:AI@ H>AE?C AEIFG:IRGFL 1> I@F
>UFG I@F H:QI I@EGIC CF:GQ @:Q TFF; I@F FNIF;I I> FNIF;I I@:I D>UFG;PF;I H>AE?EFQ ?>;IGETRIF I> <FP>3
O@E?@ QI:IF3QH>;Q>GF< B:PEAC HA:;;E;D V!'[ H>AE?EFQ DG:H@E? IGF;<Q9 I@FGF EQ : ;FF R;<FGQI:;< TFIIFG
:;< HG>DG:PPFQ :GF QR??FQQBRA E; GF<R?E;D BFGIEAEIC I@F B:?I>GQ I@:I QRHH>GI >G E;@ETEI I@F :<>HIE>; >B
:;<9 @F;?F9 I@F G:IF >B H>HRA:IE>; DG>OI@L $I EQ GFA:3 !' H>AE?EFQ :;< HG>DG:PPFQL 0>OFUFG9 AEPEIF<
IEUFAC F:QC I> <FP>;QIG:IF I@:I ?>;IG:?FHIEUF RQF EQ ?>PH:G:IEUF GFQF:G?@ @:Q QIR<EF< E; <FI:EA I@F
@ED@FG9 :;< BFGIEAEIC A>OFG9 E; ?>R;IGEFQ O@FGF D>U3 :?I>GQ :;< HG>?FQQFQ E;U>AUF< E; H>AE?C P:YE;D >;
FG;PF;IQ @:UF FQI:TAEQ@F< :?IEUF !' HG>DG:PPFQ9 !'9 :;< @>O I@EQ @:Q I:YF; HA:?F >UFG IEPF E;
I@:; E; ?>R;IGEFQ O@FGF I@EQ EQ ;>I I@F ?:QFL QHF?E]? ;:IE>;:A :;< E;IFG;:IE>;:A ?>;IFNIQ V!E;YAF9
0>OFUFG9 EI EQ P>GF <E`?RAI I> Q@>O : ?:RQ:A AE;Y J^Kde 2:GOE?Y9 J^_d[L +>;IFNIR:A B:?I>GQ9 E; H:G3
TFIOFF; I@F FQI:TAEQ@PF;I >B !' QFGUE?FQ :;< :;C IE?RA:G9 @:UF ?@:;DF< F;>GP>RQAC >UFG I@F H:QI
GF<R?IE>; E; BFGIEAEICL 1@FGF ?>;IE;RFQ I> TF AEIIAF I@EGIC CF:GQL .RGE;D I@F A:IF J^fgQ9 Q>PF ?>R;IGEFQ
?>;QF;QRQ :P>;D <FP>DG:H@FGQ9 OEI@ ?>;UE;?E;D OFGF TFDE;;E;D I> HG>P>IF !' :?IEUFAC9 TRI P:;C
Q?@>A:GQ@EH QRHH>GIE;D >HH>QE;D ?>;?ARQE>;QL *; D>UFG;PF;IQ E; I@F <FUFA>HE;D O>GA< GFP:E;F<
I@F >;F @:;<9 0FG;:;<Fa VJ^_b[ :;< 'GEI?@FII U>?:AAC >HH>QF O@:I O:Q HFG?FEUF< :Q : OFQ3
VJ^^b[ :GDRF I@:I !' HG>DG:PPFQ QR??FF< E; IFG;3<GEUF; H>AE?C :DF;<:L WC I@F PE< J^_gQ9 P>QI
?>R;IGEFQ O@FGF HF>HAF :AGF:<C OEQ@ I> AEPEI B:PEAC ?>R;IGEFQ @:< :<>HIF< !' HG>DG:PPFQ :Q H:GI >B
;:IE>;:A H>HRA:IE>; H>AE?EFQ :EPF< :I AEPEIE;D H>HR3
QEaF :;< OEAA I@RQ <FP:;< ?>;IG:?FHIEUF QFGUE?FQL
A:IE>; DG>OI@ :;< HG>P>IE;D F?>;>PE? <FUFA>H3
)>UFG;PF;IQ FQI:TAEQ@ QFGUE?FQ9 E; >I@FG O>G<Q9 E;
PF;IL $; I@F PE A:IF J^^gQ9 :;AA>OE;D I@F
GFQH>;QF I> HGFFNEQIE;D <FP:;<L *; I@F >I@FG @:;<9
$;IFG;:IE>;:A +>;BFGF;?F >; '>HRA:IE>; :;<
#:RA<E; :;< />QQ VJ^^J[ :GDRF I@:I D>UFG;PF;I
.FUFA>HPF;I V$+'.[ @FA< E; -FHIFPTFG J^^b9 I@FGF
!' QFGUE?FQ P:C :?IR:AAC QIEPRA:IF <FP:;< O@E?@
EQ TG>:< ?>;QF;QRQ TF@E;< I@F ?>;?FHI >B GFHG><R?3
AF:<Q I> QRTQFcRF;I GF<R?IE>;Q E; B:PEAC QEaFL 1@EQ
IEUF @F:AI@L $; I@EQ ?>;IFNI9 !' EQ QFF; :Q >;F >B :
EQ : <FT:IF O@E?@ P:C ;FUFG TF ?>;?AR<F< Q:IEQB:?3
G:;DF >B HGEP:GC @F:AI@ QFGUE?FQ :EPF< :I PFFIE;D
I>GEAC :P>;D <FP>DG:H@FGQ DEUF; I@F ?>PHAFNEIC >B
I@F GFHG><R?IEUF ;FF<Q >B E;<EUE<R:AQL
FQI:TAEQ@E;D ?:RQ:A AE;YQL !RGI@FGP>GF9 GF?F;I <FUFA3
>HPF;IQ E; E;IFG;:IE>;:A @F:AI@ H>AE?C @:UF GF<R?F<
I@F AFUFA >B :IIF;IE>; H:E B:PEAC QEaF :Q I@F HGE;3 !(!)*+,-!) ./!0123/4 !(5 01+6357 3. +61
?EH:A >RI?>PF >B GFHG><R?IEUF @F:AI@ HG>DG:PPFQL 7+85*
4QE;D : <E\FGF;I :;:ACIE?:A AF;Q9 H>AE?C :;:ACQEQ9
1@EQ H:HFG EQ <G:O; BG>P : QIR<C V%FF ") *+,9
I@EQ H:HFG QFFYQ I> :<<GFQQ :; EPH>GI:;I D:H E; I@F
J^^h[ >B B>RG H:EGQ >B A>O3E;?>PF ?>R;IGEFQ O@E?@
OFGF QFAF?IF PE;EPEQF <E\FGF;?FQ E; F?>;>PE?9
8"RI@>G B>G ?>GGFQH>;<F;?FL Q>?E:A :;< ?RAIRG:A B:?I>GQ OEI@E; F:?@ H:EG9 TRI
^b^
400
FTV P4 f&& !" #$%
5"A;& E4 Q&;&+,&' +/6*,)$&% "*' 3)/.)"!!& &(/), %+/)&%" EFFVI "*' $*0"*, !/),";$,7 D")/6*' EVV '&",-% 3&)
2)/.)"!!& &(/), %+/)& DU /0 !"#$!6!I EVVV A$),-% $* EFFVI4 5-& !/%, /AL$/6% '$(&)&*+&%
EFGV EFHZ EFHF A&,8&&* ,-& ,8/ +/6*,)$&% ")& ,-& -$.-&) WO2 3&)
+"3$," $* 2"=$%,"* DXQYEHSZ $* EFFVI ,-"*
:"*.;"'&%- EV TG GZ
2"=$%,"* ZG \V \H
:"*.;"'&%- DXQYHGZI> &L&* "0,&) "'[6%,$*. 0/) 36)N
@$!A"A8& O? ZG TS +-"%$*. 3/8&) 3")$,74 O/,"A;7> ,-& ,8/ +/6*,)$&%
@"!A$" V ES \F -"L& 36)%6&' L&)7 '$(&)&*, 3/36;",$/* 3/;$+$&%
5-"$;"*' ^G SE HV
2-$;$33$*&% T^ TS \F %$*+& ,-& &");7 EFGV%> "*' -"L& &#3&)$&*+&' %$.*$RN
56*$%$" \V TF SF +"*,;7 '$L&).&*, 0&),$;$,7 /6,+/!&%4 5-& &");$&)
?;.&)$" EV ZT \S '&+;$*& $* 0&),$;$,7 $* :"*.;"'&%- +/6;' */, -"L&
Q/6)+&9 C"6;'$* "*' :&)&;%/* DEFGHI "*' J/%% !" #$% DEFFZI4 A&&* 3)&'$+,&' 0)/! ,-& "A/L& %/+$/&+/*/!$+ 0"+N
,/)%> 8-$+- 8/6;' %6..&%, " 0&),$;$,7 '$(&)&*,$"; $*
,-& /33/%$,& '$)&+,$/*4
!"#$!$%& '$(&)&*+&% $* ,-& %,)&*.,-% /0 ,-&$) 12
X*,$; EFS\ @$!A"A8& "*' @"!A$" 8&)& ";%/ /*&
3)/.)"!!&%4 5-& +"%& %,6'7 +/6*,)$&% 8&)&9
+/6*,)7> Q/6,-&)* "*' O/),-&)* J-/'&%$" )&%3&+,N
:"*.;"'&%-<2"=$%,"*> 5-"$;"*'<2-$;$33$*&%> 56*$%$"<
$L&;74 :/,- ")& 3)&'/!$*"*,;7 ]-)$%,$"* +/6*,)$&%>
?;.&)$" "*' @$!A"A8&<@"!A$"4
8$,- ;$,,;& '$(&)&*+& $* 0&),$;$,7 "*' $*0"*, !/),";$,7
5-& 0/6) 3"$)% /0 +"%& %,6'7 +/6*,)$&% 8&)&
)",&% ,-)/6.-/6, ,-& EFSV% "*' EFGV%4 5-&)& -"%
+-/%&* 0/) ,-&$) )/6.-;7 %$!$;") %/+$/&+/*/!$+
";%/ A&&* " %$!$;") 3",,&)* /0 '&+;$*& $* "'6;, $;;$,&)N
+-")"+,&)$%,$+%> 7&, '$(&)&*+& $* ,-& %,)&*.,- /0
"+7 D0)/! "A/6, SVU $* EFS^ ,/ ^Û $* EFFVI>
,-&$) 3/36;",$/* 3/;$+$&% "*' 12 3)/.)"!!&%> "%
";,-/6.- WO2 3&) +"3$," $*+)&"%&' $* @$!A"A8&
!&"%6)&' A7 ,-& 3)/.)"!!& &(/), %+/)&%B /0
'6)$*. ,-& EFHV% ", " 0"%,&) )",&4 _&%3$,& ,-&$)
C"6;'$* "*' :&)&;%/* DEFGHI "*' C"6;'$* "*'
)/6.- %$!$;")$,$&%> 0&),$;$,7 )",&% $* @$!A"A8& 0&;;
J/%% DEFFEI4 K'&";;7> +/6*,)$&% 8/6;' -"L& A&&*
!/)& )"3$';7 ,-"* $* @"!A$" DZÛ L% \U %$*+& ,-&
3/;$,$+";;7 "*' %/+$/&+/*/!$+";;7 $'&*,$+"; 8$,-$*
EFSV%I> "++/!3"*$&' A7 " %,)/*.&) 3)/.)"!!&
&"+- 3"$)> &*"A;$*. "*7 '$L&).&*+& $* )&3)/'6+,$L&
&(/), %+/)& 0/) $,% 12 3)/.)"!!&%4
/6,+/!&% ,/ A& ",,)$A6,"A;& ,/ ,-&$) 3/;$+$&% "*'
56*$%$" "*' ?;.&)$" ")& *&$.-A/6)$*. +/6*,)$&%
3)/.)"!!&%4 K* )&";$,7> /0 +/6)%&> */ ,8/ +/6*,)$&%
+/!3/%&' 3)&'/!$*"*,;7 /0 ?)"A "*' C6%;$!
")& 8-/;;7 ";$=&4 K* "''$,$/*> %$!$;") +/6*,)$&% ,&*'
3/36;",$/*%4 5-& ,8/ +/6*,)$&% %-")& -$%,/)$+";
,/ &L/;L& %$!$;") 3/;$+$&% DC"6;'$* "*' :&)&;%/*>
;$*=% 8$,- 1)"*+& ";,-/6.- 56*$%$"> ;$=& @"!A$">
EFGHM C"6;'$* "*' J/%%> EFFE4 5-& %&;&+,$/* 3)/N
-"' " !/)& 3&"+&06; ,)"*%$,$/* ,/ $*'&3&*'&*+&4
+&%% 8"% ";%/ ;$!$,&' A7 "L"$;"A;& "++&%% ,/ 3/;$+7
WO2 -"% A&&* )/6.-;7 %$!$;") DXQY^VVV`^TVV $*
"*' '&!/.)"3-$+ '"," 8-$+- 06),-&) )&%,)$+,&' ,-&
EFFVI "0,&) "'[6%,$*. 0/) 36)+-"%$*. 3/8&) 3")$,74
+-/$+& /0 +/6*,)$&% D&4.4 O/),- P/)&" "*' Q/6,-
K*0"*, !/),";$,7 -"% ";8"7% A&&* %;$.-,;7 ;/8&) $*
P/)&"I4 1$*";;7> "*7 -$%,/)$+"; +/!3")$%/* /0 +"%&
56*$%$" D'&+;$*$*. 0)/! ETT ,/ ^H '&",-% 3&) EVVV
%,6'$&% *&&'&' ,/ )&+/.*$%& ,-", 3/;$+$&% "*' 3)/N
;$L& A$),-% A&,8&&* EFSV`FVI ,-"* ?;.&)$" D'&+;$*$*.
.)"!!&% +-"*.& /L&) ,$!&> "*' ,-", '$(&)&*+&% "*'
0)/! ESV ,/ S\I> "% -"% ,-& 3)/3/),$/* /0 ,-& "'6;,
%$!$;")$,$&% 8&)& )&;",$L& ,/ &"+- 3"$) )",-&) ,-"*
3/36;",$/* ,-", $% $;;$,&)",& D^TU +/!3")&' 8$,-
"A%/;6,& "+)/%% ";; +/6*,)$&%4
\Û $* EFFVI4 a/8&L&)> ,-&%& '$(&)&*+&% "33&") 0")
5-& %&;&+,&' +/6*,)$&% ")& %-/8* $* 5"A;& E> 8$,-
,// !/'&%, ,/ "$* ,-& '$L&).&*+& $* 3)/.)"!!&
,-& R)%, +/6*,)7 /0 &"+- 3"$) -"L$*. %,)/*.&) 3/36N
&(/), %+/)&% "*' 0&),$;$,7 )",&% /L&) ,-& 3"%, ,-$),7
;",$/* 3/;$+$&% "*' 12 3)/.)"!!&% ,-"* ,-& %&+/*'4
7&")%4
:"*.;"'&%- "*' 2"=$%,"* 8&)& 6*$,&' "% /*&
5-"$;"*' "*' ,-& 2-$;$33$*&% ")& 3&)-"3% ;&%%
+/6*,)7 6*,$; EFGE> ")& A/,- 3)&'/!$*"*,;7 K%;"!$+>
8&;;N!",+-&' "% +/!3")",$L& +"%& %,6'$&%> A/,- 3/;N
"*' -"L& %$!$;") ;&L&;% /0 "'6;, $;;$,&)"+7 DSTU $*
$,$+";;7 "*' +6;,6)";;7> '&%3$,& %-")$*. " )&.$/*";
3)/#$!$,74 Q&;&+,&' $*'$+",/)% /0 %/+$/&+/*/!$+
B? 3)/.)"!!& &(/), %+/)& /) )",$*. /0 ,-& %,)&*.,- /0 '&L&;/3!&*,> -/8&L&)> %6..&%, ,-", ,-& 2-$;$33$*&%
0"!$;7 3;"**$*. 3)/.)"!!& &(/), $% '&)$L&' 0)/! %+/)N 8"% "+,6";;7 !/)& 0"L/6)"A;7 3;"+&' '6)$*. ,-&
$*. ";/*. ,-$),7 $,&!% .)/63&' $*,/ 0/6) +/!3/*&*,%9
3/;$+$&% "*' %,".&N%&,,$*. "+,$L$,$&%> %&)L$+& "*' %&)L$+&N
EFSV% "*' EFGV%> " 3&)$/' 8-&* 0&),$;$,7 A&."* ,/
)&;",&' "+,$L$,$&%> )&+/)' =&&3$*. "*' &L";6",$/*> "*' '&+;$*& !/)& %-")3;7 $* 5-"$;"*'4 K* EFSV ,-&
"L"$;"A$;$,7 /0 +/*,)"+&3,$L& !&,-/'%4 5-$% $*0/)!",$/* 2-$;$33$*&% -"' " -$.-&) WO2 3&) +"3$,"> "*' ;/8&)
8"% /A,"$*&' ,-)/6.- '&,"$;&' c6&%,$/**"$)&% %&*, ,/ "'6;, $;;$,&)"+7 "*' $*0"*, !/),";$,7 )",&%4 K, $% /*;7
./L&)*!&*, /d+$";%> '/*/) ".&*+7 3&)%/**&;> =*/8;N %$*+& ,-& &");7 EFHV% ,-", 5-"$;"*'b% WO2 3&)
&'.&"A;& +$,$e&*% "*' 0/)&$.*&)%4 5-&)& ")& ,8/ +"L&",%
0/) ,-$% )"*=$*.4 1$)%,> ,-& "%%&%%!&*,% 8&)& !"'& A7 +"3$," -"% .)/8* ,/ &#+&&' $,% *&$.-A/6)> 8-$;&
$*'$L$'6";% 8-/ 8&)& 0"!$;$") 8$,- 0&),$;$,7 ,)&*'% "*' "'6;, ;$,&)"+7 $* A/,- +/6*,)$&% $% */8 ";!/%, 6*$N
;&L&;% /0 +/*,)"+&3,$L& 6%&4 5-$% +/6;' -"L& ;&' ,/ 3)&N L&)%";4
[6'.$*. 3)/.)"!!&% 0"L/6)"A;7 /) 6*0"L/6)"A;& K* %6!!")7> ,-& %&;&+,&' 3"$)% /0 +/6*,)$&% %-/8
'&3&*'$*. /* 8-&,-&) 0&),$;$,7 $% '&+;$*$*.4 Q&+/*'> ,-&
,8/ ;",&) !&"%6)&% /0 %,)&*.,- 6%&' " !/)& +/!3)&-&*N
" '$(&)&*+& $* ,$!$*. "*' )",& /0 '&+;$*& $* 0&),$;$,7
%$L& %&, /0 $*36,% 8-$+- !"7 %6..&%, " ;"+= /0 +/!3")N 6*&#3&+,&' 0)/! +/!3")$*. ,-&$) %/+$/&+/*/!$+
"A$;$,7 8$,- ,-& R)%, !&"%6)& ,"=&* $* ,-& &");7 EFGV%4 0&",6)&%4 K* /,-&) 8/)'%> ,-&)& $% ;$,,;& &L$'&*+&
4 01
>&$(20 52&++(+@ (+ 2#1 (+)#$* )#/+-"(*' H[G
!"#$ %&'() '#)(#*)#+#$() (+,()&-#"' -.&- #+* :%; 5#2()0 )#&2(-(#+' .&4* %**+ $#"* 2(3*20 -#
)#/+-"0 1#/2, %* $#"* 2(3*20 -# .&4* &+ *&"2(*" #" 5"*4&(2 #4*" -($* 1.*"* 5#2()0 !#"$/2&-(#+ .&'
$#"* "&5(, !*"-(2(-0 ,*)2(+* -.&+ -.* #-.*"6 7.(' 2*, (+4#24*, & '.&"(+@ #! -.* 5#2()0 "('3' #!-*+ &''#)(C
-.* '-/,0 -# &,,"*'' -1# $&(+ 8/*'-(#+'9 &-*, 1(-. >? 5"#@"&$$*'N &+,
:&; <.0 ,(, '#$* )#/+-"(*' ,*4*2#5 &55"#5"(&-* :); (+'-(-/-(#+&2 &+, K+&+)(&2 '-&%(2(-0 .&' %**+
&+, *=*)-(4* >? 5"#@"&$$*' 1.(2* #-.*"' ,(, +#-A ($5#"-&+- -# &).(*4(+@ $#"* *=*)-(4* ($52*$*+C
:%; B#1 )&+ -.(' (+'(@.- )#+-"(%/-* -# /+,*"C -&-(#+ #! >? 5"#@"&$$*'E &' *4(,*+)*, %0 (+)"*&'*,
'-&+,(+@ -.* "*2&-(#+'.(5 %*-1**+ >? 5"#@"&$$*' "&-*' #! )#+-"&)*5-(4* /'*6
&+, !*"-(2(-0 ,*)2(+*A Z ,(')/''(#+ #! *&). #! -.*'*E (+ "*2&-(#+ -# *&).
D+ &+'1*"(+@ -.*'* 8/*'-(#+'E -.* '-/,0 /'*, & 5&(" #! )#/+-"(*'E (' 5"#4(,*, %*2#16 7.* K+,(+@'
5#2()0 &+&20'(' !"&$*1#"36 ?#2()0 &+&20'(' (' & '/@@*'- -.&- -.*"* (' & +**, !#" 5#2()0 $&3*"' -#
%"#&, -*"$ 1.(). *+)#$5&''*' $&+0 -.*#"*-()&2 @(4* )#+'(,*"&-(#+E +#- #+20 -# -.* )#+-*+- #! 5#5/C
&+, $*-.#,#2#@()&2 &55"#&).*'6 D+ -.(' '-/,0E 5#2C 2&-(#+ 5#2()(*' &+, >? 5"#@"&$$*'E %/- -# .#1 5#2C
()0 1&' &+&20'*, !"#$ -.* 5*"'5*)-(4* #! &)-#"' &+, ()0 &)-#"'E 5"#)*''*' &+, )#+-*P-/&2 !&)-#"' )&+
5"#)*''*' :<&2- &+, F(2'#+E GHHI;E ,"&1(+@ '-"#+@20 (+O/*+)* -.* '-"*+@-. &+, *=*)-(4*+*'' #! '/).
#+ 5#2(-()&2 -.*#"(*' &+, )#+)*5-' -# /+,*"'-&+, 5#2()(*'6
.#1 5#2()0C$&3(+@ #+ 5#5/2&-(#+ &+, >? 1&' )&"C
"(*, #/- (+ *&). 5&(" #! )#/+-"(*' #4*" & 5*"(#, #!
-."** ,*)&,*'6 J*@(++(+@ 1(-. & '($52* $#,*2 #! -.* !"# $%$!$&!$'% '( )')*+&!$'% )'+$,$#- &%.
(&/$+0 )+&%%$%1 )2'12&//#-3 !"# 2'+# '(
5#2()0 5"#)*'' &' )#+'('-(+@ #! !#/" 5.&'*' :(6*6 5"#C ,'&+$!$'%- '( )'+$,0 #+$!#-
%2*$ (,*+-(K)&-(#+E 5#2()0 !#"$/2&-(#+E ($52*$*+C
-&-(#+ &+, *4&2/&-(#+; :L(+@,#+E GHMIN B&$ &+, 7.* )&'* '-/,(*' %*@&+ %0 &+&20'(+@ .#1
B(22E GHMI;E -.* '-/,0 !#)/'*, #+ -.* 3*0 &)-#"' (+ 5#5/2&-(#+ (''/*' (+ @*+*"&2E &+, >? (+ 5&"-()/2&"E
*&). )#/+-"0 &+, .#1 -.*0 (+O/*+)*, -.*'* ,(=*"C )&$* -# %* (,*+-(K*, &' & 2*@(-($&-* )#+)*"+ #+
*+- 5.&'*'6 7.* $#'- 5"#$(+*+- &)-#"' 1*"* -.* 5#2()0 &@*+,&E &+, .#1 -.(' 1&' -.*+ -&3*+
*P5*)-*, -# %* QQ5#2()0 *2(-*'RR ,*K+*, %0 F"(+,2* !#"1&", (+ 5#2()0 $&3(+@6 >#" &22 5#2()0 $&3*"'E
&+, 7.#$&' :GHHG; &' (+,(4(,/&2' #" @"#/5' QQ!#"C -.* /+(4*"'* #! 5#-*+-(&2 (''/*' -# %* &,,"*''*,
$&220 ).&"@*, 1(-. $&3(+@ &/-.#"(-&-(4* ,*)('(#+' (' 4&'-6 <.&- 2*&,' -# &--*+-(#+ %*(+@ @(4*+ -#
(+ @#4*"+$*+-RR6 D+ &,,(-(#+E -.* '-/,0 1&' )#+C #+* (''/* #4*" &+#-.*" $&0 %* (+O/*+)*, %0 -.*
)*"+*, 1(-. -.* .('-#"()&2 &+, )#+-*$5#"&"0 )#+-*P- *$*"@*+)* #! +*1 (+!#"$&-(#+E '.(!-' (+ 5/%2()
1(-.(+ 1.(). 5#2()0C$&3(+@ 1&' )&""(*, #/-6 7.* #5(+(#+E 2#%%0(+@ %0 5"*''/"* @"#/5' #" -.* )#+C
'-/,0 /'*, S*().-*"R' -05#2#@0 -# (,*+-(!0 &+, 4()-(#+' #! 5#2()0 $&3*"' -.*$'*24*'6 7.* #!-*+
*P52#"* !#/" -05*' #! )#+-*P-/&2 !&)-#" T '(-/&C -"&+'(-#"0 +&-/"* #! 5#2()0 $&3(+@ )&+ &2'# 2*&,
-(#+&2E '-"/)-/"&2E )/2-/"&2 &+, *P#@*+#/' :S*().-*"E 5#2()0 $&3*"' -# $#4* -.*(" &--*+-(#+ !"#$ #+*
GHUH;6 (''/* -# &+#-.*"6
V'(+@ -.(' &+&20-()&2 !"&$*1#"3E +&-(#+&2 7.* '-/,0 '#/@.- -# (,*+-(!0 1.&- &+, .#1
"*'*&").*"' )&""(*, #/- *(@.- ,*-&(2*, )&'* '-/,(*' #! &)-#"' 1*"* (+4#24*, (+ '/55#"-(+@ #" #55#'(+@ -.*
-.* .('-#"()&2 *4#2/-(#+ #! 5#5/2&-(#+ 5#2()(*' &+, (+(-(&-(#+ #! 5#5/2&-(#+ 5#2()(*' &+, >? 5"#C
>? 5"#@"&$$*' (+ *&). )#/+-"06 7.* $*-.#,' /'*, @"&$$*' (+ *&). )#/+-"06 Z))#",(+@ -# F"(+,2*
%0 -.* "*'*&").*"' 1*"* (+-*"4(*1' 1(-. 3*0 (+!#"C &+, 7.#$&' :GHHG;E 5#2()0 *2(-*' QQ.&4* )#+'(,*"&%2*
$&+-' T +#-&%20 5&'- &+, 5"*'*+- 5#2()0 $&3*"'E ')#5* -# (,*+-(!0 5"#%2*$'E &"-()/2&-* @#&2'E ,*K+*
#W)(&2' #! *P-*"+&2 ,#+#" &@*+)(*' &+, "*5"*'*+-&C '#2/-(#+'E &+, -.(+3 '-"&-*@()&220 &%#/- -.*(" ($C
-(4*' #! XFY' T &+, "*4(*1' #! 5"($&"0 &+, '*)C 52*$*+-&-(#+RR6 7.* '-/,0 *P52#"*, -.* *P-*+- -#
#+,&"0 $&-*"(&2' #+ 5#5/2&-(#+ 5#2()(*' &+, >? 1.().E (+ )#/+-"(*' 1.*"* %"#&,20C%&'*, #" )#.*'(4*
5"#@"&$$*'6 7.* )&'* '-/,0 "*5#"-' 1*"* -.*+ /'*, )#&2(-(#+' #! 5#2()0 *2(-*' !#"$*, (+ '/55#"- #! >?E
-# ,"&1 #/- )#$5&"('#+' 1(-.(+ *&). 5&(" #! (- 1&' $#"* 2(3*20 -.&- -.* (''/* &55*&"*, *&"2(*" #+
)#/+-"(*' )#+)*"+(+@ -.* 5#2()0 &)-#"'E 5"#)*''*' -.* 5#2()0 &@*+,& &+, 1&' -&3*+ !#"1&", $#"*
&+, )#+-*P-' ).&"&)-*"('(+@ -.*$E &+, -# ,"&1 2*'C *=*)-(4*206 D+ )#+-"&'-E 1.*"* )#&2(-(#+' ,(, +#-
'#+' "*@&",(+@ .#1 -.*'* !&)-#"' .&4* )#+'-"&(+*, !#"$E #" 1.*"* )#&2(-(#+' #55#'(+@ >? 1*"* )#$C
#" *+&%2*, 5#5/2&-(#+ 5#2()(*' &+, >? 5"#@"&$$*' 5&"&-(4*20 '-"#+@E -.(' 1&' &))#$5&+(*, %0 1*&3*"
(+ -.*(" )#/+-"(*'6 Z+ &+&20'(' #! -.* 2(+3 %*-1**+ 5#2()(*' &+, 5"#@"&$$*'6
5#2()0C$&3(+@ &- -.* +&-(#+&2 &+, @2#%&2 2*4*2' .&'
%**+ 5"*'*+-*, *2'*1.*"* :S** &+, <&2-E GHH[;6 !"#$%#&' #&' (") *"$%$++$&),
7.* '-/,0 !#/+, -.&- -.*"* &"* -."** !*&-/"*' #! J*@(++(+@ 1(-. 7.&(2&+, &+, -.* ?.(2(55(+*'E (+
5#2()0C$&3(+@ 1.(). $&0 *P52&(+ -.* ,(=*"*+)* (+ %#-. )#/+-"(*' -.* (+(-(&2 ($5*-/' %*.(+, >? )&$*
5"#@"&$$* '-"*+@-. (+ *&). 5&(" #! )#/+-"(*'9 !"#$ *P-*"+&2 ,#+#" &@*+)(*'6 D+ 7.&(2&+,E !#22#1C
:&; -.* !#"$&-(#+ #! 5#2()0 )#&2(-(#+'E '/55#"-*, (+@ & GH[M <#"2, J&+3 "*5#"- #+ -.* &,4*"'*
%0 5#2()0 *2(-*'E .&' %**+ &+ ($5#"-&+- )#+-"(%/-#" ($5&)- #! "&5(, 5#5/2&-(#+ @"#1-. #+ -.* )#/+-"0R'
-# -.* (+(-(&-(#+ #! 5#5/2&-(#+ 5#2()(*' &+, >? 5"#C *)#+#$() ,*4*2#5$*+-E .(@.C2*4*2 )#+'/2-&-(#+'
@"&$$*'N 1*"* .*2, &$#+@ 5#2()0 *2(-*' 1.(). (+4#24*,
4 02
Teb *< Z## '1 %.2
!"# "#$% &' (!$!# )*+,- .$/$ 0123 !"# 4"+#' !"#$%$&' %() !%#$"%
#5#46!+7# )89+/# /+,+(!#923 -&7#9,/#,! /+,+(!#9(3 ="# (!9#,-!" &' $ 8&:+4; 4&$:+!+&, &' #:+!#( $:(&
4+7+: (#97$,!(3 !"# $4$%#/+4 4&//6,+!;3 "#$:!" %+(!+,-6+("#( ^+/B$B@# '9&/ ^$/B+$< 0,
89&'#((+&,$:( $,% !"# /$(( /#%+$< ="+( >>"+-":; ^+/B$B@#3 #$9:; 8&:+4;D/$]+,- &, NC 4#,!9#% &,
4&::#4!+7# 89&4#((?? @$( !"#, +,(!+!6!+&,$:+(#% +,!& !"# ^+/B$B@#$, G$!+&,$: N$/+:; C:$,,+,- L&6,4+:
$ (!9&,- 4&$:+!+&, @+!" !"# 49#$!+&, &' !"# )^GNCL23 $ B&%; 49#$!#% +, !"# #$9:; ST_X( $,%
A6B4&//+!!## &, C&86:$!+&, C&:+4; $,% C:$,D B$4]#% B; "+-"D:#7#: :#$%#9("+8 '9&/ A$::; M6-$B#3
,+,- @"+4" 4&,!+,6#% !& "&:% 9#-6:$9 /##!+,-( !"# 89#(+%#,!?( @+'#3 $,% K(!#9 `&&"#,#3 "+( (+(!#9D
!& %+(46(( !"# 4&6,!9;?( 8&86:$!+&, 8&:+4;< 0!( +,D:$@< Q :$!#9 %+9#4!&9 %#(49+B#% !"# &9-$,+($!+&,?(
/#/B#9("+8 +,4:6%#% 9#89#(#,!$!+7#( &' !"# /+,+(D 9&:# $( >>:#$%+,- '9&/ B#"+,%??3 #,4&69$-+,- 4&,(6:D
!9+#( &' 86B:+4 "#$:!"3 E,$,4# $,% #%64$!+&,F !"# !$!+&, @+!" >>&8+,+&, :#$%#9(?? +,4:6%+,- 9#:+-+&6(
G$!+&,$: A!$!+(!+4$: HI4#F 9#(#$94" +,(!+!6!+&,( -9&68(3 GPH(3 (#,+&9 4+7+: (#97$,!(3 !"# B6(+,#((
+, (!$!# 6,+7#9(+!+#(F $,% !"# J#8$9!/#,! &' 4&//6,+!; $,% !"# /$(( /#%+$a< ="+( 7+#@ +( (68D
=#4",+4$: $,% K4&,&/+4 L&&8#9$!+&,3 HI4# &' 8&9!#% B; !"# G$!+&,$: Q((&4+$!+&, &' GPH(
!"# C9+/# M+,+(!#9< N+,$::;3 $ G$!+&,$: N$/+:; )GQGPH2 @"& "$7# 8$9!+4+8$!#% $4!+7#:; +, !"#
C:$,,+,- L&//+!!## @$( 49#$!#% 6,%#9 !"# ^GNCL!3 $,% B; !"# R&9:% `$,] @"+4" &B(#97#%
M+,+(!9; &' C6B:+4 O#$:!"3 4&/89+(#% &' !&8D:#7#: !"$! >>^+/B$B@# "$( '&::&@#% $ %#:+B#9$!# 4&,(6:!$D
B69#$649$!( )#<-< 8#9/$,#,! (#49#!$9+#(23 %+9#4!&9( !+7# $,% B9&$%DB$(#% 89&4#(( &' 8&86:$!+&, 8&:+4;
&' 9#(#$94" +,(!+!6!+&,(3 $,% "#$%( &' 9#:#7$,! %#7#:&8/#,! @"+4" +( +, ]##8+,- @+!" +!( 8$9!+4+8$D
GPH( (64" $( !"# C:$,,#% C$9#,!"&&% Q((&4+$!+&, !&9; (;(!#/ &' -&7#9,/#,! $,% +!( '&46( &, ,$!+&,$:
&' ="$+:$,% )CCQ=23 !& '&9/6:$!# (8#4+E4 8&:+4+#( 4&,(#,(6(?? )R&9:% `$,]3 ST_T2<0, ^$/B+$3
&, NC )R&,-B&&,(+,3 STTU2< R+!" !"+( B9&$% $,% C9#(+%#,! *#,,#!" *$6,%$?( #\&9!( !& B6+:% >>&,#
4&"#(+7# 4&$:+!+&, &' 8&:+4; #:+!#(3 ="$+:$,% @$( ,$!+&,?? %69+,- !"# STWX( %+% ,&! +,4:6%# "+-"D:#7#:
$B:# !& +,+!+$!# $,% (6(!$+, $ (!9&,- NC 89&-9$//# (688&9! '&9 NC< K$9:; #\&9!( !& B6+:% 86B:+4 (68D
%#(8+!# '9#V6#,! 4"$,-#( +, -&7#9,/#,! '9&/ !"# 8&9! '&9 NC 4$/# '9&/ !"# N$/+:; C:$,,+,-
STWX(< R#:'$9# Q((&4+$!+&, &' ^$/B+$ )NCRQ^23 $, GPH
="# 4&$:+!+&, &' 8&:+4; #:+!#( +, ="$+:$,% 4&,D '&9/#% +, ST[b @+!" 8$9!+$: '6,%+,- '9&/ !"#
!9$(!( @+!" !"# %+(6,+!; @"+4" 4"$9$4!#9+(#% !"# 0,!#9,$!+&,$: C:$,,#% C$9#,!"&&% N#%#9$!+&,
C"+:+88+,#( %69+,- !"# #$9:; (!$-#( &' !"# 8&:+4; 89&D )0CCN23 $,% '9&/ (/$:: L"9+(!+$,D96, "#$:!"
4#((< J69+,- !"# STWX(3 #5!#9,$: %&,&9 $-#,4+#( 4#,!9#(< ="+( @$( /#! @+!" 7&4$: &88&(+!+&, '9&/
(64" $( !"# Y<A< Q-#,4; '&9 0,!#9,$!+&,$: :$9-# 9#:+-+&6( -9&68(3 :#% B; !"# L$!"&:+4 L"694"3
J#7#:&8/#,! )YAQ0J23 0,!#9,$!+&,$: C:$,,#% $,% '9&/ !"# R&/#,?( Z#$-6#3 :#% B; !"# 89&/+D
C$9#,!"&&% N#%#9$!+&, )0CCN23 N&9% N&6,%$!+&, ,#,! E-69# &' L"+B#($ *$,]$($< 0! @&6:% ,&! B#
$,% C&86:$!+&, L&6,4+:3 $:(& $4!+7#:; 89&/&!#% NC 6,!+: !"# /+% ST_X(3 @"#, !"# YG C&86:$!+&, N6,%
89&-9$//#( )Z+/3 ST[W2< ="+( @$( +,+!+$::; (68D )YGNCQ2 (644#(('6::; @&, &7#9 ]#; +,%+7+%6$:(3
8&9!#% B; C9#(+%#,! M$94&( @"& $!!#/8!#% !& B9+,- +,4:6%+,- M9(< *$,]$($3 !"$! $ 8&:+4; 4&$:+!+&, +,
!&-#!"#9 %+\#9#,! -9&68( $! $ "+-"D:#7#: /##!+,- '$7&69 &' NC B#-$, !& '&9/< ="+( @$( (&&,
4&,7#,#% B; "+( K5#46!+7# A#49#!$9;3 .$'$#: A$:$(3 #5!#,%#% !& &!"#9 8&:+4; #:+!#(3 (64" $( !"#
!& %+(46(( 8&86:$!+&, +((6#(< ="+( :#% !& !"# 49#$!+&, M+,+(!#9( '&9 O#$:!" $,% G$!+&,$: L&//+((+&, '&9
&' !"# C&86:$!+&, L&//+((+&, )C&8L&/2 +, STWT3 $ J#7#:&8/#,! C:$,,+,-3 $( 9#c#4!#% +, 86B:+4 (!$!#D
4&,(6:!$!+7# B&%; 4&/89+(+,- 9#:#7$,! -&7#9,/#,! /#,!( (688&9!+,- !"# ,##% !& :+/+! 8&86:$!+&,
/+,+(!9+#( $,% 9#89#(#,!$!+7#( &' /$,; 9#:+-+&6( -9&@!" )O&8]+,( $,% A+$/@+d$3 ST_eF *$:6/B$3
-9&68(< O&@#7#93 A$:$( :#'! !"# C"+:+88+,#( (&&, STTU2<
$'!#9 !& B#4&/# !"# E9(! K5#46!+7# J+9#4!&9 &' !"#
YG C&86:$!+&, N6,% )YGNCQ23 :#$7+,- !"# !$(] &'
B6+:%+,- $ 8&:+4; 4&$:+!+&, B#"+,% NC @+!"&6! (6ID *+(","% %() -./'0"%
4+#,! :#$%#9("+8< C&8L&/ V6+4]:; B#4$/# $ "+-":; ="# '&9/$!+&, &' $ (688&9!+7# 4&$:+!+&, &'
8&:+!+4+(#% B&%;3 @+!" '9#V6#,! 4"$,-#( /$%# !& +!( 8&:+4; #:+!#( &44699#% +, =6,+(+$ #$9:+#93 $,% !& $
/#/B#9("+83 /$,%$!# $,% $6!"&9+!; +, $, #\&9! !& @+%#9 #5!#,!3 !"$, +, Q:-#9+$< 0, !"# 4&,!#5! &'
$44&//&%$!# %+\#9#,! +,!#9#(! -9&68( )L$9+,&3 @+%#D9$,-+,- (&4+$: 9#'&9/( 6,%#9 !"# :#$%#9("+8
STTU2< ="# 9#(6:! @$( $, +,#\#4!6$: B&%; @"+4" &' C9#(+%#,! `&69-6+B$ $,% !"# G#&DJ#(!&69
:$4]#% !"# 4&"#(+&, !& /&7# 8&:+4+#( '&9@$9% C$9!; B#-+,,+,- +, !"# STeX(3 86B:+4 (!$!#/#,!(
#\#4!+7#:;< B#-$, !& B# /$%# B; 8&:+4; #:+!#( &, !"# :+,]
B#!@##, #4&,&/+4 %#7#:&8/#,! $,% 4&,!9&: &'
8&86:$!+&, -9&@!"< A688&9! '&9 $ ,$!+&,$: 8&86D
:$!+&, 8&:+4; 4$/# '9&/ $ B9&$% 9$,-# &' 89&/+D
a0,!#97+#@ @+!" J9 ^+,$,-$3 J+9#4!&9 &' ^+/B$B@# ,#,! +,%+7+%6$:( $,% +,!#9#(! -9&68( +,4:6%+,- !"#
G$!+&,$: N$/+:; C:$,,+,- L&6,4+:3 O$9$9#3 SW
N#B96$9; STTU<
C9+/# M+,+(!#93 M+,+(!#9 &' C6B:+4 O#$:!"3 G$!+&,$:
!0,!#97+#@ @+!" M9(< J$,"$3 G$!+&,$: Q((&4+$!+&, &' Y,+&, &' =6,+(+$, R&/#, )YGN=23 =6,+(+$,
GPH( )GQGPH23 O$9$9#3 SW N#B96$9; STTU< Q((&4+$!+&, '&9 N$/+:; C:$,,+,- )Q=CN23 9#:+-+&6(
4 03
1#/*!9 3!#++*+6 *+ !.4 *+(./" (.-+,%*"& <@>
!"#$"%&' %"&"#%() *+&,*,-,*.+& #+$ ,)" /#&& /"$*#0 180 S/3.%,#+,!9' ,)"*% 3.&*,*.+ 4#& &-33.%,"$ :9
1.% "2#/3!"' +"4&3#3"% (.5"%#6" .7 18 *+(%"#&"$ ,)" &-:&,#+,*#! 3%"&"+(" .7 "2,"%+#! $.+.% #6"+(*"&
7%./ ,"+ #%,*(!"& 3"% 9"#% :",4""+ ;<=>?=@ ,. 4).&" %"3%"&"+,#,*5"&' *+ # (.-+,%9 &. $"3"+$"+, .+
#+ #5"%#6" .7 .+" #%,*(!" 3"% 4""A :9 ,)" /*$ 7.%"*6+ #*$' 4#& 3#%, .7 ,)" 3.!*(9 "!*,"0 E.6",)"%'
;<BC&D0 E)*& &-33.%, 4#& &,%"+6,)"+"$ 7-%,)"% :9 :.,) ,).&" 4). 3%.5*$"$ #+$ ,).&" 4). :"+"R,,"$
,)" (%"#,*.+ .7 ,)" !"#$ %&'()*&+ ,$ -&.(++$ $' 7%./ ,)*& *+_.4 .7 "2,"%+#! 7-+$*+6 F"060 KGL'
/)01+&'()* FGH18I *+ ;' #J!*#,"$ 4*,) ,)" %"&"#%() *+&,*,-,*.+&' H`G&I 7.%/"$ # &,%.+6 (.O
KG8L' #+$ ,)" L*6)"% 8.3-!#,*.+ M.-+(*! #+$ #!*,*.+ &-33.%,*+6 18 3%.6%#//"& FK#)/..$'
N"6*.+#! 8.3-!#,*.+ M.-+(*!& F)"#$"$ :9 ,)" (.-+O
;<<QI0
,%9P& 6.5"%+.%&I *+ ;<BQ -+$"% ,)" 8%*/" K*+*&,"%0
S+ 8#A*&,#+' ,)" ()#+6*+6 %"!#,*.+&)*3& #/.+6
E)" %"&-!, .7 ,)*& )*6)O!"5"! (.#!*,*.+ .7 &-33.%,'
3.!*(9 "!*,"& 4#& #!&. #+ */3.%,#+, $","%/*+#+, .7
:#(A"$ :9 3.!*(9 "!*,"&' 4#& ,)#, 18 %"/#*+"$
R%/!9 .+ ,)" 3.!*(9 #6"+$#0S+ T!6"%*#' ).4"5"%' &-(("&& 7.% 18 3%.6%#//"&0 a+$"% ,)" ("+,%#!*&"$
,)" -2)*' 3(4$2&'()* %&'()*&+$ F1UHI (#/" ,. 6.5"%+/"+, .7 T9-: [)#+ $-%*+6 ,)" ;<@C& #+$
3.4"% *+ ;<=@ 4*,).-, VV# &,%.+6 "+.-6) &.(*#! :#&" ;<=C&' .&1+&*&9 F%"!*6*.-& !"#$"%&I $*$ +., 3!#9 #+
*+ .+" &"(,*.+ .7 ,)" 3.3-!#,*.+ ,. "+#:!" *, ,. *+_-"+,*#! %.!"0 S+$""$' -&*+6 # VV&,"#/ %.!!"%
*/3.&" %#$*(#! ()#+6"& .+ ,)" %"&,PP!0 1#("$ 4*,) #33%.#()PP .7 %#3*$ 3.!*(9 *+*,*#,*.+ #+$ */3!"/"+O
3"%&*&,"+, $*&-+*,9 #/.+6 *,& 3.!*,*(#! (.+&,*,-"+(9' ,#,*.+' 1#/*!9 8!#++*+6 M.//*&&*.+"% b+5"% T$*!
#& 4"!! #& 4*,)*+ 6.5"%+/"+,' 3.!*(9 "!*,"& *+ 7#5.-% *+,%.$-("$ ,)" (.-+,%9P& R%&,' #+$ 3"%)#3& &,%.+O
.7 18 (.-!$ .+!9 /.5" &!.4!90 W9 ,)" /*$ ;<=C&' 6"&,' "].%, ,. 3%.5*$" 18 3%.6%#//"&0 W",4""+
4./"+P& 6%.-3& :"6#+ ,. (#!! 7.% */3%.5"$ 18 &"%O ;<=@?BC' ,)" 7#/*!9 3!#++*+6 &()"/" 4#& $"3!.9"$
5*("& #+$ #(("&& ,. (.+,%#("3,*.+ #& # %*6),0 E)*& +#,*.+#!!9' # VV(%#&) 3%.6%#//"PP :#(A"$ :9 )*6)O
4#& 7.!!.4"$ :9 ,)" R%&, VV:*%,) &3#(*+6PP ("+,%" *+ 3%.R!" 3-:!*(*,9 #+$ &-:&,#+,*#! #*$ 7%./ acTS\
T!6*"%& #+$ # 5&'6& F%"!*6*.-& "$*(,I &-33.%,*+6 ,)" #+$ .,)"% "2,"%+#! $.+.%&!0^)*!" ,)*& (%"#,"$ /#+9
5.!-+,#%9 -&" .7 (.+,%#("3,*.+0 X", 3.!*,*(#! !"#$"%& :"+"R(*#%*"&' $""3"% $*5*&*.+& 4"%" (%"#,"$ #/.+6
%"/#*+"$ $*5*$"$ ,)%.-6).-, ,)" ;<BC& #+$ ;<YC&0 (*5*! &"%5#+,& (./3",*+6 ,. &)#%" *+ ,)" !#%6" *+_-2
8%"&*$"+, W.-/"$*"++" (.+,*+-"$ ,. :" -+(.+O .7 %"&.-%("&"0 E)" %"&-!, 4#& ,)#,' *+&,"#$ .7 :"*+6
5*+("$ :9 ,)" +""$ ,. !*/*, 3.3-!#,*.+ 6%.4,)'
:-*!, .+ # :%.#$ (.#!*,*.+ .7 A"9 3.!*(9 #(,.%&'
6*5"+ ,)" (.-+,%9P& &*Z" #+$ .*! %"&.-%("&' #+$ &#4
*+5.!5"/"+, *+ ,)" 18 3%.6%#//" 4#& (!.&"!9 ,*"$
18 #& 3#%, .7 4"&,"%+O!"$ +".(.!.+*#!*&/
,. ,)" "2*&,*+6 &9&,"/ .7 3.!*,*(#! 7#5.-%&0 E)-&' ,)"
F[.-#.-(*' ;<<QI0
.5"%,)%.4 .7 ,)" T9-: %"6*/" *+ ;<=Y !"$ ,. *,&
$.4+7#!! #& 4"!!0 T& [)#+ F;<<QI 4%*,"&' VVS7 3.3-O
7&*8+&,$9: &*, /&;(9'&* !#,*.+ 4#& # 7#5.-%*," .7 T9-:P&' *, (.-!$ +"5"%
S+ ,)" R+#! 3#*% .7 (.-+,%*"&' *, )#& 3"%)#3& +., :"(./" 8%"&*$"+, W)-,,.P& 3", (#-&"PP0 S, 4#& +.,
:""+ &. /-() ,)" 7.%/#,*.+ .7 # 3.!*(9 (.#!*,*.+' -+,*! ,)" /*$ ;<BC& ,)#, W)-,,. :"6#+ ,. &-33.%,
:-, %#,)"% ,)" 4"#A+"&& .7 .33.&*,*.+ ,. 18' 4)*() 18 :9 &",,*+6 -3 # 6.5"%+/"+, (.//*,,"" ,. *+5"&O
$*&,*+6-*&)"& W#+6!#$"&) 7%./ 8#A*&,#+0 \-%*+6 ,)" ,*6#," ,)" 3.3-!#,*.+ 3%.6%#//"' #+$ &,#,*+6 3-:O
/*$ ;<BC&' 3.,"+,*#! .33.&*,*.+ ,. 18 3%.6%#//"& !*(!9 ,)#, VV/-() /.%" +""$& ,. :" $.+"PP0 T6#*+'
:9 K-&!*/ !"#$"%& *+ W#+6!#$"&) 4#& "]"(,*5"!9 ).4"5"%' ,)*& $*$ +., !"#$ ,. 4*$"+"$ &-33.%,0
-+$"%/*+"$ :9 ,)"*% 3.!*,*(#! /#%6*+#!*&#,*.+ #7,"% \-%*+6 ,)" ;<BB 6"+"%#! "!"(,*.+' 18 4#& #6#*+
,)"9 7#5.-%"$ -+*,9 4*,) 8#A*&,#+ $-%*+6 ,)" ^#%
(#-6), -3 *+ 3.!*,*(#! ,-%/.*!0 T& d.&"3) ^)""!"%'
.7 S+$"3"+$"+("0 E)*& .3"+"$ ,)" 4#9 7.% 3.!*(9
M)*"7 .7 ,)" acTS\ K*&&*.+ ,. 8#A*&,#+ F;<=<?BBI
"!*,"& 4*,)*+ 6.5"%+/"+, ,. ,#A" # R%/ 3.&*,*.+ .+
%"(#!!&' VVb5"%9,)*+6 7"!! #3#%, 4)"+ ,)" 3.!*,*(#!
7.%("& ,..A .5"% 3.3-!#,*.+ 3"%&.++"! #+$ 5")*(!"&
DW#&"$ .+ # &-%5"9 :9 `-"$$#+# F;<<QI .7 QY; #%,*(!"& 7.% ,)" ;<BB "!"(,*.+&PP0 ^)"+ `"+"%#! e*# $"(!#%"$
4)*() #33"#%"$ *+ ,)" /#*+ E-+*&*#+ +"4&3#3"%& 7%./
;<=>O;<BQ0
/#%,*#! !#4 *+ ;<BB' 18 4#& .+(" #6#*+ %"!"6#,"$ ,.
!E)" -2)*' ,$ 3(4$2&'()* %&'()*&+$ F1UHI 4#& # (.O # /#%6*+#! 3!#(" .+ ,)" 3.!*(9 #6"+$#0
#!*,*.+ .7 5#%*.-& 7#(,*.+& #+$ (!#&&"& *+ T!6"%*# $-%*+6 S+ #!! 7.-% 3#*%& .7 (.-+,%*"&' ,)"%"7.%"' 18 3%.O
,)" #+,*O(.!.+*#! &,%-66!"0 8.!*,*(#! $*]"%"+("& :",4""+ 6%#//"& 4"%" *+*,*#,"$ "#%!*"% #+$ /.%" "]"(,*5"!9
%"6*.+& 4"%" .5"%(./" $-%*+6 ,)" 4#% #6#*+&, ,)"
1%"+() :-, 7#(,*.+#!*&/ &-%7#("$ #6#*+ .+(" *+$"3"+O 4)"%" (.#!*,*.+& .7 3.!*(9 "!*,"& &-33.%,"$ ,)"/0 S+
$"+(" 4#& #()*"5"$0 1.% # $*&(-&&*.+ &"" N.:"%,& E)#*!#+$' e*/:#:4" #+$ W#+6!#$"&)' "2,"%+#!
F;<YQI0 $.+.% #6"+(*"& 7.%/"$ 3#%, .7 ,)*& 3.!*(9 "!*," 6*5"+
!S+,"%5*"4 4*,) [#:*%' f*(" 8%"&*$"+, F;<=C?BgI #+$
8%"&*$"+, F;<B=?3%"&"+,I .7 ,)" 1#/*!9 8!#++*+6 ,)" */3.%,#+(" .7 7.%"*6+ #*$0 L.4"5"%' *, 4#& ,)"*%
T&&.(*#,*.+ .7 W#+6!#$"&)' \)#A#' g K#%() ;<<Q0 %.!" *+ (.+,%*:-,*+6 ,. # (.)"&*5" 3.!*(9 (.#!*,*.+
"1.% "2#/3!"' 18 3%.6%#//" &,#] %"("*5"$ &#!#%*"& 4*,) +#,*.+#! 3.!*(9 "!*,"&' %#,)"% ,)#+ ,)"*% /"%"
&-:&,#+,*#!!9 )*6)"% ,)#+ .,)"% 6.5"%+/"+, "/3!.9""&' 3%"&"+(" #& *+ ,)" 8)*!*33*+"&' e#/:*# #+$
)#$ -&" .7 # _"", .7 5")*(!"& ,. 5*&*, !.(#! (.//-+*,*"&'
#+$ 4"%" *+5*,"$ ,. (.+7"%"+("& #+$ ,%#*+*+6 (.-%&"& 8#A*&,#+' 4)*() )#& :""+ ,)" */3.%,#+, 7#(,.% *+
#:%.#$0 6",,*+6 #+$ A""3*+6 18 .+ ,)" 3.!*(9 #6"+$#0
404
INV g5 h++ !" #$%
!"#$%&' ()*%+, $%!- %& .#/%*, (*#&&%&' )*%= :$) -. 9)$7)0##+% %&+##+2 :)$# *&% 0*# &$
!" #$%& '$("&)*+%, -. /0% 1++" 0"2 )+#0*"% 0 )+3+)%+ &8$ '+"&()*+% $: 9)$#$&*"7 40)7+ :0#*4*+%
'$"&)$3+)%*04 *%%(+5 6+04*"7 8*&/ &/+ #$%& 9)*30&+ 0"2 9$9(40&*$" 7)$8&/ :$) PP&/+ 7)+0&"+%% $: &/+
$: /(#0" 1+/03*$()%, ;+& 8*&/ 8*2+%9)+02 0"2 "0&*$"QQ5 B/+%+ 3*+8% 8+)+ #0*"&0*"+2 1; &8$ '$#?
#0<$) *#94*'0&*$"% :$) 9(14*' 9$4*';, -. #(%& 1+ #*&&++% $: &/+ R0&*$"04 S+%+0)'/ T$("'*4 8/*'/, *"
099)$0'/+2 1; 9$4*'; #0=+)% 8*&/ 2(+ '0)+ 0"2 0 )+9$)& %(1#*&&+2 &$ &/+ T01*"+& *" HIUO, *"%*%&+2
0&&+"&*$"5 >"+ 80; $: 099)$0'/*"7 %('/ 0 9$&+"? &/0& B/0*40"2Q% 9$9(40&*$" %/$(42 '$"&*"(+ &$ 7)$8
&*044; '$"&+"&*$(% *%%(+ *% &$ %9)+02 &/+ 9$4*'; )*%=5 0% 0" +"7*"+ :$) +'$"$#*' 2+3+4$9#+"&5
.$4*'; )*%= *% 2+@"+2 0% &/+ 9$%%*14+ "+70&*3+ '$"%+? -()&/+)#$)+, &/+ )+9$)& 0)7(+2 &/0& 0 -. 9$4*';
A(+"'+% &/0& '0" 0)*%+ :)$# 9()%(*"7 0 90)&*'(40) #*7/& 4+02 &$ 0 2+&+)*$)0&*$" *" 9(14*' #$)04*&;
9$4*';5 B/+ 9$&+"&*04 *#90'& #0; 1+ #0&+)*04, %('/ GM$"71$$"%*", HIIVJ5 B$ 022)+%% &/*% 9$4*'; )*%=,
0% 0 4$%% $: &0C )+3+"(+, $) #$)+ *"&0"7*14+ 1(& 9$4? %(99$)&+)% $: -. G&/+ '$04*&*$" $: 9$4*'; +4*&+%
*&*'044; *#9$)&0"&, %('/ 0% 0 4$%% $: 9(14*' '$"@? 2+%')*1+2 01$3+J *"3$43+2 2*L+)+"& *"%&*&(&*$"%
2+"'+5 D+& 9$4*'; #0=+)% #(%& :)+A(+"&4; &0=+ %('/ 8*&/*" 0"2 $(&%*2+ $: 7$3+)"#+"& 1; (%*"7 0
)*%=%5 E''$)2*"7 &$ F)*"24+ 0"2 B/$#0% GHIIHJ, 1)$024; 099+04*"7 )0&*$"04+ :$) -. 9)$7)0##+%5
S+'$"'*4*"7 -. 8*&/ &/+ '$("&);Q% W(22/*%& '(4&()+,
%$4(&*$"% &$ 0"; 7*3+" %+& $: 9$4*'; 9)$14+#% 0)+ "$& "$&014; &/+ %&)$"7 )+3+)+"'+ :$) 4*:+ *" 044 :$)#%,
$13*$(% 1+'0(%+ &/+ *#90'& $: 9$4*'; '0""$& 0480;% 1+ %&)$"7 +#9/0%*% 80% 940'+2 $" &/+ "++2 &$ )+2('+
="$8" *" 0230"'+, 1+'0(%+ &/+ 4$7*' $: +'$"$#*'% 0"2 &/+
4$7*' $: 9$4*&*'% :)+A(+"&4; 2$ "$& '$*"'*2+, 0"2 1+'0(%+ 9$9(40&*$" 7)$8&/ *" $)2+) &$ *#9)$3+ &/+ PPA(04*&;
)+04 '$%&% 0)+ *#9$%+2 $" %9+'*@' 7)$(9% *" %$'*+&; 8/+" $: 4*:+QQ $: 044 B/0*%5 !" &/*% 80;, 9$4*'; )*%= 80%
9$4*'*+% 0"2 *"%&*&(&*$"% 0)+ 04&+)+25 E44 9$4*'; '/$*'+% &/(% &+#9+)+2 1; 0)7(*"7 &/0& -. 9)$7)0##+% 8+)+ *"
*"3$43+ ("'+)&0*"&; 0"2 )*%=5 &/+ '$44+'&*3+ *"&+)+%& $: 044 '*&*X+"%5
!" '$"&)0%&, 9$4*'; )*%= )+#0*"+2 '$"'+"&)0&+2 *"
K9+'*@' 9$4*'*+% 30); *" &/+ 0#$("& $: 9$4*'; )*%=
&/+ /0"2% $: )+40&*3+4; :+8 *"2*3*2(04% *" &/+
*"3$43+2 2+9+"2*"7 $" &/+ "0&()+ $: &/+ *%%(+
./*4*99*"+%, "0#+4; &/+ .)+%*2+"& 0"2 /*%Y/+) %+"*$)
022)+%%+2, %&0=+/$42+)% 0L+'&+2 0"2 9$4*'; +"3*)$"?
9$4*&*'04 099$*"&++%5 M/*4+ &/+ 7$3+)"#+"& ')+0&+2
#+"&5 !" 7+"+)04, 9$4*'*+% 9$%*"7 7)+0&+) )*%= *"'4(2+
0 %+"%+ $: %/0)+2 )+%9$"%*1*4*&; :$) 9$9(40&*$" 9$4?
&/$%+ 8/*'/ 0L+'& 0 8*2+) )0"7+ $: %&0=+/$42+)%,
*'*+% *" B/0*40"2 &/)$(7/ 0 1)$024; 099+04*"7
*"3$43+ 0 )+40&*3+4; 40)7+ 0#$("& $: )+%$()'+%, 0"2
)0&*$"04+, 8/+" .)+%*2+"& Z0)'$% 80% )+?+4+'&+2 *"
/03+ 0 7)+0&+) 40'= $: *":$)#0&*$" $) ="$84+27+
HIUI, /+ :+4& %+'()+ +"$(7/ &$ 0%%(#+ #('/ $: &/+
G*5+5 ("'+)&0*"&;J5
9$4*'; )*%= /*#%+4:5 -. 9)$7)0##+% 8+)+ 3+); #('/
>"+ %&)0&+7; :$) 2+04*"7 8*&/ 9$4*'; )*%= *% &$
/*% 02#*"*%&)0&*$"Q% *"*&*0&*3+, 8*&/ 9$4*'; :$)#(?
%9)+02 )*%= 0')$%% &*#+ $) 940'+5 B/*% '$(42 1+
40&*$" *"3$43*"7 :+8 *"2*3*2(04% $&/+) &/0" 9$4*&*'04
0'/*+3+2, :$) +C0#94+, 1; +"'$()07*"7 1)$02 90)?
099$*"&++%5 M/+" /*% /$42 $" 9$8+) 8+0=+"+2 *"
&*'*90&*$" *" 9$4*'; #0=*"7 1; 0 8*2+) )0"7+ $: %&0?
&/+ HI[O%, +C&+)"04 2$"$) 07+"'*+% #0*"&0*"+2
=+/$42+)%, ')+0&*"7 0 '$44+'&*3+ )0&*$"04+ :$) -.
#$#+"&(# 1+/*"2 -. 1; 9)$3*2*"7 0 40)7+ 9)$?
'$"7)(+"& 8*&/ 1)$024; /+42 %$'*04 304(+% 0"2
9$)&*$" $: &/+ :("2*"7 :$) 9)$7)0##+%\ 1(&, *"
1+4*+:% G+575 1*)&/ %90'*"7, +'$"$#*' 2+3+4$9#+"&J,
2$*"7 %$, 1+'0#+ &/+ #0*" )*%= /$42+)%5 E% T0)*"$
$) *"&)$2('*"7 &/+ 9$4*'; #$)+ 7)02(044; $3+) &*#+5
GHIIVJ 8)*&+%, /+03; )+4*0"'+ $" +C&+)"04 :("2*"7
!" &/*% 80;, &/+ )*%= 0%%$'*0&+2 8*&/ -. "++2 "$& 1+
'$"&)*1(&+2 &$ 0 1+4*+: &/0& &/+ 9$4*'; 80% PP0
*"'())+2 1; 0 90)&*'(40) *"2*3*2(04 $) 7)$(9 8/$
:$)+*7" *#9$%*&*$" 8/*'/ /0% "+3+) 2+3+4$9+2 *&%
#*7/&, *" &()", 1+ 3(4"+)014+ &$ $99$"+"&% $: &/+
$8" 4$'04 '$"%&*&(+"';QQ5 ]$8+3+), &/+ 40'= $: 0
9$4*';5
1)$02 2$#+%&*' '$"%&*&(+"'; :$) -. '$"&*"(+25 B/+
B/+ %&(2; %$(7/& &$ +C94$)+ &/+ +C&+"& &$ 8/*'/
'$"'+"&)0&*$" $: 9$4*'; )*%= *" &/+ +C+'(&*3+, 0"2
-. 9)$7)0##+% 8+)+ #$)+ 4*=+4; &$ 1+ %(%&0*"+2
'$"%+A(+"& 3(4"+)01*4*&; $: -. &$ 9$4*&*'04 '/0"7+,
8/+)+ 9$4*'; )*%=% 8+)+ %(''+%%:(44; %9)+025 M/*4+
1+'0#+ 0990)+"& 8/+" EA(*"$ 1+'0#+ 9)+%*2+"& *"
&/+ 2+7)++ $: 9$4*'; )*%= 0%%$'*0&+2 8*&/ -. 30)*+2
HIÛ5 >99$%+2 &$ &/+ %&)$"7 *"_(+"'+ $: &/+ `5K5
:)$# '$("&); &$ '$("&);, 0% 2+%')*1+2 1+4$8, &/+
$" &/+ '$("&);Q% :$)+*7" 0"2 2$#+%&*' 9$4*'*+%, 0"2
'$#90)0&*3+ +L+'&*3+"+%% 8*&/ 8/*'/ 9$4*'; +4*&+%
2)08*"7 #('/ $: /+) %(99$)& :)$# &/+ T0&/$4*'
%9)+02 &/+ %9+'*@' )*%= *" +0'/ 90*) $: '$("&)*+% 80%
T/()'/, %/+ %08 -. 0% '4$%+4; *2+"&*@+2 8*&/ &/+
+C9+'&+2 &$ 2*%&*"7(*%/ %&)$"7+) :)$# 8+0=+) -.
9)+3*$(% )+7*#+ $) 0% :$)+*7"?2)*3+" GT0)*"$, HIIVJ5
9)$7)0##+%5
B/*% 80% 0''$#90"*+2 1; &/+ 40'= $: $"+ '4+0)
!" B/0*40"2, 8/+)+ &/+ 7$3+)"#+"& 9()%(+2 0
)0&*$"04+ :$) -. 9)$7)0##+% 8*&/ 1)$02 0"2 %(%?
9)$"0&04*%& 9$4*'; ("&*4 &/+ 40&+ HINO%, &/+ 9$4*';
&0*"+2 099+04, *"%&+02 02$9&*"7 2*L+)+"& )0&*$"04+%
0% *"&+)"0&*$"04 $9*"*$" '/0"7+2!5
\bC&+)"04 2$"$) 07+"'*+% 9)$3*2+2 cNd $: :("2*"7 :$) -. !" a*#1018+, &/+ 7$3+)"#+"& :0'+2 0 /*7/ 4+3+4
9)$7)0##+% *" &/+ ./*4*99*"+% 1+&8++" HI[Oe^^5 $: 9$4*'; )*%= *" %(99$)&*"7 -. 0:&+) *"2+9+"2+"'+
!S0&*$"04+% 9(& :$)&/ 1; &/+ -*4*9*"$ 7$3+)"#+"& /03+
*"'4(2+2 %$'*04 0"2 +'$"$#*' 2+3+4$9#+"& GHIUIJ,
*" HIÔ5 !& "++2+2 &$ &)0"%:$)# 9(14*' 9+)'+9&*$"%
:0#*4; 8+4:0)+ GHI[^J, A(04*&; $: 4*:+ GHI^[J, %(%&0*"014+ $: -., :)$# 1+*"7 0 )+#"0"& $: &/+ '$4$"*04 90%&,
2+3+4$9#+"& GHIIfJ 0"2 )+9)$2('&*3+ /+04&/ GHIIcJ5 &$ 0 %+)3*'+ 9+)'+*3+2 0% )(" 1; 0"2 :$) 140'=
405
:%#"./ B.%(("(5 "( .3& "(03#' 03A(-?"') H\\
!"#$%$&'%()* +,") &%) "("-"%../ %0,"'1'2 $/ %..3&4 ('(- ?3.' "( :; D?3# -,' #"2 GHUV)9 %(2 "- &%) )'-
"(5 %( 6789 -,' :%#"./ ;.%(("(5 <))30"%-"3( %B%?- D?3# 3-,'? )'0-"3() 3D -,' P8R* R3&'1'?9
=:;<>9 -3 $' -,' #%"( ?")@ -%@'?9 -,A) BA--"(5 :; #%?5"(%.")%-"3( 3D -,' B?35?%##' &%) B?'1'(-'2 $/
"()-"-A-"3(%../ %- %?#C) .'(5-,* <- -,' )%#' -"#'9 B3."0/ '."-')9 )A0, %) ;?')"2'(- [%,#%(9 &,3 ),%?'2
-,' 531'?(#'(- 03(-"(A'2 -3 DA(2 )3#' 3D -,' -,' B3."0/ ?")@ $/ BA$."0./ 2'0.%?"(5 -,' (''2 -3
:;<C) %0-"1"-"') &"-, %22"-"3(%. )ABB3?- D?3# 'E-'?4 03(-?3. -,' 03A(-?/C) B3BA.%-"3( 5?3&-, %) NN-,'
(%. 23(3?) =)'' $'.3&>* F"-, -,' $%(("(5 3D (A#$'? 3(' B?3$.'#CC =P%,#3329 GHHQ>* ;3."0/
;?31'?% "( GHIG9 D3..3&'2 $/ -,' ?')"5(%-"3( 3D ?")@ &%) %.)3 ),%?'2 $/ -,' 'E-'()"1' $A?'%A0?%0/
#%(/ .3(54)'?1"(5 =&,"-'> )-%J 3D -,' :;<9 -,' &,"0, $'0%#' "(13.1'2 "( :;9 %(2 $/ -,' ('-&3?@
531'?(#'(- ?'(%#'2 -,' 3?5%(")%-"3( -,' K,".2 3D 678) &,3)' %0-"1"-"') &'?' "()-?A#'(-%. "( 2'."4
LB%0"(5 %(2 :'?-"."-/ <))30"%-"3( =KL:<>9 &"-, %( 1'?"(5 .30%. )'?1"0')*
'#B,%)") 3( 0,".2 )B%0"(5 ?%-,'? -,%( D'?-"."-/ 03(4 O()-"-A-"3(%. "()-%$"."-/ "( ;%@")-%( ,%) 03"(0"2'2
-?3.* <) -,' %2#"(")-?%-3? 3D -,' KL:< )-%-'2 "( &"-, % ."#"-'2 ?3.' D3? 678)* :?3# -,' .%-' GH\V)9
GHIM9 NNO( %( <D?"0%( 03(-'E-9 /3A 0%((3- -%.@ 3D -,' :;<; -33@ % .'%2 ?3.' "( 5A"2"(5 -,' 531'?(4
."#"-"(5 0,".2?'(* K,".2 )B%0"(5 ,3&'1'? ") #3?' #'(- 3( B3BA.%-"3(* W/ -,' #"2 GHUV)9 ,3&'1'?9
03(1"(0"(5 "( -,%- /3A %?' (3- )-3BB"(5 %(/3(' -,' 531'?(#'(- 2")03A?%5'2 "(0?'%)'2 678 %0-"1"-/
D?3# ,%1"(5 0,".2?'( $A- ,'.B"(5 -,'# B.%( -,'"? %) % B3-'(-"%. -,?'%- -3 0'(-?%.")'2 B3."-"0%. B3&'?*
D%#"."')CC =P%1'('@%9 GHHQ>* 7?%2A%../9 &"-, 5?'%-'? O( GHI\9 % $?"'D %--'#B- &%) #%2' -3 "()-"-A-"3(%.")'
%00'B-%(0' $/ $.%0@ !"#$%$&'%() %0,"'1'29 -,' -,' ?3.' 3D 678)9 B%?-./ %- -,' $',')- 3D -,'
531'?(#'(- #%2' -,' KL:< % B%?%)-%-%. "( GHIQ 23(3?)9 "( 3?2'? -3 5%.1%(")' 678) %(2 B?31"2'
%(2 ?'(%#'2 "- -,' !"#$%$&' 6%-"3(%. :%#"./ %(3-,'? 0,%(('. D3? DA(2"(5* +,") .'2 -3 -,' 0?'4
;.%(("(5 K3A(0". =!6:;K>*O( !%#$"%9 B3."0/ ?")@) %-"3( 3D -,' 63(4731'?(#'(- 8?5%(")%-"3()
&'?' %))A#'2 $/ 678) %(2 % )#%.. (A#$'? 3D K33?2"(%-"(5 K3A(0". =678KK> &,3)' 03(-?3.
'E-'?(%. 23(3?)* <) % ?')A.-9 :; &%) )''( %) % )33( $'0%#' % )3A?0' 3D 03(-'(-"3( $'-&''( -,'
1'?-"0%. B?35?%##' "( -,' P8R 3&('2 $/ 3A-4 :;<;9 -,' 3(./ 678 &"-, 'E-'()"1' 'EB'?"'(0' "(
)"2'?)9 ?%-,'? -,%( %( "(-'5?%-'2 B%?- 3D -,' :;9 %(2 -,' 531'?(#'(-* <) Z,%( =GHHQ> &?"-')9
531'?(#'(-C) 3&( ,'%.-, B3."0"')* F"-, ."--.' )AB4 ('53-"%-"3() B?31'2 % NN-')-"#3(/ -3 2''B'? ?"D-)
B3?- D?3# 531'?(#'(- "()-"-A-"3()9 -,' B?35?%##' -,%- &3A.2 (3- $' ?')3.1'2CC* < )-?A55.' '()A'2
"-)'.D ,%2 -3 B'?)A%2' .30%. %A-,3?"-"') -3 "(-?32A0' D?3# GHI\]H^9 &"-, -,' 531'?(#'(- &"),"(5 -3
)'?1"0')* ?'-%"( 03(-?3. 3D 3$_'0-"1')9 #3("-3?"(5 %(2 '1%.A4
O( +A(")"%9 "()-"-A-"3(%. #3$"."-/ %BB'%?) -3 ,%1' %-"3( 3D %0-"1"-"')9 %) &'.. %) DA(2"(5* O( GHH^9 &,'(
$''( $%.%(0'2 $/ -,' )B?'%2 3D B3."0/ ?")@ &"2'./ -,' #%(2%-' 3D -,' 678KK 'EB"?'29 B3."0/ #%@'?)
-,?3A5,3A- 531'?(#'(-9 5"1"(5 :; % S?#'? "()-"4 $'0%#' '#$?3".'2 31'? &,'-,'? %(2 "( &,%- D3?#
-A-"3(%. )ABB3?- $%)' -,%( "( <.5'?"%* TA?"(5 -,' -,' 678KK ),3A.2 $' ?'03()-"-A-'2* O( GHH\ "- &%)
GHUV)9 -,' 86:; &%) .30%-'2 &"-,"( -,' P8;R S(%../ ?'4')-%$."),'2 %) -,' 6%-"3(%. +?A)- D3?
%(2 #%(%5'2 $/ % W3%?2 3D T"?'0-3?) ?'B?')'(-"(5 ;3BA.%-"3( F'.D%?' =6<+;8F>*
'"5,- #"(")-?"') %(2 -,?'' (%-"3(%. 3?5%(")%-"3() O( %.. D3A? B%"?) 3D 03A(-?"')9 -,'?'D3?'9 -,' 31'?4
=7A'22%(%9 GHHQ>* O( GHIM -,' 86:; $'0%#' 03(0'(-?%-"3( 3D B3."0/ ?")@ "( ?'.%-"1'./ D'& ,%(2)
%X."%-'2 &"-, -,' P"(")-?/ 3D F3#'( %(2 :%#"./ #%2' :; B?35?%##') #3?' 1A.('?%$.' -3 B3."-"0%.
<J%"?)9 %(2 "( GHIY "- #31'2 $%0@ %5%"( -3 -,' 0,%(5'* O- %BB'%?) -,%-9 %.3(5 &"-, B3."0/ 034
P8;R* +,?3A5,3A- -,") B'?"329 -,' ?%-"3(%.' 3D %."-"3()9 )-?3(5 :; B?35?%##') (''2 -3 $' %04
"#B?31"(5 -,' )-%-A) 3D &3#'( &%) 03()")-'(-./ 03#B%("'2 $/ % ?%-"3(%.' -,%- $3(2) B3."0/ '."-')
5"1'( D3? -,' :; B?35?%##'* O( <.5'?"% :; -35'-,'? &"-, % $?3%2'? 03()-"-A'(0/* O( )3#'
?'#%"('2 2'S('2 %(2 "()-"-A-"3(%../ SE'2 %) % 03#4 03A(-?"')9 %0,"'1"(5 )A0, 03()'()A) &%) #3?' 2"X4
B3('(- 3D #%-'?(%. %(2 0,".2 ,'%.-, =PKR> &"-,"( 0A.- -,%( 3-,'?) 5"1'( 'E")-"(5 B3."-"0%. 2"1")"3() 3?
-,' P8R* W'/3(2 -,' #'2"0%. B?3D'))"3(9 %(2 -3 % "()-%$"."-/* `'-9 -3 -,' 'E-'(- -,%- B3."0/ #%@'?)
.'))'? 'E-'(- &3#'(C) 3?5%(")%-"3()9 B3."0/ ?")@ &%) &'?' %$.' -3 2'%. 'J'0-"1'./ &"-, ?")@9 '1'( "( B34
)B?'%2 %#3(5 D'& 3-,'?)* O(2''29 B3."-"0%. .'%2'?) -'(-"%../ 13.%-".' B3."0/ '(1"?3(#'(-)9 :; B?34
2"2 (3- )ABB3?- -,' B3."0/9 %(2 3B'(./ 0?"-"0")'2 "- 5?%##') &'?' %23B-'2 %(2 #%"(-%"('2* T'%."(5
%- -,' "(-'?(%-"3(%. .'1'. 2A?"(5 -,' GHUV) &"-, B3."0/ ?")@ ") 0.'%?./ (3- % B?'0")' )0"'(0'9 $A-
=Z3A%3A0"9 GHHQ>* +,") #'2"0%.")%-"3( 3D :; .'D- -,' B%?- 3D -,' %?- 3D 'J'0-"1' 531'?(#'(-*
B3."0/ ?")@ ?'.%-"1'./ A(),%?'29 .'%2"(5 "- -3 $'03#'
#%?5"(%.")'2 %#3(5 531'?(#'(- B3."0"')*
+,' )%#' "('J'0-"1'(')) %- 2'%."(5 &"-, B3."0/ !"##$%& !'($)* $%#' !+,)#$)-. #/- +'(- '0
$%1#$#"#$'%,( ,%2 0$%,%)$,( 1#,3$($#* $%
?")@ &%) D3A(2 "( ;%@")-%( &,'?' "- ,%) $''( 03(4 $4!(-4-%#,#$'%
0'(-?%-'2 "( B?3#"('(- "(2"1"2A%.) 3? 'E-'?(%.
23(3?)9 ?%-,'? -,%( )B?'%2 %0?3)) "()-"-A-"3()* +,") < -,"?2 S(2"(5 3D -,' )-A2/ ") -,%- -,'?' &'?' "#4
.'2 :; -3 $'03#' % 1"0-"# 3D B3."-"0%. 0,%(5' &,'( B3?-%(- 2"J'?'(0') "( ,3& :; B?35?%##') &'?' "#4
B3."0/ '."-') D'.. D?3# B3&'? =Z,%(9 GHHQ>* O( B.'#'(-'2 "( -,' '"5,- 03A(-?"')* F%.- =GHH\> &?"-')
W%(5.%2'),9 D3?'"5( 23(3?) %.)3 %))A#'2 % B?3#"4 -,%- #3)- %--'(-"3( "( -,' %(%./)") 3D ,'%.-, B3."0/
406
A[Z e6 c%% !" #$%
!"# $%%& '()*#%+ (& ,!% '(-.*/",0(& (' 1(/0)23 40,! .";0&83 0&)/*+0&8 " '"0/*-% ,( +%#08&",% )/%"-
,!% "##*.1,0(& ,!", "+(1,%+ 1(/0)0%# 40// $% 0.5 -%#1(&#0$0/0,2 '(- 0.1/%.%&,0&8 ,!% 1-(8-"..%6
1/%.%&,%+ "# +%#0-%+6 7!%-% 0# 0&)-%"#0&8 -%)(85 E*&+0&8 (' EF 1-(8-"..%# 0& ,!% ,4( )(*&,-0%#
&0,0(& 0& ,!% !%"/,! #%),(-3 !(4%9%-3 ,!", #,-",%80%# "/#( +0:%-%+6 K& "$#(/*,% ,%-.#3 0, 0# 1%-!"1# #*-5
'(- ,!% 0.1/%.%&,",0(& (' 1(/0)0%# "-% "/#( 0&,%8-"/ 1-0#0&8 ,!", ,!% F!0/0110&%# -%)%09%+ )(.1"-"$/%3 0'
,( %:%),09% 1(/0)2 .";0&86 <& 0&)-%"#0&8 ".(*&, (' &(, /"-8%-3 ".(*&,# (' 1(1*/",0(& "##0#,"&)% ,!"&
-%#%"-)! !"# #(*8!, ,( 0+%&,0'2 9"-0(*# 1-%)(&+05 7!"0/"&+ (9%- ,!% #".% 1%-0(+3 40,! ,!% /"-8%#,
,0(&# *&+%- 4!0)! %:%),09% 0.1/%.%&,",0(& ())*-# %M,%-&"/ +(&(-# $%0&8 ,!% Y(-/+ N"&; "&+ _O<K\6
=>(84((+ "&+ ?*&&3 @ABCD6 K& 7!"0/"&+3 !(4%9%-3 ,!%-% !"# $%%& 8-%",%- #,"$5
7!% #,*+2 '(*&+ ,!", #,-(&8%- EF 1-(8-"..%# 0/0,2 (' '*&+0&8 (9%- ,0.%6 <# %M,%-&"/ "0+ '%// '-(.
4%-% 8%&%-"//2 )!"-"),%-0#%+ $2 8-%",%- #,"$0/0,2 (' TB` 0& @AB@ ,( @a6Z` 0& @AA@ (' ,(,"/ '*&+0&8 '(-
0&#,0,*,0(&"/ !(.% "&+G(- '*&+0&86 H& ,!% '(-.%-3 EF3 ,!% 8(9%-&.%&, 8-"+*"//2 "##*.%+ 8-%",%-
0.1/%.%&,",0(& %&)(*&,%-%+ +0I)*/,0%# 4!%-% EF L&"&)0"/ -%#1(&#0$0/0,2 =Y(&8$((&#0&3 @AACD6 7!0#
1-(8-"..%# 4%-% "))(.1"&0%+ $2 '-%J*%&, )!"&8%# L&"&)0"/ ,-"&#0,0(& +0+ &(, ())*- 0& ,!% F!0/0110&%#
0& 0&#,0,*,0(&"/ /()",0(& "&+G(- #,-*),*-%6 4!%-% 1(/0,0)"/ )(&,-(9%-#2 (9%- EF3 "&+ ,!% )(*&5
F-(8-"..%# 4%-% %#1%)0"//2 %:%),09% 4!%-% ,!% 0&5 ,-2Q# +%1%&+%&)% (& %M,%-&"/ "0+3 .%"&, " !%"92
#,0,*,0(&# -%#1(&#0$/% '(- 1(/0)2 '(-.*/",0(& "&+ +%1%&+%&)% (& %M,%-&"/ +(&(-#6 7!0# /%', ,!% 1-(5
0.1/%.%&,",0(& 4%-% )/(#%/2 /0&;%+ 0& ,!% 1(/0)2 8-"..% 9*/&%-"$/% 4!%& F-%#0+%&, <J*0&( "+(1,%+
1-()%##3 "&+ 4!%-% -%#1(&#0$0/0,2 '(- 0.1/%.%&5 " 1(/0)2 (' #%/'5-%/0"&)% 4!0)! /%+ ,( " #*++%&
,",0(& 4"# )/%"-/2 +%."-)",%+6 K& "++0,0(&3 -%/0"$/% +%)/0&% 0& %M,%-&"/ "0+3 *&"))(.1"&0%+ $2 &( )(--%5
L&"&)0"/ #*11(-, 4"# "/#( " &%)%##"-23 "/$%0, &(, #1(&+0&8 0&)-%"#% 0& 8(9%-&.%&, '*&+0&86 K& ,!%
#*I)0%&,3 '"),(- 0& )-%",0&8 #,-(&8 EF 1-(8-"..%#6 /",% @ABU#3 %M,%-&"/ "0+ $%8"& ,( 0&)-%"#% "8"0&
7!% )"#% #,*+0%# #!(4%+ ,!", 1-%+0),"$0/0,2 (' '*&+5 *&+%- F-%#0+%&, R".(#6 >(4%9%-3 ,!% 8(9%-&.%&,
0&8 )(..0,.%&, (9%- ,0.%3 "))(.1"&0%+ $2 0&#,05 )(&,0&*%+ ,( 1-(90+% (&/2 " #."// '-"),0(& (' '*&+5
,*,0(&"/ #,"$0/0,23 4"# .(-% 0.1(-,"&, ,!"& 0&8 '(- EF3 40,! _O<K\ =ZC`D "&+ _VEF< =SC`D
#08&0L)"&,3 $*, 9(/",0/%3 /%9%/# (' '*&+0&8 $2 8(9%-&5 )(&,-0$*,0&8 ,!% /"-8%#, #!"-%# =W"-0&(3 @AACD6
.%&,# "&+ %M,%-&"/ +(&(-#6 K& b0.$"$4%Gb".$0" "&+ 7*&0#0"G</8%-0"3 " #%15
N(,! 0&#,0,*,0(&"/ "&+ '*&+0&8 #,"$0/0,2 +0#,0&5 "-",0(& (' ,!% 0&#,0,*,0(&# -%#1(&#0$/% '(- 1(/0)2 '(-5
8*0#!%+ ,!% EF 1-(8-"..% 0& 7!"0/"&+ '-(. .*/",0(& "&+ 0.1/%.%&,",0(& )!"-"),%-0#%+ ,!%
,!% F!0/0110&%#6 K& 7!"0/"&+3 ,!% O*$)(..0,,%% 4%";%- EF 1-(8-"..%#6 K& b".$0" ,!%-% 4"# .0&05
'(- F(1*/",0(& F(/0)2 "&+ F/"&&0&8 -%."0&%+ ."/ 0&#,0,*,0(&"/ #*11(-, '(- EF #%-90)%# *&,0/ ,!%
PP,!% '()"/ 1(0&, '(- ,!% )(5(-+0&",0(& (' "),090,0%# /",% @ABU#3 40,! ,!% L-#, 1*$/0) EF )/0&0) %#,"$/0#!%+
(' "// *&0,# %&8"8%+ 0& (1%-",0(&#3 -%#%"-)!3 0& @ABB6 7!% #,*+2 '(*&+ ,!", ,!0# ."2 !"9% $%%&
"&+ -%#(*-)% "//()",0(& 666 "&"/2#0# (' &",0(&"/ +*% ,( ,!% '-"8.%&,",0(& (' ,!% 1(/0)2 1-()%##
+%.(8-"1!0) ,-%&+#3 "&+ ,!% .(&0,(-0&8 "&+ ".(&8 +0:%-%&, 0&#,0,*,0(&#6 7!% V",0(&"/
%9"/*",0(& (' 1-(8-"..% 0.1/%.%&,",0(& 0& "))(-+5 W(..0##0(& '(- \%9%/(1.%&, F/"&&0&8 4"# -%#1(&5
"&)% 40,! ,!% 1(1*/",0(& 1/"&QQ =R($0&#(& "&+ #0$/% '(- 1(/0)2 '(-.*/",0(&3 4!0/% ,!% XH> 0.5
R")!"1"%,"2";(.3 @AASD #0&)% ,!% %"-/2 @ATU#6 E(- 1/%.%&,%+ 1(/0)2 +%)0#0(&#6 c0.0,%+ EF #%-90)%#
,!% EF 1-(8-"..%3 0& 1"-,0)*/"-3 ,!% &",0(&"/ 4%-% 1-(90+%+ ", ,!% /()"/ /%9%/ $2 #."//5#)"/%
'".0/2 1/"&&0&8 1-(8-"..% !"# -%."0&%+ 40,!0& V?H# $*, ,!0# !"# &(, $%%& )((-+0&",%+ 40,! 8(95
,!% &",0(&"/ '".0/2 1/"&&0&8 )(..0,,%% =VEFWD %-&.%&, 1(/0)26 K& b0.$"$4%3 ,!% #,*+2 '(*&+ "
(' ,!% XHF> #0&)% 0,# )-%",0(& 0& @ATU3 "&+ !"# .(-% *&0L%+ 0&#,0,*,0(&"/ #,-*),*-%3 L-./2 /0&;%+ ,(
$%%& ,!% PP'()"/ 1(0&, '(- (1%-",0(&# "&+ )((-+05 ,!% 1-0."-2 !%"/,! )"-% 0&'-"#,-*),*-%3 "&+ $"#%+
&",0(& (' EF "),090,0%# ,!-(*8!(*, ,!% )(*&,-2QQ (& " )/(#% -%/",0(&#!01 $%,4%%& ,!% XH> "&+
=Y(&8$((&#0&3 @AACD6 7!0# -%/",09%/2 *&)!"&80&8 bVEFW6 <# " 1"-"#,","/ (-8"&0#",0(& /()",%+ *&+%-
0&#,0,*,0(&"/ #,-*),*-% )(&,-"#,# 40,! ,!% F!0/0110&%# ,!% XH>3 ,!% bVEFW ."0&,"0&%+ )/(#% )(..*&05
4!%-% ,!% F(1W(. !"# $%%& '-%J*%&,/2 )!"&8%+3 )",0(& 40,! ,!% 8(9%-&.%&, (& "// #,"8%# (' ,!% 1(/5
'-(. " -%#%"-)! 0&#,0,*,0(& 0& @AZA3 ,( )((-+0&",0&8 0)2 1-()%##6 V(- +0+ 0, &%%+ ,( )(.1%,% 40,! (,!%-
$(+2 0& @ATU3 ,( 0.1/%.%&,%- )*. )((-+0&",(- 0& #%-90)% 1-(90+%-#6 7!% bVEFW !"+ L-. (4&%-#!01
@AT[3 "&+ L&"//2 ,( )((-+0&",(- $*, &(, 0.1/%.%&5 (' ,!% EF 1-(8-"..% "&+ !"+ )/%"- -%#1(&#0$0/0,2
,%- 0& @ABT6 \*-0&8 ,!0# 1%-0(+3 ,!% 0&#,0,*,0(&"/ '(- 0,# 0.1/%.%&,",0(&6 N2 @ABB3 "& %#,0.",%+ T[d
!(.% (' ,!% F(1W(. #!0',%+ ".(&8 ,!% HI)% (' Aa` (' )(&,-")%1,09%# 4%-% ($,"0&%+ ,!-(*8! 1*$5
,!% F-%#0+%&,3 V",0(&"/ ])(&(.0) "&+ \%9%/(1.%&, /0) #%),(- #%-90)%# =e"/*.$"3 @AACD6
<*,!(-0,2 =V]\<D3 \%1"-,.%&, (' >%"/,! "&+ ?-%",%- #,"$0/0,2 (' '*&+0&8 "/#( +0:%-%&,0",%+ ,!%
\%1"-,.%&, (' O()0"/ Y%/'"-% "&+ \%9%/(1.%&,3 ,4( 1-(8-"..%#6 K& b0.$"$4%3 1*$/0) '*&+0&8 ('
"&+ ,!% "11(0&,.%&, (' &0&% %M%)*,09% +0-%),(-# EF !"# $%%& #*11/%.%&,%+ $2 L&"&)0&8 '-(.
40,! "& "9%-"8% ,%&*-% (' ,4( ,( ,!-%% 2%"-# _O<K\3 ,!% Y(-/+ N"&;3 _VEF< "&+ V?H#
=W"-0&(3 @AACD6 Y!0/% ,!0# 0&#,0,*,0(&"/ 0&#,"$0/0,2 =X"9%&%;"3 @AACD6 7!% EF 1-(8-"..% 0& b".$0"
4"# " -%^%),0(& (' 4%"; )(..0,.%&, ,( EF $2 ,!% -%)%09%+ .0&0."/ 8(9%-&.%&, '*&+0&8 +*-0&8 ,!0#
8(9%-&.%&,3 0, )(&,-0$*,%+ ,( +0I)*/,0%# 0& 1(/0)2 #".% 1%-0(+3 "&+ '"- /%## "##0#,"&)% '-(. %M,%-&"/
4 07
;+/6*? K*+##6#, 6# *"> 6#8"/) 8".#-$6)% D^X
!"#"$%& '() *+$,)%- %./ 01234&5 /6**6"#7 8+/) -"-+**)! 12344 /6**6"# H)->))# CDXCa5E +#! 123SC
9$"/ 1:;<= >(68( 9.#!)! -() 8$)+-6"# "9 -() /6**6"# H)->))# CD5Ca5U& 26#8) -() /6! CD5E%Q
9+/6*? ()+*-( .#6- 6# -() @AB 6# CD5E& @+F"$ )T-)$#+* +6! -" '.#6%6+b% ;< K$",$+//) (+%
!"#"$% -" G6/H+H>) >)$) #"-68)+H*? +H%)#- 9$"/ !)8*6#)!Q 9$"/ S[Y "9 -() A:;<b% H.!,)- 6# CD5S
"$ )$$+-68 6# -()6$ +%%6%-+#8) -" G+/H6+& I- >+% #"- -" CEY 6# CDD[& I/K"$-+#-*?Q -() ,"L)$#/)#- (+%
.#-6* CDD4 -(+- 12=IJ +,$))! -" K$"L6!) + %.H%-+#M -+N)# .K -(6% %*+8N >6-( K.H*68 9.#!6#, 0W.)!!+#+Q
-6+* K+8N+,) "9 %.KK"$- >6-( -() O"$*! P+#NQ AJ= CDDS7&
01R7 +#! 2IJ= 0R+*./H+Q CDDS7& '(6% 8"/H6#)! I# <+N6%-+# +#! P+#,*+!)%( +% "#) 8".#-$?Q -()
*+8N "9 ,"L)$#/)#- +#! !"#"$ 8"//6-/)#- -" ;< @AB ()*! *"> %-+-.% +/"#, ,"L)$#/)#- /6#6%M
(+% $)%.*-)! 6# *"> +#! .#%-+H*) *)L)*% "9 9.#!6#,& -$6)%Q +#! -() ()+*-( 6#9$+%-$.8-.$) $)/+6#)! K""$*?
'.#6%6+ +#! =*,)$6+ %("> + %6/6*+$ 8"#-$+%- 6# 6#M !)L)*"K)!& =*-(".,( + :+-6"#+* ;+/6*? <*+##6#,
%-6-.-6"#+* *6#N+,)%& '() '.#6%6+# K.H*68 ()+*-( 6#M ]".#86* 0:;<]7 >+% 8$)+-)! .#!)$ -() @AB 6#
9$+%-$.8-.$) >+% )TK+#!)! 9$"/ -() CDUE%Q >6-( ;< -() /6! CDUE%Q -() ()+*-( %?%-)/ >+% +*$)+!? "L)$M
8*"%)*? 6#-),$+-)! 6#-" -(6% %-$.8-.$) -($".,( -() H.$!)#)! +#! 6**M)V.6KK)! -" -+N) "# + cc8$+%( K$"M
A:;<Q + K+$+%-+-+* $)%K"#%6H*) 9"$ H"-( -() 9"$/.M ,$+//)bb& J)%K6-) -(6% >)+N#)%%Q + /+%%6L) %8+*6#,
*+-6"# +#! 6/K*)/)#-+-6"# "9 ;< K"*68?& J)%K6-) .K "9 -() ;< K$",$+//) 9"**">)! >(68( ,)#)$+-)!
9$)V.)#- 8(+#,)% -" 6-% 6#%-6-.-6"#+* ("/) +#! + + *+$,) 6#d"> "9 )T-)$#+* 9.#!6#,& I/K*)/)#-+-6"#
8"/K*)T 6#%-6-.-6"#+* %-$.8-.$) 9"$ %)$L68) !)*6L)$? %.e)$)! /+#? K$"H*)/% +% + $)%.*- 6#8*.!6#, K""$
8"/K+$)! -" -() "-()$ 8+%) %-.!? 8".#-$6)%Q -() +!/6#6%-$+-6"#Q $)8"$! N))K6#, +#! )L+*.+-6"#Q +#!
A:;< (+% $)/+6#)! -() N)? "$,+#6%+-6"# "L)$ +/H6-6".% -+$,)- %)--6#, 0R(+#Q CDDS7& '()
0W.)!!+#+Q CDDS7& I- (+% H))# /+6#*? -($".,( -() K$",$+//) 8"**+K%)! >6-( -() 8(+#,) "9 ,"L)$#M
K.H*68 %)8-"$ -(+- 8"#-$+8)K-6L)% (+L) H))# !6%-$6H.M /)#- +#! >+$ >6-( P+#,*+!)%( 6# CDXC& 1#!)$
-)! 0XXY 6# CD557 0Z"%% !" #$%Q CDD[7& '() ;< K$"M <$)%6!)#- P(.--"Q -() ,"L)$#/)#- "9 <+N6%-+#
,$+//) 6# =*,)$6+ (+% H))# *"8+-)! 6# -() @AB H)8+/) N))# -" )TK+#! -() #+-6"#+* ;< K$",$+//)
%6#8) -() )+$*? CDXE% >()# -() \$%- P6$-( 9$"/ -() /6! CDXE%Q +,+6# -($".,( + $+K6! %8+*6#,M
Z),.*+-6"# ])#-$) "K)#)! 6# =*,6)$%& P? CDX5 -()$) .K "9 %.88)%%9.* *"8+* K$",$+//)%& =% + $)%.*- "9
>)$) CUE 8)#-$)% 6# "K)$+-6"#Q +#! 6# CD5^ ^DD K"*6-68+* -.,M"9M>+$% H)->))# 9)!)$+* +#! K$"L6#86+*
8)#-$)%Q )+8( K$"L6!6#, 8"#-$+8)K-6L)% +% K+$- "9 -() +.-("$6-6)%Q (">)L)$Q -() ;< K$",$+//) /"L)!
#+-6"#+* ()+*-( %)$L68) 02-)K()#Q CDD[7& J)%K6-) -() 9$"/ -() <"K.*+-6"# O)*9+$) J6L6%6"# "9 -() @AB
)T6%-)#8) "9 -()%) 9+86*6-6)%Q (">)L)$Q /"%- >"/)# 6# CDXUQ -" -() @6#6%-$? "9 <*+##6#, +#!
0UEY7 "H-+6#)! -()6$ "$+* 8"#-$+8)K-6L)% 0-() /"%- J)L)*"K/)#- 6# CD5EQ -" K$"L6#86+* F.$6%!68-6"# 6#
9$)V.)#-*? .%)! /)-("!7 9$"/ -() K$6L+-) %)8-"$ CD54Q -" -() @6#6%-$? "9 <"K.*+-6"# O)*9+$) 6# CD5D
!.$6#, -() /6! CD5E% 0R".+".86Q CDDS7& A#) $)+%"# 0Z"H6#%"# !" #$%Q CD5C7& =/"#, -() 6#6-6+-6L)% K.-
9"$ -(6% /+? H) -() K""$ 8"L)$+,) "9 K$6/+$? 8+$) 9"$-( !.$6#, -(6% K)$6"! >)$) -() 8"#-6#.".% /"-6LM
9+86*6-6)% K+$-68.*+$*? 6# $.$+* +$)+%& '() )/K(+%6% +-6"# %8()/)Q 8"#-$+8)K-6L) 6#.#!+-6"# %8()/)
H? -() ,"L)$#/)#- "# 8.$+-6L) 8+$)Q H.6*!6#, "9 0CDX4aXX7Q %"86+* /+$N)-6#, 0CD5S7 +#! %"86+* +8-6"#
.$H+# ("%K6-+*% +#! -$+6#6#, "9 ()+*-( K)$%"##)* K$",$+//) 0CDD^7&
!.$6#, -() CDXE% /+? (+L) /)+#- *6/6-)! 6/M = K""$*? !)L)*"K)! ()+*-( 6#9$+%-$.8-.$) >+% +*%"
K*)/)#-+-6"# "9 ;< %)$L68)% +- -() &'$#(#) +#! +# 6/K"$-+#- 9)+-.$) 6# P+#,*+!)%(Q +#! )+$*?
*#'+#) *)L)*%& I- 6% #"-+H*) -(+- :WA% (+L) #"- +--)/K-% +- 6/K*)/)#-6#, ;< K$",$+//)% >)$) #"-
"K)$+-)! ;< %)$L68)% 6# =*,)$6+& %.88)%%9.*& '" "L)$8"/) -(6% K$"H*)/Q -()
J)%K6-) 6-% ,$)+-)$ >)+*-( 9$"/ "6* $)L)#.)% <"K.*+-6"# ]"#-$"* +#! ;+/6*? <*+##6#, J6L6%6"#
!.$6#, -() CDXE%Q 9.#!6#, *)L)*% 9"$ ;< 6# =*,)$6+ >+% 8$)+-)! 6# CDXS >6-(6# -() @6#6%-$? "9 B)+*-(
(+% H))# 9+$ *">)$ -(+# 6# '.#6%6+& O(6*) 6- 6% !6_M +#! <"K.*+-6"# ]"#-$"*& 2.KK"$-)! H? )T-)$#+*
8.*- -" )%-6/+-) -() +/".#- "9 ,"L)$#/)#- 9.#!6#,Q 9.#!6#,Q -() !6L6%6"# >+% (6,(*? L)$-68+* 6# %-$.8-.$)
H)8+.%) "9 -() *+8N "9 + %)K+$+-) H.!,)- 9"$ ;< ?)- >6-( -() +6/ "9 K$"L6!6#, ;< +% K+$- "9 )T6%-6#,
>6-(6# @]B .#-6* -() CD5E%Q L"8+* "KK"%6-6"# H? @]B %)$L68)%& I/K"$-+#-*?Q .#*6N) <+N6%-+#Q +
K"*6-68+* *)+!)$% -" -() 8"#-$"* "9 K"K.*+-6"# ,$">-( %-$"#, $"*) >+% ,6L)# -" :WA% +- -(6% )+$*? %-+,)
!.$6#, -(6% K)$6"! %.,,)%-% -(+- -() +/".#-% >)$) -" 6/K*)/)#- ;< %)$L68)% *"8+**? +#! \** ,+K% 6# -()
#"- %.H%-+#-6+*& 2/+** +/".#-% "9 K"K.*+-6"# +%%6%-M >)+N ()+*-( %?%-)/& P? CDDEQ +$".#! C[E :WA%
+#8) >)$) K)$6"!68+**? K$"L6!)! 9$"/ )T-)$#+* >)$) K$"L6!6#, ;< %)$L68)% 6# P+#,*+!)%(Q /+#? 6#
!"#"$%Q #"-+H*? 1:;<=Q 9"$ !)/",$+K(68 %-.!6)% 8"**+H"$+-6"# >6-( -() @ABQ +#! >)$) )%-6/+-)! -"
+#! 8)#%.%)%& I- >+% #"- .#-6* CD5DQ (">)L)$Q -(+- H) %.KK*?6#, [EY "9 8"#-$+8)K-6L) .%)$%
+#? %6`)+H*) %./ 0123D /6**6"#7 >+% ,6L)#Q -"KK)! 0@+(/""!Q CDDS7&
.K >6-( +# +!!6-6"#+* CEY "9 -(6% %./ H? -() ,"LM J6e)$)#8)% 6# 9.#!6#, (+% +*%" H))# )L6!)#- 6#
)$#/)#- 0R".+".86Q CDDS7& I# 8"/K+$6%"#Q 9.#!6#, -() ->" 8".#-$6)%& I# <+N6%-+# 9.#!6#, (+% H))#
9"$ -() A:;< 6# '.#6%6+ (+% H))# %.H%-+#-6+**? $)*+-6L)*? .#%-+H*) +#! (+% #)L)$ $)+8()! -() *)L)*%
(6,()$ 9$"/ H"-( ,"L)$#/)#- +#! )T-)$#+* %".$8)%Q +8(6)L)! 6# P+#,*+!)%(& P)->))# CDXXa5[Q -() ,"LM
#"-+H*? 9$"/ -() O"$*! P+#NQ 1:;<=Q <"K.*+-6"# )$#/)#- %(">)! *6--*) \#+#86+* 8"//6-/)#- -"
]".#86* +#! I<<;& ;"$ )T+/K*)Q O"$*! P+#N *"+#% )6-()$ ;< "$ -() ()+*-( %)8-"$ +% + >("*)& =%
408
(Q) L@ a// (, 0'I
!"#$% &'())* +,-./01 2234/ $55$66-%7 $66"8$.-"%0 5"6-8B $%$6B0-0 ,/0/$,84 -0 %//;/; ." A/../, :%;/,\
9", .4/ 0"8-$6 0/8.",0 &-%86:;-%7 ,/5,";:8.-</ 4/$6.4* 0.$%; 4"+ 5"6-8B #$?-%71 9", E> $%; ".4/, $,/$0 "9
$,/ $ ,/=/8.-"% "9 5"6-.-8$6 ,/5,/0/%.$.-"% -% 4/$6.41 8$% A/ 9:,.4/, 0.,/%7.4/%/;@ K/05-./ .4/ -%\
>$?-0.$%@ >:A6-8 5"6-8B -0 9",#:6$./; AB1 $%; -% .4/ ./,%$.-"%$6 5"6-8B 04-9. 9,"# 5"5:6$.-"% 8"%.,"6 ."
-%./,/0.0 "91 $ .-%B /6-./ +4-84 ;/C%/0 5,-",-.-/0 $%; ,/5,";:8.-</ 4/$6.4 0-%8/ .4/ '((M J%./,%$.-"%$6
$55,"5,-$./0 ,/0":,8/0 $88",;-%76BDD@ 34/ #$-% `"%9/,/%8/ "% >"5:6$.-"% $%; K/</6"5#/%. -%
0":,8/ "9 9:%;-%7 9", E> 4$0 8"#/ 9,"# /F./,%$6 `$-,"1 .4/0/ -00:/0 ,/#$-% ,/6/<$%.@ J%8,/$0-%7 5"6\
0":,8/01 %".$A6B 9,"# GHIJK &L4$%1 '((M*@ J% -.-8$6 0/%0-.-<-.B ." $ +-;/, ,$%7/ "9 4/$6.4 8$,/
N$%76$;/041 0/5$,$./ 5:A6-8 $88":%.0 +/,/ 0/. :5 -00:/01 8":56/; +-.4 .4/ 7,/$./, ./84%-8$6 8"#56/F-.B
9", E> $%; 4/$6.4 -% '(OMPOQ1 +-.4 .4/ 7"</,%#/%. "9 8"#5,/4/%0-</ ,/5,";:8.-</ 4/$6.4 0/,<-8/0 .4$%
05/%;-%7 $A":. RS "9 -.0 ;/</6"5#/%. A:;7/. "% .,$;-.-"%$6 E>1 8$% "%6B 0/,</ ." 4-746-74. .4/ %//;
5"5:6$.-"% $%; E> :%.-6 '()OP))1 ,-0-%7 ." M@MS -% 9", 8$,/9:6 $../%.-"% ." .4/ 5,"8/00/0 $%; $8.",0
,/8/%. B/$,0 &T$4#"";1 '((M*@ J% 8"#5$,-0"%1 9", -%<"6</; -% 5"6-8B ;/</6"5#/%.@
/F$#56/1 J%;"%/0-$ 4$0 05/%. $,":%; "%/ 5/, 8/%.@
J#5",.$%.6B1 .4/0/ 9:%;0 4$</ A//% 0:55",./; AB !"#$%&'()*(+($,-_34-0 ,/0/$,84 +$0 9:%;/; AB .4/
6$,7/, $#":%.0 "9 /F./,%$6 $-; .4$% ,/8/-</; AB G%-./; !$.-"%0 >"5:6$.-"% E:%; $%; a"%;"% H84""6 "9
>$?-0.$%1 -%86:;-%7 9":, U",6; N$%? 6"$%0 &9", [B7-/%/ $%; 3,"5-8$6 T/;-8-%/@ 34/ $:.4",0 +":6; 6-?/
." $8?%"+6/;7/ .4/ 8"%.,-A:.-"% "9 ;/.$-6/; 8$0/ 0.:;-/0
/F$#56/1 GHVWX' #-66-"% 9", '((YP(W*@ $%; <$6:$A6/ -%0-74.0 AB .4/ 9"66"+-%7 %$.-"%$6 ,/0/$,84\
34/ 9":, 5$-,0 "9 8":%.,-/0 04"+0 .4$.1 +4-6/ /,0b a/;-<-%$ `$,-%" &>4-6-55-%/0*1 !/A-4$ c:/;;$%$
.4/,/ -0 %" 0-%76/ ",7$%-0$.-"%$6 9",#:6$ 9", E> &3:%-0-$*1 IB/04$ L4$% &>$?-0.$%*1 I6- L":$":8-
5,"7,$##/0 .4$. /%0:,/0 /Z/8.-</ -#56/#/%.$.-"%1 &I67/,-$*1 a:#A$ L$6:#A$ &d$#A-$*1 e$-0:6 T$4#"";
-%0.-.:.-"%$6 $%; C%$%8-$6 0.$A-6-.B ;-0.-%7:-04/; &N$%76$;/04*1 a/"%$,; T$</%/?$ &d-#A$A+/* $%; L:$
U"%7A""%0-% &34$-6$%;*@
0.,"%7 9,"# +/$? 5,"7,$##/0@ I 86/$,6B ;/0-7%$./;
-%0.-.:.-"%$6 4"#/ "</, .-#/ $%; 6"%7 ./,# C%$%8-$6
8"##-.#/%. +/,/1 "9 8":,0/1 $ ,/=/8.-"% "9 4-74 ()*)()#!)&
6/</6 5"6-.-8$6 8"##-.#/%.@ ["+/</,1 -%0.-.:.-"%$6
`$,-%"1 a@ &'((M* >"5:6$.-"% >"6-8B -% .4/ >4-6-55-%/0@
$%; C%$%8-$6 0.$A-6-.B1 -% .:,%1 9$8-6-.$./; .4/ 5"6-8B G%5:A6-04/; 8$0/ 0.:;B@
5,"8/00 .4,":74 +4-84 E> 5,"7,$##/0 8":6; A/ E-%?6/1 f@ &'(OY* 34/ 5"6-.-8$6 /%<-,"%#/%. "9 5"5:6$.-"%
-%.,";:8/; $%; 0:0.$-%/;@ 8"%.,"6 -% J%;-$ $%; >$?-0.$%@ J% .%'/,/"0' 1"/($"( 0$)
.%23'0,/%$ 1,3)/(-1 /;0 e@ `6-%."%1 U@ E6$04 $%; e@
c";+-%@ a/F-%7."% N""?01 a/F-%7."%@
!"#!$%&'"# c,-%;6/1 T@ $%; 34"#$01 f@ &'(('* .34'/" 56%/"(- 0$)
.%'/"7 560$*(8 96( .%'/,/"0' :"%$%+7 %; <(;%=+ /$
34-0 5$5/, 4$0 0":74. ." 0:556/#/%. .4/ "%7"-%7 >(?('%2/$* 5%3$,=/(-@ f"4%0 ["5?-%0 G%-</,0-.B >,/001
;/A$./ $#"%7 ;/#"7,$54/,0 "% .4/ -#5$8. "9 N$6.-#",/@
5"5:6$.-"% 5"6-8-/0 $%; E> 5,"7,$##/0 "% 9/,.-6-.B c:/;;$%$1 !@ &'((M* >"5:6$.-"% >"6-8-/0 $%;
>,"7,$##/0b K/./,#-%$%.0 $%; `"%0/^:/%8/0 -%
;/86-%/ AB -;/%.-9B-%7 9$8.",0 .4$. #$B 0:55",. ", 3:%-0-$ &'(QWP'((Y*@ G%5:A6-04/; 8$0/ 0.:;B@
-%4-A-. .4/ $;"5.-"% "9 E> 5"6-8-/0 $%; 5,"\ [$#1 `@ $%; [-661 T@ &'()M* 96( 2%'/"7 2=%"(-- /$ ,6(
7,$##/0@ G0-%7 8"#5$,$.-</ 5"6-8B $%$6B0/0 -% +%)(=$ "02/,0'/-, -,0,(@ [$,</0./, U4/$.04/$91 N,-74."%@
9":, 5$-,0 "9 8":%.,-/01 -. 4$0 A//% 04"+% .4$. .4/ [/,%$%;/g1 K@f@ &'()M* 13""(-- %= ;0/'3=(@ A0+/'7 2'0$$/$*
2=%*=0++(- /$ ,6( 96/=) B%=')@ c,//%+""; >,/001
0.,/%7.4 "9 8"##-.#/%. AB 7"</,%#/%.0 ." E> 4$0 a"%;"%@
A//% 04$5/; AB ;-Z/,/%8/0 -% .4/ 5,"8/00 AB +4-84 ["7+"";1 N@ $%; c:%%1 a@ &'()M* .%'/"7 0$0'7-/- ;%= ,6(
E> 5"6-8-/0 4$</ A//% -%-.-$./;1 9",#:6$./; $%; -#\ =(0' &%=')@ hF9",; G%-</,0-.B >,/001 hF9",;@
56/#/%./;] .4/ 05/8-C8 $8.",0 -%<"6</; -% .4-0 5,"\ ["5?-%01 3@ $%; H-$#+-g$1 e@ &'()Q* `"%</,7-%7 9",8/0b
>,"8/00/0 6/$;-%7 ." .4/ ,/.4-%?-%7 "9 5"5:6$.-"% 5"6-8B
8/00] $%; .4/ 8"%./F. +-.4-% +4-84 .4-0 5,"8/00 4$0
-% d$#A-$@ J% C--3(- /$ D0+4/0$ >(?('%2+($,1 /;0 L@
.$?/% 56$8/@ h0/-\[+/;-/ $%; T@ !;:6"1 55@ OYP(W@ h#/%$%$ >,/001
34/ "</,$66 8"%86:0-"% "9 .4/ 0.:;B -0 .4$. 4"+ d$-,/@
5"6-8-/0 $,/ #$;/1 $%; +4" #$?/0 .4/#1 $,/ /^:$66B L$6:#A$1 a@ &'((M* K/./,#-%$%.0 $%; `"%0/^:/%8/0 "9
-#5",.$%. ." +4$. 5"6-8-/0 $,/ #$;/@ T:84 "9 .4/ >"5:6$.-"% >"6-8-/0 $%; >,"7,$##/0 -% d$#A-$@
G%5:A6-04/; 8$0/ 0.:;B@
5"5:6$.-"% 5"6-8B 6-./,$.:,/ 4$0 9"8:0/; "% .4/ 8"%\ L-%7;"%1 f@ &'()M* !*($)0-E 0',(=$0,/?(- 0$) 234'/" 2%'F
./%. "9 5"6-8-/0 _ 9", /F$#56/1 +4$. 8"%.,$8/5.-"%0 /"/(-@ a-..6/ N,"+% $%; `"#5$%B1 N"0."%@
." 5,"<-;/1 +4-84 0"8-$6 7,":50 ." .$,7/.1 4"+ ." L4$%1 I@ &'((M* >"5:6$.-"% >"6-8-/0 $%; >,"7,$##/0 -%
;/6-</, ,/5,";:8.-</ 4/$6.4 0/,<-8/0 $%; 0" "%@ >$?-0.$%@ G%5:A6-04/; 8$0/ 0.:;B@
34/0/ $,/ 86/$,6B <-.$6 5"6-8B ^:/0.-"%0@ ["+/</,1 L":$":8-1 I@ &'((M* 34/ I67/,-$% >"5:6$.-"% >"6-8B@
G%5:A6-04/; 8$0/ 0.:;B@
.4/,/ ,/#$-%0 $% -#5",.$%. 7$5 -% .4/ 5"5:6$.-"% a//1 L@1 a:041 a@1 U$6.1 c@ $%; `6/6$%;1 f@ &'((Q*
$%; 4/$6.4 5"6-8B 6-./,$.:,/ "% .4/ 0.,$./7-/0 .4$. .%23'0,/%$ .%'/"/(- 0$) .=%*=0++(-8 >(,(=+/$0$,- 0$)
5"6-8B #$?/,0 8$% :0/ ." -%.,";:8/1 ;/</6"5 $%; 5%$-(G3($"(- /$ :/*6, >(?('%2/$* 5%3$,=/(-@ a"%;"%
8$,,B ":. %/+ ", 84$%7/0 -% 5"6-8B@ 34-0 5$5/, 4$0 H84""6 "9 [B7-/%/ $%; 3,"5-8$6 T/;-8-%/1 G!
>"5:6$.-"% E:%;1 a"%;"%@
;/08,-A/; .4,// 5"./%.-$6 9/$.:,/0 "9 0:84 $ 0.,$./7B a//1 L@ $%; U$6.1 c@ &'((Q* a-%?-%7 %$.-"%$6 $%; 76"A$6
_ 8"$6-.-"%0 "9 5"6-8B /6-./01 05,/$;-%7 "9 5"6-8B 5"5:6$.-"% $7/%;$0b `$0/ 0.:;-/0 9,"# /-74. ;/</6"5-%7
,-0?1 $%; -%0.-.:.-"%$6 $%; C%$%8-$6 0.$A-6-.B@ T",/ 8":%.,-/0@ 96/=) B%=') H30=,(='7 +,&Y*1 YQOPYOY@
409
U12#A: BA177#75 #7 A?J #7$?2" $?L7&'#"9 -S-
!"#$%&"'( )* +,-.-/ ! "#$%&'&()*+ &%%'#&", (# %#-)". &/6 I$%-)"&()#/( "49 M* !""&" 174 C* GA12* 0A1'"74?7
&/&-.0)01 2+&-(, "&'+ %#-)". )/ 3#4' /&()#/0* 0123'#45" ;'"99( _H@?'4*
67#8"'9#&: ;'"99( 0123'#45"* M?3#79?7( P*( D%1%( )* 174 D%1%( I* +,-Q,/ ]%"
!#2( <* +,-.=/ >#9&?': ?@ @12#A: BA177#75 #7 &%" @12#A: BA177#75 B'?5'122" #7 ;1\#9&17V P%1& J"7&
;%#A#BB#7"9* C7 5'++6#$ &/6 7,)-)%%)/+0 7#%4-&()#/ J'?75b I/(+'/&()#/&- 5&$)-. 7-&//)/: 7+'0%+"()*+0 %$ QST
8#/('#-( "4* D*E* F#&1A#17? <?'?9B"( GBB"74#H G* I"J -Y*
K1: ;L3A#9%"'9( )17#A1* M?99( E*( )1LA4#7( P*;*( <'""7( D*M* 174 0??\"( W*
)1%2??4( M* +,--N/ K"&"'2#717&9 ?@ ;?BLA1&#?7 ;?A#$#"9 +,--Y/ 5&$)-. 7-&//)/: &/6 8,)-6 94'*)*&- 7'#:'&$$+0
174 ;'?5'122"9 #7 O175A14"9%* 67BL3A#9%"4 $19" J&0 &00+00+6 )/ >??>K( ]13A" ,[* ;?BLA1&#?7 0?L7$#A(
9&L4:* I"J c?'\*
)1LA4#7( P*;* 174 O"'"A9?7( O* +,-.Q/ 0?74#&#?79 ?@ @"'&#R D&"B%"7( P* +,--Y/ 7')$&'. 2+&-(, 8&'+ )/ (,+ !'&L
A#&: 4"$A#7" #7 4"8"A?B#75 $?L7&'#"9( ,-=ST.S* 9(46)+0 )/ M#'-6* D?2"'9"& >?L9"( D?2"'9"&*
P1A&( <* +,--S/ 2+&-(, 7#-)".H !/ )/('#64"()#/ (# 7'#"+00
5&$)-. 7-&//)/: !+S/( Q-T,NQ*
&/6 7#G+'* ^"4 O??\9( !?74?7*
)1LA4#7( P*;* 174 M?99( E* +,--,/ U12#A: BA177#75 B'?R
P1A&( <* 174 <#A9?7( !* +,--N/ M"@?'2#75 &%" %"1A&% 9"$R
5'122"9V WX?'&9 174 '"9LA&9( ,-QYTQ-* 9(46)+0 )/ 5&$)-.
&?' #7 4"8"A?B#75 $?L7&'#"9V ]%" $"7&'1A '?A" ?@ B?A#$:
7-&//)/: ""+=/( ZS[TZ=.* 171A:9#9* 2+&-(, 7#-)". &/6 7-&//)/: !$ ZSZTZ.[*
)18"7"\1( !* +,--N/ U12#A: ;A177#75 ;?A#$: P1'J#$\( K* +,-QY/ N)((+' 7)--01 7#%4-&()#/ 7#-)")+0 &/6
K"&"'2#717&9V ]%" ^#2313J" 019"* 67BL3A#9%"4 $19" ;,+)' I$%-+$+/(&()#/ )/ <):,( E+*+-#%)/: 8#4/(')+0*
9&L4:* 0123'#45" 67#8"'9#&: ;'"99( 0123'#45"*
I?217( _* +,-QQ/ ;,+ 7#-)()"&- <"#/#$. #3 7&=)0(&/ P?753??79#7( `* +,--N/ ;?BLA1&#?7 ;?A#$: 174
>?@ABCD* `;C ;'"99( !?74?7* ;'?5'122"9 #7 ]%1#A174* 67BL3A#9%"4 $19" 9&L4:*
;'#&$%"&&( !*>* +,--N/ K"9#'"4 @"'&#A#&: 174 &%" #2B1$& ?@ P?'A4 O17\ +,-Q-/( O)$L&LG+ 7#%4-&()#/ 9+"(#' F+%#'(*
B?BLA1&#?7 B?A#$#"9* 7#%4-&()#/ &/6 E+*+-#%$+/( F+*)+G CMOK( P19%#75&?7( K0*
"#$ ,TS=* M?3"'&9( >* +,-QN/ ]%" B?A#&#$9 ?@ GA5"'#17 9?$#1A#92* C7
M?3#79?7( P* 174 M1$%1B1"&1:1\?2( E* +,--Z/ ]%" '?A" P#'(, !3')"&1 8#/(+$%#'&'. 7#-)()"0 &/6 <"#/#$)"
?@ 5?8"'72"7& BA177#75 #7 ]%1#A174a9 @"'&#A#&: 4"$A#7"* E+*+-#%$+/(( "49 M* !1JA"99 174 G* U#74A1:* 0'??2
C7 ;,+ F+*#-4()#/ )/ !0)&/ 5+'()-)(.1 E)$+/0)#/H 8&40+0 >"A2( !?74?7*
410
The Incidence of Public Spending on Healthcare:
Comparative Evidence from Asia
Owen O’Donnell, Eddy van Doorslaer, Ravi P. Rannan-Eliya,

Aparnaa Somanathan, Shiva Raj Adhikari, Deni Harbianto,
Charu C. Garg, Piya Hanvoravongchai, Mohammed N. Huq,
Anup Karan, Gabriel M. Leung, Chiu Wan Ng, Badri Raj Pande,
Keith Tin, Kanjana Tisayaticom, Laksono Trisnantoro,
Yuhui Zhang, and Yuxin Zhao
The article compares the incidence of public healthcare across 11 Asian countries and
provinces, testing the dominance of healthcare concentration curves against an equal
distribution and Lorenz curves and across countries. The analysis reveals that the
distribution of public healthcare is prorich in most developing countries. That distri-
bution is avoidable, but a propoor incidence is easier to realize at higher national
incomes. The experiences of Malaysia, Sri Lanka, and Thailand suggest that increas-
ing the incidence of propoor healthcare requires limiting the use of user fees, or
protecting the poor effectively from them, and building a wide network of health
facilities. Economic growth may not only relax the government budget constraint on
propoor policies but also increase propoor incidence indirectly by raising richer
individuals’ demand for private sector alternatives. JEL Codes: H22, H42, H51.
Owen O’Donnell (corresponding author) is an assistant professor of quantitative methods at the

University of Macedonia, Greece; his email address is ood@uom.gr. Eddy van Doorslaer is a professor
of health economics at Erasmus University, the Netherlands; his email address is vandoorslaer@few.eur.
nl. Ravi P. Rannan-Eliya is director of the Institute for Health Policy in Sri Lanka; his email address is
ravi@ihp.lk. Aparnaa Somanathan is a fellow at the Institute for Health Policy, Sri Lanka; her email
address is aparanaa@ihp.lk. Badri Raj Pande is the director of the Nepal Health Economics Association;
his email address is neil@info.com.np. Shiva Raj Adhikari is a researcher at the Nepal Health
Economics Association; his email address is sssadhikari@yahoo.com. Laksono Trisnantoro is a
professor of health policy at Gadjah Mada University, Indonesia; his email address is trisnantoro@
yahoo.com. Deni Harbianto is a researcher at Gadjah Mada University, Indonesia; his email address is
d_harbianto@yahoo.com. Charu C. Garg is a health economist at the World Health Organization; her
email address is gargc@who.int. Piya Hanvoravongchai is a researcher at the International Health Policy
Programme, Thailand; his email address is piyaorn@ihpp.thaigov.net. Kanjana Tisayaticom is a
researcher at the International Health Policy Programme, Thailand; her email address is kanjana@ihpp.
thaigov.net. Mohammed N. Huq is a lecturer at Jahangirnagar University, Bangladesh; his email
address is m_nazmul@proshikanet.com. Anup Karan is a Takemi Fellow at the Harvard School of
Public Health; his email address is akaran@hsph.harvard.edu. Gabriel M. Leung is a professor of
THE WORLD BANK ECONOMIC REVIEW, VOL. 21, NO. 1, pp. 93 – 123 doi:10.1093/wber/lhl009
Advance Access Publication 24 January 2007
# The Author 2007. Published by Oxford University Press on behalf of the International Bank
for Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions,
please e-mail: journals.permissions@oxfordjournals.org
93
411
94 THE WORLD BANK ECONOMIC REVIEW
Propoor public spending on healthcare and other services is a stated objective

of national governments and international agencies. It is central to the mission
of the World Bank and is a key component of the Heavily Indebted Poor
Countries Initiative and the International Monetary Fund’s Poverty Reduction
and Growth Facility. Motivations include redressing inequity in the distribution
of healthcare, reducing health inequality, and raising the human capital of the
poor and thereby the growth potential of the economy. In low-income
countries, where administrative constraints on redistribution through cash
transfers are particularly binding, a subsidiary justification for public spending
on healthcare may be the alleviation of poverty and the reduction of inequality
(Besley and Coate 1991). The validity of these arguments for public spending
on healthcare rests on the empirical question of whether the spending is in fact
targeted to the poor.
Benefit incidence analysis identifies the recipients of public spending in
relation to their position in the income distribution. Benefit incidence studies,
many conducted by the World Bank, generally find that public spending on
healthcare in developing countries is not concentrated on the poor (van de
Walle 1995; Castro-Leal and others 2000; Mahal and others 2000; Sahn and
Younger 2000; Filmer 2003). Most of these studies have been conducted on an
ad hoc basis, with relatively little attention to consistency in methods.
Limitations in the comparability of the evidence make it difficult to draw
lessons about the economic, political, and health system characteristics that
explain greater and lesser success in targeting health spending to the poor.
This article presents comparable evidence on the incidence of public health
spending using consistent methods across eight Asian countries (Bangladesh,
India, Indonesia, Malaysia, Nepal, Sri Lanka, Thailand, and Vietnam) and
three Chinese provinces or regions (Gansu, Heilongjiang, and Hong Kong
Special Administrative Region). Dominance tests are used to determine
whether the distribution of public healthcare deviates significantly from perfect
equality. Many indicators show that poorer individuals are generally less
healthy (Gwatkin and others 2003) and, one may presume, in greater need of
healthcare. From an egalitarian perspective an equitable distribution of health-
care demands that resources be concentrated on the poor. Evidence that the
translational public health at the University of Hong Kong; his email address is gmleung@hku.hk. Keith
Tin is a researcher at the University of Hong Kong; his email address is tinyiukei@hkusua.hku.hk. Chiu
Wan Ng is a lecturer at the University of Malaya, Malaysia; her email address is chiuwan.ng@ummc.
edu.my. Yuxin Zhao is a professor of health economics at the National Health Economics Institute,
China; her email address is yuxin.zhao@cnhei.edu.cn. Yuhui Zhang is a researcher at the National
Health Economics Institute, China; his email address is zyh@nhei.cn. The authors thank three
anonymous referees and the editor for valuable comments. The European Commission International
Research Cooperation with Developing Countries (INCO-DEV) program (ICA4-CT-2001-10015)
funded the Equity in Asia–Pacific Health Systems (Equitap) project from which this article derives. The
Health, Welfare, and Food Bureau of the government of the Hong Kong Special Administrative Region
funded the analysis for Hong Kong. A supplemental appendix to this article is available at http://wber.
oxfordjournals.org/.
412
O’Donnell and others 95
poor do not receive their population share of health spending would be suffi-
cient to reject equity in the allocation of public healthcare. While the main jus-
tification for public provision of healthcare is likely to be its impact on the
level and distribution of population health, redistribution of living standards
may be a further motivation in largely informal economies that are constrained
in the execution of tax and cash transfer policies.1 To assess the redistributive
impact of public health spending, its distribution is compared with the Lorenz
curve of household income.
One limitation of many previous benefit incidence studies is the crudeness of
the unit cost data used to value services (van de Walle 1998; Sahn and
Younger 2000). This study derives costs from detailed health accounts, avail-
able for most of the countries and provinces, which document public expendi-
tures across health services, facilities, and regions. This allows examination of
whether conclusions about the incidence of public healthcare are sensitive to
analysis of use or expenditure data.
Data and methods are described in the next section and results are presented
and discussed in section II. The findings are summarized in section III.
I. DATA AND METHODS
The objective is to estimate and assess the distribution of public healthcare in

relation to economic status. For each country data are from recent health or
socioeconomic surveys that provide information on both use of public health-
care and a suitable measure of living standards (see table S-1 in the supplemen-
tal appendix, available at http://wber.oxfordjournals.org/). All are nationally
representative except for the surveys of Chinese provinces. The preferred proxy
for living standards is household ( per adult equivalent) consumption, which
includes the value of goods produced by the household for its own consump-
tion and a use-value of housing and durable goods.2 Household expenditure,
rather than consumption, is used for Hong Kong SAR, where household
production is much less significant. For Malaysia the only available measure of
living standards included in the health survey is household income, which is
likely to understate the living standards of rural households. It is, however, the
measure that has been used in previous incidence studies of Malaysia
(Meerman 1979; Hammer, Nabi, and Cercone 1995).
Distributions of three categories of public healthcare—hospital inpatient
care, hospital outpatient care, and nonhospital care—are examined.
1. In Latin America cash transfers are increasingly used to affect the distribution of income, as well
as that of health and education services, but this is less so in the low-income economies of Asia, where
in-kind transfers, such as healthcare, continue to predominate.
2. The equivalence scale used is eh ¼ (Ah þ 0.5Kh)0.75, where Ah is the number of adults in
household h, and Kh is the number of children 0–14 years old. Parameter values were set on the basis
of estimates summarized in Deaton (1997, pp. 241– 70).
413
Nonhospital care is an aggregate of visits to doctors, polyclinics, health

centers, and antenatal care (table S-2). For inpatient care the recall period is 12
months, except in Bangladesh (3 months) and Sri Lanka (2 weeks). For all
other care the recall period is generally 2 weeks to 1 month, except in
Bangladesh where it is 3 months.
Use data do not capture variations in the quality of services received across
facilities and geographic locations. This is a potentially important deficiency
given evidence of marked quality differences favoring richer neighborhoods
even within a single city, such as Delhi, India (Das and Hammer 2005). The
service-specific non-negative public subsidy received by an individual can be
defined as:
ð1Þ Ski ¼ maxð0; qki ckj # fki Þ
where qki is the quantity of service k used by individual i, ckj is the unit cost of
providing k in region j where i resides, and fki is the amount paid for k by i.
Where possible, variations in costs by facility (local, district, teaching hospital)
and service (inpatient/outpatient) are taken into account. Unit costs are com-
puted as:
TREkj
ð2Þ ckj ¼ P
qki wi
i[j
where TREkj is total recurrent public expenditure and wi is an expansion factor

that inflates sample use to population Puse. The total public subsidy received by
an individual is computed as Si ¼ k ak Ski , where the ak terms are scaling
factors that standardize use recall periods across services.
National health accounts, available for Bangladesh, the Chinese provinces,
Hong Kong SAR, Sri Lanka, and Thailand, are used to disaggregate expenditure
figures by facility, service, and region. Full accounts are not available for India,
Indonesia, Malaysia, Nepal, and Vietnam. For India unit subsidies computed for
another benefit incidence study are used (Mahal and others 2000). These are
specific to 960 subgroups (three facilities, 16 major states, urban–rural
residence, gender, and five income quintiles). For Indonesia public health expen-
diture review figures allow expenditures to be disaggregated for each of 30
provinces. For Malaysia expenditure data were disaggregated to five levels of
public hospital care, but geographic disaggregation was not undertaken since the
use data could not be analyzed by this dimension. Incomplete health accounts
for Nepal allow disaggregation by hospital and nonhospital care by region. For
Vietnam public accounts and hospital costing estimates were used to compute
unit costs by service and facility but not by region (World Bank 2001).
Subtraction of the user payment from equation (1) to get the net benefit of
the service is appropriate provided that quality is not responsive to the
414
payment. This is an untestable assumption with the available data. For China,
India, Indonesia, Malaysia, Nepal, and Sri Lanka either the survey data do not
contain information on payments made by individuals for public health ser-
vices or the data are not considered sufficiently reliable, for example, because
payments for public and other care are likely to be confused. For these
countries it is assumed that all users in a particular region pay the same
charge for a given service. Waiting and travel time also reduce the net benefit
from care and should, in principle, be valued and subtracted in computing the
subsidy. The survey data do not permit this, however. As a consequence,
benefits to the rural poor, in particular, may be overstated to the extent
that they travel long distances to access better quality care. By contrast, the
cost of waiting time will be less for the poor if time is valued according to
wage rates.
The incidence of public healthcare is described by its concentration curve,
which plots the cumulative proportion of healthcare use and subsidy against
the cumulative proportion of the population ranked by household consumption
per adult equivalent. To establish whether the subsidy is propoor, in the sense
that lower income individuals receive more of the subsidy than the better-off, a
test is conducted of whether the concentration curve dominates (lies above) the
458 line. Whether the poorest 20 percent of individuals consume more than 20
percent of healthcare is also tested. Dominance of the concentration curve over
the Lorenz curve of household consumption is tested to establish whether
spending on public healthcare reduces inequality.
For the dominance tests standard errors of the ordinates of curves and of
differences in ordinates are computed, allowing for dependence between curves
where appropriate (Bishop, Chow, and Formby 1994; Davidson and Duclos
1997).3 A multiple comparison approach to testing is adopted (Beach and
Richmond 1985; Bishop, Formby, and Thistle 1992), with the null defined as
curves being indistinguishable. This is tested against both dominance and cross-
ing of curves (Dardanoni and Forcina 1999). The null is rejected in favor of
dominance if there is at least one significant difference between the ordinates
of two curves in one direction and no significant difference in the other direc-
tion across 19 evenly spaced quantile points from 0.05 to 0.95. The null is
rejected in favor of crossing if there is at least one significant difference in each
direction. The 5 percent level of significance is used with critical values from
the studentized maximum modulus distribution to allow for the joint nature of
the test (Beach and Richmond 1985).4
An alternative dominance test consistent with the intersection–union prin-
ciple (Kaur, Rao, and Singh 1994; Howes 1996), which has been used in the
3. The computation is carried out in Stata.

4. Dardanoni and Forcina (1999) show that the probability that this test will falsely reject the null
in favor of dominance does not exceed the significance level and report Monte Carlo evidence
suggesting that the actual significance level is well below its nominal value.
415
benefit incidence literature (Sahn and Younger 2000; Sahn, Younger, and
Simler 2000), takes nondominance as the null and tests this against the alterna-
tive of strict dominance. This is a conservative test that requires statistically
significant differences in ordinates at all points of comparison for the null to be
rejected. Dardanoni and Forcina (1999) present Monte Carlo evidence showing
that while this test reduces the probability of falsely rejecting nondominance to
a negligible value, compared with the multiple comparison approach it has
greatly reduced power of detecting dominance when true. Given these results,
most weight in the discussion below is given to the results from the multiple
comparison tests, but discrepancies with the more conservative intersection–
union test are pointed out.
I I. RE S U LTS
In Hong Kong SAR, Malaysia, and Thailand the concentration curve of the
total public health subsidy dominates both the Lorenz curve and the 458 line of
equality (table 1, final column), indicating that the subsidy is both
inequality-reducing and propoor. With the exception of the comparison with
the 458 line in the case of Thailand, these dominance results are robust to use
of the stricter test. In Sri Lanka an equal distribution of the total subsidy is not
rejected. In relative terms this shifts the distribution of living standards toward
the poor, as the concentration curve dominates the Lorenz curve. In the
remaining countries and provinces the concentration curve of the total subsidy
is dominated by the 458 line but, with the exceptions of India and Nepal, dom-
inates the Lorenz curve. That is, the subsidy is prorich but inequality reducing.
For Bangladesh and the two Chinese provinces nondominance relative to both
the Lorenz curve and the 458 line cannot be rejected when the more conserva-
tive intersection–union test is employed.5
The degree to which the public health subsidy is targeted to the poor can be
seen more explicitly by examining the share of the subsidy received by the
poorest 20 percent of individuals (table 2). Public healthcare is clearly most
propoor in Hong Kong SAR, with the poorest fifth of the population receiving
almost two-fifths of the total subsidy (table 2, final column). In Malaysia the
poorest quintile also receives significantly more than 20 percent of the total
subsidy, but the propoor bias is much less than it is in Hong Kong SAR. In Sri
Lanka and Thailand the poorest quintile’s share of the total subsidy does not
differ significantly from 20 percent. In the remainder of countries and pro-
vinces, with the exception of Bangladesh, the poorest 20 percent of individuals
receive significantly less than 20 percent of the public health subsidy. The share
going to the poorest 20 percent of individuals is lowest in Nepal, at less than 7
percent, followed by the two Chinese provinces, at 8 –10 percent. In these
5. Concentration and Kakwani indices, which provide summary measures of the magnitude by
which the concentration curve deviates from the 458 line and the Lorenz curve, are given in table S-3.
416
T A B L E 1 . Tests of Dominance of Concentration Curves for Public Health Service Use and Subsidy against the Lorenz Curve
and the 45 Degree Line of Equality
Use Subsidy
Hospital Hospital Hospital Hospital

inpatient outpatient Nonhospital inpatient outpatient Nonhospital Total
8 8 8 8 8 8
Country, province, or region Lorenz 45 Lorenz 45 Lorenz 45 Lorenz 45 Lorenz 45 Lorenz 45 Lorenz 458
Bangladesh 2 þ 2 þ þ 2 þ þ –
Gansu, China þ 2* þ* 2 n.a. n.a. þ 2 þ 2 n.a. n.a. þ 2
Heilongjiang, China 2* þ* 2 n.a. n.a. þ 2 þ 2 n.a. n.a. þ 2
Hong Kong SAR þ* þ* þ* þ* þ* þ* þ* þ* þ* þ* þ* þ* þ* þ*
India 2* þ* 2* þ* þ 2* þ* þ* þ* 2*
Indonesia 2 2* x 2* þ* þ 2* 2* 2 2* þ* þ þ* 2*
Malaysia þ* þ þ* þ* þ* þ* þ þ* þ* þ* þ* þ*
Nepala þ 2* n.a. n.a. þ* 2 2* n.a. n.a. x 2* x 2*
Sri Lanka þ* þ* þ þ* þ* þ* n.a. n.a. þ*
Thailand þ* þ* x þ* þ* þ* þ* þ þ* þ* þ* þ
Vietnam þ 2* 2* þ* þ þ* 2 þ 2* þ* þ þ* 2
Blank cell indicates failure to reject the null hypothesis that curves are indistinguishable using the multiple comparison test (Bishop, Formby, and
Thistle 1992) at the 5 percent significance level.
x indicates rejection of the null hypothesis that curves are indistinguishable in favor of curves crossing using the same test.
þ /2 indicates rejection of the same null hypothesis in favor of dominance using the same test. A þ indicates that healthcare is more concentrated on
the poor than is household consumption per adult (Lorenz) or equal per capita distribution (458), while a – indicates that it is less concentrated.
*indicates rejection of the null hypothesis of nondominance in favor of an alternative of strict dominance using the intersection – union test (Howes
1996) and a 5 percent significance level. Dominance is in the direction indicated by the þ or – , as above.
n.a. means that data were not available to conduct the test.
a
The results in the hospital inpatient columns refer to both inpatient and outpatient.
O’Donnell and others
Source: Authors’ calculations based on survey data documented in table S.1 (see supplemental appendix available at http://wber.oxfordjournals.org/).
99
417
418
100
T A B L E 2 . Share of Total Household Consumption and Public Healthcare Subsidy Received by Poorest Quintile
of Individuals (percent)
Hospital care
Country, province, Household consumption
or region per adult equivalent Inpatient Outpatient Nonhospital care Total subsidy
Bangladesh 7.25* (0.0437) 15.20 (6.3732) 11.60* (1.8853) 24.42 (5.5695) 16.78 (3.4916)
Gansu, Chinaa 5.24* (0.0695) 7.27* (1.5331) 9.57* (1.6473) n.a. 8.17* (1.2265)
Heilongjiang, Chinaa 5.98* (0.0759) 6.57* (1.8184) 12.32* (2.5677) n.a. 10.47* (1.8729)
Hong Kong SAR 6.82* (0.0377) 38.77* (3.2580) 38.68* (2.2048) 38.19* (1.7718) 38.73* (2.7463)
India 10.50* (0.0083) 10.70* (1.1086) 18.59 (1.6219) 26.23* (1.5471) 12.49* (0.9553)
Indonesia 9.77* (0.0078) 3.80* (0.3762) 5.77* (0.4857) 19.73 (0.3199) 13.46* (0.2582)
THE WORLD BANK ECONOMIC REVIEW
Malaysia 7.20* (0.0370) 21.19 (0.8807) 18.72 (1.1208) 32.25* (1.3422) 22.95* (0.6921)
Nepalb 8.05* (0.0534) 3.52* (1.4851) 3.52* (1.4851) 9.04* (1.7220) 6.64* (1.1780)
Sri Lankac 8.31* (0.0725) 20.76 (2.6013) 21.11 (1.9418) n.a. 20.88 (1.8367)
Thailand 6.94* (0.0589) 21.26 (1.4144) 17.70* (1.0278) 31.16* (1.9137) 20.06 (0.8963)
Vietnam 8.78* (0.0429) 13.64* (1.9209) 11.55* (1.7049) 19.73 (1.7346) 14.79* (1.5416)
*Significantly different from 20 percent at the 5 percent significance level. Bold indicates that the subsidy share is significantly different from the
household consumption share.
n.a. means that data were not available to conduct the test.
Note: Numbers in parentheses are standard errors.
a
There are no data on nonhospital care, but low-level hospitals, equivalent to polyclinics and health centers, are included.
b
It is not possible to distinguish between hospital inpatient and outpatient visits.
c
The subsidy specific to nonhospital care cannot be computed.
Source: Authors’ calculations based on data documented in table S.1 (see supplemental appendix available at http://wber.oxfordjournals.org/).
cases, and in Bangladesh, India, and Indonesia, the richest quintile receives
more than 30 percent of the total subsidy (not shown in table). In all cases but
Nepal the share of the subsidy going to the poorest quintile is significantly
greater than its share of total household consumption.
Differences in Incidence across Health Services
Only in Hong Kong SAR does the concentration curve dominate the 458 line
for both hospital inpatient and outpatient care and for nonhospital care (see
table 1), with the poorest quintile receiving about 39 percent of the subsidy to
all three services (see table 2). In Malaysia the concentration curves for inpati-
ent and nonhospital care lie above the 458 line, but the outpatient care curve
does not deviate significantly from the line of equality (see table 1). In
Thailand it is inpatient care that is equally distributed, while the concentration
curves for the other types of care dominate the diagonal, at least using the less
stringent test criteria. However, in both Malaysia and Thailand the poorest
quintile receives significantly more than 20 percent of the subsidy only for non-
hospital care (see table 2). In Sri Lanka there is equality in the distributions of
all services except for a propoor distribution of outpatient care as measured by
use (see table 1). In the remainder of countries and provinces, concentration
curves for hospital care tend to lie below the diagonal—meaning that the
better-off consume more—while the curves for nonhospital care lie above it.
The poorest quintile fairly consistently receives less than 20 percent of the
subsidy for hospital care and significantly more than 20 percent of the subsidy
for nonhospital care only in India (see table 2).
For most countries and provinces the distribution of nonhospital care domi-
nates that of hospital inpatient and outpatient care (table 3), confirming that
nonhospital care is generally more targeted to the poor than is hospital care.
Comparison of Use and Subsidy Distributions
Estimating the incidence of the public healthcare subsidy requires much more
information than that of raw use. Unit costs must be estimated at the facility
and regional levels and, where appropriate and possible, fees paid by individ-
uals must be identified. The effort involved to obtain this extra information is
worthwhile only if there is significant variation in unit costs or fees with the
indicator of household living standards and if this covariance is sufficiently
large relative to that for use. The dominance tests reported in table 1 display a
considerable consistency across the use and subsidy measures. Only in 10 of 58
pairwise comparisons do the conclusions of the test differ depending on
whether the distribution of use or the subsidy is examined. This is not an
insubstantial degree of disagreement, but it suggests that the results of domi-
nance tests are generally robust to the measure over which incidence is exami-
ned and that variation in use, not unit subsidies, is the main driver of the
public subsidy distribution. This increases the confidence that can be placed in
studies that look only at use. It is consistent with the findings of Sahn and
419
420
102
T A B L E 3 . Tests of Dominance between Concentration Curves for Different Public Health Services and between Use and
Subsidy Distributions
Use Use and subsidy
Country, province, Hospital inpatient Hospital inpatient Hospital outpatient Hospital Hospital
or region versus outpatient versus nonhospital versus nonhospital inpatient outpatient Nonhospital
Bangladesh op.ip
Gansu, China op.ip* n.a. n.a. use.subsidy n.a.
Heilongjiang, China op.ip n.a. n.a. n.a.
Hong Kong SAR use.subsidy
India op.ip* non-h.ip* non-h.op* subsidy.use subsidy.use
Indonesia op.ip non-h.ip* non-h.op use.subsidy* use.subsidy* use.subsidy
Malaysia non-h.ip non-h.op use.subsidy* use.subsidy* n.a.
Nepal non-h.(ip þ op)a non-h.(ip þ op)a use.subsidy

Sri Lanka op.non-h use.subsidy use.subsidy n.a.
Thailand non-h.ip non-h.op* subsidy.use
Vietnam non-h.ip* non-h.op* subsidy.use*
ip is inpatient, op is outpatient, non-h is non hospital.

Blank cell indicates failure to reject the null hypothesis that curves are indistinguishable using the multiple comparison test at the 5 percent
significance level.
. indicates that the null hypothesis is rejected in favor of dominance, for example, op . ip indicates that outpatient care is more propoor than
inpaticare and use . subsidy indicates that the use distribution is more propoor than the subsidy distribution.
*indicates rejection of the null hypothesis of nondominance in favor of an alternative strict dominance in the direction indicated by ., as above, using
the intersection –union test and a 5 percent significance level.
a
Test is between all hospital care (inpatient and outpatient) and all nonhospital care.
Source: Authors’ calculations based on survey data documented in table S.1 (see supplemental appendix S.1 available at http://wber.oxfordjournals.
org/).
Younger (2000) but somewhat stronger, since the current study allows for
more sources of heterogeneity in unit subsidies.
Notwithstanding this result, there are significant differences between the
distributions of use and subsidy. In Indonesia, Malaysia, and Sri Lanka the
use distributions dominate—they are more propoor than the subsidy
distributions—for all services, and in Gansu, Hong Kong SAR, and Nepal this
is true for some services (see table 3). Dominance is not always found using the
more conservative test, however. Urban–rural and regional differences in the
quality of care are the most likely reason that the subsidy is less propoor than
use. Only in India, Thailand, and Vietnam does the subsidy distribution domi-
nate the use distribution for certain services, indicating that the subsidy per
unit of care falls as household consumption rises. This is likely due to user
payments rising with household consumption, whether because of exemptions
granted to the poor or because richer households are paying for higher quality
care that is not reflected in the unit cost figures.
Cross-Country Comparisons
As would be expected from the results already presented, the subsidy concen-
tration curve of Hong Kong SAR dominates that of all other countries and
provinces (table 4).6 The incidence of public care is so skewed toward the poor
that the distribution of total healthcare ( public and private) in Hong Kong
SAR is propoor (Leung, Tin, and O’Donnell 2005).7 While this is in striking
contrast with the distribution of healthcare in the low- and middle-income
countries examined in this article, it is consistent with the distribution that pre-
vails in most high-income economies (Van Doorslaer, Masseria, and Koolman
2006).
There are no significant differences between the concentration curves of
Malaysia, Sri Lanka, and Thailand, where the subsidies range from slightly
propoor to evenly distributed. On the less strict test the Vietnamese distri-
bution is dominated by that of Hong Kong SAR, Malaysia, and Thailand and
it is indistinguishable from that of Sri Lanka. It dominates the subsidy distri-
butions of all the remaining countries and provinces using the less stringent
test.8 For most pairwise comparisons the subsidy concentration curves of
Bangladesh, Gansu, Heilongjing, India, Indonesia, and Nepal are indistinguish-
able. Exceptions are that India and Indonesia dominate Gansu and Nepal
using the less strict test. In all these countries and provinces the public health
subsidy is significantly and substantially prorich (see tables 1 and 2). This is
6. See table S-4 for cross-country dominance tests for each type of health service subsidy.
7. Some 43.5 percent of total expenditure on health in Hong Kong SAR is funded from private
sources (Hong Kong Domestic Health Accounts 1999–2000).
8. This is not due simply to the fact that unit subsidies are negatively correlated with household
consumption in Vietnam, unlike in most other countries and provinces. Only one cross-country
dominance result for Vietnam becomes insignificant when use of each service rather than the subsidy to
each service is examined.
421
422
104
T A B L E 4 . Cross-Country Dominance of Public Health Subsidy Concentration Curves

Malaysia Thailand Sri Lanka Vietnam Bangladesh Indonesia India Gansu Heilongjiang Nepal
Hong Kong SAR D* D* D* D D* D* D* D* D* D*

Malaysia n.s. n.s. D D D* D* D* D* D*
Thailand n.s. D D D* D* D D* D*
Sri Lanka n.s. ns D D D D* D*
Vietnam D D* D D* D D*
Bangladesh ns ns ns ns ns
Indonesia ns D ns D
India D ns D
Gansu, China ns ns
Heilongjiang, China ns
n.s. indicates failure to reject the null hypothesis that the curves are indistinguishable using the multiple comparison test at the 5 percent significance
level.
D indicates rejection of the null in favor of dominance (more propoor) of the row country over the column country by the same test.
*indicates that the intersection – union test rejects the null of nondominance against the alternative of strict dominance at the 5 percent significance
level.
Source: Authors’ calculations based on survey data documented in table S.1 (see supplemental appendix available at http://wber.oxfordjournals.org/).
consistent with the findings of the majority of benefit incidence studies con-
ducted in developing countries (van de Walle 1995; Castro-Leal and others
2000; Mahal and others 2000; Sahn and Younger 2000; Filmer 2003). But
Malaysia, Thailand, Sri Lanka, and to a lesser extent Vietnam stand out as
exceptions to this norm of prorich bias. Why is it that public healthcare is
more propoor in these four countries than it is in other developing countries of
Asia and elsewhere?
National income is an obvious candidate to explain cross-country variation
in the targeting of public health spending. Public healthcare is strongly targeted
to the poor in Hong Kong SAR in large part because Hong Kong is rich
enough to afford a dual system of universal public healthcare funded from
general taxation and a private healthcare system used predominantly by the
better-off to bypass the bottlenecks and inconveniences of the public system. It
is surely no coincidence that Malaysia and Thailand are the only other two
countries where public health spending is significantly propoor. While they are
not nearly as rich as Hong Kong SAR, they are considerably better off than the
other countries included in this study (see table S-5).
Economic development is not the sole explanation for cross-country differ-
ences in the incidence of public healthcare. It does not explain why Sri Lanka,
despite a lower GDP per capita than Indonesia, achieves a distribution of
health resources that is much more favorable to the poor. Levels of public
spending on health and health system characteristics might be expected to
explain part of the residual cross-country variation in targeting of the poor. In
per capita terms Sri Lanka spends 2.5 times as much as Indonesia on public
healthcare (table S-5). The scale of public spending may influence its incidence
by affording a wider geographic distribution of public health facilities and so
bring services closer to poor, rural populations.
There may also be a trickle-down effect. At low levels of spending the politi-
cally powerful, higher income urban elite may be more successful than the
rural poor in capturing spending for programs that meet their own needs. As
spending levels rise and more of the health needs of higher income groups are
satisfied, additional programs can be better targeted to the needs of the poor
(Lanjouw and Ravallion 1999). Countering this tendency, the pressure from
higher income groups for prioritization of tertiary-level city hospitals may be
maintained by the attraction of continuing advances in medical technology
(Victora and others 2000).
The extent to which higher income groups claim the benefits from public
healthcare will depend on whether an attractive private sector alternative
exists. Income-elastic demand for healthcare quality, in particular amenities
and convenience of service, will lead to greater substitution of private for
public care by an expanding middle-class as the economy grows. Hammer,
Nabi, and Cercone (1995) argue that this mechanism was largely responsible
for the increased propoor incidence of public health spending in Malaysia
between the mid-1970s and the mid-1980s. The private sector continues to
423
grow in Malaysia, driven in part by dissatisfaction with the responsiveness of

the public system (Shepard, Savedoff, and Phua 2002). In Thailand, which
has also achieved impressive economic growth in recent years, the private
sector is also expanding rapidly (Towse, Mills, and Tangcharoensathien
2004).
The combination of (near) universal public provision, a private sector
offering an attractive alternative, and incomes that make demand for this
alternative effective leads to redistribution through public provision in the
way that theory predicts (Besley and Coate 1991). This mechanism implies a
possibly uncomfortable tradeoff between the quality of public healthcare and
the extent to which it is targeted to the poor. In lower income countries,
such as Bangladesh, India, and Indonesia, separation of low- and high-
income groups into the public and private sectors is constrained not only by
the limited purchasing power of the middle class but also by marked intra-
sectoral quality differentials. There is evidence of pronounced income gradi-
ents in the quality of private sector care used in India (Das and Hammer
2005). There, as in Bangladesh, the poor make extensive use of unqualified
private providers.
This discussion suggests that economic development, the scale of public
health spending, and the availability and quality of private sector alternatives
may each help explain cross-country variation in the incidence of public health
spending. Regression analysis is used to examine whether this is the case across
the study countries and provinces and others for which benefit incidence results
are available from other studies (Filmer 2003). Only 24 observations are avail-
able for this analysis, and so the results (table 5) should be treated with due
caution. It is an exploratory exercise and not an empirical test of hypotheses.
The dependent variable is the (log) percentage of the total public subsidy
received by the poorest quintile. This share increases significantly with GDP
per capita, with an elasticity of about 0.3. At a lower level of significance (10
percent), the poorest quintile’s share is also increasing with public health
spending as a percentage of GDP, with an elasticity of about 0.5. So, for a
given GDP there is some evidence that the share of the subsidy going to the
poor is increasing with the scale of public health spending.
To examine whether, for a given level of public expenditure, the share of the
subsidy going to the poor increases with use of private sector alternatives,
public spending as a percentage of total expenditure on health is included in
the regression. Consistent with the hypothesis, the coefficient is negative but
does not reach conventional levels of significance. The regression residuals are
largest, in absolute value, for the two Chinese provinces. Public health spend-
ing in these provinces is much less targeted on the poor than would be
expected given GDP and the scale of public spending and its share of total
health financing. This is most likely due to the extensive imposition of user
charges with no income-related exemptions. Excluding these two provinces
increases the magnitude and significance of the coefficients. In particular, the
424
T A B L E 5 . Cross-Country Regression Analysis of Targeting of the Public

Health Subsidy (Dependent variable: log of percentage of public health subsidy
received by poorest quintile)
Excluding Gansu and
Full sample Heilongjiang
Robust standard Robust standard

Coefficient errora Coefficient errora
Log of gdp per capitab 0.3214*** 0.1002 0.3426*** 0.0889

Public health expenditure 0.2337* 0.1190 0.2971*** 0.0884
as percent of gdp
Public health expenditure 20.0080 0.0049 20.0110** 0.0043
as percent of total health
expenditure
Eastern Europe and 20.3308 0.2091 20.4895** 0.1889
Central Asia
Latin America and 20.2478 0.3535 20.4338 0.2990
Carribean
Sub-Saharan Africa 20.8630*** 0.3004 21.0750*** 0.2093
Constant 0.0691 0.7465 0.0294 0.7118
Sample size 24 22
R2 0.5712 p-value 0.7421 p-value
RESET (F3,n-k-3) 0.76 0.5371 0.71 0.5671
*Significant at the 10 percent level; **significant at the 5 percent level; ***significant at the 1
percent level.
Note: Observations are the 11 countries and provinces for the years of this study plus those
from Filmer (2003): Armenia (1999) Bangladesh (1995), Bulgaria (1995), Costa Rice (1992),
Cote d’Ivoire (1995), Ecuador (1998), Georgia (2000), Ghana (1994), Guinea (1994), Honduras
(1995), Nicaragua (1996), South Africa (1994), and Vietnam (1993).
a
Robust to heteroscedasticity of general form.
b
Gross domestic product per capita in purchasing power parity dollars at constant 2000
prices.
source: Dependent variable, authors’ calculations based on data in table S. 1 (see
supplemental appendix available at http://wber.oxfordjournals.org/) and that reported in Filmer
(2003). GDP, World Bank, various years, World Development Indicators. Health expenditure,
WHO, various years, National Health Accounts and World Health Report Statistical Annexes.
negative coefficient on the public health financing share becomes significant at

5 percent.9 Although this study has found that the public health subsidy is not
targeted on the poor in the majority of the 11 Asian countries and provinces
examined, the distribution appears to be even more skewed toward the better-
off in Eastern Europe and Central Asia and in Sub-Saharan Africa.10
9. The results are similar if the weight given to observations with large absolute residuals is reduced,
but not set to zero, using robust regression. The results are also robust to the exclusion of Hong Kong
SAR, where the subsidy is much more propoor and GDP is much higher than in the other countries and
provinces.
10. Other potential explanatory factors, including the Gini coefficient, the urbanization rate, and
the doctor supply rate, were not found to be significant.
425
These regression results tell only of associations in a fairly small sample of

countries and should not be interpreted as causal effects. GDP may be acting as
a proxy for a number of primary determinants of incidence, such as the quality
of governance and preferences for redistribution. Through human capital
acquisition, assuming that the marginal product of investments in health is
higher for poorer (and sicker) individuals, GDP may itself be responsive to the
targeting of healthcare to the poor. Polices are of course endogenous. The
positive correlation between the scale and the propoor incidence of public
spending may derive from the degree of political commitment to reaching the
poor. Reducing racial conflict in post-independence Malaysia was a major
motivation for the expansion in access to healthcare and the channeling of
public resources to the rural Malay population (Hammer, Nabi, and Cercone
1995). The early adoption of democracy and female suffrage in Sri Lanka
contributed to the high priority given to healthcare and the wide geographical
distribution of health resources in response to the lobbying of local politicians
(McNay, Keith, and Penrose 2004). In fact, a 1928 commission proposed
the full enfranchisement of women at the same time as men as a means of
securing a political lobby for the prioritization of healthcare (Rannan-Eliya
2001). High rates of female literacy and a relatively high degree of female
autonomy have raised awareness of maternal and child health problems,
leading to high rates of use of modern health facilities and medicines (Caldwell
1986).
Political and economic circumstances determine the motivation and
resources for the pursuit of propoor public healthcare, but realization of the
objective depends on the specific health sector policies adopted. One policy has
been to minimize charges for poor patients in accessing care. There are
virtually no fees for public health services in Sri Lanka, and fees are minimal in
both Hong Kong SAR and Malaysia (table S-6). In all three cases fees are
not retained by facilities or even by the health sector, but accrue to general
revenues, thus undermining providers’ incentives for generating fee revenue.
The near avoidance of user fees in resource-poor Sri Lanka has been feasible
only by driving down unit costs (Rannan-Eliya 2001). Nonmonetary
incentives, such as professional development and opportunities to work
simultaneously in the private sector, help maintain high levels of staff
productivity. In Thailand fees have been much higher. Prior to the introduction
of universal coverage in 2001, public hospitals received 20–50 percent of
their revenue from user fees (Towse, Mills, and Tangcharoensathien 2004).
But the disincentive effect on use by the poor was limited through a fairly
effective healthcard scheme that covered about two-thirds of the poor.
Crucially, this scheme compensated providers for fee exemptions from a
designated budget.
A geographically dispersed network of health facilities close to the rural
population also appears to contribute to the propoor targeting of health spend-
ing. In Malaysia half the population lives within 10 kilometers of a public
426
hospital and within 4.6 kilometers of a public clinic.11 In Sri Lanka most of
the population has lived within 5 kilometers of a healthcare facility since the
early 1970s, and most of the rural population is within 5 –10 kilometers of a
peripheral facility (Hsiao 2000). In Thailand, although beds and doctors are
highly concentrated in Bangkok, an extensive rural infrastructure has been
developed over decades. There are primary care health centers in all subdis-
tricts and community hospitals in all districts (Towse, Mills, and
Tangcharoensathien 2004). The introduction of universal coverage has initiated
a major shift of resources from urban hospitals to primary care. Vietnam also
has a relatively high level of provision in rural areas through a comprehensive
network of commune health centers.
But the contribution of primary care to propoor public health spending
should not be exaggerated. Public health spending is better targeted on the
poor in Hong Kong SAR, Malaysia, Thailand, Sri Lanka, and Vietnam because
the distribution of hospital care is more favorable to the poor and not because
more resources are devoted to nonhospital care (see table S-3). Of course, hos-
pitals differ. In Malaysia and Sri Lanka many hospitals are small in scale and
not particularly well equipped. But their wide geographic distribution makes
them accessible to the rural poor. In many other low-income countries, such as
Bangladesh, resources are more concentrated in large, well-equipped hospitals
in urban centers that are inaccessible to the poor.
III. CONCLUSION
The analysis reveals substantial variation across Asia in the incidence of public
subsidies for healthcare. Public spending is strongly propoor in high-income
Hong Kong SAR. The total public health subsidy is more moderately propoor
in low- to middle-income Malaysia and Thailand and it is evenly distributed in
low-income Sri Lanka. At a still lower level of national income the subsidy is
mildly prorich in Vietnam. In the remainder of the low-income countries and
provinces examined, which account for the far greater share of the Asian popu-
lation, the better-off receive substantially more of the subsidy than do the
poor. In most cases there is prorich bias in the distribution of hospital care,
while nonhospital care is propoor. A greater share of the healthcare subsidy
goes to hospital care, and so this dominates the overall distribution. While
public health subsidies are typically not propoor, they are inequality reducing
in all cases except India and Nepal.
Most within- and between-country dominance tests are robust to whether
the distribution of healthcare use or the value of the subsidy is examined. This
is a reassuring result since the health accounts data required for analysis of
subsidy incidence are often unavailable and raw use data must be relied on.
There are, however, significant differences between the distribution of
11. Authors’ calculations from the 1996 National Health and Morbidity Survey.
427
healthcare use and healthcare subsidies, with use often more propoor. Where
this occurs, the likely explanation is urban–rural and interregional differences
in the nature and funding of facilities.
The analysis shows that the prorich distribution of public healthcare subsi-
dies that is pervasive in most developing countries is avoidable but that effec-
tive targeting is easier to realize at higher levels of national incomes. The
experiences of Malaysia, Sri Lanka, Thailand, and Vietnam suggest that achiev-
ing a more propoor incidence of public health spending requires limiting the
use of user fees, or at least effectively protecting the poor from them; building
a wide geographic network of health facilities; and ensuring that hospital care,
which absorbs most spending, is sufficiently targeted at the poor.
428
APPENDIX: THE INCIDENCE OF PUBLIC SPENDING ON H E A LT H CA R E : CO M PA R AT I V E E V I D E N C E F RO M A S I A
T A B L E S 1 . Description of sample surveys
Institution
Survey Survey conducting Survey Sampling Response Sample size
Country year name survey coverage Survey design unit rate individuals
Bangladesh 1999– Health and Bangladesh Bureau National Stratified Household and 99% 56,010
2000 Demographic of Statistics Individual
Survey (HDS) (BBS)
2000
Gansu 2003 National Health Ministry of Health Gansu province Stratified, cluster Household 100% 15,535
(China) Household (poor in sample. Self-
Interview west China) weighting
Surveys
Heilongjiang 2003 Heilongjiang Health bureau of Heilongjiang Stratified, cluster Household 100% 11,572
(China) Health Heilongjiang province sample. Self-
Household province (north-east weighting
Interview Survey China)
Hong Kong April– Thematic Census and National Stratified. Household 78.4% 31,672
SAR June Household Statistics Sample (noninstitutional; (noninstitutional);
2002 Survey in the Department, weights individual 97.2%
second quarter Government of applied (institutional) (institutional)
of 2002 Hong Kong
SAR
India 1995–96 National Sample National Sample National Stratified, cluster Household 100% 629,024
Survey 52nd Survey sample.
round Organisation Weights
applied
Indonesia 2001 Socioeconomic National Board of National Stratified, cluster Household 98% 889,413
Survey Statistics sampling. Self-
(SUSENAS) weighted
(Continued )
111
429
430
112
TABLE S1. Continued

Institution
Survey Survey conducting Survey Sampling Response Sample size
Country year name survey coverage Survey design unit rate individuals
Malaysia 1996 National Health Public Health National Stratified, cluster Household 86.90% 59,903
and Morbidity Institute, sample.
Survey II Ministry of Weights
Health applied
Nepal 1995/96 Nepal Living Central Bureau of National Stratified, cluster Household 96.60% 18,855
Standards Statistics (CBS) sample.

Survey Weights
applied
Sri Lanka 1996/97 Consumer Finance Central Bank Excluded Stratified Household 98% 399,28
Survey Northern
Province due
to civil war.
Thailand Jan –June Socioeconomic National Statistical National Stratified Household 80% 17,489
2002 Survey Office
Vietnam 1998 Living Standards General Statistical National Stratified, cluster Household 70% 28,623
Survey Office sample.
Weights
applied
T A B L E S 2 . Measures of healthcare utilisation
Hospital care Nonhospital care
Polyclinic/health
Inpatients outpatients Doctor visits center Antenatal care Comments
Bangladesh
Reference period last episode in previous 3 months last episode in previous 3 months 3months Care at satellite and community
Measurement unit Number of days Number of Number of clinics also included but not
visits visits child immunisation
Gansu and Heilongjiang (China)
Reference period 12months 2 weeks n.a. 2 weeks n.a. Data on hospital care only. Five
Measurement unit Number of days Number of Number of levels of hospital are
visits visits distinguished, the lowest of
which are equivalent to
polyclinics.
Hong Kong SAR
Reference period 12 months 30 days 30 days n.a. n.a. Hospital outpatient includes
Measurement unit Number of days Number of Number of visits to specialist and A&E.
visits visits Doctor visits is general
outpatient visits.
India
Reference period 12 months 2 weeks 2 weeks 2 weeks 2 weeks
Measurement unit Number of days any visits any treatment period any visits
Indonesia
Reference period 12 months 1 month n.a. 1 month 1 month Puskesmas (inpatients and
Measurement unit Number of days Number of n.a. Number of visits, Number of outpatients) and
visits Number of days for visits supplementary Puskesmas
inpatient (outpatients) included in
health centre/polyclinic.
Polindes and Posyandu in
antenatal care.
(Continued )
113
431
432
TABLE S2. Continued
114
Hospital care Nonhospital care

Polyclinic/health
Inpatients outpatients Doctor visits center Antenatal care Comments
Malaysia
Reference period 12 months 2 weeks n.a. 2 weeks n.a.
Measurement unit Number of Number of Number of visits
admissions visits
Nepal
Reference period 30 days n.a. 30 days n.a. Data does not allow distinction
Measurement unit Number of n.a. Number of visits n.a. between hospital IP and OP
visits
Sri Lanka
Reference period 2 weeks 2 weeks 2 weeks 2 weeks n.a.
Measurement unit Any admission Any visit Any visit Any visit
Thailand
Reference period 12 months 1 month n.a. 1 month n.a. A distinction is made between
Measurement unit Number of Number of Number of visits public and private care only
admissions visits for the last 2 IP admissions
and the last episode of other
care. Assumed all care
received in same sector
Vietnam
Reference period 12 months 4 weeks n.a. 4 weeks n.a. No distinction between public
Measurement unit Number of days Number of Number of visits and private sector for IP care.
visits Since vast majority of
hospitals were public,
assumed all IP is public
IP inpatient.
OP outpatient.
n.a. not applicable.
T A B L E S 3 . Summary indices of incidence of incidence of the public healthcare subsidy
Hospital care
Inpatient Outpatient Non-hospital care Total public subsidy
Bangladesh
Concentration index 0.2325 (0.1154) 0.1356 (0.0360) 0.0474 (0.0838) 0.1588 (0.0609)
Kakwani index 20.1338 (0.0909) 20.2388 (0.0372) 20.3358 (0.0692) 20.2244 (0.0499)
Subsidy share 47.99% 25.33% 26.69% 100%
Gansu (China)
Kakwani index 20.2286 (0.0439) 20.3529 (0.0360) 20.3529 (0.0360) 20.2758 (0.0332)
Subsidy share 65.42% 34.58% 34.58% 100%
Heilongjiang (China)
Kakwani index 20.1242 (0.0652) 20.2281 (0.0510) 20.2281 (0.0510) 20.1946 (0.0424)
Subsidy share 60.09% 39.91% 39.91% 100%
Hong Kong SAR
Kakwani index 20.6919 (0.0356) 20.6491 (0.0265) 20.6173 (0.0232) 20.6831 (0.0301)
Subsidy share 82.47% 13.36% 4.17% 100%
India
Kakwani index 0.0122 (0.01928) 20.2476 (0.02113) 20.3830 (0.03281) 20.0390 (0.0165)
Subsidy share 83.68% 9.62% 6.65% 100%
Indonesia
Kakwani index 0.1752 (0.0248) 0.0880 (0.0187) 20.3142 (0.0047 20.1245 (0.0080)
Subsidy share 26.54% 14.86% 58.59% 100%
Malaysia
Kakwani index 20.4100 (0.0131) 20.3863 (0.0235) 20.3863 (0.0235) 20.4493 (0.0123)
115
(Continued )
433
434
116
TABLE S3. Continued

Hospital care
Inpatient Outpatient Non-hospital care Total public subsidy
Subsidy share 37.02% 38.53% 24.45% 100%

Nepal
Kakwani index 0.1268 (0.0605) 0.1268 (0.0605) 20.0677 (0.0487) 0.0384 (0.405)
Subsidy share 54.58% 54.58% 45.24% 100%
Sri Lanka
Kakwani index 20.3313 (0.0252) 20.4042 (0.0172) 20.4042 (0.0172) 20.3561 (0.0284)
Subsidy share 68.00% 32.00% 32.00% 100%
Thailand
Kakwani index 20.4199 (0.0317) 20.4348 (0.0242) 20.6463 (0.0335) 20.4361 (0.0210)
Subsidy share 50.74% 45.16% 4.18% 100%
Vietnam
Kakwani index 20.1495 (0.0471) 20.0599 (0.0667) 20.4623 20.2573 (0.0458)
Subsidy share 86.88% 2.13% 10.98% 100%
Robust standard errors in parentheses.

Source: Authors’ calculations from data documented in table S-1.
T A B L E S 4 . Cross-country Dominance of Public Health Subsidy Concentration Curves
Total subsidy Malaysia Thailand Sri Lanka Vietnam Bangladesh Indonesia Gansu India Heilongjiang Nepal
Hong Kong SAR D* D* D* D D* D* D* D* D* D*

Malaysia n.s. n.s. D D D* D* D* D* D*
Thailand n.s. D D D* D D* D* D*
Sri Lanka n.s. n.s. D D D* D* D*
Vietnam D D* D* D D D*
Bangladesh n.s. n.s. n.s. n.s. n.s.
Indonesia D n.s. n.s. D
Gansu (China) n.s. n.s.
India D n.s. D
Heilongjiang (China) n.s.
Hospital inpatient subsidy Malaysia Thailand Sri Lanka Vietnam Bangladesh Gansu India Heilongjiang Nepal Indonesia
Hong Kong SAR D* D* D* D D D* D* D* D* D*

Malaysia n.s. n.s. D n.s. D* D* D* D* D*
Thailand n.s. D n.s. D D* D D* D*
Sri Lanka n.s. n.s. D D* D D* D*
Vietnam n.s. D D D D* D
Bangladesh n.s. n.s. n.s. n.s. n.s.
Gansu (China) n.s. n.s. n.s. n.s.
India n.s. D D*
Heilongjiang (China) n.s. n.s.
Nepala
Indonesia D
Hospital outpatient subsidy Sri Lanka Thailand Malaysia India Gansu Bangladesh Vietnam Heilongjiang Indonesia
Hong Kong SAR D* D* D D* D* D D* D* D*

Sri Lanka n.s. n.s. n.s. D D D* D D*

Thailand n.s. n.s. D D D D D*
117
(Continued )
435
436
118
TABLE S4. Continued

Total subsidy Malaysia Thailand Sri Lanka Vietnam Bangladesh Indonesia Gansu India Heilongjiang Nepal
Malaysia n.s. D D D* D D*
India D D D* D D*
Gansu (China) n.s. n.s. n.s. D
Bangladesh n.s. n.s. D
Vietnam n.s. D
Heilongjiang (China) D
Non-hospital subsidy Hong Kong Malaysia India Vietnam Indonesia Bangladesh Nepal
Thailand D D D D D* D D*
Hong Kong SAR n.s. D D D* D* D*

Malaysia D D D D D*
India n.s. D* D D*
Vietnam D D D*
Indonesia n.s. D*
Bangladesh n.s.
Note: Countries/provinces are ranked from most to least propoor according to values of concentration indices.
Tests follow the multiple comparison approach with the null hypothesis defined as curves being indistinguishable. n.s. indicates failure to reject the
null at 5% significance.
D indicates that the subsidy concentration curve of the row country/province dominates (is more pro-poor) than that of the column country/province.
There are no cases of crossing concentration curves.
*indicates that the intersection union principle test rejects the (different) null of nondominance against the alternative of strict dominance at 5%. If
no *appears, then this test does not reject its null.
a
comparison with Nepal are for the aggregate of inpatient and outpatient subsidies.
T A B L E S 5 . National Income and Government Expenditure on Health
General government General government General government expenditure
GDP per capita, expenditure on expenditure on health on health as % total
Territory Yeara PPP $b health as % GDPc per capital, PPP $ expenditure on health
Bangladesh 1999 1495 0.98 15 27

China 2002 4568 2.26 103 42
Gansu (China) 2002 2661 2.38 63 42
Heilongjiang 2002 5434 1.48 80 36
(China)
Hong Kong 2001/02 26049 3.26 849 57
SAR
India 1996 1994 0.81 16 16
Indonesia 2001 3146 0.57 18 36
Malaysia 1996 8254 1.34 111 58
Nepal 1995/96 1179 1.20 14 24
Sri Lanka 1996/97 2951 1.63 48 50
Thailand 2000 6740 2.04 138 61
Vietnam 1998 1854 1.44 27 33
a
Year of survey used for distributional analysis.
b
GDP per capita in international $ using purchasing power parity (PPP) exchange rates. Constant year 2000 prices.
c
General government expenditure on health including social insurance.
Source: GDP per capita—World Development Indicators, World Bank. Health expenditures—National health accounts estimates, except: India,
Malaysia and Vietnam from World Health Report, Stastistical Annexes, WHO, and Nepal from (HMG/Nepal 2000 and Hotchkiss, Rous and others
1998).
119
437
438
120
T A B L E S 6 . Charges and exemptions for public healthcare

Nonpoor groups exempt from
Charged services Free Services Income/poverty related fee waivers charges
Bangladesh Secondary services (nominal Most primary care (or local Poor exempt or pay lower charge Civil servants (selected services)
registration fee for inpatient/ services); medicines within
outpatient); Inpatient care in facility; immunization; some
major hospitals reproductive healthcare
China Inpatient (including etc Family planning None Old Red Army soldiers and
medicines); Outpatient Retirees
(including medicines);
Immunisation
Hong Kong Inpatient (including medicines); Accident and emergency (until Welfare recipients exempt Civil servants and dependents
SAR outpatient (including December 2002) (reduced rate for Inpatients);
medicines); dental hospital staff and dependents
India Inpatient bed charge; Hospital consultation and certain None formally. Indirect relation to Civil sevants
outpatient registration medicines. Primary care/health income through price
charge; certain medicines; center/polyclinic consultation differentiation in inpatient care.
tests/x-rays; dental and medicines. Family planning. Informally, “poor” can be
Vaccinations and immunizations exempted partially or fully from
charges
Indonesia All medical care and medicines None Poor exempt from all charges. Charges determined at local
Indirect relation of inpatient government level. Some better
charges to income through price off local govts. Provide free
discrimination health centre care
Malaysia Hospital inpatient and Family planning and vaccinations/ Hospital directors have discretion Infants less than 1 year
outpatient. Primary care. immunizations. Outpatient ante to waive fees for destitute. (outpatient). State rulers,
Dental care. Diagnostics and and postnatal care. Treatment of Upper limit on charges for third Governors and families. Civil
x-rays infectious diseases on third class class ward patients servants (including retired)
wards. Dental care for pregnant and dependents. Local
women and pre school children authority employees and
dependents
Nepal All medical care and medicines. Emergency services; selected Poor either exempt or pay reduced None
Nominal charge for vaccines, immunization and charge but not fully
outpatient varying with reproductive health services. implemented.
facility. 60% subsidy for medicines at
Health Posts and Primary Care
centres.
Sri Lanka Family planning services. All medical and medicines except No official exemptions, but limited None
Patients occasionally asked family planning. survey evidence suggests that
to buy medicines/supplies facility staff tend to avoid asking
from private retailers when the poorest patients to self-
out of stock at facility. purchase medicines and supplies,
or ration available stocks to
them.
Thailand All medical care and medicines. Nonpersonal healthcare; EPI Poor exempted from user fees and children ,12; elderly .60;
After Oct 2001, fixed fee (30 vaccination co-payments. Informally, those public health volunteers;
Baht) UC scheme means very “unable to pay” are exempted. monks.
minimal co-payment.
Vietnam Fees for most services Outpatient services at commune Fee exemptions for individuals Families of health personnel,
introduced in 1989. health centres. who have certification of certain classes of patients (like
Medicines rarely provided indigency from neighbourhood handicapped, TB), orphans.
free of charge. or village People’s Committee.
121
439
REFERENCES
Beach, Charles M., and James Richmond. 1985. “Joint Confidence Intervals for Income Shares and
Lorenz Curves.” International Economic Review 26:439– 50.
Besley, Timothy, and Stephen Coate. 1991. “Public Provision of Private Goods and the Redistribution
of Income.” American Economic Review 81(4):979– 84.
Bishop, John A., K. Victor Chow, and John P. Formby. 1994. “Testing for Marginal Changes in Income
Distributions with Lorenz and Concentration Curves.” International Economic Review 35(2):479–88.
Bishop, John A., John P. Formby, and Paul D. Thistle. 1992. “Convergence of the South and
Non-South Income Distributions, 1969–1979.” American Economic Review 82(1):262– 72.
Caldwell, John C. 1986. “Routes to Low Mortality in Poor Countries.” Population and Development
Review 12(2):171 – 220.
Castro-Leal, Florencia, Julia Dayton, Lionel Demery, and Kalpana Mehra. 2000. “Public Spending
on Health Care in Africa: Do the Poor Benefit?” Bulletin of the World Health Organisation
78(1):66– 74.
Dardanoni, Valentino, and Antonio Forcina. 1999. “Inference for Lorenz Curve Orderings.”
Econometrics Journal 2(1):49– 75.
Das, Jishnu, and Jeffrey Hammer. 2005. “Which Doctor? Combining Vignettes and Item Response to
Measure Clinical Competence.” Journal of Development Economics 78(2):348– 83.
Davidson, Russell, and Jean-Yves Duclos. 1997. “Statistical Inference for the Measurement of the
Incidence of Taxes and Transfers.” Econometrica 65(6):1453– 65.
Deaton, Angus. 1997. The Analysis of Household Surveys: A Microeconometric Approach to
Development Policy. Baltimore, MD: Johns Hopkins University Press.
Filmer, Deon. 2003. “The Incidence of Public Expenditures on Health and Education.” Background
note for the World Development Report 2004: Making Services Work for the Poor. Washington,
D.C.: World Bank.
Gwatkin, Davidson R., Shea Rustein, Kiersten Johnson, Roshni Pande, and Adam Wagstaff. 2003.
Initial Country-Level Information about Socio-Economic Differentials in Health, Nutrition and
Population, vols.I and II. Washington, D.C.: World Bank, Health, Population, and Nutrition
Division.
Hammer, Jeffrey S., Ijaz Nabi, and James A. Cercone. 1995. “Distributional Effects of Social Sector
Expenditures in Malaysia, 1974–89.” In Walle Domenique van de, and Nead Kimberly eds., Public
Spending and the Poor. Baltimore, MD: Johns Hopkins University Press.
Howes, Stephen. 1996. “A New Test for Inferring Dominance from Sample Data.” World Bank,
Washington, D.C.
Hsiao, William. 2000. “A Preliminary Assessment of Sri Lanka’s Health Sector and Steps Forward.”
Cambridge, Mass.: Harvard School of Public Health.
Kaur, Aamrjot, B. L. S. Prakasa Rao, and Harshinder Singh. 1994. “Testing for Second-Order
Stochastic Dominance of Two Distributions.” Econometric Theory 10(5):849–66.
Lanjouw, Peter, and Martin Ravallion. 1999. “Benefit Incidence, Public Spending Reforms and the
Timing of Program Capture.” World Bank Economic Review 13(2):257–73.
Leung, Gabriel M., Keith Tin, and Owen O’Donnell. 2005. “Redistribution or Horizontal Equity in
Hong Kong’s Mixed Public-Private Health System: A Policy Conundrum.” EQUITAP Working
Paper 22, Erasmus, Rotterdam, Netherlands, and Institute for Policy Studies, Colombo.
Mahal, Ajay, Janmejaya Sing, Farzana Afridi, Vikram Lamba, Anil Gumber, and V Selvaraju. 2000.
“Who Benefits from Public Health Spending in India?” New Delhi: National Council for Applied
Economic Research.
McNay, Kirsty, Regina Keith, and Angela Penrose. 2004. Bucking the Trend: How Sri Lanka Has
Achieved Good Health at Low Cost—Challenges and Policy Lessons for the 21st Century. London:
Save the Children.
440
Meerman, Jacob. 1979. Public Expenditure in Malaysia: Who Benefits and Why? New York: Oxford
University Press.
Rannan-Eliya, Ravindra P. 2001. “Strategies for Improving the Health of the Poor—the Sri Lankan
Experience.” Health Policy Programme, Institute of Policy Studies, Colombo.
Sahn, David E., and Stephen D. Younger. 2000. “Expenditure Incidence in Africa: Microeconomic
Evidence.” Fiscal Studies 21(3):321–48.
Sahn, David E., Stephen D. Younger, and Kenneth R. Simler. 2000. “Dominance Testing of Transfers in
Romania.” Review of Income and Wealth 46(3):309–27.
Shepard, Donald S., William Savedoff, and Kai Hong Phua. 2002. Health Care Reform Initiatives in
Malaysia: Report of a Consultation with the Planning and Development Division, Ministry of
Health, Malaysia. Waltham, Mass.: Brandeis University, Shneider Insitute for Health Policy.
Towse, Adrian, Anne Mills, and Viroj Tangcharoensathien. 2004. “Learning from Thailand’s Health
Reforms.” British Medical Journal 328:103– 05.
van de Walle, Domenique. 1995. “The Distribution of Subsidies through Public Health Services in
Indonesia, 1978–87.” In Dominique van de Walle, and Nead Kimberly eds., Public Spending and
the Poor. Baltimore, MD: Johns Hopkins University Press.
van de Walle, Domenique. 1998. “Assessing the Welfare Impacts of Public Spending.” World
Development 26(3):365–79.
Van Doorslaer, Eddy, and Cristina Masseria, Xander Koolman and the OECD Health Group Equity
Research. 2006. “Inequalities in Access to Medical Care by Income in Developed Countries.”
Canadian Medical Association Journal 174(2):177 –83.
Victora, C. G., J. P. Vaughan, F. C. Barros, A. C. Silva, and E Tomasi. 2000. “Explaining Trends in
Inequalities: Evidence from Brazilian Child Health Studies.” The Lancet 356(9235):1093– 98.
World Bank. 2001. Vietnam: Growing Healthy—A Review of Vietnam’s Health Sector. Hanoi.
World Bank. Various years. World Development Indicators. Washington, D.C.
World Health Organization. Various years. World Health Report: Statistical Annexes. Geneva.
4 41
7. Action research Rigour in action research
Given the features of action research and the active role
Lucy ilson of the researcher in the process, the three key approaches
to ensuring rigour, particularly addressing the possibility
reat ritain and orthern reland of researcher bias (Meyer, 2001), are:
triangulation across data sources and rich contextu-
n
Action research is one form of emancipatory research.
alization of experience
It has a long tradition in community and organizational
researcher reflexivity
n
development work, for example, including work that
adopts a systems thinking approach (for example, Luckett member checking, that is the feedback of findings to
n
& Grossenbacher, 2003). It is also increasingly being used participants for their review and reflection.
in quality improvement work in low- and middle-income
countries (see for example, work supported by the
References
Institute for Health Improvement http://www.ihi.org/IHI/ Loewenson R et al. (2010). TARSC/ E UINET, CEGSS,
Programs/StrategicInitiatives/DevelopingCountries.htm ) SATHI-CEHAT, Experiences of participatory action research
and in health policy and systems-related work with com- in building people centred health systems and approaches
munities – such as the work on governance issues to universal coverage: Report of the Sessions at the Global
supported by the Regional Network on Equity in Health Symposium on Health Systems Research, Montreux,
in Southern and Eastern Africa (Loewenson et al., 2010). Switzerland, TARSC, Harare (http://www.equinetafrica.org
However, there are still relatively few published action /bibl/docs/GSHSR 20PRA 20report 20Dec 20
research studies. 2010.pdf, accessed 16 September 2011).
Action research is an overarching approach to research. Luckett S, Grossenbacher K (2003). A critical systems
”Essentially action research is concerned with gene- intervention to improve the implementation of a District
rating knowledge about a social system, while, at the Health System in Kwa Zulu-Natal. Systems Research and
same time, attempting to change it” (Meyer, 2001:173). ehavioural Science, 20:147–162.
Sometimes the researchers are those whose practices
Meyer J (2001). Action research. In: Fulop N et al., eds.
and actions are the subject of inquiry; sometimes exter-
Studying the organisation and delivery of health services:
nal researchers can support participants to examine their
research methods. London, Routledge:172–187.
practices and experiences, and also act as facilitators to
support the introduction of new practices or interven-
tions. Such research is always flexible in character and
responsive to participants’ changing needs as findings
are repeatedly fed back to them, reflected on and, perhaps,
acted on. Action research studies always involve multiple
methods, but are mainly qualitative in nature and are
often written up as case studies.

Overview of selected papers
Two papers were chosen for this section as they together
illuminate the approach of action research, based on the
same study.
Khresheh & Barclay (2008) report on the findings of their

action research study supporting the implementation of a
new birth record system in three Jordanian hospitals.
Subsequently, they report their reflections on their
experience in conducting this study (Khresheh & Barclay,
2007).

Khresheh R, Barclay L (2007). Practice research
engagement (PRE): Jordanian experience in three
Ministry of Health hospitals. ction Research, 5:123.
http://dx.doi.org/10.1177/1476750307077313
Reproduced by permission of the SAGE
n
Publications. Copyright SAGE Publications, 2007.
Khresheh R, Barclay L (2008). Implementation of a new

birth record in three hospitals in Jordan: a study of health
system improvement. Health Policy and Planning,
23:76–82.
n
Action Research
Volume 5(2): 123–138
Copyright© 2007 SAGE Publications
Los Angeles, London, New Delhi and Singapore
www.sagepublications.com
DOI: 10.1177/1476750307077313
ARTICLE Practice–research engagement

(PRE)
Jordanian experience in three Ministry of

Health hospitals
Reham Khresheh
Mutah University, Jordan
Lesley Barclay
Charles Darwin University, Australia
ABSTRACT
This article describes the practice–research engagement (PRE)

that occurred during an action research project conducted in
three hospitals in Jordan. The project aimed to develop and test
the feasibility of an improved clinical record-keeping system.
This article focuses on how relationships were built and evolved
over time with national and local leaders and practitioners to
facilitate the study, and how this led to a process of health sys-
tem improvement. The article draws on outcomes and analyses
from data collected in field notes, recorded interviews and focus
groups. Results showed that the PRE approach assisted people to
change as they undertook a process of clinical improvement and
KEY WORDS health systems development.
• action research
• clinical
improvement
• partnerships
• practitioners–
researchers
engagement
• system
development
123
4 45
124 • Action Research 5(2)
Background
This article describes the details of the practitioner–researcher engagement

process, undertaken by the first author and field researcher (Khresheh), to imple-
ment a shared, consolidated clinical record (the Jordanian Consolidated Birth
Record [JCBR]) within three hospitals in Jordan. We took a pragmatic quality-
improvement approach to the research. In the article we describe how practice–
research engagement was employed within a number of action research cycles to
contribute to health service improvement. The JCBR was tested in three different
hospitals in Jordan with the new record being completed by staff in addition to
the normal records for the duration of the study. In particular, we analyse
the importance and complexity of engaging project participants and building
relationships, which deepened as staff and researcher worked together towards
the ultimate success of the project. Himmelman’s (2001) framework is used to
analyse the researcher-managed process of engagement. Change processes within
action research cycles (Brown, 2001) were undertaken jointly by the field
researcher and practitioners to achieve mutual goals around practical system
improvement (Batliwala, 2003; Brown, Bammer, Batliwala & Kunreuther, 2003;
Reason & McArdle, 2006).
Problem identification and need for change

Jordan is a small developing country, located in the Middle East. The Jordanian
population is approximately 5.5 million with a birth rate of 29/1000, a death rate
of 5/1000 and a fertility rate of 3.7 (Department of Statistics, 2004). The Maternal
Mortality Rate (MMR) in Jordan is 41/100,000 live births with 82 percent of
these identified as being preventable (Nsheiwat & Al-Khalidi, 1997). Preliminary
research suggested improvements could be made in maternal and child health
data systems. Initial field visits to Jordan by both authors and their meetings with
key stakeholders in the Ministry of Health in October 2003 confirmed the need
to improve maternity care records and to develop ways of monitoring perform-
ance.
The research reported in this article was part of a larger study aimed at
testing the introduction of a new clinical record in Jordan. This was designed
to inform planning, and demonstrate accountability from the local level to a
national system of monitoring perinatal mortality and morbidity, as well as
improve clinical outcomes and organizational efficiency. The study was con-
ceived and planned by both authors in conjunction with Jordanian health leaders,
with the field research carried out by the first author who is a Jordanian national.
Other publications are in preparation including an article describing the process
of implementation of the new record as a first step toward system improvement
and a second article reporting on the testing of the new standardized record to
4 46
Khresheh & Barclay Practice–research engagement (PRE) • 125
provide regular and reliable data around birth services and allow comparison
with evidence-based practice.
There are two information systems for birthing women in Jordan held in
two parallel national records. The antenatal and postnatal records are held in
the Maternal Child Health clinics in the community where care is provided by
obstetricians, midwives and nurses. Labour and birth records are held in the
hospital where other obstetricians, midwives and nurses provide care. Currently
in Jordan there is no opportunity for women to retain copies of these records in
the form of hand-held records. As a result clinicians work without vital informa-
tion and there are no opportunities to ‘benchmark’ performance of clinicians, or
by one hospital against another hospital, or to compare Jordanian outcomes of
maternity care with international standards.
The study
The study used an action research process underpinned by Practice–Research

Engagement (PRE) (Brown, 2001). Brown describes PRE processes as: working on
a problem that requires the resources of both practitioners and researchers;
recruiting participants appropriate to the problem and the PRE process; establish-
ing shared values, goals and expectations for joint work and diagnosing institu-
tional arrangements that support or retard PRE, organizing the engagement
process to use participants’ resources effectively and learning from the process
about the issues and PRE. These principles were integrated into each action
research cycle and influenced the relationships built with participants. The action
research was also guided by literature that acknowledges that research aimed at
practical systems change cannot generate knowledge or improvement without
engaging with practitioners (Batliwala, 2003; Brown et al., 2003; Lindsey, Sheilds
& Stajduhar, 1999; Reason & McArdle, 2006).
The Ministry of Health gave approval to conduct the study in three selected
hospitals in three different areas of the country. The Ethics Committee of the
University of Technology, Sydney (UTS), also provided ethical clearance for the
research to be undertaken.
Prior to commencement of the project we conducted a baseline audit that
investigated the quality of the data kept in the regular clinical record. We then
engaged policy-makers and clinicians in the design and then the implementation
of the JCBR (Brown et al., 2003). In this process, which is described below, we
applied change theories within action research cycles during the planning, imple-
mentation and evaluation of the new clinical record (Brown, 2001).
The audit of 180 medical records of mothers from the three participating
hospitals confirmed the poor quality of the data collected in hospitals. Only
50 percent of these records were adequately completed with documentation
4 47
occurring in 18 different places throughout the record. It took an average of 90

minutes to complete the full record for each woman. Further, the records were
often completed retrospectively by a person who had not provided the care that
was described, as clinicians would complete the information when they had time
or leave this task to their senior colleagues to complete (Khresheh, 2006).
In keeping with the first principle of PRE (Brown, 2001) the record audit
data were presented at preliminary meetings with practitioners in the field.
Health leaders in the Ministry of Health, managers of the three hospitals, and
health professionals working in the maternity departments in participating hospi-
tals all agreed that high quality clinical data for the care of mothers should be
a priority and that the current poor records needed to be improved. The new con-
solidated and linked birth record, the JCBR, was considered to be a solution
that could improve the quality of record-keeping systems around birth, inform
planning, and demonstrate accountability from the local level to a national level
through monitoring perinatal mortality and morbidity. It was believed the record
would improve clinical outcomes and organizational efficiency. This new record
was based on a similar record used in the Australian health system (NSW
Department of Health, 2004).
This high level of agreement around the new record as a solution to the
problem of poor clinical records supported the rationale for the study. Partici-
pants were enthusiastic about the potential for an integrated, better designed,
shared record to assist clinicians, managers and policy-makers to improve a range
of outcomes for which they were responsible: the clinicians for improved care
of individual women, the managers for more accountable performance in their
hospital and the policy-makers for a system that reviewed services and worked to
improve health care. In exchange for the assistance of the researchers in design-
ing, implementing and evaluating the new record, participants agreed to facilitate
and contribute to finding solutions and to generate the data needed to investigate
and understand the quality improvement process.
Initial interviews and focus groups were undertaken with 36 people. These
included staff in medical records departments (n = 3), nurses and midwives (n =
15), medical directors and senior staff in hospitals (n = 12) and Ministry of Health
officials (n = 6). These initial interactions with participants in the interviews and
focus groups at national and hospital level helped identify the practitioners in
the various departments who were able to work with researchers to improve the
quality of records and manage the change process. The formation of ‘communica-
tive spaces’ through focus groups (Reason, 2004) encouraged interaction that led
to active participation. New forms of communication also developed among par-
ticipants from different disciplines, for example nursing, medical and midwifery
professionals came together for the first time, with the research providing them
with the opportunity for mutual understanding and a means to reach a shared
agreement about actions (Reason, 2004). These discussions raised the practi-
4 48
tioners’ awareness of the problem by presenting them with an analysis of their

own baseline data. This helped to build motivation for change as they realized
their own data was of poor quality.
Once the strategy for improvement was identified and agreed with health
leaders and hospital staff, practice–research engagement groups were established
at different levels of the health system and in the three different hospital settings.
These were entitled the ‘National Steering Group’, with separate groups: the
‘Local Leadership Group’ and ‘Local Action Group’. The National Steering
Group comprised experts from the Ministry of Health and included those respon-
sible for maternal and child health, quality assurance and nursing leadership. The
National Steering Group provided the field researcher with the authority to
conduct her work, assistance in planning, guidance and recommendations that
were helpful to the study. The Local Leadership Groups consisted of the man-
agers of the three selected hospitals, directors of nursing, medicine and medical
records departments. The Local Leadership Groups provided the researcher with
the authority to work with their staff, guided and assisted in her work and were
linked at a policy and professional level with the national steering group. The
Local Action Groups consisted of the health professionals who were working in
the maternity departments of the selected hospitals, including obstetricians, resi-
dent doctors, midwives and nurses, and included medical records department
workers. The Local Action Groups were supported and guided by their directors
in their work with the researcher in the implementation of the JCBR. The process
of practitioners–researcher engagement in this study and the role of the researcher
in relation to the National Steering Group, Local Leadership Groups and Local
Action Groups is described in Figure 1.
The action research cycles
Planning
Overlapping cycles of action research, diagnosing, planning, implementing and
evaluating activities were used in each setting and guided the researcher’s inter-
actions with participants (Davison, Martinsons & Kock, 2004; Meyer, 1993).
Interviews and focus groups conducted with staff from three Maternal Child
Health clinics, linked to the participating hospitals, allowed additional data to be
collected to investigate the changed record system and its impact outside hospi-
tals and were also fed back into research cycles and conclusions.
Health policy leaders, hospital managers, clinical and medical directors and
clinicians were all included in planning to ensure their co-operation and commit-
ment in achieving the aims of the study (Brown, 2001; Brown et al., 2003; Evans,
2003; Larrabee, 2004). Open communication and co-operative interactions
between researcher and practitioners on each level produced valuable feedback
4 49
National Steering
Group-MOH
Maternal & Child Quality Assurance Nursing Leaders

Health Leaders Leaders
3 Local Leadership Managers of

Groups in hospitals Hospitals
(A, B & C)
Director of Nursing Director of Medical Director of Registration

Department Department Department
3 Local Action Groups in hospitals

(A, B & C)
Midwives & Doctors Registration

Nurses Workers
Figure 1 A description of the practitioners–researcher engagement process used in this study
on the final draft of the JCBR and the process of its implementation in the field.
This included modifications of the items in the record, adding new items, specify-
ing who should complete the JCBR, the training of the health professionals in the
use of the JCBR and the process of its implementation. The items, the design of
the record, and identifying who should complete the form were decided and
agreed. The draft JCBR was reviewed, discussed and revised many times during
group meetings. All groups from the national to local levels were then invited to
provide feedback on the final draft of the JCBR before this was implemented.
‘Training’ sessions that were conducted to enable practitioners to use the
new JCBR were also motivating, engaging and consultative in the PRE sense
(Brown et al., 2003). They were planned jointly by the first researcher and the
action research groups and conducted based on their recommendations. This
included the number of training sessions to be conducted in each hospital, the
45 0
knowledge and practice that needed to be provided to health professionals and

the best time for conducting training sessions.
Implementation
The National Steering Group led the study, with members identifying participat-
ing hospitals, providing formal approval for the study and encouraging hospitals
to participate. Local Leadership Groups supported the implementation of the
JCBR in the three hospitals, facilitating the involvement of their staff in the
process of the implementation. The Local Action Groups working in the mater-
nity departments at the three hospitals were involved in the implementation of
the JCBR. The staff of the registration office, admission unit, labour room and
postnatal department in the three hospitals shared this responsibility with the
researcher. The local director encouraged staff to become actively involved in
the implementation process. This helped increase the staff’s commitment to the
implementation process and enabled continuous feedback to be included in the
evolving, shared process of the study.
The flexibility of the PRE approach and the overlapping action research
cycles assisted the researcher and practitioners in dealing with problems that
arose during the implementation of the JCBR. The frequent interaction between
the researcher and staff during the fieldwork and the co-operative relationship
that shaped this interaction created opportunities for the researchers and whole
team to reflect, analyse and make change during the implementation process.
This resulted in rapid problem solving and was used to keep staff informed and
provide supportive feedback to them. For example, the researcher found during
earlier field visits that some health professionals did not complete their sections in
the record as had been agreed. The researcher, with the co-operation of the direc-
tor of each department, conducted additional meetings with staff providing more
explanation and clarification of the process, resulting in improved compliance in
record-keeping.
The engagement between the support groups, practitioners and researcher
was organized, managed and sustained by the field researcher, balancing the
different values, goals, perspectives and capacities of the researcher and practi-
tioners. This helped limit any negative impact of unequal levels of participation,
and maintained the co-operative relationship between researcher and practi-
tioners (Brown, 2001; Brown et al., 2003). The organization of the practice
research engagement process into national and local action groups helped solve
the problems of power differences that would have arisen if these groups had
been integrated (Brown, 2001; Brown et al., 2003). This enabled open and frank
exchanges within each group unhampered by issues of status and power as group-
ings included similar levels of authority, experience and participation. Action
research groups at each level were provided with different types of support to
451
manage the change processes of the project ranging from the use of authority and
guidance to the personal involvement by clinicians as staff used the new record on
a daily basis.
Evaluation
The evaluation of the implementation of the JCBR in the three participating
hospitals began at a local level in September 2004 and finished nationally with a
meeting of leaders in May 2005. Immediate outcomes as well as longer-term
evaluations were assessed using record audits, interviews and focus groups. The
data obtained were analysed into themes on the basis of frequency and strength
of responses. Findings from the evaluation of the JCBR were fed back to groups
at each level and discussed in meetings. Initially findings were also shared with
the Local Action Groups in the three hospitals at meetings held during field visits
to each hospital. These findings and the staff reflections were then shared with
Local Leadership Groups in each hospital. Finally the findings and the reflections
of both these groups were reported to the National Steering Group in the
Ministry of Health. The ultimate results of the project were presented, by invita-
tion, at a national public forum hosted by the National Steering Group. This
inclusive process allowed the project to become the jointly owned work of the
Ministry of Health and the researcher, and indicated the ultimate success of the
practice–research engagement process (Brown, 2001).
The engagement between the researcher and the practitioners challenged
current practices, identified effective improvements and developed a tool, the
JCBR, which was based on both research and practice (Brown, 2001). The
members of all the action research groups and the researchers found the results of
the study important and promising. Relationships had been built between the
research team and the action research groups that have evolved into long-term
collaborations on national and local levels. These have been sustained subsequent
to the research being completed. One of us, the first author, has been invited to
help in a new project for the Ministry of Health that aims to improve the quality
of records in maternity care in all the hospitals in Jordan. Additionally practi-
tioners originally involved in the research are still collecting data on the JCBR and
sending this to the first author for analysis.
Outcomes and analysis
The outcomes of this study are on multiple levels and exist in the real world as the
Jordanian maternal child health system continues to evolve informed by our work.
The learning that we experienced as researchers during the study is explicated
below, as we believe it is valuable to share.
452
Formation of the practice–research engagement groups

The PRE group formation phase started in January 2004 and continued until
June 2004 during the diagnosing, or planning, phase of the early action research
cycles. It was important to have a co-operative inquiry group that consisted of
people who shared a common concern for developing understanding and practice
in a specific field. It was the field researcher’s role to create the conditions for
democratic dialogue among participants within each group (Reason & McArdle,
2006). Data generated during interviews and focus groups was also fed back into
the system and informed and stimulated representatives of the whole system in
thinking through and planning change.
Key people were identified and engaged in the process during the initial
interviews and discussions and continued to participate throughout the project.
The researcher targeted her efforts to establish, build and strengthen relationships
with the partners in the study and used networking to maximum effect in the
early stages (Himmelman, 2001). One health leader in the Ministry of Health, for
example, directed the researcher to other key people who could help; this ‘snow-
balling’ technique of recruitment became the starting point for building action
research teams. Frequent, informal meetings and discussions with key people
identified their interests and capacity to engage in the research process. Focus
groups helped to identify other appropriate people for the practice–research
engagement at the local level who were then invited to participate in the project
subject to completion of normal formal consent processes (Brown, 2001).
The nature and level of the working relationship that developed between
the field researcher and her colleagues in each setting became clearer over the
duration of the project and differed according to the nature of tasks undertaken
together. Different strategies for engagement were also used with different groups
and with individuals. These were influenced by the Jordanian culture, gender and
professional role and type of involvement. This was of particular importance
when applied to the relationship between the field-based researcher and partici-
pants who were doctors, nurses and midwives. This became played out very
overtly because the field researcher was a woman and a nurse, making it chal-
lenging in the early stages for her to achieve a collegial or leadership role with
male medical directors who were of higher gender and professional status within
this culture. For example, in one of the participating hospitals, the director of the
medical department, a doctor, initially completely refused to co-operate. While
there were other complicating factors also operating, relating to hierarchical
disputes within the hospital, additional efforts were required to gain his co-
operation in the research. The researcher took advice from other health pro-
fessionals working in the hospital on the best way to get the co-operation of the
hospital directors, and was ultimately successful in developing a personal and
friendly relationship with each separately, and over time earning their respect as
a researcher.
453
Establishing shared goals and objectives

The researchers and action research groups all shared a common goal of improv-
ing the quality of care provided to birthing women and increasing the account-
ability for services by the health professionals. However, another unpredicted
level of goal sharing and team work developed through this study. Training
sessions, focus groups and meetings helped doctors, nurses and midwives in the
three hospitals to interact positively, find common interests and begin to work in
teams focused on this project’s goal rather than as different status individuals
(Reason, 2004). Shared objectives and frameworks developed as the content of
the JCBR was renegotiated, tested and modified based on group suggestions
during discussions in meetings (Brown, 2001). These negotiations resulted in the
strategies and methods suitable for the implementation of the JCBR within the
Jordanian health system and its hospitals. Further discussions and negotiations
resulted in agreement on the level and degree of commitment of participants and
how their own interests would be served by their participation in the research
(Batliwala, 2003; Brown, 2001; Brown et al., 2003; Lindsey et al., 1999).
The field researcher actively applied the principle of reciprocation confirm-
ing the notion that in PRE, research does not just ‘take’ but also contributes
(Redelmeier & Cialdini, 2002). She found that she could meet the needs of some
participants in ways that enhanced their working relationship. For example, one
hospital director was interested in becoming more up-to-date with normal birth
and evidence-based practice, and needed a source for this information. The
researcher provided her with copies of articles about evidence-based practice and
normal delivery and also recommended a contact person within the WHO office
in Jordan.
The changing nature of the partnership in PRE

While the commitment to developing partnerships between the researcher and
practitioners in this study was based on PRE principles (Brown, 2001), it can also
be explained using the definitions of networking, co-ordination, co-operation and
collaboration strategies identified by Himmelman (2001). These definitions
describe the transformation of power relations necessary to achieve coalitions
between organizations to solve problems. They can be usefully applied in relation
to the different levels of PRE achieved over the duration of the study and the ever
developing relationships in this project.
Networking involves the exchange of information for mutual benefit. It
initiated the relationship and began to build trust between researcher and practi-
tioners (Himmelman, 2001). Coordinating involves the exchange of information
for mutual benefit and altering activities for a common purpose. This requires
time to develop. In this study the establishment of trust between researcher and
454
practitioners was demonstrated, for example, in the relationship built between

the researcher and the hospitals’ managers, leading to joint planning of the study
and implementation of the training program. Co-operating involves the exchange
of information, altering activities, and sharing resources for mutual benefit and a
common purpose (Himmelman, 2001). This also requires significant amounts of
time, high levels of trust, and a significant sharing of ‘turf’. This took at least 10
months to develop and considerable targeted effort by the field researcher and
shaped, for example, the relationship between the researcher and the National
Steering Group. This co-operation resulted in letters authorizing the research
being sent to hospitals and generated the co-operation, guidance and support of
Local Leadership Groups and department directors in facilitating their staff’s
involvement in the study. The co-operation of the directors of departments was
not only logistically desirable, but also increased the motivation of their staff to
attend the two hour training sessions. Directors informed the participants about
the activities of the training sessions and some promised the nurses and midwives
on their staff a day ‘off’ if they attended the training sessions. The field researcher
also provided small incentives during the training sessions, such as food, drinks
and small gifts in keeping with local Jordanian customs.
A collaborative strategy operates at the peak level of Himmelman’s hier-
archy. It involves exchange of information, altering activities, sharing resources
and enhancing the capacity of practitioners for mutual benefits and a common
goal (Himmelman, 2001). Again this requires the highest level of trust, consider-
able amounts of time, and extensive sharing undertaken for the good of the
research and its potential outcomes. The shared goals of improving the health
care for mothers and babies and the maternal and child health system in Jordan
guided the researcher and practitioners in their joint work and created commit-
ment for the considerable effort needed for the research to succeed. For example,
involvement of health care professionals (Local Action Groups) in the imple-
mentation of the birth record over a period of time, and their willingness to con-
tribute to the improvement of the quality of care in their health systems, required
them to complete two sets of records for the duration of the study. Figure 2
describes the nature and the level of partnership development with the three
action research groups.
Institutional arrangements
Institutional arrangements may affect the practice–research engagement work
and, as Brown (2001) describes, researchers need to learn how to interact within
institutional requirements. The first step was obtaining permission from leaders
in the Jordanian health systems for the study. The Ministry of Health’s interest
and subsequent permission for the study helped provide managers of the hospi-
tals with the flexibility to engage in the research and to use their own authority to
455
LAG
LLG
NSG
Collaborating
Co-operating
Co-ordinating
Networking
0-month 18-month
Figure 2 The nature and the level of partnerships development with the three action
research groups, the National Steering Group (NSG), Local Leadership Groups (LLG) and
Local Action Groups (LAG), over the duration of the study and how these changed against
Himmelman's hierarchy
facilitate the implementation process. This high level approval enabled the
engagement of practitioners throughout the health system and reduced institu-
tional constraints regarding their participation. It also helped the researcher to
interact with practitioners in the field in a flexible and authoritative manner.
The second level of institutional participation was required at the hospital
level. The manager’s permission for the research allowed directors of each depart-
ment (medical, nursing, registration) to engage in the process and use their
authority similarly with their staff to facilitate the research process. This provided
doctors, nurses, midwives and other workers, who implemented and used the
JCBR, with the flexibility to participate and reflect on the process of the imple-
mentation with the researcher.
Frequent discussions between the researcher and practitioners helped iden-
tify the challenges and/or constraints that an institution might impose on the
practice–research engagement within the field. These challenges and constraints
were documented during fieldwork and discussed in PRE meetings. We found, as
have others, that organizational development and action research can be strongly
emancipatory, creating processes and structures for collaborative inquiry (Reason
& McArdle, 2006). These processes encourage values of inquiry and learning and
mutual respect for other people (Reason & Bradbury, 2001; Reason & McArdle,
2006).
456
Conclusion
Practice–research engagement was effective in merging the insights of practice

with the analytic tools of research to generate new knowledge and improvement
in practice. It also helped us learn about managing a process of change that ulti-
mately could improve a health system (Brown, 2001).
Good communication skills, skilful listening, flexibility and respect are
some of the strategies the change agent should use to build trust and close rela-
tionships with key persons (Buonocore, 2004). The field researcher attempted to
model these characteristics and apply them with leaders in the system and
appeared to be highly successful in doing so.
Baseline data from interviews and focus groups showed that the time for
training, shortage of staff and workload, and resistance to change, were potential
barriers to the implementation of the JCBR and constituted risks to the study.
Strategies to overcome these were identified early and key people at national and
local levels assisted us in implementing these strategies. We found the work of
Reason and McArdle (2006), stating how action research can contribute to
organizational development through more effective work practices and better
understanding of processes of organizational change, to hold strongly in this
study. Information collection and feedback to staff led to joint problem solving so
that organizational development became not only a process of organizational
improvement but also a process of mutual and liberating inquiry. In this study,
for example, practitioners learned together that each discipline needed to improve
their record-keeping behaviours and work together as a team to improve care for
women.
The researchers identified resistance to change by health professionals as
one of the barriers to the implementation of the JCBR. Most resistance to change
occurs due to lack of knowledge about the change and fear of the unknown.
Understanding the key areas of change management and how to avoid obstacles
are critical to project success. Professionals may feel threatened, especially if there
is no clear positive benefit of change apparent to them immediately. They will
assume negative consequences and act accordingly to stop or delay the change
process (Handly, Grubb & Keefe, 2003; Howardell, 2006; Linton, 2002). In this
study, health professionals working in the maternity departments in the three
hospitals were involved in the implementation process of the JCBR. In addition,
key persons at national and local levels actively assisted and were actively
engaged. Effective communication, clear and shared goals and establishment of
joint involvement and shared ownership proved effective strategies that were
adopted to enhance change and were successful in preventing resistance.
Co-operation from health professionals was essential for the implementa-
tion of the JCBR. One of the participating hospitals was an institution where the
field researcher had previously worked. She was well known in this hospital with
457
most of the staff having previously been colleagues and they readily accepted and
co-operated with her. While creating some bias, this confirmed the importance of
close relationships between researcher and practitioners in the process of imple-
menting the required change. She worked hard to build this type of relationship in
the other two hospitals, where she was not known initially, and while not achiev-
ing the same depth or duration of relationship, she was ultimately successful.
Our research verified findings reported in the literature that careful
structured planning of the change process helps overcome barriers to change
(Buonocore, 2004). Also that preventing resistance to change is better than over-
coming it, with involvement and communication being the best strategies to
prevent resistance to change (Szocska, Rethelyi & Normand, 2005). Achieving
change in a public-sector organization requires more than minimizing resistance
however and is difficult because the complexity is overwhelming. Success depends
on the quality of the implementation, on the sensitivity to different points of view,
the degree of support from key persons in the organization and the reliability of
principles of the change approach adopted (Byram, 2000; Iles & Sutherland,
2001; Winkelman, 2003).
We found that developing effective practice–research engagement and
using action research at different levels of the system concurrently enabled us to
achieve substantial health system change. Our work has confirmed that a PRE
approach can facilitate complex health system change associated with quality
improvement.
References
Batliwala, S. (2003). Bridging divides for social change: Practice–research interactions

in South Asia. Organization, 10(3), 595–615.
Brown, L. D. (2001). Practice–research engagement and civil society in a globalizing
world. Washington, DC: The Houser Center for Non-Profit Organizations.
Brown, L. D., Bammer, G., Batliwala, S., & Kunreuther, F. (2003). Framing practice–
research engagement for democratizing knowledge. Action Research, 1(1), 81–
101.
Buonocore, D. (2004). Leadership in action: Creating a change in practice. AACN
Clinical Issues, 15(2), 170–181.
Byram, D. (2000). Leadership: A skill not a role. AACN Clinical Issues, 11(3), 463–
469.
Davison, R. M., Martinsons, M. G., & Kock, N. (2004). Principles of canonical
action research. Information Systems Journal, 14(1), 65–89.
Department of Statistics (2004). Jordan in Figures 2003. Amman: Department of
Statistics.
Evans, M. G. (2003). Evaluating computerised health information systems: We are
still getting information technology wrong. British Medical Journal, 326(7394),
860–863.
Handly, M. J., Grubb, S. K., & Keefe, N. (2003). Essential activities for implementing
458
a clinical information system in public health nursing. Journal of Nursing

Administration, 33(1), 14–16.
Himmelman, A. T. (2001). On coalition and the transformation of power relations:
Collaborative betterment and collaborative empowerment. American Journal
of Community Psychology, 29(2), 277–284.
Howardell, D. (2006). Overcoming people’s fear of change. Retrieved 7 July 2006
from: http://www.theacagroup.com/overcome.html
Iles, V., & Sutherland, K. (2001). Organisational change: A review for health care
managers, professionals and researchers. London: NCCSDO.
Khresheh, R. (2006). Investigating the feasibility of introducing a consolidated birth
record: The first step in an information system that identifies outcomes and
enables improvement. Unpublished PhD thesis, University of Technology,
Sydney.
Larrabee, J.H. (2004). Advancing quality improvement through using the best evi-
dence to change practice. Journal of Nursing Care Quality, 19(1), 10–13.
Lindsey, E., Sheilds, L., & Stajduhar, K. (1999). Creating effective nursing partner-
ship: Relating community development to participatory action research.
Journal of Advanced Nursing, 29(5), 1238–1245.
Linton, J. D. (2002). Implementation research: State of art and future directions.
Technovation, 22, 65–79.
Meyer, J. (1993). New Paradigm research in practice: The trials and tribulations of
action research. Journal of Advanced Nursing, 18, 1066–1072.
Nsheiwat, M., & Al-Khalidi, Z. (1997). Maternal mortality study. Ministry of Health,
Research and Studies Directorate. Jordan. Retrieved from: http://www.moh.
gov.jo/research/biostatistics/mortality.htm
NSW Department of Health (2004). New South Wales mothers and babies 2003.
Centre for Epidemiology and Research. NSW Public Health Bulletin.
Reason, P., & Bradbury, H. (2001). Handbook of action research: Participative
inquiry and practice. London: Sage.
Reason, P. (2004). Action research: Forming communicative space for many ways of
knowing. Response to MD. Anisurr Rahman International Workshop on
Participatory Action Research Dhaka. Retrieved 18 January 2006 from:
http://www.bath.ac.uk/~mnspwr/paperslist.htm
Reason, P., & McArdle, K. L. (2006). Action research and organization development.
Retrieved 18 January 2006 from: http.://www.bath.ac.uk/~mnspwr/paper-
slist.htm
Redelmeier, D. A., & Cialdini, R. B. (2002). Problems for clinical judgment: 5.
Principles of influence in medical practice. Canadian Medical Association
Journal, 166(13), 1680–1684.
Szocska, M. K., Rethelyi, J. M., & Normand, C. (2005). Managing healthcare reform
in Hungary: Challenges and opportunities. British Medical Journal, 331,
231–233.
Winkelman, W. J. (2003). Evaluating computerised health information systems.
British Medical Journal, 327 (7407), 162–163.
Reham Khresheh is Dean Assistant for Student Affairs and Lecturer at Mutah
University, Karak, Jordan. She received her PhD from University of Technology,
Sydney, Australia in 2006. Her research interests include improving knowledge and
459
health services that support the health of women and their families in Jordan.
Particularly, the focus is on maternal child health and health services research and
systems improvement. Address: Faculty of Nursing, Mutah University, Karak, Jordan
61710.
Lesley Barclay is Foundation Chair of the Graduate School for Health Practice at
Charles Darwin University, Northern Territory, Australia. She has worked in inter-
national development for nearly 20 years with AusAID, World Bank and WHO,
providing advice in primary health care, maternal infant/child health and capacity
building in health worker education systems in Asia, Melanesia and the Pacific
Islands. Her research focus is now on system change to improve health service
delivery. Her role as leader and mentor has been instrumental in improving mater-
nity services in Australia and internationally. This work is characterized by strong
partnerships between professions and a respect for traditional birthing practices.
Address: Graduate School for Health Practice, Institute of Advanced Studies, Charles
Darwin University, Darwin, Northern Territory 0909, Australia.
46 0
! The Author 2007; all rights reserved. Advance Access publication 27 October 2007 doi:10.1093/heapol/czm039
Implementation of a new birth record in

three hospitals in Jordan: a study of health
system improvement
Reham Khresheh1 and Lesley Barclay2*
Accepted 13 August 2007

This study tested the introduction of a new integrated clinical record in Jordan
where currently no clinical report links antenatal, birth and postnatal care for
women. As a result, no continuity of information is provided to clinicians nor
are there national statistics on trends, or performance of hospitals around birth.
Our study was conducted in the Jordanian Ministry of Health, the maternity
wards and registration departments of three hospitals in Jordan and in the
Maternal and Child Health Centres located near these hospitals. We used an
exploratory, descriptive design and practice-research engagement to investigate
and report on the process of change to improve and implement the new birth
record. Through engaging practitioners in research, care improved, the quality of
reporting changed, managers developed more effective measures of hospital
performance and policy makers were provided with information that could form
the basis of a national maternity data monitoring system. Quantitative and
qualitative audit data demonstrated improved clinical reporting, organizational
development and sustained commitment to the new record from clinicians,
managers and policy leaders.
Keywords Maternal health services, practice-research engagement, quality assurance,
health care quality
KEY MESSAGES
! Clinical information can extend beyond individual patient care to include quality review and improvement processes
within health information systems.
! Identifying a shared goal and engaging practitioners and researchers in practical activity to achieve this goal can bring
about sustained clinical improvement.
! A Practice Research Engagement process led by a skilled researcher can play a key role in quality improvement beyond
the immediate aims of the research project.
Introduction
Efforts to develop a Perinatal National Minimum Data Set have
been undertaken in many countries, led by the World Health
1
Lecturer, Faculty of Nursing, Mutah University, PO Box 7, Karak, Jordan.
Organization (World Health Organization 2004). The aim of
2
Professor of Health Services Development, Institute of Advanced Studies,
Charles Darwin University, Darwin, NT, Australia.
such data collection is to improve the health of mothers and
* Corresponding author. Institute of Advanced Studies, Charles Darwin
babies by monitoring perinatal health, as well as providing
University, Darwin, NT 0909, Australia. E-mail: Lesley.Barclay@cdu.edu.au ongoing information to service providers and policy makers
76
461
THE IMPACT OF A NEW BIRTH RECORD 77
about trends and patterns in the health status of mothers and records and clinicians often not reporting on their own care.
babies. Although perinatal surveillance systems are feasible in In addition, pregnant women had no access to their own
developed countries, they have still not been implemented records and there was no peer review of performance in the
widely (Beck et al. 2003; Laws and Sullivan 2004). Often where health team. Since data were not used for analysis or planning
elements of such record systems exist, they are described as there were no opportunities to ‘benchmark’ performance
fragmented, incompatible, uncoordinated and not comprehen- individually by clinicians or by the hospital against other
sive, and a concerted effort is needed to enable regular hospitals or with international evidence or standards (Khresheh
monitoring of maternal morbidity and mortality (AbouZahr 2006). No published studies or reports are available in Jordan
and Wardlaw 2001). Researchers have stressed that practice describing professional practice during labour that could be
could be adversely affected by inaccurate clinical information linked to morbidity outcomes, and efforts have been made
and that there is an urgent need for the development of through research, rather than routine data collections, to
standard data-collection tools for collection of high-quality data explore the causes of mortality (Nsheiwat and Al-Khalidi
(Wyatt and Wright 1998; M’kumbuzi et al. 2004). 1997; Khouri and Masaad 2002). However, a National
Jordan is geographically small and classified as a developing Information System now being introduced provides an oppor-
country with a population of 5.5 million people (World Health tunity to combine an ‘on line’ clinical data entry system with
Organization 2001). The birth rate is 29/1000 and the fertility one that can report trends in the safety and quality of birthing
rate is high at 3.7 (Department of Statistics 2004). In Jordan, services through aggregation of clinical data (National
27 Ministry of Health hospitals provide birth services, with Information Centre 2001).
nearby Maternal and Child Health Centres providing antenatal The study reported here aimed to investigate the feasibility
and postnatal care. The record-keeping system in these and outcomes of introducing a new birth record shared
hospitals and Maternal and Child Health Centres is controlled between hospital and community. The new record, the
nationally, resulting in two separately located information Jordanian Consolidated Birth Record (JCBR), is based on the
systems. Pregnant women’s antenatal and postnatal records NSW Perinatal Data Collection Form (NSW Department of
are held in the community at the Maternal and Child Health Health 2004) which is part of the Australian perinatal data
Centres, while labour and birth records are in the hospital. collection system of national reporting and benchmarking
There is no system linking the antenatal, birth and postnatal (Laws and Sullivan 2004). It was modified by Jordanian
record to provide continuity of information to clinicians, experts to meet Jordanian needs. This Australian tool was
national statistics on trends, or performance of hospitals chosen for pragmatic reasons because of its accessibility to the
around birth. Incorrect reporting of maternal and infant researchers. The JCBR consists of eight basic areas, many
mortality and morbidity is likely as there are difficulties in modified from the Australian tool to suit the Jordanian context.
collection and aggregation of poor quality data, with no Items include demographic data and information on maternal
validation of this or quality controls in place. health, the pregnancy, labour, delivery and perinatal outcomes
Available data are usually based on ‘snap shot’ research or (see Box 1). Modifications suggested by the Ministry of
surveys which are costly to conduct (Nsheiwat and Al-Khalidi Health officials included removal of data not applicable to the
1997; Shihadeh and Al-Najdawi 2001), while large-scale Jordanian community such as ‘aboriginality’ and addition of
demographic and reproductive health surveys are carried out data such as the woman’s nationality, occupation, husband’s
intermittently (Department of Statistics and Macro occupation and income, type of health insurance, final
International Inc. 1998; Department of Statistics and ORC diagnosis upon discharge, haemoglobin level, blood grouping
Macro 2003). Our baseline study confirmed that recording of and Rh of mother and infant etc.
birth data collected in hospital records was of poor quality, Our research objectives were to improve the quality of the
recorded in 18 different places in the hospital record and that clinical reporting system, to enhance organizational develop-
insufficient hospital data were returned to the community to ment through teamwork around data collection and to improve
inform postnatal care. There were other concerns including the quality of care by linking community and hospital
duplication of data that was time consuming for the recorder, antenatal, birth and postnatal records. We explored whether
with frequent gaps in information, retrospective completion of the JCBR could be the basis of a national maternity data system
Box 1 Items collected in the new record (JCBR)

Demographic: Record number, national number, hospital name, woman’s full name, age, address, date of admission,
nationality, occupation for woman and husband, and health insurance.
Maternal health: Number of previous pregnancies, live births and abortions; type of last delivery and number of previous
caesarean sections.
Pregnancy: Date of last menstrual period, antenatal care, medical conditions, smoking status.
Labour and delivery: Onset of labour, pain relief or anaesthetics, presentation at birth, type of delivery, perineal status,
surgical repair of vagina or perineum, immediate post-natal complications.
Baby: Place of birth, date of birth, sex, plurality, birth weight, estimated gestational age, Apgar scores, resuscitation of baby,
breastfeeding initiation, birth defect, admission to NICU.
Discharge notes: Final diagnosis, mother discharge status, baby discharge status, date of discharge for mother and baby,
laboratory results for mother, physician treatment and advice.
462
that would monitor and benchmark maternity care services in opportunistic audit of 42 records, randomly selected from
Jordan. The results are reported against these objectives. women who gave birth 7 months after implementation, was
estimated to be sufficient to assess the sustainability of the
effect of the new record on the quality of reporting and
recording of data about mothers and babies. Quantitative data
Methods were collected for record audits and analysed using Excel
Design spreadsheets and frequency tabulation.
The study, begun in January 2004 and completed in October Nine interviews and three focus groups (spread equally across
2005, used an exploratory, descriptive design and an each of three participating hospitals) were undertaken before
action research approach; practice-research engagement. This the implementation of the JCBR with a total of 36 people
approach was chosen because the design allows investigation of who played key roles at different levels of the health system.
a quality improvement process that simultaneously supports This established their opinions of the record. A second round of
change management (Brown 2001). The study drew on a interviews and focus groups was undertaken immediately after
combination of quantitative and qualitative data to compare the implementation of the JCBR with clinicians and medical
baseline data with implementation data produced from the new record staff (n ¼ 40), with mothers (n ¼ 15) who received their
record. own copy of the JCBR and with Maternal and Child Health
Centre staff who provided antenatal and postnatal care
Setting (n ¼ 21). A third round of interviews with hospital staff
(n ¼ 23) occurred concurrently with the 7 months post-
The study was conducted in three Ministry of Health hospitals
implementation audit. Questions explored how the JCBR
(A, B, and C) purposively selected by the Ministry in different
affected the reporting and recording of data and how staff
regions of Jordan. Two hospitals are peripheral hospitals and
used the data subsequently. Audio-recorded data from inter-
one is a tertiary centre. These hospitals provide birthing services
views and focus groups were transcribed in Arabic and content
for approximately 14 530 women annually (Department of
analysis was used to identify repetitive themes which are
Statistics 2001). The Ministry of Health in Jordan and the
reported qualitatively using text and quotes. Field notes
Ethics Committee of the University of Technology, Sydney,
recorded observations made during visits to hospitals, to clinics
approved the study. De-identified hospital data were used for
or during appointments, and were updated daily. Field notes
the record audit and all participants in the action research
were also analysed for themes, quotes and observations, which
study formally consented to participate.
have been extracted to exemplify findings reported here.
The field researcher, Khresheh, who is Jordanian and a
Data collection and analysis clinician, worked alongside staff during the implementation in
The study consisted of three standardized record audits, pre- each hospital. During random weekly checks at each hospital
implementation baseline data, immediate post-implementation the accuracy of data being recorded in the JCBR was assessed,
data and a smaller opportunistic evaluation conducted 7 enabling us to see if records had been fully completed, and if
months post-implementation. Qualitative data that describe not, the type and number of incomplete items, the accuracy of
and explain the change process were collected in each setting these records compared with the notes in the medical records
from managers, clinicians and medical records staff using focus and the differential participation rates of groups of staff.
groups, interviews and standardized questions. Field notes were Consistent with a methodology that promotes change, these
also kept. visits maintained, or built, the commitment of the participating
A training programme was conducted in each hospital to health professionals as the researcher and participants worked
prepare health professionals to use the JCBR. This was also an together to analyse and improve results. At each visit, the
important part of the action research methodology as sharing researcher randomly drew 10 records of participating women.
information from the outset, especially the baseline data which The participation rates of different groups of staff were
confirmed the nature of the problem, was helpful in motivating identified through their signatures, enabling the researcher to
participants to work together to improve their record keeping. identify whether a doctor or midwife or registration worker had
Clinicians were also encouraged to complete the JCBR completed his or her section. The percentages of completed
contemporaneously, rather than retrospectively. sections for each different group’s signatures were calculated by
A detailed coding sheet was developed for manually coding the researcher (see Figure 3 below).
audit data from records. This sheet collected completeness of
record, errors, persons/role of person completing the record as
well as clinical data. This coding sheet was tested pre-baseline,
modified slightly and then used for baseline assessments, post-
Results
implementation and for the longer term follow-up. Quality of clinical records
The first record audit was with a random sample of 180 In pre-implementation audit data, up to 50% of the records
records, 60 from each hospital, of women who gave birth in were inadequately completed with important clinical informa-
2003. This sample was considered of sufficient size to convin- tion unrecorded. It was also difficult to find this information
cingly describe the size and nature of the problem. This was since records were duplicated and recorded in 18 different
compared with results of a second audit post-implementation places and by up to four different care-providers. In contrast,
(n ¼ 1254) that sampled records of all women who gave birth post-implementation data showed that although completion
during 2 months use of the new record. The size of the third rates of the JCBR continued to improve over time, even at the
463
100 100
80 80
Complete
60
60 Audit 1 %
40 Audit 1
40 Audit 2
20 Audit 3
20 0
Pain relief Birth Type of Surgical
0
presentation delivery repair
Gestational Apgar Admission Discharge
age scores to NICU status Type of data
Type of data
Figure 2 Improvements in a sample of clinical data before (Audit 1,
Figure 1 Improvement across a sample of data measuring complete- n ¼ 180) and 7 months after the implementation of the JCBR (Audit 3,
ness of record before (Audit 1, n ¼ 180) and after the implementation of n ¼ 42) in the three hospitals
the JCBR (Audit 2, n ¼ 1254) in the three hospitals
100
Table 1 Improvements in a sample of aggregated data from three 80
hospitals before implementation (Audit 1), after implementation
60 Midwives
(Audit 2) and at 7 months post-implementation of the JCBR (Audit 3) %
40 Doctors
Audit 1 Audit 2 Audit 3 Registration workers
20
(n ¼ 180) (n ¼ 1254) (n ¼ 42)
0
Type of data No. % No. % No. %
2 weeks 4 weeks 6 weeks 8 weeks
Gestation age 101 56 992 79 31 74
Figure 3 Improvement of staff commitment toward the implementa-
Apgar scores 55 31 940 75 25 60
tion of the JCBR over 2 months
Admission to NICU 75 42 1072 86 24 57
Baby discharge status – – 1153 92 7 17
Table 2 Improvements in a sample of clinical data aggregated across

three hospitals before implementation (Audit 1), after implementation
(Audit 2) and 7 months post-implementation of the JCBR (Audit 3)
beginning of implementation, quality and completeness were
Audit 1 Audit 2 Audit 3
better. By the end of the implementation period, more than (n ¼ 180) (n ¼ 1254) (n ¼ 42)
75% of records were fully completed. Hospital A had the best Type of data No. % No. % No. %
results with more than 78% of the record items completed at
Pain relief and anaesthesia 65 36 1074 86 28 67
the end of implementation, followed by Hospital B then
Presentation at birth 113 63 1163 93 32 76
Hospital C. Overall percentages of improvement on a sample
of data are shown in Figure 1 and Table 1.1 Type of delivery 145 81 1172 93 42 100
The quality and completeness of the recording of important Surgical repair of the 132 73 1111 89 35 83
clinical information such as Apgar scores, gestational age, vagina or perineum
admission to NICU and discharge status improved. Additional
data which had not been sufficiently well recorded previously,
or were absent from the mother’s medical records, were now
being collected. This included socio-economic status of the Organizational and individual performance
mother; type of last delivery; information on the current The action research and practice-research engagement process
pregnancy, for example the date of the first visit to the was designed to engage staff in a quality improvement process.
antenatal clinic and smoking status; complications after deli- Initially we found there was a relatively low commitment by
very; information about breast-feeding initiation and advice for staff to the new system in all departments in each hospital,
the mother about postnatal care. Now, all this data could be although commitment by midwives was generally higher than
found in one place in the record. Results from the third record for doctors. Registration workers were less committed in two of
audit showed sustained improvement in the completeness rate the three hospitals. However, commitment and enthusiasm for
and the quality of data recorded in the women’s medical record change gradually improved over time for all groups of health
(Figure 2 and Table 2). professionals (Figure 3).
Qualitative post-implementation data also showed that the The use of JCBR increased the health professionals’ sense
JCBR was considered to be useful and valuable for hospital of responsibility toward the care they provided, their respect
staff, mothers and Maternal and Child Health Centre staff. for others in the team and the accuracy of data they recorded
Hospital staff wanted the JCBR to continue because documen- in the mother’s records. As a midwife said, ‘everyone had
tation was easy to complete, took less time and effort, and to write and record and sign off the exact care provided to a
provided valuable data that assisted their clinical decision- woman so the next care-provider could continue with the
making. For example, one obstetrician said, ‘the information on suitable care.’
the baby after delivery is very important. I don’t know why we The managers of the hospitals and clinical directors supported
didn’t record these data before.’ both staff and the field researcher, and contributed to the
464
positive effect the JCBR had on promoting teamwork and beyond individual patient care to include quality review and
enhancing relationships among health team members. For improvement processes. The data produced and their improved
example, a midwife commented, ‘it was team work; we all quality confirms the claim by others that this information can
participated in completing the JCBR.’ While a doctor said, assist with allocation of resources, budgetary and long-term
‘when I was not sure of something recorded in the JCBR usually planning, and productivity measurement (Slagle 1999; World
I went back to the responsible care-provider for more clarifica- Health Organization 2004).
tion.’ The JCBR also improved the health professionals’ record- The process of quality improvement in clinical practice and
keeping habits including the timely and accurate completion of health system development is complex and challenging. Quality
the important clinical data record and recording of new clinical improvement should focus on areas of real importance, the
data unrecorded previously. The accessibility of data from organization should have capable leadership and be prepared to
the JCBR also encouraged some of the health professionals change, and the external environment should encourage change
to review their practices for the first time, motivating them to (Shortell et al. 1998). In this study, action research, which
question the current situation. One resident doctor demon- emphasizes practice-research engagement and is based on
strated the enhanced clinical leadership and accountability theories of change management (Brown 2001), was successful
when he stated, ‘at first I found this new record unnecessary, in introducing and managing the change identified by the
but with time I valued its importance; we noticed that the researcher-practitioner team, as well as investigating this
majority of mothers have low haemoglobin.’ process and its outcomes.
The implementation phase of the research was completed in Effective leadership is necessary to manage improvement
2 months. During this time staff undertook a double load as in clinical practice settings. This leadership involves influencing
they completed their routine documentation as well as the others to contribute to positive outcomes (Redelmeier and
JCBR. Staff from all three hospitals have continued using the Cialdini 2002). As a result of the researcher working with
JCBR and are still collecting the statistical summary copies them, health workers demonstrated increased professionalism,
of the new record and sending them to the researcher for while managers and clinical directors were supportive in
analysis. creating a simple change that enhanced the working environ-
ment in a way that appears to be sustained.
Creating links between services As well as solving the immediate practical clinical record
Health professionals in the Maternal and Child Health Centres problems, a significant outcome was that the research process
were highly motivated to link community and hospital records. helped initiate, develop and maintain new opportunities for
They were aware that information about the course of labour, professional dialogue as doctors, nurses and midwives worked
details of birth and health of the baby influences the quality of towards the common goal of improving health care for mothers
postnatal care. For example, one obstetrician said, ‘these and babies. This process helped in building a team in a
information systems are of no benefit if they stay like this, hierarchal environment where professionals were not used to
without connection. We need complete information if we want this mode of operating. Practitioners were given the opportunity
to make real improvement.’ In interviews and focus groups, to work in new ways with medical record workers, nurses and
staff suggested that client-held records would facilitate integra- midwives who are usually low status within the system.
tion and that a simple computerized system based on three Obstetricians, at first somewhat sceptical, ultimately responded
hard copies of the antenatal record, birth and postnatal record positively, also finding that teamwork produced better results
would be feasible. The immediate post-implementation and for their work. Providing women with their own copy of their
longer term evaluations, record audits and interviews confirmed clinical record facilitated their communication with health
the potential of developing a sustainable national hospital- professionals.
based perinatal information system using the new record and Proper staff preparation was important and is necessary in
connecting all hospitals and nearby Maternal and Child Health any major quality improvement process (USAID 1999).
Centres. There is national commitment in Jordan to achieve During training, health professionals were educated about
these links. the purpose of the study and became committed to the new
record. This enabled them to maintain a sense of control,
built further support during implementation, and also
minimized resistance to change (Henry 1997; Moody et al.
Discussion
2001). Commitment of staff to the process of implementation
The study had a number of limitations. This included the varied across hospitals and among the health professionals
positive bias that was introduced by purposive selection of the themselves. From the beginning, Hospital A showed the
hospitals and the researcher’s attention to the quality of highest commitment of staff while Hospitals B and C
interaction with people. This was intentional and an element
began with less commitment but improved over time. It
of the design. The evaluation conducted after 7 months was
was likely that the relationship between the researcher and
opportunistic rather than ideally situated in scope or time from
the hospital staff contributed, as the researcher was already
completion to convincingly measure long-term sustainability.
known to colleagues in Hospital A at the beginning of the
study, and she was able to spend more time in the field there
System improvement because of its close location to her home. Despite this,
The study confirms that clinical information and health hospitals B and C also showed significant and sustained
information systems can be used for purposes that extend improvement.
465
Midwives and registration workers were initially more Beck LF, Johnson CH, Morrow B. 2003. PRAMS 1999 Surveillance Report:
committed than doctors to improvement processes. They may Pregnancy Risk Assessment Monitoring System. Atlanta, GA: Division of
have accepted the leadership of the researcher, a female nurse, Reproductive Health, National Center for Chronic Disease
Prevention and Health Promotion, Centers for Disease Control
more readily than the doctors, but this may also be explained
and Prevention.
by gender issues common in some Arab countries. Overall,
Brown LD. 2001. Practice- Research Engagement and Civil Society in a
because of the strategy and effort of building strong researcher-
Globalizing World. Washington, DC: The Houser Center for Non-
practitioner relationships, the commitment of all staff improved
profit Organizations.
over this time, including that of male doctors.
Brydon-Miller M, Greenwood D. 2003. Why action research? Action
The JCBR helped improve the accuracy of data recorded
Research 1: 9–27.
by health professionals, with data now recorded once only
Department of Statistics. 2001. Statistical Yearbook: Number of Births by
by the person providing care, reducing risks of transcription
Hospitals. Amman: Ministry of Health.
errors. The increased accuracy of data in the JCBR may also be
Department of Statistics. 2004. Jordan in Figures 2003, 6th edn. Amman:
explained by the increased accountability of health professionals
Department of Statistics.
toward their data. Important data that were previously not
Department of Statistics (Jordan) and Macro International Inc. 1998.
collected such as Apgar scores at birth were now available,
Jordan Population and Family Health Survey 1997. Calverton, MD:
were able to be aggregated and could be used for a national Department of Statistics and Macro International Inc.
reporting system.
Department of Statistics (Jordan) and ORC Macro. 2003. Jordan
Similar to other improvement studies, interaction with Population and Family Health Survey 2002. Calverton, MD:
practitioners in the field helped to identify problems and Department of Statistics and ORC Macro.
plan and initiate solutions during the implementation period Henry P. 1997. Overcoming resistance to organisational change. Journal
(Webb 1990; Nolan and Grant 1993; Brydon-Miller and of the American Dietetic Association 97: S145–7.
Greenwood 2003). The reciprocal process, in which the Khouri S, Masaad D. 2002. Causes of infant mortality in Jordan. Saudi
researcher and practitioners informed each other, established Medical Journal 23: 432–5.
new knowledge and effective problem-solving actions. In turn Khresheh R. 2006. Investigating the feasibility of introducing a
this appeared to develop collaborative relationships with consolidated birth record: the first step in an information system
practitioners in the field, to build motivation and to provide that identifies outcomes and enables improvement. PhD thesis,
practical support. Local and national leaders helped manage University of Technology, Sydney.
and lead this project. Laws PJ, Sullivan EA. 2004. Australia’s mothers and babies 2001. AIHW
Cat. No. PER 25. Sydney: AIHW National Perinatal Statistics Unit
(Perinatal Statistics Series No. 13).
M’kumbuzi VRP, Amosun SL, Stewart AV. 2004. Retrieving physiother-
Conclusion apy patient records in selected health care facilities in South
Our study showed that identifying a shared goal and engaging Africa- is record keeping compromised?. Disability and Rehabilitation
practitioners and researchers in practical activity to achieve this 26: 1110–6.
goal can bring about sustained clinical improvement. Clinicians Moody G, Choong Y, Greenwood D. 2001. An action research approach
were recording better quality, more useful data collected with to the development of a clinical pathway for women requiring
increased professionalism following the study. The health Caesarean sections. Contemporary Nurse 11: 195–205.
professionals were using aggregated information to evaluate National Information Centre. 2001. Information age policies and strategies.
their performance and the hospitals could use the data in Amman, Jordan: National Information Centre.
planning for improvements. The policy makers who supported Nolan M, Grant G. 1993. Action Research and quality of care: a
and guided the study as partners now have a basis to apply the mechanism for agreeing basic values as a precursor to change.
results nationally. They are closer to their goals of consolidating Journal of Advanced Nursing 18: 305–11.
data into electronic records that can be analysed automatically, Nsheiwat M, Al-Khalidi Z. 1997. Maternal Mortality Study in Jordan’s
which provides the capacity to monitor the national maternity Health Status: Findings from the epidemiologcal studies and strategies for
future surveillance system. Amman, Jordan: Ministry of Health and
system.
Health Care.
NSW Department of Health. 2004. New South Wales Mothers and
Babies 2003. NSW Public Health Bulletin Supplement 15, S-5. Sydney:
Endnote Centre for Epidemiology and Research, NSW Department of
Health.
1
While all items mentioned in Table 1 were analysed, there were too Redelmeier DA, Cialdini RB. 2002. Problems for clinical judgment: 5.
many items to be presented here or in Tables 2 and 3. Therefore, a Principles of influence in medical practice. Canadian Medical
sample of items that represent particularly important clinical data Association Journal 166: 1680–4.
that were not collected or inadequately collected previously have
been presented. Shihadeh A, Al-Najdawi W. 2001. Forceps or Vacuum extraction: a
comparison of maternal and neonatal morbidity. Eastern
Mediterranean Health Journal 7: 106–14.
Shortell SM, Bennett CL, Byck GR. 1998. Assessing the impact of
References continuous quality improvement on clinical practice: what it will
AbouZahr C, Wardlaw T. 2001. Maternal mortality at the end of the take to accelerate progress. Milbank Quarterly 76: 593–624.
decade: signs of progress? Bulletin of the World Health Organization Slagle TA. 1999. Perinatal information systems for quality improvement:
79: 561–73. vision for today. Pediatrics 103: 266–76.
466
USAID. 1999. The Comprehensive Postpartum Project, Jordan. Amman, World Health Organization. 2004. Developing Health Management
Jordan: USAID. Information Systems. Manila, Philippines: WHO Regional Office for
Webb C. 1990. Partners in research. Nursing Times 82: 40–4. the Western Pacific.
World Health Organization. 2001. Country profiles, Jordan. Online at Wyatt JC, Wright P. 1998. Design should help use of patient’s data.
http://www.who.int/countries/jor/en/, accessed 2 March 2006. The Lancet 352: 1375–8.
467
Part
Re le t o o ealth ol y
a d y te Re ear h
This final section of the Reader aims to stimulate broader References
thinking about key methodological and other issues when
doing Health Policy and Systems Research (HPSR). Some de Savigny D et al. (2009). Systems thinking: Applying
of the papers presented here focus on research strategy a systems perspective to design and evaluate health
issues, including critical papers that address weaker areas systems interventions. In: de Savigny D, Adam T, eds.
of current HPSR practice in low- and middle-income coun- Systems thinking for health systems strengthening. Geneva,
tries. Other papers report researchers’ own reflections on World Health Organization:49–71.
their experience. http://www.who.int/alliance-hpsr/resources/97892
41563895/en/index.html
In addition to this selection of papers, we encourage
readers to draw on the ‘How to do ...’ series of papers in Rationale for selection: to stimulate thinking about
n
new approaches to intervention evaluation that allow
the journal Health Policy and Planning as they can inform
for systems
and guide the use of particular methods in HPSR.
English M et al. (2008). Health systems research in a
low-income country: easier said than done. rchives of
iseases in Childhood, 93:540-544.
http://dx.doi.org/10.1136/adc.2007.126466
Rationale for selection: to provoke critical reflection
n
on the practical and methodological challenges
of doing intervention and evaluation work in LMIC
health settings
Erasmus E, Gilson L (2008). How to start thinking about

investigating power in the organizational settings of
policy implementation. Health Policy and Planning,
23(5):361–8.
http://dx.doi.org/10.1093/heapol/czn021
Rationale for selection: practical introduction to
n
investigating power in implementation
Hanson K et al. (2008). Vouchers for scaling up

insecticide-treated nets in Tanzania: methods for
monitoring and evaluation of a national health system
intervention. MC Public Health 8:205.
http://dx.doi.org/10.1186/1471-2458-8-205
Rationale for selection: practical guide to develop-
n
ment and Monitoring and Evaluation study, with
strong focus on feedback to support implementation
Part 5 Reflections on Health Policy and Systems Research 471

Hyder A et al. (2010). Stakeholder analysis for health van der Geest S, Sarkodie S (1998). The fake patient: a
research: case studies from low- and middle-income research experiment in a Ghanaian hospital. Social
countries. Public Health 124:159–166. Science Medicine, 47(9), 1373–1381.
http://dx.doi.org/10.1016/j.puhe.2009.12.006 http://dx.doi.org/10.1016/S0277-9536(98)00179-8
Rationale for selection: provides insights into appli-
n Rationale for selection: reflective paper on research
n
cation of stakeholder analysis with conclusions for approaches, addressing ethical issues
how to use research to influence policy
Walt G et al. (2008). ‘Doing’ health policy analysis:
Molyneux CS et al. (2009). Conducting health-related methodological and conceptual reflections and chal-
social science research in low income settings: ethical lenges. Health Policy and Planning, 23:308–317.
dilemmas faced in Kenya and South Africa. ournal of http://dx.doi.org/10.1093/heapol/czn024
nternational evelopment, 21(2):309–326. Rationale for selection: to provoke critical reflection
n
http://dx.doi.org/10.1002/jid.1548 on how to do health policy analysis work
Rationale for selection: o provoke critical reflection
n
on ethical issues for household level HPSR
Ridde V (2008). Equity and heath policy in Africa: using

concept mapping in Moore (Burkina Faso). MC Health
Services Research, 8:90.
http://dx.doi.org/10.1186/1472-6963-8-90
Rationale for selection: reports important method for
n
investigating actor understandings and perceptions
Sakyi EK (2008). A retrospective content analysis of

studies of factors constraining the implementation
of health sector reform in Ghana. nternational ournal
of Health Planning and Management, 23:259–285.
Rationale for selection: provides an e ample of a
n
strong document review
Schneider H, Palmer N (2002). Getting to the truth? Re-

searching user views of primary health care. Health Policy
and Planning, 17(1):32–41.
http://dx.doi.org/10.1093/heapol/17.1.32
Rationale for selection: to provoke critical reflection
n
on how to assess patient and user perspectives
Health Policy and What does this Reader offer?
Systems Research Health Policy and Systems Research (HPSR) is often criticized for lacking
A Methodology Reader rigour, providing a weak basis for generalization of its findings and,
therefore, offering limited value for policy-makers. This Reader aims to ad-
dress these concerns through supporting action to strengthen the quality
of HPSR.
The Reader is primarily for researchers and research users, teachers and
students, particularly those working in low- and middle-income countries
(LMICs). It provides guidance on the defining features of HPSR and the
critical steps in conducting research in this field. It showcases the diverse
range of research strategies and methods encompassed by HPSR, and it
provides examples of good quality and innovative HPSR papers.
“Health Policy and Systems Research is a rapidly developing and critically
important field of health research, but has lacked any coherent
presentation of its nature, scope and methods. This Reader remedies
this gap, and will be an indispensable source of guidance for anyone
conducting Health Policy and Systems Research or wishing to learn
about it,” said Anne Mills, Professor of Health Economics and Policy and
Vice-Director, London School of Hygiene and Tropical Medicine.
The production of the Reader was commissioned by the Alliance for Health
Policy and Systems Research and it will complement its other investments
in methodology development and postgraduate training.
World Health Organization

Avenue Appia 20
CH-1211 en ve 27
Switzerland
ISB 978 92 4 150313 6
Tel.: 41 22 791 29 73
Fax: 41 22 791 41 69
alliance-hpsr who.int
http: www.who.int alliance-hpsr

Diplo

Uploaded by

Copyright:

Available Formats

Diplo

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Diplo

Uploaded by

Copyright:

Available Formats

Health Policy and

ISBN 978 92 4 150313 6 (NLM classification: W 84.3)

© World Health Organization 2012

Printed by the WHO Document Production Services, Geneva, Switzerland

2. The case study approach .......................................................................................................................... 161

3. The ethnographic lens ............................................................................................................................... 235

5. Investigating policy and system change over time ............................................................................. 317

6. Cross-national analysis .............................................................................................................................. 361

7. Action research ........................................................................................................................................... 443

Why is the Reader needed?

About this Reader 13

The Reader is mainly for use by:

How is the Reader structured?

Part 2 outlines key steps to follow when conducting HPSR studies.

together demonstrate the breadth and scope of HPSR work

Three broader texts of use to those doing HPSR are:

About this Reader 15

Lydia Bendib managed the overall production of the Reader.

About this Reader 17

Key points from this section

Features that define HPSR are:

Key characteristics of HPSR

Part 1 ntroduction to Health Policy and Systems Research 21

Key points from the following four sections

Four central elements in HPSR are:

Part 1 ntroduction to Health Policy and Systems Research 23

Figure 1 The different levels of health systems

Figure 2 The interconnections among the health system building blocks

Part 1 ntroduction to Health Policy and Systems Research 25

Some disease or programme-specific interventions also

Constraint Disease-specific response Health-system response

Financial inaccessibility: Exemptions/reduced prices Development of risk-pooling

Physical inaccessibility: Outreach for focal diseases Reconsideration of long-term

Poorly motivated staff Financial incentives to reward Institution of proper performance

Part 1 ntroduction to Health Policy and Systems Research 27

not only is policy designed to change a given

Policy actors evaluated, including how researchers influence policy-

Part 1 ntroduction to Health Policy and Systems Research 29

Key points from this section

Figure 3 The terrain of HPSR

Health Systems Health Policy

Hardware: Content &

Box 1: Suggested topics for health systems research

Financial and human resources:

Source: Sanders Haines,

Part 1 ntroduction to Health Policy and Systems Research 31

What HPSR is not

Part 1 ntroduction to Health Policy and Systems Research 33

7. Understanding the nature of social and

Key points from this section

However, unlike the dominant health research traditions

Related disciplinary Epidemiology Policy analysis Anthropology

Key research Deductive: Hypothesis driven Deductive and inductive Inductive

Part 1 ntroduction to Health Policy and Systems Research 35

Critical realism Causality

In comparative case study analysis, generalization is not

Part 1 ntroduction to Health Policy and Systems Research 37

Part 1 ntroduction to Health Policy and Systems Research 39

Shiffman J et al. (2004). The emergence of political

Smith R, Hanson K (2011). What is a health system.