犬只数量管理监测与评估指南英文版

ARE WE MAKING
A DIFFERENCE?
A Guide to Monitoring and Evaluating
Dog Population Management
Interventions
March 2015
All Photos © IFAW unless otherwise noted.
Preface
International
Companion Animal
Management Coalition
Almost every country invests in dog population management (DPM) in some form; however
there is no agreed measure to establish whether an intervention is successful. Through this
guidance document, the International Companion Animal Management (ICAM) Coalition
aims to provide encouragement and advice on impact assessment of DPM. Our aim is to
enable academics, practitioners and funders to track progress, learn and subsequently
improve their DPM impact through the use of measurable objective indicators. Indeed, for us
to achieve our long term goals in DPM we need an evidence base for our future decisions, in
other words “success depends on knowing what works” Bill Gates, cited by Savedoff et al.
(2006), pp iv.
Our focus is on applying scientific solutions to real world problems and encouraging
an increase in scientific research on DPM. Our scope is international, with a particular
interest in simple methods and meaningful indicators for communities searching for cost-
effective impact assessment. We do not pretend to have identified the ‘gold standard’ for
impact assessment; instead we have recommended the best of current practice and also
suggested adaptations of indicators and methods used in other contexts. We therefore
strongly encourage innovation and testing of this guidance and would welcome all feedback
through our website www.icam-coalition.org.
In this guidance, we have defined DPM as an intervention that includes activities directly
involving dogs; for example sterilisation, vaccination, parasite control, sheltering, adoption or
euthanasia. We also recognise that it may include targeted education or campaign activities
to enhance dog owner knowledge and alter their behaviour.
This guidance does not tell you how to plan or run an intervention; for information
in designing and implementing interventions see our previous publication ‘Humane
management of dog populations’ (ICAM Coalition, 2008). Instead, this guidance aims to
complement the previous publication by focussing on how to measure the impact of a
DPM intervention, whatever activities it may include. DPM interventions are not driven by a
single common desired impact; rather each intervention may have one or more of a range
of impacts. We have included guidance on measuring and evaluating eight of the most
common impacts which we feel are relevant to most DPM interventions.
The ICAM Coalition is grateful for the many people who have been involved in developing
this guidance. Not least our collaborating partners at the Boyd Orr Centre for Population and
Ecosystem Health at the University of Glasgow, the Jeanne Marchig International Centre for
Animal Welfare Education (JMICAWE) at the University of Edinburgh, the Animal Welfare and
Behaviour group at University of Bristol, and the Department of Veterinary Tropical Diseases
at University of Pretoria. In addition, our thanks goes to the many experts and hardworking
DPM implementers in the field who gave up their time and knowledge so willingly to help
others; many but not all of whom are listed in the acknowledgements. Through your insights
we hope to help others make the world a better place for dogs and the communities
amongst which they live.
1
TABLE OF CONTENTS
Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Background. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Aim of the guidance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Process of developing the guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
How to use this document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
What is Monitoring and Evaluation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Why invest in monitoring and evaluation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Identifying Dog Population impacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Recommended and suggested indicators by impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
■ Impact 1: Improve dog welfare (animal based indicators) . . . . . . . . . . . . . . . . . . . . . . 14
Physical health indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Recommended indicator - Body condition score. . . . . . . . . . . . . . . . . . . . . . . . . . 14
Recommended indicator – Skin condition score. . . . . . . . . . . . . . . . . . . . . . . . . . 15
Suggested indicator – Specific disease and injury, e.g. tethering related
injuries and canine transmissible venereal tumours (TVTs). . . . . . . . . . . . . . . . . . . 16
Suggested indicator – Female : male ratios. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Suggested indicator – Culling of dogs by authorities. . . . . . . . . . . . . . . . . . . . . . . 17
Emotional well-being indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Suggested indicator – Dog-dog interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Suggested indicator – Human-dog interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . 18
■ Impact 2: Improve care provided to dogs (resource based indicators) . . . . . . . . . . . . 20
Suggested indicator – Dog care-giving behaviours in adults. . . . . . . . . . . . . . . . . . . . . 21
Suggested indicator – Dog care-giving behaviours in children. . . . . . . . . . . . . . . . . . . 22
Suggested indicator – Owner engagement with intervention . . . . . . . . . . . . . . . . . . . . 23
■ Impact 3: Reduce dog population density/Stabilise population turnover . . . . . . . . . . 24
Recommended indicator – Dog density along streets . . . . . . . . . . . . . . . . . . . . . . . . . 24
Reduction in dog population turnover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Recommended indicator - Lactating females . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Suggested indicator – Pregnant females. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Suggested indicator – Litters per female. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Suggested indicator – Mortality and age structure. . . . . . . . . . . . . . . . . . . . . . . . . 26
■ Impact 4: Reduce risks to public health. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Recommended and suggested indicators – Dog bites. . . . . . . . . . . . . . . . . . . . . . . . . 29
Recommended indicators - Impact on rabies risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Recommended indicator – Dog rabies cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Recommended indicator – Suspect rabid dog bites . . . . . . . . . . . . . . . . . . . . . . . 33
Recommended indicator – Human rabies cases. . . . . . . . . . . . . . . . . . . . . . . . . . 34
Recommended – Vaccination coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Recommended indicators – Impact on echinococcosis risk. . . . . . . . . . . . . . . . . . . . . 36
Recommended indicator – Infected livestock offal. . . . . . . . . . . . . . . . . . . . . . . . . 37
Recommended indicator – Human cystic echinococcosis. . . . . . . . . . . . . . . . . . . 37
Suggested indicator – Infection in dogs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Recommended indicators – Impact on leishmaniasis risk. . . . . . . . . . . . . . . . . . . . . . . 39
Recommended indicator – Human disease and infection . . . . . . . . . . . . . . . . . . . 40
Recommended indicator – Dog disease and infection. . . . . . . . . . . . . . . . . . . . . . 41
International Companion Animal Management Coalition 2

Table of Contents
■ Impact 5: Improve public perception. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Recommended indicator – Adoption of dogs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Recommended indicator – Attitudes towards dogs. . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Suggested indicator – Dog-related complaints. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Suggested indicator – Human-dog interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Suggested indicator – Cruelty towards dogs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
■ Impact 6: Improve rehoming/adoption centre performance. . . . . . . . . . . . . . . . . . . . . 45
Recommended indicator – Annual live release rate . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Recommended indicators – Intake, net rehoming, footfall and time in shelter. . . . . . . . 45
■ Impact 7: Reduce negative impact of dogs on wildlife. . . . . . . . . . . . . . . . . . . . . . . . . 47
Recommended indicator – Presence of dogs in wildlife areas . . . . . . . . . . . . . . . . . . . 47
Recommended indicator – Predation events and predation impacts . . . . . . . . . . . . . . 47
Recommended indicator – Disease incidence in dogs and wildlife. . . . . . . . . . . . . . . . 48
■ Impact 8: Reduce negative impact of dogs on livestock . . . . . . . . . . . . . . . . . . . . . . . 50
Suggested indicator – Livestock predation by dogs. . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Suggested indicator – Livestock disease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Methods of measurement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
■ Questionnaire surveys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Relevant Impacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Sampling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Recruitment of respondents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Interviewer bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Health and safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Agreement with attitude statements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Sample attitude statements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Analysis of changes in attitudes over time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Using questionnaires to estimate fecundity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Using questionnaires to estimate survival. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Tools available to implement and analyse questionnaires. . . . . . . . . . . . . . . . . . . . . . . 61
■ Participatory research methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Relevant impacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Participatory tools for impact assessment in DPM. . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Creating groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Facilitation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Participatory excercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
■ Street surveys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Route selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Available tools for street surveys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
■ Secondary sources of information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Surveillance effort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Frequency versus incidence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Geographical resolution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

Table of Contents
■ Clinic records. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Biased sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Data to collect for each dog. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Tools available for data recording. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Survival estimates of roaming dogs utilising clinic data . . . . . . . . . . . . . . . . . . . . . 77
■ Behavioural observation method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Assumptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Protocol. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Pre-observation data collection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
■ Street surveys and questionnaires for measuring vaccination coverage. . . . . . . . . . . 83
Marking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Sample size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Street surveys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Questionnaires. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Comparison between methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Making your impact assessment robust. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
■ Ethical review. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
■ Attribution and measuring your intervention effort . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Elements of robust experimental design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Measuring intervention effort. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
■ Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
■ Consistency in method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
■ Increasing and testing observer reliability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Body condition score training and agreement test. . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Using your results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Annex A – Body condition scoring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Annex B – Example of recording sheet for behavioural observations. . . . . . . . . . . . . . 107
Annex C – Six criteria for rabies diagnosis in living dogs (Tepsumethanon et al. (2005)). . . . 109
Annex D – Calculating dog population size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Which intensive survey methods to use? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Roaming owned dogs - questionnaires. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Unowned dogs – mark-resight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Annex E – Sample questionnaire. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Attitude statements used in Colombo, Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Attitude statements used in Tanzania . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Dog attitude Questions, used in Japan and UK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Attitude statements used with 4 year old children in UK, Italy and Spain. . . . . . . . . . . 129

Background
International
Companion Animal
AIM OF THE GUIDANCE

The International Companion Animal Management (ICAM) Coalition was formed in 2006 to
support the development and use of humane and effective companion animal population
management worldwide1. Through our work to achieve this mission, we realised that
effectiveness, or an answer to the question “are we making a difference?”, was often a
subjective assessment of how well an intervention had worked and was not commonly
based on objective scientific measurement. However there were notable exceptions to this
and some excellent innovations in monitoring (regular data collection to measure important
indicators) and evaluation (thoughtful assessment of what the data shows regarding targeted
impacts) occurring around the world that could form a foundation for guidance.
This guidance therefore aims to recommend valid, reliable, practical, and feasible ways
of assessing the impact of domestic dog population interventions; impact assessment is
another term for the learning that can be achieved through monitoring and evaluation. We
hope that this will support academics, practitioners and funders to track progress, learn
and subsequently improve their DPM impact through the use of measurable indicators. The
focus is on applying scientific solutions to real world problems and encouraging an increase
in scientific research on DPM. Our scope is international, with a particular interest in simple,
repeatable methods and meaningful indicators for communities searching for cost-effective
impact assessment.
PROCESS OF DEVELOPING THE GUIDANCE

The process of developing this guidance included an initial literature review, interviews with
experts and practitioners in the field, testing of some of the novel methods of measurement
and indicators, and extensive reviews and consultations with all the ICAM Coalition members
and project collaborating partners.
HOW TO USE THIS DOCUMENT

The following sections include an overview of the process of monitoring and evaluation,
accompanied by explanation and introduction to the key terms used throughout the
document. The main benefits of monitoring and evaluation to dog population management
are also outlined here. They also present an important first step to monitoring and evaluation;
identification of the impacts which we hope to see change. In other words “the difference we
are trying to make through our intervention”. Once the targeted impacts of the intervention
have been identified, navigating through the rest of the document becomes simple and
bespoke to your intervention. Not all interventions will be targeting the same impacts. Select
those that are most relevant to your intervention and turn to the related section(s) of the
guidance document. Select the indicators that are most suitable for you to evaluate the
impact in your specific situation. We recommend that you choose more than one indicator
to allow the change in the impact to be explored, and potentially validated, by more than
one method; also known as ‘triangulation’. Once selected, follow the guidance under each
indicator to the method of measurement that is most practical and feasible for your particular
1
Current members include the International Fund for Animal Welfare (IFAW), World Animal Protection, Humane
Society International (HSI), Royal Society for the Prevention of Cruelty to Animals (RSPCA) International, World
Small Animal Veterinary Association (WSAVA) and Global Alliance for Rabies Control (GARC).
5
Background
dog population. In most cases, addition information on how to implement the method of
measurement is given in a subsequent section. See figure 1 for a flow diagram that shows
this process.
Which impacts would you like to achieve through your intervention?

Impact Indicators Methods of measurement
Impact 1 Body condition score Questionnaire surveys
Improve dog welfare Skin condition
Street surveys
(animal based measures) Specific illness or injury e.g.
tethering injuries and TVTs Clinic records
Female:male ratio
Culling of dogs by authorities Secondary sources of info
Dog/dog interactions
Human/dog interactions Behavioural observation
Impact 2 Dog care-giving behaviours in adults Questionnaire surveys

Improve care provided Dog care-giving behaviours in children
to dogs (resource based measures) Owner engagement with intervention Clinic records
Impact 3 Dog density along surveyed streets Street surveys

Reduce dog density/ Lactating females
Stabilise turnover Indicators Pregnant females Clinic records
Litters per female
Mortality Age structure Questionnaire surveys
Impact 4 Dog bites

Reduce risks to Secondary sources of info
Impact on rabies risk: Dog rabies cases
public health Suspect rabid dog bites Human rabies cases
Vaccination coverage
Impact on echinococcosis risk: Infected livestock offal Street surveys and questionnaires
for vacc coverage
Human cystic echinococcosis Infection in dogs
Impact on leishmaniasis risk:
Human leishmaniasis disease and infection Secondary sources of info
Dog leishmaniasis disease and infection
Questionnaire surveys
Impact 5 Adoption of dogs
Improve public perception Attitude toward dogs Participatory research
Dog-related complaints
Human/dog interactions Behavioural observation
Cruelty towards dogs Secondary sources of info
Impact 6 Annual live release date

Improve rehoming Intake
centre performance Net rehoming
Net rehoming : footfall
Time in shelter
Impact 7 Presence of dogs in wildlife areas

Reduce negative impacts Predation events and impacts
of dogs on wildlife Disease incidence in dogs and wildlife
Impact 8 Livestock predation by dogs

Reduce negative impacts Livestock disease:
of dogs on livestock Infected livestock offal Secondary sources of info
Livestock rabies cases
Figure 1
In the final two Sections ‘Making your impact assessment robust’ and ‘Using your results’,
the guidance covers key ways of ensuring best possible data collection, the basics of
analysis and interpretation, and how to use the results to improve your intervention or
communicate your successes and reasons why your intervention needed to be altered.
We strongly encourage communicating necessary changes to interventions as well as
successes, learning what doesn’t work and needs to be changed is as important as
knowing what does work.

Background
WHAT IS MONITORING AND EVALUATION?

An intervention is a set of activities that aims to make a targeted change or impact to a set
of people, animals or environment.
Example: an intervention that catches, neuters and returns a number of stray dogs
every month in an Asian city. The desired impact of this intervention is to reduce the
density of dogs and to improve the welfare of the stray dogs.
Monitoring requires systematic and routine data collection. Monitoring an intervention

includes measuring the progress of the intervention itself; the intervention effort. Monitoring
also includes regular measurement of indicators that reflect changes in the targeted
impacts, as well as relevant factors in the environment that may also influence the same
impacts as the intervention. Indicators (also known as metrics) are measurable signs of
impacts; they are the things we would see or hear if our desired impact was occurring.
Methods of measurement describe how data relating to the indicators was collected.
In our Asian city example, for the impact of reducing dog density a suitable indicator
may be the number of dogs seen on a set of standard routes along public roads.
The method of measurement for this indicator may be a street survey once every
6 months conducted following a consistent protocol (e.g. same routes, same time of
day and same observation process) for observing dogs on public property. For the
impact of improving dog welfare, we may select the indicator of proportion of the
stray dog population that is emaciated. The method of measurement would again
be the 6 monthly street survey including body condition scoring of all dogs observed.
Monitoring would also include recording the number and location of all dogs
neutered and returned; this represents the intervention effort.
Evaluation of an intervention uses data collected through monitoring, sometimes combined

with other data sourced infrequently and specifically for the evaluation, to answer questions
about “what difference did this intervention make?”; in particular in relation to the targeted
impacts, although unexpected impacts are also important. Evaluation explores the difference
made by the intervention and compares it to what would have happened without the
intervention, also known as the counterfactual (Savedoff et al., 2006).
In our dog population management example, an evaluation may look at the data
relating to dog density in the city where the intervention took place and compare it to
a limited number of routes used in another city where no intervention was used, over
the same time frame. In this case, the question is “has dog density decreased over
time in the city where the intervention took place?” and “how does this compare to
the change in density in a city where no intervention was used?”

Background
Evaluation may also ask whether the intervention could have been more efficient and cost
effective overall, by comparing the cost of the intervention against any savings created
through the impacts.
In summary:
Definition DPM example 1 DPM example 2

An Intervention is a combined Catch, neuter and release of Annual rabies vaccination of dogs in
set of activities with specific roaming dogs in an Asian city a semi-rural region of sub-Saharan
changes or impacts in mind Africa
Impacts are the changes we • Reduce dog density Reduce rabies in dogs and therefore
hope to contribute towards • Improve the welfare of in people
through our interventions roaming dogs
Indicators are measurable signs • Number of dogs seen on a Number of reported dog rabies
of impacts (also known as set of routes along public cases, dog bites and human rabies
metrics); they are the things we roads deaths. People saying “I have not
would see or hear if our desired • The percentage of roaming heard of a rabies case in my village
impact was occurring dogs with emaciated body for many years, it used to happen
condition almost every year”
• People saying “puppies dying
on the streets is a rare sight
these days”
Methods of measurement Observation of the number Quarterly meetings with Municipal
are the techniques we use to and body condition score of all Veterinary Department and General
measure our indicators roaming dogs observed during Hospital to access data on dog
a six monthly ‘street’ survey rabies cases, dog bites and human
rabies deaths
Effort is the immediate result of The number of dogs caught, The number of dogs vaccinated,
your activities neutered and released subsequent vaccination coverage
(% of population vaccinated)
Input is the time and resources Financial costs per dog plus Financial costs per dog plus capital
put into implementing the capital costs of intervention costs of intervention infrastructure
intervention infrastructure
Before an intervention is launched it is necessary to measure the baseline of indicators

selected to reflect the impacts; although note that some methods of measurement are
part of the intervention itself (e.g. the collection of data relating to dogs that pass through
the intervention clinic, see Section Clinic records) and so the baseline would be measured
through the first phase of the intervention. By establishing a baseline for each indicator, the
change in the indicators after the intervention was launched can be measured. Establishing
a baseline may also allow you to state a target within a specific time span and define clear
goals from the outset. For example, a potential target could be to reduce the percentage
of dogs with emaciated body condition from 20% to below 10% within 3 years of the
intervention starting.

Background
WHY INVEST IN MONITORING AND EVALUATION?

Monitoring and evaluation has many purposes: to inform donors of the impacts created
through their funding; to inform the public of any impacts on them and/or their dogs as
beneficiaries of the intervention; to provide evidence for lobbying to sustain or replicate an
intervention; and to compare intervention and their relative impacts. But the most important
purpose is for improvement of the current intervention and subsequent interventions via
learning what is, and is not, successful and the dissemination of these findings to a wider
audience. The potential for learning and subsequent improvement can be visualised in an
intervention or ‘project’ cycle (blue text and boxes indicate monitoring and evaluation
activities):
Test assumptions, Develop concept

refine concept of dog population
and evolve the situation and
intervention intervention
Initial assessment
and analysis
Intervention planning
Set impacts and

Evaluation and
identify indicators
learning
Develop monitoring
and evaluation plan
Implementation
Monitor progress
of intervention and
indicators

Background
The importance of monitoring and evaluation to learning cannot be overstated. Many

interventions begin with a very basic understanding of the system they are hoping to impact
upon, relying on assumptions about the root causes of problems experienced by dogs,
and the communities amongst which they live. By using monitoring and evaluation, these
interventions can test their assumptions about how their activities affect dogs and people
using objective data. This will provide them with evidence of what is working and what
needs to be changed. It is therefore essential that intervention staff and donors remain
flexible and open minded to the evidence produced through monitoring and evaluation,
ready to implement changes when needed.
Evaluating the impact of interventions and therefore ensuring our policies and intervention
designs are based on the best available evidence is a concern for all organisations looking
to make the world a better place. The human development movement has been striving for
improvement in evaluating impact for decades. In the Centre for Global Development 2006
report ‘When will we ever learn? Improving lives through impact evaluation’ (Savedoff et al.,
2006), Bill Gates is quoted as saying “Success depends on knowing what works” (pp iv). It
is heartening to see that we are not alone in the struggle to evaluate impact and therefore
develop evidence-based understanding about what works and what doesn’t. However,
human development-related impact evaluations are numerous, and systematic reviews
involving many tens of evaluations can be conducted to rigorously assess the impact of
a policy or intervention across a number of different contexts. Furthermore, the human
development field is supported by a shared understanding of what indicators are important;
for example, there are 60 official indicators for the 10 Millennium Development Goals.
However, systematic reviews of a wide body of impact assessments and internationally
agreed and standardised indicators, are only an aspiration for the dog population
management field at present. By developing a set of recommended and suggested
indicators, and ways of measuring these for dog population management, we hope to
provide a framework and inspiration for future evaluations that will further our understanding.
IDENTIFYING DOG POPULATION IMPACTS

Monitoring and evaluation requires an understanding of what impacts the intervention is
striving to achieve. If you don’t know where you are going, how will you know when you get
there?
In the project cycle described previously, the cycle begins with an intervention concept; the
desire to intervene to reduce a threat (such as zoonotic disease) or improve the situation
for a group of beneficiaries (such as the welfare of roaming dogs). This desire is turned into
an intervention plan with clear impacts and associated indicators, as well as budgeted and
timed activities suitable for the dog population dynamics and dog ownership pattern of
the location. A critical stage of intervention concept and planning is initial assessment and
analysis. This process explores and understands the root causes of the visible problems
in the specific location, including the sources of dogs causing or experiencing these
problems, in order to inform bespoke intervention planning. This stage is described in detail
in the ICAM Coalition Guidance on Humane Dog Population Management (available at
www.icam-coalition.org). This includes in-depth consultation with all relevant stakeholders
to establish an agreed, comprehensive understanding of the local dog population and a

Background
realistic set of impacts for the intervention. Ensuring these impacts are realistic requires a
further stage of establishing the logical steps describing how the intervention will achieve
the desired impacts with your specific dog population and community in mind. This is
also termed a ‘theory of change’ and is sometimes expressed as a ‘logical framework’ or
‘logic model’2. This stage will help test whether your intervention is truly suited to achieving
the desired impacts and will explicitly state the intervening objectives that should also be
monitored to establish attribution and whether the intervention is going to plan.
The intervention itself may include a range of activities, selected to suit the problems and
root causes of the location. The ICAM Coalition Guidance on Humane Dog Population
Management describes several of these potential activities; education, legislation,
registration and identification, sterilisation and contraception, holding facilities and rehoming
centres, euthanasia, vaccination and parasite control and controlling access to resources.
In this current guidance we have identified indicators suitable for reflecting change in eight
of the most common impacts resulting from interventions that include one or more of these
activities. Most interventions will have a subset of these impacts in mind as opposed to all
eight; these impacts may be worded slightly differently but we hope similar enough that
they can be matched against one of the impacts described here. Selection of indicators will
depend both on what indicators appear most relevant for your local dog population and
your intervention theory of change, and also on what methods of measurement you can
practically perform with the resources available.
We appreciate that this section has described an ideal situation where clear root causes to
problems have been established, building a strong foundation for planning an intervention
with identifiable impacts and indicators. In many situations, interventions work with a range
of assumptions about the impacts they will be able to influence. For example: interventions
that include sterilisation of dogs may hope that this reduction in reproduction will improve
dog welfare, thoughtfully designed monitoring and evaluation will provide the evidence
required to test such assumptions; questions about the source of unowned dogs (is this
population self-sustaining or maintained by recruitment from owned dog populations?) can
also be explored by monitoring and evaluating how interventions impact on the density and
stability of these different populations. Further, some interventions will experience unintended
consequences and monitoring and evaluation will need to be conducted with receptiveness
to such unplanned effects. In short, although a clear plan for the intervention and how to
assess its impacts is ideal, in reality, monitoring and evaluation requires flexibility and an
open mind to what we can learn.
2
Potential sources for further guidance on developing theories of change and/or logical frameworks include
the definition, tools and resources available at www.theoryofchange.org, the Open Standards self-paced online
tutorials from the Conservation Measures Partnership available at http://cmp-openstandards.org/
and ‘Sharpening the development process’ book by INTRAC available
from http://www.intrac.org/resources.php?action=resource&id=345

Recommended and Suggested
Indicators by Impact
Photo: © World Animal Protection
Recommended and Suggested

Indicators by Impact
International
Companion Animal
This section presents indicators for reflecting change in 8 common impacts targeted by dog
population management interventions, for each impact more than one indicator is presented;
1. Improve dog welfare (animal based indicators)
2. Improve care provided to dogs (resource based indicators)
3. Reduce dog density/stabilise turnover
4. Reduce risks to public health
5. Improve public perception
6. Improve rehoming centre performance
7. Reduce negative impact of dogs on wildlife
8. Reduce negative impact of dogs on livestock
Some indicators are relatively well tested and we recommend these to measure changes
in the stated impact. Others are relatively novel to DPM and appear valuable for measuring
change in the stated impact but are not yet well tested; these are suggested indicators. We
would value any feedback you have on the use of these suggested indicators, with the aim
of promoting these to recommended if they prove valid (able to truly measure change in the
impact they were supposed to be reflecting), reliable (repeated measures would produce the
same result) and feasible (this indicator can be measured with methods that are possible to
perform in most locations).
Following each recommended or suggested indicator is a description of the methods of

measurement that can be used to collect data on these indicators. Refer to the Methods
of measurement section for further guidance on how to conduct these methods.
The indicators and methods of measurement included in this guidance document have
been selected because they can be meaningful reflections of change in important impacts
and are affordable options that can be implemented by most DPM interventions. However
we suggest that further support from scientific stakeholders, such as universities, would be
beneficial; including for planning data collection, analysis of data, objective interpretation of
results and subsequent publication in peer-reviewed journals (preferably open access) to
support credibility and dissemination of findings to other DPM interventions.
Note that the indicators included here are those relating to impact (e.g. improvements
in dog welfare or reduction in public health risks) and not effort (e.g. the number of dogs
vaccinated, sterilised or otherwise intervened). See section ‘What is Monitoring and
Evaluation?’ for further explanation of these terms.
13
Impact 1: Improve Dog Welfare
(Animal Based Indicators)
International
Companion Animal
Dog welfare can be defined as how well a dog is coping with its environment (adapted from
Broom, 1991). Failure to cope will result in the dog suffering, but a dog that copes well with
the challenges presented by the environment can have acceptable or even good welfare.
IMPACT 1
Dog population management interventions may aim to improve dog welfare either by
adapting the environment to make it easier for dogs to cope, for example how people treat
or care for dogs, and/or by the intervention improving the coping mechanisms of the dogs
themselves, for example vaccination helps dogs to mount an immune response to fight the IMPACT 2
challenge of specific diseases. Furthermore, a dog’s welfare includes not only its physical
health but also its emotional well-being (adapted from Dawkins (2006)). Hence assessing
dog welfare is best achieved by selecting indicators that reflect both physical health and how
the dog feels, as expressed through its behaviour.
IMPACT 3
This impact of improving dog welfare focuses on ‘animal based’ indicators, these require
measuring the welfare status of the dogs themselves. This is closely related to Impact 2 –
‘Improve care provided to dogs’, which focuses on ‘resource based’ indicators, focused on
what is provided to dogs to influence their welfare. These impacts are clearly closely linked
and we suggest that measuring both is ideal.
IMPACT 4
Physical Health Indicators IMPACT 5

Recommended indicator - Body condition score
Body condition can be scored through observation alone without the need for physical
examination of the dog and hence is relatively safe and quick to conduct. The scores IMPACT 6
are awarded on the basis of body fat coverage and not on coat health or injuries. Body
condition can range from emaciated to obese (1 – 5); hence it reflects the quality and
quantity of food resources and is affected by concurrent underlying conditions such as
disease and parasite load. Several studies have shown an increase in body condition score IMPACT 7
following interventions that included surgical neutering and/or basic veterinary health care
(e.g. Sankey et al., 2012; Steinberger, 2012; Totton et al., 2011; Yoak et al., 2014).
There are several scoring systems available. We recommend a 5-point scoring system
(1 = Emaciated; 2 = Thin; 3 = Ideal; 4 = Overweight; 5 = Obese, see Annex A) as this is
IMPACT 8
quick to learn and establish good inter-observer reliability (see Section ‘Increasing and
testing observer reliability’). When using body condition score for monitoring, use only the
scores for adults and exclude both puppies and lactating females. Body condition score
systems for puppies are different to those of adults, although systems do exist and could
be used in addition, as puppies are difficult to observe when surveying and tend to occur in
clusters, they provide less reliable data than adults. Females, even if they start in good body
condition, may lose condition when lactating. This can be quickly regained when puppies
are weaned, so their condition is not a reliable reflection of general population health.
The recommended indicator is the percentage of adults (excluding lactating females)

with a body condition score of 1 (emaciated). If a very low proportion of the population
is already at body condition score 1 it will be difficult to show significant change over time
because the level is already very low, hence the indicator can instead be the percentage
14
of the population at body condition scores 1 and 2 (emaciated and thin). Note that
although only dogs with poor body condition are used as the indicator, all dogs should be
scored for body condition, as scoring is more likely to be subconsciously skewed when only
focusing on a proportion of the population.
This indicator can be measured through street surveys. With this method of measurement
the indicator will reflect the welfare of the roaming dog population. Alternatively body
condition score data can be collected from dogs as they pass through an intervention IMPACT 1
(see section on ‘Clinic’ records). With this method of measurement the indicator will reflect
the welfare of the population accessed by the intervention; this will be owned dogs if the
intervention encourages owners to bring their animals to the clinic, representing a select
subset of dogs which may have a different body condition to the general dog population. IMPACT 2
Recommended indicator – Skin condition score
Dogs can suffer from skin conditions with a range of causes, including fungal pathogens,
parasites and allergies. In the context of using skin condition as an indicator of dog welfare
at the population level, a diagnosis of the cause of the skin condition is not required; a skin
IMPACT 3
condition indicates poor welfare both due to the discomfort of the skin condition itself and
also potentially reflecting an underlying health problem. Importantly, a visible skin condition,
without diagnosis of cause, can be scored through observation alone without the need for
physical examination. A visible skin condition includes any sign of hair loss or scaly, inflamed IMPACT 4
or sore skin but does not include dirty fur, elbow keratosis (thickened skin at the elbows),
skin tumours or hernias.
5
The simplest scoring system is presence or absence of a visible skin condition. This has
been used successfully in evaluation of dog population management interventions in several IMPACT
locations (e.g. Garde et al. (2012) in Chile; Sankey et al. (2012) in Sri Lanka; and Totton et al.
(2011) in India). The indicator is the percentage of adults with a visible skin condition.
It should be noted that prevalence of skin conditions may change in across seasons
(e.g. fungal infections and allergic responses may have seasonal fluctuations); hence
comparisons should be made with data collected at the same time of year.
IMPACT 6
It is possible to develop an indicator for skin condition that includes some measure of
7
severity. For example no skin condition,
mild skin condition affecting <20% of the IMPACT
body and severe skin condition affecting
>20% of the body; which is potentially
beneficial if a more significant negative
welfare impact is assumed for a severe
skin condition as compared to a mild
IMPACT 8
condition. However, this requires more
training effort with observers to ensure they
can reliably score a mild versus a severe
case. In addition, a reduction in severe
cases will probably occur in parallel to a
reduction in all visible skin conditions and
hence including this measure of severity
may not improve the sensitivity of the
indicator to change but it will be more

difficult to establish inter-observer reliability. With this in mind, we recommend using simply
presence/absence of a skin condition.
As with body condition score, this indicator of visible skin conditions can be measured both
through street surveys, which will reflect the welfare of the roaming dog population or from
dogs as they pass through an intervention (‘clinic’ records), which will reflect the welfare of
the population accessed by the intervention.
IMPACT 1
Suggested indicator – specific disease and injury,
e.g. tethering related injuries and canine transmissible
venereal tumours (TVTs)
Dog populations are generally susceptible to the same disease and injury risks, however
IMPACT 2
there may be particular diseases or injuries that are especially prevalent, or even relatively
unique, to some locations and these may be targeted by the intervention for reduction.
For example, in some locations, tethering dogs is common practice and is associated with
particular injuries such as wounds around the neck. In other locations canine transmissible IMPACT 3
venereal tumours (TVTs) are relatively common. These two examples are described here in
more detail, but the principles can be applied to any specific diseases or injuries that the
intervention is targeting and therefore would like to monitor.
Tethered dogs that have experienced injuries may be brought to intervention clinics for
IMPACT 4
treatment and hence clinic records may be a suitable method of measuring a change in
the prevalence of these injuries (see Section ‘Clinic’ records for more details). However,
dogs brought to the clinic by their owners could represent a particular biased sample,
further these biases could change over time. A potentially more unbiased measure of IMPACT 5
the prevalence of these injuries could be collected during a questionnaire survey. When
conducting a questionnaire survey at the household, the owner could be asked to show the
interviewer their dogs at which time they could be assessed (either using a clinical exam or
just visibly; the method used would need to be kept consistent) for injuries. Each dog could
also be photographed to help with later more detailed analysis, although this must be done IMPACT 6
with the owner’s consent. This could provide data on the indicator of percentage of dogs
with tethering related injuries. This indicator could be further developed with categories of
7
injury type or levels of severity.
IMPACT
Canine transmissible venereal tumours (TVTs) are transmitted between dogs during
copulation (mating), licking, biting and sniffing tumour affected areas. The tumour cells are
themselves the infectious agents. Although copulation is not the only route of transmission,
it is a common route as the tumours are often sited on and within the genitalia. Interventions
that include sterilisation of dogs may therefore affect the prevalence of these tumours in
IMPACT 8
the population, not only in the sterilised dog population but also unsterilised dogs that may
have a lower risk of contracting TVTs if fewer dogs are engaging in reproductive activity.
Interventions may also include treatment or euthanasia of infected dogs and these may
also reduce prevalence in the wider population over time as the reservoir of infected dogs
decreases. Moreover, reduction in roaming dogs and increases in confinement (without
concurrent increases in sterilisation or treatment) also seemed to be linked to reductions in
TVT prevalence in the UK, presumably due to reductions in transmission risk as the number
of dogs engaging in reproductive behaviour is reduced. TVTs can present a welfare problem,
in particular in the case of secondary bacterial infections, myiasis (maggot infestation) or

when the tumour becomes large enough to cause an obstruction or impede movement.
Despite this theoretical link between TVTs and welfare, the prevalence of TVTs as an
indicator of dog population welfare has not been widely reported, hence it is presented here
as a suggested indicator.
Although TVTs can grow to a size that makes them easily visible, the majority will only be
apparent on clinical exam or even during surgical sterilisation. Hence street surveys are
not advised as a method of measurement, as the prevalence will be very low. Instead, the IMPACT 1
presence/absence of TVTs should be recorded as the dogs pass through an intervention
that allows for clinical examination or surgical sterilisation (see section ‘Clinic’ records).
The indicator is therefore the percentage of dogs with a TVT. This should be reported
separately for dogs that were assessed during clinical exam and found to have a TVT and
were not subsequently sterilised (e.g. in the case of euthanasia), to those that were observed
IMPACT 2
to have a TVT during surgical sterilisation (which will include those dogs that were identified
has having a TVT and subsequently sterilised); the prevalence may be different between
these two populations as surgical sterilisation and clinical examination may have a different
chance of exposing the presence of a TVT. IMPACT 3
Suggested indicator – Female : male ratios
A change in the ratio of female : male dogs from male skewed to equal ratios over time
may function as an animal based indicator for dog welfare; the process behind this is IMPACT 4
assumed to be a change in how people are treating dogs of different genders. Dogs
produce, on average, equal ratios at birth, however we often observe a skew towards more
males than females in owned and roaming dog populations, especially in communities
where sterilisation or other forms of reproduction control are not widely accessible. This is
presumably because females present more of a management concern to owners; unwanted
IMPACT 5
litters and regular oestrus leading to males fighting to gain access to the female are reasons
that male dogs may be preferred. In these communities, owners may be preferentially caring
for males, adopting/buying male dogs or purposefully killing female puppies leading to a sex
skew towards males. Interventions that provide improved access to spaying females may IMPACT 6
reduce this bias against females and hence a more equal ratio of females:males will become
apparent over time. The sex ratio is measured using either/both street surveys of roaming
dogs or questionnaire surveys of owners; both these methods are covered in the Sections
‘Street surveys’ and ‘Questionnaires’. IMPACT 7
Suggested indicator – Culling of dogs by authorities
This indicator is regarding the culling of roaming dogs in situ, i.e. culling in the streets, with
no holding period opportunity for reuniting or rehoming the dog (note that systems including IMPACT 8
a holding period for reuniting/rehoming before euthanasia of dogs that cannot be rehomed
is included in the section on rehoming/adoption centre performance). This assumes that the
method used is inhumane and hence carries a welfare cost to the dogs. Some interventions
are developed as an alternative to widespread culling and hence a reduction or cessation
of culling roaming dogs will be seen as important indicators of an improvement in
dog welfare. Note that culling of dogs may have been halted to allow the intervention
to commence and so more suitable indicators may include halting any resumption of
culling in the intervention area and geographical expansion of replacing culling with
the alternative intervention. Accessing data on culling by authorities will probably require
close collaboration with the relevant authority as this data may not be immediately publicly
accessible. Use of secondary sources of information is covered in the Section ‘Secondary
sources of information’.

Emotional well-being indicators
Suggested indicator – Dog-dog interactions

The social behaviour of animals can be an indicator of their welfare; it can both reflect their
1
underlying emotional state, for example fear underlying aggression or relaxation allowing
expression of play behaviour, and can in itself lead to a welfare problem, for example in the IMPACT
case of injuries caused by fighting. Social behaviour has been used for the assessment
of welfare in many other species (for example, social behaviour forms part of the Welfare
Quality protocols for pigs, cows and poultry welfare assessment, www.welfarequality.net)
and dogs in a shelter or laboratory setting. It is assumed that healthy, stable social groups
will show more affiliative and less agonistic interactions. However it has not been used
IMPACT 2
previously for assessing the welfare of roaming dogs and so it is presented here as a
suggested indicator.
Observation of roaming dogs can be performed using a standard protocol and all social
interactions between dogs scored according to their ‘outcome’: amicable, neutral, mating or
IMPACT 3
aggressive. The behaviour of the dogs at the outset of the interaction may imply a different
intent however it is the final outcome of the interaction that is recorded. The indicators are
4
therefore the percentage of amicable interactions and the percentage of aggressive
interactions out of the total of all dog-dog interactions recorded. Only dogs over four IMPACT
months of age should be scored for social behaviour. The social interactions between
puppies, and the behaviour between puppies and adults, are potentially different from those
between adults; arguably, interactions involving puppies follow a more hardwired pattern of
behaviour that is less affected by concurrent stress than behaviour between adult animals
and so may be a less sensitive indicator of the underlying welfare state of the population.
IMPACT 5
The social behaviour indicators are measured through direct behaviour observation of
roaming dogs at a sample of sites selected for their high frequency of interactions between
dogs. Section ‘Behavioural observation method’ details this method in full. It should be
noted that this method of observing interactions will be biased towards noisy or more
IMPACT 6
obvious social behaviours and observers are likely to miss more subtle social signals.
However, so long as the method of observation remains the same over time this effect will
be consistent and hence the indicators will still reflect changes in social behaviour, albeit the
more noisy or obvious ones.
IMPACT 7
Suggested Indicator – Human-Dog Interactions
The way that people behave towards animals can impact significantly on their welfare
(Hemsworth, 2003). People may intend to simply increase the distance between themselves
IMPACT 8
and dogs, for example by shouting or throwing stones, but repeated use of this behaviour
can result in dogs becoming fearful of people. As roaming dogs are almost continuously in
the presence of people, this high level of fear may result in prolonged stress which in turn will
impact negatively on their welfare. Alternatively, repeated acts of kindness between people
and dogs, including feeding and petting, may reduce fear, alleviate stress and improve
welfare. There has been significant research into measuring the behaviour of stockpersons
around farm animals (Hemsworth, 2003) but very little into the behaviour of people towards
roaming dogs, hence this indicator is presented here as a suggested indicator.
Recording all behaviours of people towards dogs in a busy scene of both roaming dogs
and people would be difficult to perform without using videos and later transcription of
behaviours during reduced speed playback. Hence for efficiency, we suggest using the

method of measurement described in section ‘Behavioural observation method’ which

requires only recording those extremes of human behaviour, both positive (such as feeding
a dog) and negative (such as hitting a dog). This allows for real-time recording of behaviour
which is more efficient (no need for later transcribing), less obtrusive and so less likely to
attract attention or alter peoples’ behaviour towards dogs whilst you are observing. The
indicators are therefore the percentage of positive human behaviours and the percentage
of negative human behaviours out of total of all ‘extremes’ of human-dog interactions.
IMPACT 1
IMPACT 2
IMPACT 3
IMPACT 4
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Impact 2: Improve Care Provided To
Dogs (Resource Based Indicators)
International
Companion Animal
The care provided to dogs can subsequently affect dog welfare and public health, but
improving the way people care for their dogs can be a desired impact in its own right.
Establishing indicators for the care of dogs requires a concept of what care is desired.
IMPACT 1
The International Fund for Animal Welfare (IFAW) have developed the term ‘adequate
guardianship’, defined as ‘the resources, environmental conditions and social interactions
2
necessary to meet an individual animal’s physiological and psychological needs necessary
to maintain an acceptable level of health and well-being’. This encompasses provision of the IMPACT
following:
Resources:
• Food
Environmental conditions:
• Safe, appropriate shelter
Social Interactions:
• Companionship of
IMPACT 3
• Water • Avoidance of injury and people and other
cruelty dogs as suited to
• Basic preventative and the individual dog
curative veterinary care • Opportunity for exercise
IMPACT 4
Importantly, adequate guardianship also requires a guardian to ensure these conditions
persist; hence this impact of improving care provided to dogs is focused on the behaviour of
people towards their dogs (note their caring behaviour should be reflected in the condition of
IMPACT 5
their dog; this is included under the impact of improving dog welfare). This guardian needs
to at least provide the basics of suitable food/water, shelter, basic veterinary care, prevent
intentional cruelty and behave in a way consistent with community health and safety.
The precise behaviours required will depend on the location and what a dog needs to remain
IMPACT 6
in a good state of welfare, taking into account local environmental conditions and diseases.
For example in northern Canada, dogs may require shelter from the cold and unfrozen
water to be provided daily by guardians, whilst in sub-Saharan Africa access to shelter from
the sun, water and regular deworming for Echinococcus granulosus might be the priority
IMPACT 7
guardianship behaviours. In particular, ensuring the consistent provision of basic veterinary
care may be difficult in those countries where veterinary care is difficult to access, and so the
definition of ‘basic’ will necessarily need to change dependent on location and local disease
risks. Note that confinement on private property and under guardian control when on public IMPACT 8
property (leash or otherwise) is not stated in the aforementioned requirements for adequate
guardianship. Roaming on public property may not necessarily lead to welfare problems,
may allow the dog to meet some of its welfare needs (e.g. accessing companionship and
exercise) and may also be acceptable in some cultures. In other cultures this may not be
acceptable and may even be legislated against, in which case confinement will be part of
adequate guardianship in that country. There may also be specific human behaviours that
an intervention aims to reduce, such as permanent confinement of dogs on tethers or killing
female puppies as a form of population management, which could be replaced by alternative
management strategies made accessible by the intervention.
20
Impact 2: Improve Care Provided to Dogs
As the breadth of potential indicators is wide, and dependent on the characteristics of the
location, only a selection of indicators are mentioned here and innovation to develop the
most meaningful indicators for the location is particularly encouraged.
This impact of improving care provided to dogs focuses on ‘resource based’ indicators;
what is provided to dogs to influence their welfare. The assumption is that improved care
will lead to improvements in dog welfare. However we suggest it is ideal to also measure
changes in the ‘animal based’ indicators covered by Impact 1 – ‘Improve dog welfare’. IMPACT 1
Suggested indicator – Dog care-giving behaviours in adults
2
Measuring changes in guardianship behaviours will require asking people about their
behaviours. This is commonly done by using a questionnaire; see Section ‘Questionnaire IMPACT
surveys’. Indicators relating to dog care that can be measured by data collected through a
questionnaire include: percentage of dogs that are sterilised; percentage of dogs that were
vaccinated in the last 12 months; percentage of dogs that were dewormed or treated
for ectoparasites in a time period suitable for the local conditions; percentage of dogs
fed at least once daily; percentage of dogs given water daily; and percentage of dogs
IMPACT 3
with permeant access to shelter.
Note that conducting a questionnaire face-to-face at the household also provides an

opportunity to observe and record the condition of any dogs owned by the household. This
can provide an additional source of data on dog welfare and also allows for confirmation
IMPACT 4
of care-provision behaviours reported by the respondent (e.g. ‘my dog has access to
shade’ can be checked through observation of the dog). The sample questionnaire includes
5
space to record body condition score and skin condition of any of the household’s dogs
observable by the interviewer. IMPACT
A novel indicator was used to reflect a
change in investment in dog care on
a Lakota reservation in the US. The
intervention managers observed an IMPACT 6
increase in the sales of commercial dog
food, despite a decrease in the dog
7
population size across the same period
of time (Steinberger, 2012). Measuring IMPACT
changes in commercial dog food
sales requires contacting dog food
sales points and may only be suitable
where communities are reliant on a few
known sales outlets and where this is
IMPACT 8
relevant to your intervention.

Suggested indicator – Dog care-giving behaviours in children

Where an intervention includes school programmes to improve children’s dog care
behaviours, a questionnaire can be used to assess changes in knowledge and attitudes
towards dogs; importantly this assumes that a change in knowledge and attitude will lead
to behaviour change. This questionnaire can be delivered to a class of children before,
immediately after and several months later (ideally also just 2 weeks later) to assess whether
knowledge is increased and also retained through the programme. Not all classes need IMPACT 1
to be monitored intensively in this way; a sample of 2 or more classes per age group or
across schools can provide an indication of how successful the education programme is at
changing knowledge and attitudes of children towards dogs. Note that in some countries
conducting questionnaires of children in schools will be regulated and appropriate approvals
will need to be sought.
IMPACT 2
The questionnaire will need to be designed to suit the aims of the education programme.
If expanding knowledge of good dog care is the aim, the children could be presented with
multiple-choice questions about relevant care behaviours, hence the indicator is the % of
correct answers to dog care questions. If a change in attitudes is also the aim, a set of
IMPACT 3
attitude questions can also be asked and the indicator is a change in the average attitude
score. A review of relevant literature, and a list of validated attitude scales for children, can
be found in the report ‘Promoting a ‘duty of care’ towards animals among children and
young people’ (Muldoon et al., 2009). One example of a validated attitude scale for 4-year IMPACT 4
old children, relating specifically to dogs, is a 9-item list developed by Lakestani et al. (2011)
which can be found in Annex E .
The children are asked to answer the questionnaire individually without talking to one
another. The questionnaire should be kept very short, taking no more than 5 minutes to
IMPACT 5
complete. It’s important that the children are reassured that this is not a test that matters to
them individually and they need not worry about the results.
Such an approach will test changes in knowledge and attitudes following an education
programme but does not measure actual behaviour towards dogs. Testing children’s
IMPACT 6
behaviour with real dogs would be difficult to achieve whilst ensuring health and safety of
both children and dogs, but puppets or vignettes (short stories that set-up a scenario) can
be used to ask children how they would respond in certain situations. For example, a story
that builds up the scenario of walking home from school and finding a dog on the path that IMPACT 7
you do not recognise: What would you do? Or for dog-care, you wake up in the morning,
get dressed, go downstairs and there is your dog: What does he need this morning? The
8
indicator would be the % of correct behavioural responses described for a set of dog-
related situations. For young children, this would need to be run with just small groups of IMPACT
children or individuals and the scenarios verbally acted out, however older children could
read the vignettes and write their responses and so could be done with entire classes in a
test set-up.

Suggested indicator – Owner engagement with intervention

For interventions that include veterinary clinics or delivery of basic health care through
field stations, an indicator of improved care may be reflected by an increase in owner
engagement with the intervention. Indicators that could measure this increase in owner
engagement include an increase in the number or proportion of dogs brought to a
clinic/field station by their owner/carer (proportion calculated by comparing the number
of dogs brought by an owner or carer compared to the number of dogs caught and brought IMPACT 1
in by intervention staff, where this approach is used in the intervention). Where relevant, the
amount donated or paid by owners for the intervention services may be measured over
time to reflect changes in financial investment by owners. These indicators are all measured
through clinic records, covered in the methods of measurement section. IMPACT 2
In some locations, the use of local veterinary services
could also be used to reflect changes in dog care,
with the number of dog appointments per unit
time at local veterinary surgeries as the indicator.
A change in such an indicator may result from
IMPACT 3
interventions that encourage veterinary care through
campaigning and education programmes, although
an increase in the use of local veterinary services
may also occur when the intervention itself is offering IMPACT 4
access to sterilisation/vet care. An increase in both
intervention services and the use of local veterinary
services was found in several counties in the US
after an intervention and was hypothesised to be
due to social positive reinforcement and widespread
IMPACT 5
marketing/publicity (Frank and Carlisle-Frank, 2007).
IMPACT 6
IMPACT 7
IMPACT 8

Impact 3: Reduce Dog Population Density
/Stabilise Population Turnover
International
Companion Animal
Reducing dog population size or density is commonly stated as a desired impact of DPM.
This is targeted towards roaming or ‘stray’ dogs, as opposed to wanting to reduce the
overall dog population. Stabilising the population, also termed reducing population turnover,
IMPACT 1
may also be desired (reducing births and deaths, with each dog living longer on average).
Reduced turnover could confer welfare benefits (e.g. fewer puppies being born and dying)
and it may also be useful for disease control; if vaccinated dogs live longer, and fewer
(naturally susceptible) puppies are born, the proportion of the population that is immune to IMPACT 2
the disease (termed herd immunity) will remain higher for longer, providing a better barrier to
disease transmission.
IMPACT 3
Recommended indicator – Dog density
along streets
The indicator of the number of roaming dogs per km (or mile) of street surveyed is an
IMPACT 4
indicator of dog density and is preferable to an estimate of total roaming dog population
size (also known as abundance) or density estimates based on area. Firstly, it may be an
ideal reflection of the public perception of the roaming dog ‘problem’; although the average
citizen has no concept of the total number of dogs roaming in their town, they have a very IMPACT 5
real experience of the number of dogs encountered on the way to work or on their children’s
journey to school. Further, urban areas often both expand and become more dense (losing
open spaces to more streets and associated housing) leading to changes in the total dog
population that was beyond the influence of any DPM intervention and may be imperceptible
to the average citizen. However the average number of roaming dogs along streets will
IMPACT 6
correlate with the chance of an average citizen coming across a roaming dog as they travel
along the streets and hence remains a valid indicator of the impact of the intervention.
Comparisons can also be made of the average number of roaming dogs per km of street
surveyed between locations and perhaps most importantly how this number is changing IMPACT 7
over time, allowing interventions in different locations to be compared in terms of how they
impact on dog density. Finally, measuring the number of dogs per km of street surveyed
can be done relatively easily compared to establishing an accurate estimate of total
population size. IMPACT 8
The Section ‘Street surveys’ describes a method of
observing dogs along a set of standard routes. These
surveys can measure the number of dogs seen and
their visible welfare state using the indicators introduced
elsewhere in these guidelines (i.e. body condition
score and skin condition score). Street surveys are
then repeated over time (recommended every 6 or 12
months), using exactly the same routes and the same
counting protocols, to establish how this indicator of
number of dogs is changing. It is important that data
from the same time of year is compared, as roaming
dog numbers and welfare can vary with season.
24
Impact 3: Reduce Dog Population Density/Stabilise Population Turnover
Keeping the time of day of observation consistent is also very important, as this indicator is
actually the number of dogs per km of street surveyed at a particular time of day, and the
number will change throughout the day as dogs respond to movements of people, traffic
and change in environmental temperatures. The best time of day for street surveys is peak
roaming time, usually at dawn when traffic is lightest.
In some cases, an estimate of total roaming dog population size will be needed, perhaps
most commonly ahead of planning a new intervention. An estimate of population size is not IMPACT 1
required for monitoring and evaluation of intervention impact and so is not discussed in any
detail here, but see Annex D for more information on conducting population size estimates.
IMPACT 2
Reduction in dog population turnover
Recommended indicator - Lactating females IMPACT 3

An important component of dog population turnover is fecundity, i.e. the rate of
reproduction. Puppies only have time-limited passive immunity from their mothers for a
short period after birth, hence they are more susceptible to disease and its subsequent
transmission, making them an important factor in disease control. This limited immunity also
contributes to their high morbidity and mortality, so their welfare is also often compromised.
IMPACT 4
However, reliably measuring the number of puppies in a roaming dog population can be
challenging. Puppies are difficult to spot due to both their small size and also because
they spend some of their time confined to a den; in addition when they are spotted
they tend to occur in clusters with their litter mates. These factors combined means the IMPACT 5
observable percentage of puppies in a population vary considerably from survey to survey.
In comparison, the percentage of lactating females in the roaming dog population is
a more reliable statistic because lactating females are easier to spot and do not appear in
clusters. Hence lactating females are an indicator of fecundity in the roaming dog population
and a proxy for the number of puppies.
IMPACT 6
The percentage of lactating females can be measured as dogs come through an intervention
clinic by using clinic records, although this may depend on the intervention; some veterinary
surgeons may discourage owners and catching teams from bringing lactating females for
sterilisation. However the percentage of lactating females in the roaming dog population can
IMPACT 7
also be efficiently measured during the same street surveys used to measure the number of
dogs per km of street surveyed and associated welfare indicators. More details are provided
in the Section ‘Street surveys’ under methods of measurement.
IMPACT 8
The percentage of lactating females should be calculated as the percentage of all females
that are visibly lactating as this represents the reproductive activity of the roaming dog
population. As opposed to the percentage of only unsterilised females that are lactating,
which is unlikely to change with the intervention unless there is a significant change in
the resources available to the remaining unsterilised females. The percentage of lactating
unsterilised females can be monitored in addition if breeding is suspected to be resource
limited in the intervention location, although this would require sterilisation status to be
clearly visible, e.g. where ear notches have been used by the intervention to mark sterilised
dogs.
A female is defined as lactating if her mammary glands are visibly swollen. Teat size may not
be a reliable sign of lactating as females that have had previous litters may show enlarged

teats. Females in the very late stages of pregnancy may also show enlarged mammary
glands just before giving birth; because this is also a sign of active breeding these females
can be included in the category of lactation to simplify surveying.
Suggested indicator – Pregnant females

The percentage of pregnant females in the intervened population may be used as a
potential indicator of fecundity, an important component of population turnover. However, no IMPACT 1
examples of its use could be found and hence it is presented here as a suggested indicator.
It is possible that as the health of the population improves so the proportion of unsterilised
females that become pregnant and carry a litter to term will increase, similarly the age
at which dogs become pregnant may also decrease as their health status improves. In
populations of dogs that show strongly seasonal breeding (e.g. in northern India where dogs
IMPACT 2
show a peak in whelping in November; Reece et al. 2008) an additional peak in seasonal
breeding may become apparent as either dogs become recruited to the breeding population
earlier than one year old or perhaps start to have second litters within one year.
Assessing pregnancy by sight alone may be unreliable; however pregnancy can be assessed
IMPACT 3
at the time of clinical examination (depending on the stage of pregnancy) or during surgical
sterilisation. Hence a review of clinic records will provide data for estimating the percentage
of females pregnant per month and how this has changed over time. See Section ‘‘Clinic’
records’. IMPACT 4
Suggested indicator – Litters per female
The number of litters per female dog per year is an indicator of fecundity that can
be estimated for the owned dog population by using a questionnaire survey. Not all
interventions will be aiming to affect fecundity of the owned dog population, as the puppies
IMPACT 5
produced by owned dogs may well be wanted, however this could be a relevant indicator
for locations where unwanted litters of owned dogs are a problem. As described for the
indicator of pregnant females, litters per female may change with an improvement in the
health status of females, with more litters carried to term and females having litters at a IMPACT 6
younger age.
Section ‘Questionnaire surveys’ describes this method of measurement in more detail,

including a section on how to calculate this indicator from responses to the sample
questionnaire in Annex E.
IMPACT 7
Suggested indicator – Mortality and age structure
Population turnover also includes the component of mortality. For disease control, increased
longevity of vaccinated dogs can help maintain herd immunity, whilst for animal welfare
IMPACT 8
short lifespans are often accompanied by high mortality, morbidity and associated suffering.
The indicator of mortality is usually expressed as the opposite statistic, annual survival.
Examples of using changes in annual survival as an indicator of intervention impact could
not be found hence it is described here as a suggested indicator only. Measuring mortality/
survival is possible using different methods depending on the ownership status of the dogs.
When considering a population of owned dogs, owners can be asked via a questionnaire
the ages of their dog (leading to an ‘age structure’ of the population), how many dogs have
left their household in the previous 12 months and what was their fate (e.g. given away,
died, disappeared, etc.). Questionnaires and the analysis required for calculating survival are
included in Section ‘Questionnaire surveys’.

If the target population is comprised mostly of unowned dogs, calculating mortality

requires a longitudinal study and marking of dogs with individual marks such as tattoos
or microchips, usually applied during a sterilisation intervention. A sample of dogs are
re-caught, ideally again as part of the intervention, for example for booster vaccination or
deworming, and their individual marks read. These marks are then used to look back into
clinic records to the date when each dogs was sterilised. This provides a random sample of
minimum times for which dogs survive in the population following the first intervention and
marking event, and hence provides information on their survival rate. Using clinical records
IMPACT 1
and re-catch data to calculate survival is described in the Section ‘‘Clinic’ records’.
A potential alternative to measuring mortality is to measure age structure of the population

and specifically the proportion of the population that is comprised of older dogs (these
can be defined as 5 years old and above, but can be adjusted to local dog demography).
IMPACT 2
When a dog population has high turnover it will be comprised of a large number of young
dogs and relatively few older dogs, as turnover reduces the proportion of older dogs
should increase. Data relating to age structure is commonly presented as a histogram of
the number of dogs falling into age classes (this may also be called an age pyramid when IMPACT 3
data from males and females are presented back to back). These graphs can show sharp
declines from the youngest age class down to the oldest class when turnover is high and
a flatter shape with more equality between age classes as the population stabilises. For
example, in the following graphs the same population size of dogs are shown with different
age structures - example A shows a steep decline from the youngest age class down to the
IMPACT 4
oldest with very few dogs, whilst example B shows a flatter shape with more equality
between the age classes:
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Kii (1982) suggests that the regression coefficient, a number which represents the slope
of the regression line created by age pyramids can be used as an indicator of how a
population is changing in age over time. (In humans in the developed world, the histogram
can actually start to reverse with relatively large groups in the older classes and relatively
few people in the young age classes; sometimes termed an inverted pyramid). With an
owned dog population, the number of dogs falling into each age class can be found using
a questionnaire, allowing the use of relatively small age classes with ranges of 1 year. Note
that there is some evidence that reliability of age reporting reduces as dogs get older (Chris
IMPACT 1
Baker, pers comm.) and so the oldest dogs may be best combined into one larger group
of 5 year olds and older. However, where street surveys are used, the age classes could
2
be as broad as puppies, adults and old adults, with old adults defined by several physical
characteristics including grey muzzle, thickened skin, heavy eyebrows/sunken eyes, hairless IMPACT
patches and stiff gait. This would require discussion and agreement between the survey
team as this category is particularly subjective. No examples using age structure, or more
specifically the proportion of old dogs in the population, to assess the impact of
an intervention could be found, hence this is described at this stage as a suggested
indicator only.
IMPACT 3
Where questionnaires have led to detailed age structures with age classes of 1 year
increments, the median age of the population can also be calculated. The median age
of dogs can then be compared over time, or between treatment and control groups, and
tested for significance using a Mann-Whitney Test, which looks both for differences in median
IMPACT 4
and spread, essentially for whether one sample tends to have higher values than the other.
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Impact 4: Reduce Risks
To Public Health
International
Companion Animal
Public health risks associated with dogs can vary with location, both with regards to the
pathogen involved and the severity or likelihood of risk. In this section, we highlight some
indicators relating to the most common public health risks that could be targeted by dog
IMPACT 1
population management , namely dog bites, rabies, echinococcosis and leishmaniasis.3
Recommended and suggested

IMPACT 2
indicators – Dog bites
Dog bites, whether associated with subsequent disease or not, can inflict serious injury and IMPACT 3
can cumulatively represent a high cost to human health services, hence they are commonly
stated as a priority concern for citizens and governments alike. Incidence of dog bites can
also be high compared to other public health risks associated with dogs. For example, in
the US 4.5 million people are reported to be bitten each year equating to 1,500 bites per
100,000 people; with 1 in 5 of these requiring medical attention for the bite (Gilchrist et al.,
IMPACT 4
2008).
To measure the impact of an intervention over time we recommend the indicator of the
change in frequency of bites per unit time (often per month or per year). Reece et al.
(2013) used the frequency of dog bites per year to assess the impact of an Animal Birth
IMPACT 5
Control (ABC) intervention, which sterilised and vaccinated a high proportion of the roaming
dog population in Jaipur, India. They found a significant decrease in dog bites during the
6
intervention as compared to an increase in bites in the period before the intervention.
Although human population was not used as a denominator for dog bites, there was a IMPACT
concurrent human population growth of nearly 5% per year in Jaipur, thereby strengthening
this finding. Although dog bites may not be expected to increase at the same rate as the
human population it can be assumed that they would at least change in the same direction.
Reece et als (2013) findings that bites went in the opposite direction to the change in people
was particularly strong evidence of the positive impact of the ABC intervention on dog bites
IMPACT 7
in Jaipur.
Some studies have used the indicator of dog bite incidence to assess the impact of an
intervention using human population size as the denominator. However, this requires IMPACT 8
accurate data relating to the human population size served by the hospital or health centre
reporting the bites; this is not always straightforward especially when several years have
elapsed since the last human census and is a potential weakness of incidence data.
Hence using the number of dog bites per 100,000 people per unit time (often per
month or per year) is only recommended when there is accurate data available about the
human population served by the health service reporting the bites and when it is deemed
necessary; for example when trying to compare bites between locations in order to assess
impact, such as the comparison between control and treatment sites.
For a fuller discussion of the public health risks relating to dog populations refer to Dogs Zoonoses
3
and Public Health (2013) edited by CNL Macpherson, FX Meslin and AI Wandeler.
29
Impact 4: Reduce Risks To Public Health
Some interventions aim to reduce the risk of a bite from individual dogs (as opposed to
impacting on bites through reducing the number of dogs), for example, vaccinating dogs
against rabies to reduce the chances a dog will become rabid and bite; sterilising female
dogs to reduce maternal aggression (e.g. Reece et al., 2013); education programmes
to promote safe interactions with dogs; and appropriate socialisation of young dogs. If a
measure of dog density or population size is available for the same unit of time as the bite
data over the intervention period this can be used as a denominator i.e. an indicator of the
‘bite propensity’ of dogs; for example bites per year/dogs per km of street surveyed for
IMPACT 1
roaming dogs, where dog density has also been measured annually for the sub-population
of dogs targeted by the intervention. This is a suggested indicator that has not been
2
reported in the literature.
IMPACT
Indicators for dog bites depend on the use of secondary sources of data including data from
both official government and private health providers and are described in more detail in
the Section ‘Secondary sources of information’. However, some specific considerations for
dog bite data are expanded upon here. Sources of data on dog bites will differ with country,
location and rabies status: IMPACT 3
 T
he number of dog bites assessed as suspect rabid and treated with post exposure
prophylaxis (PEP), see Section ‘Recommended indicator – Suspect rabid dog bites’
for more details.
IMPACT 4
 T
he number of dog bites treated by local medical centres or Emergency Rooms in
hospitals that are not necessarily due to rabid animals; this will be the vast majority
of dog bites.
 T
he number of dog bite injuries requiring surgical reconstruction in hospital. This may IMPACT 5
include some cases included in the data relating to bites treated with PEP.
Access to both government and private health provider data requires support and
cooperation from the medical community. Ease of access may depend on whether dog
bites are officially required to be reported, as is often the case when PEP is provided by IMPACT 6
government for free or subsidised cost, and whether these data are presented publicly.
There will also be important considerations relating to the quality of data. For instance, it is
important to be able to distinguish the following factors: (1), including whether it is clear that
the dog was suspect rabid; (2), whether it was an owned or a roaming unknown dog, and IMPACT 7
(3) where the person was located when bitten; these qualifiers are relevant when analysing
the data and in particular when assessing attribution of any impact to the intervention.
An alternative method of measuring dog bites is through questionnaires, asking people

to report their personal or family experience with dog bites. Questionnaires are described
IMPACT 8
in more detail in the Section ‘Questionnaire surveys’. One specific consideration when
collecting data on dog bites through questionnaires is to use short time spans, such as:
‘Have you or anyone in your household been bitten by a dog in the past 12 months?’ By
using short time spans, as opposed to several years or ‘in your life time’, you decrease the
time required in which a change in dog bites could be revealed. However, an alternative is
to ask: “Have you been bitten by a dog in your lifetime? If yes, please state in what year this
occurred”. By asking in what year the bite occurred the frequency of bites per year across
the years would become apparent and the pre-intervention period would provide a temporal
control against which the frequency of bites per year in the intervention period could be
compared. Asking people to remember the year in which they were bitten may appear to
be a challenging task for accurate recall, but people tend to remember very vividly being
bitten by a dog, especially if this was a suspect rabid dog. It would be important to clarify

whether the person was living in the intervention area when the bite occurred, as they may
have moved across their lifetime and hence be reporting bites from other locations. Note
that an example of this approach could not be found in the literature and so this can only be
suggested at this time.
A potential additional indicator for interventions that include bite prevention programmes
for children would be a change in the number of children bitten by dogs. Both control
and intervention classes could be asked to raise their hands if they have ever been bitten, IMPACT 1
the process could then be repeated every 6 months and the rate of increase in number of
children bitten compared between control and intervention classes. Control classes could
also receive a lesson on how to treat dog bites (i.e. washing with soap and water and
going to the doctor) whilst intervention classes receive the full bite prevention programme.
Should the indicators reveal that the programme is effective at reducing dog bites it would
IMPACT 2
be necessary to follow-up all control classes with the full programme also. No published
examples of this indicator could be found and so is included here as a suggested indicator.
These methods of measuring bites will differ in the resulting bite incidence or frequency.
Hence it is essential to be consistent in the method used to collect dog bite data over
IMPACT 3
time and to be mindful of potential changes in reporting such as bites becoming officially
reportable or changes to health centre/hospital reporting systems and policies.
IMPACT 4
Recommended indicators - Impact on rabies risk
Rabies is perhaps the most feared public health risk from dogs. It is an almost invariably IMPACT 5
fatal viral disease, with over 99% of all human cases transmitted via dogs (WHO, 2013).
Some rabid dogs can show quite terrifying clinical signs and inflict serious injury. Hence in
countries where canine rabies is present it is usual for DPM to include activities with the
aim of reducing or even eliminating rabies risk from the intervention area. When assessing
the impact of an intervention on rabies risk it is ideal to use a number of indicators in IMPACT 6
combination. These include dog rabies cases, suspect or confirmed rabid dog bites and
human rabies cases. Each of these indicators are discussed in this Section with a final
7
Section ‘Vaccination coverage’, which although not an indicator of impact is an important
consideration for assessing attribution of a dog vaccination intervention. IMPACT
Recommended indicator – Dog rabies cases

The World Health Organisation (WHO) states that effective rabies surveillance should be
based on laboratory confirmed cases (WHO, 2013), however laboratory facilities are by IMPACT 8
no means ubiquitous and efforts to conduct surveillance of rabies cases based on clinical
diagnosis alone are also valuable. The effectiveness of clinical diagnosis is supported by
the recognisable signs of rabies exhibited by most dogs (see Annex C) for a process of
diagnosing rabies in dogs from clinical signs by Tepsumethanon et al. (2005)). In a study
of rabies in the Serengeti, it was found that more than 74% of clinically diagnosed cases
(recognised by villagers, livestock field officers, park veterinarians or research personnel)
were later confirmed as positive by the gold standard fluorescent antibody laboratory test
(Lembo et al., 2008); this 74% is the minimum percentage as some brain samples likely
degraded before reliable laboratory testing was possible. In summary, it is ideal to use
the indicator of the number of laboratory-confirmed dog rabies cases per unit time
(usually per month), although the number of clinically-diagnosed dog rabies cases per

unit time is also a valid indicator of rabies risk and can be particularly useful for increasing
case detection when laboratory infrastructure is weak.
If surveillance efforts or methods change part way through the period under assessment
(e.g. introduction of laboratory confirmation) the differences this makes to the indicator of
dog rabies cases needs to be reconciled. This may be achieved by including a period when
both the old and new methods are used simultaneously to allow comparison of number of
cases exposed by each method during the same time period. IMPACT 1
Note that incidence of dog rabies cases (per unit of dog population size) is not often used,
as the denominator of dog population size can be difficult to estimate accurately. However,
where the dog population is known to have changed significantly, and the population size
can be reliably estimated for the same time unit as used for the number of dog rabies
IMPACT 2
cases, the incidence per unit of dog population size (such as 100,000) may be suitable.
One alternative would be to use a measure of dog density as the denominator as opposed
to dog population size, for example dogs per km of street surveyed. It may be preferable
to use incidence where impact assessment involves comparisons between locations,
such as treatment sites compared with controls. An example of this is given by Kitala et
IMPACT 3
al. (2000) who reported incidence of dog rabies cases per 100,000 dogs for sites where
active surveillance had been introduced (a method that utilised reports from key informants),
compared to the incidence reported from surrounding areas where existing passive
surveillance continued. As a result of introducing a more active method of surveillance, 72 IMPACT 4
times more rabies cases were reported.
Data on dog rabies cases is usually sourced from veterinary or public health authorities;
a general discussion of secondary data sources can be found in the Section ‘Secondary
sources of information’. Indicators of dog rabies cases are particularly sensitive to
IMPACT 5
surveillance effort, as evidence by the study previously described (Kitala et al., 2000). To
reliably eliminate rabies from an area, Townsend et al. (2013) estimated that you need to
detect at least 5% of dog rabies cases but ideally at least 10%; any less and you may
mistakenly reduce control measures too early on the assumption that elimination has been IMPACT 6
reached when cases may be circulating undetected. Detecting at least 1 in 10 dog rabies
cases may seem like a low number but this equates to a relatively efficient surveillance
system. In an environment where most dogs roam, even significant surveillance efforts may
not detect more than 10% of rabid dogs, because roaming rabid dogs may well die outside
the home and be undetected by their owners.
IMPACT 7
When using any indicator of rabies, but in particular dog cases, intervention managers and
8
impact assessors will need to be aware of the many stages involved between recognising
a dog as potentially rabid through to a diagnosis being reported in official records. If any IMPACT
stage of this process changes in its effectiveness, the number of reported dog cases may
change irrespective of a change in disease incidence and hence this must be considered
when analysing the impact. Improvements to surveillance are likely to occur as a result of
changes in methods and processes, such as (1) the introduction of field kits for testing
(e.g. lateral flow kits; six currently available kits tested by the WHO Collaborating Centre for
Rabies Surveillance & Research in Germany were found to have low sensitivity and not fit
for purpose (Thomas Muller pers comm), however there is potential for improvement and
reliable kits are likely to be available soon); (2) improved targeting to high-risk animals only
(i.e. that are biting, behaving strangely, moribund or found dead) as opposed to random
sampling of dogs; (3) greater communication and collaboration between human health and
veterinary services; and (4) recruitment of field based key informants (e.g. as described by
Kitala et al. (2000) in Kenya). Townsend et al. (2013) provide a fuller discussion of ways to
improve effective surveillance when aiming for rabies elimination. To account for this highly
influential confounding variable of surveillance effort and to maximise the chances of effective
rabies control, it is ideal to establish a surveillance system that is as effective and consistent
as possible at the outset of any intervention. Where this has not been possible, changes
in surveillance effort during the period of data collection will need to be considered during
analysis and interpretation.
Recommended indicator – Suspect rabid dog bites IMPACT 1

Indicators relating to dog bites are also discussed in the previous Section ‘Recommended
and suggested indicator – dog bites’, however the key points most pertinent to the indicator
of suspect rabid dog bites of people (not of other animals) are summarised here.
The indicator of number of suspect rabid dog bites per unit time (usually per month or IMPACT 2
per year) can be used directly or converted into an incidence per 100,000 people per unit
time. Establishing incidence, however, requires an accurate estimate of human population
within the catchment area of the hospital or health centre reporting the bites and this may
prove challenging if the period since the last human census has been long and/or if the rate
of human population growth is unknown. Hence using frequency of suspect rabid dog bites
IMPACT 3
is recommended. Note incidence may be necessary when comparing locations, for example
when comparing intervention (treatment) sites with non-intervention (control) sites (e.g.
comparison of vaccination villages with control villages in Tanzania reported by Cleaveland
et al. (2003)), or when sites have undergone a period of large and measurable human IMPACT 4
population growth, as may have occurred over a period of 10 years or more.
A general discussion of secondary sources for dog bite data can be found in the Section
5
‘Secondary sources of information’. Specific considerations include being able to separate
suspect rabid bites from non-suspect, as vaccination alone cannot be assumed to impact IMPACT
on non-rabid bites. When looking to separate suspect rabid and non-suspect rabid bites,
it is important to consider the availability of post-exposure vaccines during the period, as
during vaccine shortages some bites may have been incorrectly recorded as non-PEP due
to lack of vaccine as opposed to the lack of suspect signs in the biting dog. This will be
particularly relevant where the number of dog bites is not recorded but instead the number
IMPACT 6
of PEP doses or courses delivered, i.e. where PEP use is a proxy for dog bites. This may be
the case in most situations and is particularly susceptible to changes in PEP availability.
Where the number of bites from suspect rabid dogs is recorded, as opposed to only PEP IMPACT 7
delivery, the definition of a suspect case will need to be known in advance, and any changes
to that definition recorded. Defining a case as suspect may include the following: (1) Was
the bite provoked or unprovoked? (2) Is the dog known and is the vaccination history
known? (3) Is the dog still alive, or has it died or disappeared? (4) Where there any particular
behavioural signs shown by the dog (as listed in Tepsumethanon et al. (2005))? However, the
IMPACT 8
circumstances of the bite may not be sufficiently clear to define a bite as non-suspect and
hence PEP will be used as a precaution.
In addition, the location of the person when bitten (not just the location of the hospital or
health centre where treated) will be needed so that suspect bites can be assigned to within
or outside the intervention area. Access to dog bite data and ensuring adequate quality for
reliable interpretation, will require support from the medical community. One novel approach
used by Tenzin et al. (2012) in Bhutan was the number of doses of human vaccine
imported per year as a proxy for the number of dog bites, which was not available, on the
assumption that vaccine was not stockpiled. This indicator only works where all the sources
of PEP are known and measured. In many countries multiple sources of PEP may exist and
may be difficult to track.

Recommended indicator – Human rabies cases

A reduction in human deaths from rabies is the most important indicator for measuring
impact on human health. However, in countries where PEP provision is widespread and
rabies prevention is well practiced by the public the number of deaths will be thankfully
low and so the number of human deaths, although still important, will not be as sensitive
for measuring change compared to dog rabies cases or suspect bites which will be more
numerous. Hence the indicator of the number of human rabies cases (equates to IMPACT 1
number of deaths, as rabies is almost invariably fatal) per unit time (usually per year)
will be most useful in those countries where PEP provision is not complete and human
deaths unfortunately occur relatively frequently.
The number of human rabies cases is likely to be low even where PEP provision is not
IMPACT 2
complete and hence is most meaningful over large geographical areas such as countries or
regions as opposed to smaller geographical areas such as individual cities or districts. When
human cases are very infrequent (<10) the presence or absence of human rabies cases
per time period may be a more useful indicator. IMPACT 3
Data on human cases can also be converted into an incidence per 100,000 people. Again
this will rely on access to good human population estimates and may be most suitable when
comparing locations (such as comparing different hospital catchment areas in Bhutan by
Tenzin et al. (2011)) or when comparing across periods of time where significant changes IMPACT 4
in the human population are known. Where a single location is being evaluated and the
time period is relatively short, the frequency of human rabies cases per unit time may be a
sufficient indicator that is not subject to error in human population estimates (e.g. as used
in Lima, Peru for the evaluation of a dog vaccination campaign where the number of human
cases reduced to zero following the campaign; Chomel et al. (1988)).
IMPACT 5
The method of measurement for the indicator of human rabies cases uses secondary
sources of data from health authorities, see Section ‘Secondary sources of information’.
As with dog rabies cases, the number of human cases will be affected by a number of
factors other than disease incidence. For example, surveillance effort may change over
IMPACT 6
time, and changes from clinical diagnosis to laboratory testing will impact on the number of
cases reported. Rabies in people is also known to be severely underreported (e.g. 100 fold
underreporting in Tanzania; Cleaveland et al. (2002)). This may be because people do not
seek hospital treatment when dying of rabies, hence their death is not reflected in hospital IMPACT 7
records nor reported to central authorities. In addition, human rabies deaths may have
been mistakenly attributed to a different cause (e.g. 11% of rabies cases were mistakenly
8
diagnosed as cerebral malaria in Malawi; Mallewa et al. (2007)). Moreover, many of these
reasons for underreporting can be affected by changes in surveillance effort, hence close IMPACT
collaboration with the health authorities will be necessary to monitor and incorporate any
changes in surveillance into interpretation of indicator changes. In addition, improvements
to surveillance systems and diagnostics would be best implemented at the outset of the
intervention or in advance to establish a more accurate baseline.
Perhaps most importantly for DPM interventions is that the number of human cases will be
significantly affected by provision of PEP; both related to its availability through the health
system and also people’s behaviour in searching out appropriate treatment. Any change
in PEP provisioning may make the number of human cases unreliable as an indicator of
the impact of a DPM intervention. Again this requires close collaboration with the health
authorities to ensure changes in PEP provision can be considered.

We strongly recommend the use of combined sources such as dog rabies cases, dog bites
and human cases to reveal changes in disease incidence.
Recommended – Vaccination coverage

The indicator of the percentage of the dog population vaccinated against rabies is
not an indicator of intervention impact (such as human or dog rabies cases) but instead
an indicator of intervention effectiveness, resulting from a combination of intervention effort IMPACT 1
and response to the intervention from the authorities and public. However for assessing the
impact of reducing rabies risk, vaccination coverage is essential for judging the attribution
of the intervention to any changes. Rabies incidence can change over time in the absence
of an intervention and a core indicator of intervention effectiveness such as vaccination
coverage will allow for more thorough analysis, including testing for correlations between
IMPACT 2
effectiveness and impact. This is the only indicator of intervention effectiveness discussed in
detail in this guidance because measurement of this indicator requires some preparation.
The most appropriate method for assessing vaccination coverage will depend on whether
dogs are usually confined or roam freely. If most of the dog population is roaming freely
IMPACT 3
(either unowned dogs or owned roaming), resource-light street surveys can be used to
assess the proportion of vaccinated (and marked) dogs. However if the majority of the
dog population is confined it will be necessary to use house-to-house questionnaires. The
vaccination campaign provides an opportunity to collect information on the proportion of IMPACT 4
confined versus roaming dogs; owners can be asked whether their dog is normally confined
at the vaccination point (or doorstep if using a door-to-door vaccination approach), whilst
dogs caught for vaccination are presumably at least sometimes roaming. Both street
surveys and questionnaires are described in more detail in the Sections ‘Street surveys’ and
‘Questionnaires’, respectively.
IMPACT 5
Measuring vaccination coverage by either street surveys or questionnaires does require
investment of resources. Where resources are limited and the methodology described in the
‘Methods of measurement’ section is not feasible, it is advisable to record only whether a
vaccination campaign was conducted or not in every location within the intervention area,
IMPACT 6
and then measure vaccination coverage in just a sample of locations. This is because leaving
some villages or wards completely unvaccinated within the intervention zone can have a very
serious impact on rabies control, as dogs in these remaining unvaccinated areas can act as
reservoir for the virus and seed or prolong outbreaks (Townsend et al. 2013).
IMPACT 7
Seroprevalence/blood testing for rabies antibodies in dogs following a vaccination campaign
is not considered suitable for monitoring vaccination coverage or levels of immunity. The
circulating antibody response to rabies vaccination is relatively short and highly variable
between individuals. Immunity to rabies involves mechanisms in addition to circulating
IMPACT 8
antibodies and a titre below the perceived ‘protective’ level does not necessarily mean the
dog is not immune. For these reasons, and perhaps also with the costs of antibody testing
in mind, WHO advise “measurement of rabies-specific antibodies is not recommended for
routine rabies surveillance” (WHO 2013; pp 93).

Recommended indicators – Impact on

echinococcosis risk
Human cystic echinococcosis is a disease caused by the tapeworm Echinococcus
granulosus that leads to hydatid cysts developing in the liver and lungs. Whilst the disease
can be treated, often necessitating surgery, it can also be fatal. There are an estimated IMPACT 1
1.2 million human cases worldwide with 200,000 new cases diagnosed annually (WHO,
2010). Dogs are the primary final hosts in the cycle of E. granulosus and the sheep strain
of E. granulosus accounts for the majority of human cystic echinococcosis cases (Eckert
and Deplazes 2004). Sheep are the primary intermediate host of the sheep strain of E.
granulosus, however other livestock species can function as intermediate hosts: specifically
IMPACT 2
goats, pigs and macropods (cattle can also act as intermediate hosts but often do not
produce fertile hydatid cysts and hence are less relevant to the E. granulosus life cycle).
The route of infection for people is through contact with infected dog faeces through which
they become accidental intermediate hosts and the life-cycle stops. Livestock are infected IMPACT 3
through ingesting eggs from pasture contaminated by infected dog faeces and dogs are
infected through eating the cysts in the offal (usually liver and lungs) of infected livestock.
Control of echinococcosis is through regular deworming of dogs with praziquantal and

preventing their access to infected offal through inspection and proper disposal of infected IMPACT 4
offal from slaughter houses and during home slaughter. This requires consistent education
and cooperation of dog owners, livestock herders and slaughter staff.
5
Echinococcus multilocularis is another species of tapeworm that causes alveolar
echinococcosis in humans, a rarer but much more serious condition. The life cycle of E. IMPACT
multilocularis is perpetuated by wildlife reservoirs (e.g. red fox, coyote, artic fox), hence dog
interventions will not eliminate the worm but they will reduce human cases as the dog is a
transmission vector of the worm to people. As control and surveillance of E. multilocularis is
usually focused on wildlife, and not dogs, the rest of this section will focus on E. granulosus.
IMPACT 6
Surveillance for echinococcosis prevalence needs to be conducted over a prolonged period
of time because the disease is asymptomatic in dogs and livestock and so clinical signs in
live animals cannot be used as an indicator. It can also be asymptomatic for many years
in people before clinical signs emerge. The prevalence of cysts in livestock at the time of IMPACT 7
slaughter is a very accessible measure of echinococcosis prevalence (see next section)
but may take 5 years for changes in prevalence to become apparent and at least 10 years
in people; as those animals and people with cysts today may have been exposed many
years previously and so changes to current transmission risk will not be visible until several
years later. As there are a limited number of significant hosts involved in the life cycle of
IMPACT 8
the sheep strain of E. granulosus, and subsequent cystic echinococcosis in humans, just
using the change in the prevalence of infected livestock will be sufficient as an indicator of
echinococcosis control. However two further indicators are described here, human surgical
cases and infection in dogs which would provide a fuller data set for impact assessment.

Recommended indicator – Infected livestock offal

The key indicator for changes in echinococcosis will be the number of livestock with
liver or lungs infected with E. granulosus cysts at the time of slaughter per unit time
(usually per month or year) and per age group of livestock, this is presented as a
percentage of livestock slaughtered in that same time period, i.e. as a prevalence. In most
locations, sheep will be the most appropriate livestock species to monitor for sheep strain
E. granulosus cysts, and in the rest of this section sheep are used as the example livestock IMPACT 1
species. However in some locations, where sheep are a minority, the prevalence in other
livestock will be more relevant, such as goats or pigs or macropods.
The age of the sheep at slaughter is important as prevalence and infectiveness of cysts
increases with age. The size of cysts also increase with the age of sheep, so although the
IMPACT 2
prevalence of cysts in lambs may reflect the impact of the intervention over a prior short
period (i.e. the life of the lamb) these cysts will be less than 3mm in size and are difficult to
detect during visual inspection of the carcass (Lloyd et al. (1998) is an example of using
sentinel lambs for surveillance but these required inspection by experienced parasitologists).
Cysts in sheep at least 2 years of age will be easier to identify during inspection and hence
IMPACT 3
the prevalence will be more reliable as the age of the sheep increases. Note this prevalence
will represent disease control over the 2 previous years, hence the need to conduct impact
assessment over several years as it will take a minimum of 2 years from launching a
deworming intervention with dogs to see a reduction in cysts in these older sheep. IMPACT 4
Accessing data relating to the indicator of prevalence of infected sheep will require
collaboration with the veterinary services and use of secondary/official sources of data
(discussed further in the Section ‘Secondary sources of information’). Meat and offal
inspection at the time of slaughter is common practice and legislated for in most countries,
IMPACT 5
however recording of the type of infection is not always done (sometimes only a weight
of infected offal is recorded) and hence working collaboratively with veterinary authorities
and slaughterhouse inspectors may be needed to ensure that infection of liver and lungs
with E. granulosus is recorded separately for each sheep along with its geographical IMPACT 6
origin (to establish whether the sheep came from the intervention area or outside) and
age. In cases where E. granulosus control is relatively new, it may be necessary to provide
training to slaughterhouse staff in identification of E. granulosus cysts using the services
of a parasitologist. One challenge to this indicator will be where home slaughter is the
predominant method of slaughter. In this case, a sample of sheep could be inspected during
IMPACT 7
home slaughter, perhaps focusing on times of year or religious festivals where a high number
of animals will be slaughtered, but this would clearly require significantly more resources than
inspection at a slaughterhouse.
IMPACT 8
Recommended indicator – Human cystic echinococcosis
Changes in the prevalence of cystic echinococcosis in people may be considered the most
relevant indicator for those interested in public health. Cystic echinococcosis in people can
be diagnosed using a range of techniques, including ultrasound and serology, however the
most straightforward indicator for surveillance will be the number of cases of surgical
intervention for the treatment of human cystic echinococcosis per unit time and
per age group. This can be used as simply the frequency of cases or as prevalence per
100,000 people; as the time span for impact assessment will need to be long (at least 10
years). Prevalence may be most sensible to incorporate changes in human population size,
and will be most relevant where comparisons between locations are made (e.g. comparing

treatment and control areas). Age group is important because cases can be expected to
disappear from younger people first, while older people may have picked up E. granulosus
decades previously and not shown clinical signs until much older.
Data relating to the number of surgical cases will need to be accessed from health
authorities; for example, Acosta-Jamett et al. (2010) accessed the number of surgical cases
from 3 provinces in Chile from the Regional Health Service case reports. However, it is
likely that underreporting does occur in many locations and hence it may be ideal to work IMPACT 1
collaboratively with hospital departments conducting liver surgery to access data on the
number of surgical cases. One benefit of the fact that E. granulosus cysts can lie dormant
for so long is that at the outset of an intervention the prevalence of disease in older people
can provide a reflection of disease risk for the previous several years. The disease in younger
people also often progresses much faster and so prevalence in children may be a more
IMPACT 2
sensitive indicator for the impact of an intervention in the early years.
Suggested indicator – Infection in dogs

The presence of an E. granulosus infection in dogs was historically identified by ‘purging’
IMPACT 3
(dosing) dogs with arecoline hydrobromide, resulting in diarrhoea and expulsion of the worm
burden. This method is considered risky due to expulsion of live worms (with a potential
to reinfect), and the deleterious and even fatal effects on dogs, particularly a risk for young
and pregnant dogs. Currently, there are more effective and safe deworming treatments (e.g. IMPACT 4
praziquantal) that kill the infective worms before expulsion, and have little or no side effects
to the dogs themselves.
5
Alternative methods of detecting E. granulosus infection include microscopic egg and
proglottid detection in faecal samples, serum antibody testing, polymerase chain reaction IMPACT
(PCR) tests for parasite DNA in faecal samples and ELISA (enzyme-linked immunosorbent
assay) tests for E. granulosus antigens in faecal samples. However, each of these tests face
different challenges: microscopic egg detection does not allow for definitive diagnosis due
to the similarity with other Taenia species eggs; diagnosis using proglottids requires these to
be found in good condition in the faecal sample; serum antibody testing has low sensitivity
IMPACT 6
and antibodies persist after infection has been cleared; and PCR tests have high cost and
relatively low sensitivity. The relatively recent development of ELISA tests for the presence
of E. granulosus antigens shows the most promise in terms of sensitivity and specificity,
low cost and ease of handling samples (faecal samples can be collected from the ground,
IMPACT 7
samples are best fresh but can be tested after up to 4 days on the ground). However this
test currently requires production of antibodies from rabbits kept in animal facilities, hopefully
this production will be moved to in vitro production in future.
Taking note of the challenges presented by

IMPACT 8
tests for E. granulosus infection in dogs, we
recommend instead focusing on the two
indicators of E. granulosus cysts in livestock
and echinococcus in humans for assessing
intervention impact on this disease. Thankfully,
the unique role of dogs in maintaining the sheep
strain E. granulosus life cycle means that even
without monitoring a change in infection rates
in dogs, a reduction in infection of livestock
and people following a dog intervention can be
confidently attributed to that intervention.

Recommended indicators – Impact on

leishmaniasis risk
Leishmaniasis is a disease of both dogs and humans caused by infection with protozoan
parasites of the Leishmania genus. Transmission between people or/and between animals
and people, is via the bite of an infected female Phlebotomine sandfly. There are two major IMPACT 1
categories of leishmaniasis in humans, the less severe cutaneous leishmaniasis (associated
with skin lesions) with 1,000,000 new human cases per year and the potentially fatal visceral
leishmaniasis (associated with anaemia, liver and spleen damage in people and kidney
damage in dogs) with 300,000 new human cases and 20,000-40,000 deaths per year
worldwide (Accessed at www.who.int/leishmaniasis/en/); although it should be noted that
IMPACT 2
the majority of deaths are caused by Leishmania donovani for which the principle reservoir
is humans and not dogs. Control measures include reducing bites from sandflies using bed
nets and insect repellent for people, insecticide impregnated collars or ‘pour-ons’ for dogs,
and killing sandflies through peridomestic spraying; or by reducing the infected dog reservoir IMPACT 3
through culling (the success of this last measure is particularly widely debated, e.g. Nunes et
al. (2010)). There are also several vaccine candidates under evaluation and recently vaccines
for dogs have become commercially available in Brazil and Europe.
Measuring the impact of interventions to control leishmaniasis would benefit from long-term IMPACT 4
monitoring as, in theory, interventions would build in their effectiveness over time as the
prevalence of infected individuals and therefore transmission decreases. For example, if
insecticide (deltamethrin) -impregnated dog collars are used in an intervention, initially some
dogs will already be infected and although the collars will reduce transmission from these
dogs they will remain infected. As the intervention continues, infected dogs will die due
IMPACT 5
to the disease and other causes and be replaced with young dogs that, if protected from
sandfly bites from near birth, would contribute to a continued reduction in the prevalence of
infected dogs which could lead to decreasing transmission to people. However, examples
of this long-term monitoring and increasing effectiveness are not currently available in the IMPACT 6
literature, and hence this hypothesis remains to be tested.
The extent of the intervention area and the size of the area used for monitoring the impact
of the intervention will need to be carefully considered. Sandflies are mobile vectors and so
the incidence of leishmaniasis on the borders of the intervention area may not be the best
IMPACT 7
reflection of intervention impact. A buffer zone can be used to minimise this; sandflies have
relatively limited average daily flight ranges (e.g. an average of <60m and maximum of 128m
found by Casanova et al. (2005)) and so just a few 100m should be sufficient. However, they
have been known to fly several 100m and hence even where buffer zones have been used, IMPACT 8
edge effects should be considered at analysis and interpretation, i.e. is there a difference in
intervention impact at the edge of the impact area as compared to the centre?

Recommended indicator – Human disease and infection

Arguably the most important public health indicator for the impact of a leishmaniasis
intervention would be the incidence or number of newly diagnosed human cases of
leishmaniasis disease per unit time. Because leishmaniasis is a difficult infection to
clear, the incidence of new cases is used as opposed to the prevalence (prevalence is the
proportion of people currently with the disease). A case of human leishmaniasis requires a
range of tests for definitive diagnosis as the clinical symptoms are similar to other diseases, IMPACT 1
for example a clinical suspicion may be followed by testing for antibodies, searching for
visible parasites in tissue smears and testing samples by polymerase chain reaction (PCR)
to detect parasite DNA, and to establish the species of Leishmania parasite. Data for this
indicator would clearly require accessing human health providers, potentially from specialist
hospitals or units where leishmaniasis is being diagnosed and treated.
IMPACT 2
Because the conversion rate from Leishmania infection to clinical disease is relative low
(often less than 1 in 20 infected people show clinical disease; Orin Courtney pers comm),
the challenge in assessing the impact of a Leishmaniasis control intervention is that the
intervention has to be very large to detect a significant clinical effect. Thus intervention
IMPACT 3
studies to detect protection against the disease are relatively rare. The alternative indicator
is the incidence or number of new cases of people infected with Leishmaniasis per
unit time. This would include those people without clinical disease but with a parasite
infection, i.e. they are asymptomatic. Dipstick tests for leishmaniasis serology are very easy IMPACT 4
to use in the field (e.g. the rK39 antigen-based immune-chromatographic test) however
they are most suited to exposing disease and tend to be less sensitive to detecting
asymptomatic infection where parasite loads are typically low. Note the WHO provide a
comparison of 5 dipstick tests for serological samples in different regions to aid selection
of the appropriate test dependent on location (the results can be found in WHO (2011)
IMPACT 5
‘Diagnostic Evaluation Series No. 4 Visceral leishmaniasis rapid diagnostic test performance’
at www.who.int/tdr/publications/documents/vl-rdt-evaluation.pdf). For testing for infection or
past exposure, an ELISA or direct agglutination test (DAT) that also looks for the presence of
antibodies in blood samples may be more sensitive. Another potential test is the leishmanin IMPACT 6
skin test which exposes the presence of a cellular immune response to Leishmania infection;
a very small amount of antigen is injected intradermally into the skin on the forearm and
7
the diameter of induration is measured 48-72 hours later. A positive response implies that
the person has previously encountered leishmania (note that clinically sick or immune IMPACT
compromised people may not show a positive response even if infected). People who were
previously negative for this test can be retested after a period of intervention to see if they
are now positive and have therefore encountered the parasite since the first test or whether
they have still avoided infection. However, it takes many months for the test to show as
positive following a new infection, hence it is most suitable for interventions that are relatively
IMPACT 8
long term, i.e. several years as opposed to several months.
As infection is usually not evenly distributed in a population, sampling all people in the
intervention area is ideal. If the intervention area is too large those people living closest to
recently diagnosed cases can be sampled, or those at greatest risk of new infection such
as children. Sampling all children in the intervention area was the approach used in Iran by
Mazloumi Gavgani et al. (2002) when assessing the impact of insecticide-impregnated dog
collars in a matched-clustered randomised controlled trial. The results showed a reduced
risk of leishmaniasis infection in children in the villages using insecticide collars as compared

with control villages where no collars were used. The tests used to establish whether there
had been new cases of infection in children were the DAT and the leishmanin skin test.
When testing people for Leishmania infection there must be a clear protocol agreed at
the outset with the local health services for what to do in the case of a positive result. The
standard treatments for leishmaniasis are complicated to administer, expensive and can
have serious side effects, and since most infections do not lead to clinical disease, treatment
is not applied to all infections but is instead based on the individual case. IMPACT 1
Recommended indicator – Dog disease and infection
2
The number of dogs newly diagnosed with clinical leishmaniasis per unit time can
be used as an indicator of impact. As with human infection, curing an infection is difficult IMPACT
and hence it is the incidence of new cases that is used as opposed to the prevalence of
all current cases. However, less than half of infected dogs show clinical signs of disease
and asymptomatic dogs have been found to be a transmission risk. Hence the number of
dogs newly infected with Leishmania per unit time will be a more sensitive indicator of
intervention impact; measuring this incidence rate requires tracking individual dogs over time
IMPACT 3
and identifying the number of dogs that previously tested negative that now test positive;
this could be practically done with owned dogs, even if they are usually free roaming, but
following the same unowned dogs over time may be challenging without the engagement of
local people who feed/care for them. Testing for infection in a large population of dogs, as IMPACT 4
with people, can be done by testing for antibodies in blood samples (e.g. by using the rK39
antigen-based immune-chromatographic test described previously for people, although
this is less sensitive in dogs, hence alternatively an ELISA or direct agglutination test (DAT)).
However serology tests for antibodies are not as sensitive nor as specific as detecting
parasite DNA in blood or tissue samples using polymerase chain reaction (PCR) and hence
IMPACT 5
most current scientific studies use PCR when testing for infection. PCR will identify dogs
that are asymptomatic; this is a test for the parasite itself as opposed to the dog’s immune
response to the parasite. Infected but asymptomatic dogs can be missed when testing for
antibodies in blood samples, which are best suited for exposing the prevalence of disease. IMPACT 6
The choice of test, including potential combinations such as initial antibody screening
followed by PCR in seronegative dogs, will depend on availability of resources (note PCR
tests are more expensive than antibody tests) but the differing sensitivities must be taken
into account when interpreting the data. PCR tests are required when evaluating the impact
of leishmaniasis vaccine interventions; vaccines will also lead to antibody production and so
IMPACT 7
separating vaccinated dogs from infected dogs using only tests for antibodies cannot be
reliably achieved.
As with testing for leishmaniasis in people, there needs to be an agreed protocol in place for IMPACT 8
with the local veterinary services for responding to positive results in dogs.

Impact 5: Improve Public

Perception
International
Companion Animal
Public perceptions of dogs will vary both between locations and between people living in the
same location. There are a few indicators that will be widely applicable, however it is likely
that perceptions will need to be explored in each location and key perceptions selected as
IMPACT 1
locally relevant indicators.
Recommended indicator – Adoption of dogs

The proportion of dogs that are acquired through adoption as opposed to other IMPACT 2
sources may be used as an indicator of increased empathy towards/positive perceptions
of street and shelter dogs. This indicator will be affected by supply, so changes in availability
of street or shelter dogs and ease of purchasing dogs will also need to be measured and
considered during interpretation. As well as changes in attitudes towards shelters which
could also have a significant impact on adoption rates. This indicator can be measured
IMPACT 3
through a questionnaire (see Section ‘Questionnaires’) that includes a question about the
source of owned dogs and/or by monitoring changes in shelter adoption rates over time. An
increase in adoption was detected over 20 months by using a questionnaire on the island of
Koh Tao, Thailand where an affordable veterinary clinic based intervention was run by a local IMPACT 4
NGO, a concurrent reduction in mortality was also observed leading to a general increase in
the size of the owned dog population (Lee 2013, unpublished data).
Recommended indicator – Attitudes towards dogs IMPACT 5

Two methods of measuring attitudes are proposed in the ‘Methods of measurement’
section; (1) questionnaires asking respondents for their levels of agreement with a range
of attitude statements and (2) participatory exercises where groups are asked to rank dog
nuisances against other public nuisances.
IMPACT 6
Attitude statements in questionnaires can be used to develop three main classes of
indicators: change in level of agreement with key attitude statements, selected because
they are particularly relevant to the local situation and intervention to be evaluated such as
“street dogs pose a danger to people”; change in summative attitude scores that combine IMPACT 7
levels of agreement with a range of attitude statements into a single score for ‘acceptance’
of dogs; and changes in ‘factors’ that underlie attitudes, such as the factor ‘dislike of stray
dogs’ comprised of 4 attitude statements (example from Miura et al. 2000), exposed through
the statistical method of factor analysis. The processes leading to these different indicators
are discussed in more detail in the Section ‘Questionnaire surveys’. A summative score for
IMPACT 8
acceptance of dogs was seen to change markedly over time following a comprehensive
dog intervention (involving several activities, including sterilisation, vaccination, education of
adults and in schools) in Colombo, Sri Lanka (Sankey et al., 2012), hence this is included as
a recommended indicator.
42
Impact 5: Improve Public Perception
Participatory exercises allow groups of local people to discuss, distil and express their
opinions on certain subjects. These should provide a richer picture of local people’s
perceptions, and reasons for these perceptions, than those gathered using questionnaires.
They can also be faster to implement, although they can suffer from lack of representation
of the wider population, dependent on how well groups are selected and facilitated
through the exercises. Three exercises are described in the Section ‘Participatory research
methods’, with the final two exercises providing a ranking of dog related nuisances
against other neighbourhood nuisances (which can be repeated over time to expose
IMPACT 1
changes in how dog related nuisances compare to other problems) and an assessment of
whether dog problems and dog benefits have increased, decreased or stayed the same
2
over time. Participatory approaches are widely used in human development and could be
very beneficial to implementing the intervention as well as for evaluation. For example, by IMPACT
improving an intervention manager’s understanding of why people hold certain perceptions
and what they would perceive as success and also by providing local people an opportunity
to explore dog issues and expose potential solutions in which they could play a role.
However, they are a relatively new concept to DPM and so the related indicators should be
considered as suggested at this stage.
IMPACT 3
Suggested indicator – Dog-related complaints
The number of complaints relating to dogs reported to local government authorities
may also be an indicator of how public perception is changing over time. There may also be
IMPACT 4
a change in the ‘nature’ of these complaints; some focusing on nuisances caused by dogs
and others expressing concern over dog welfare, including reporting of cruelty towards dog
(which may be reported to a different department to other complaints or to local NGOs).
Hence ideally complaints are split into different categories allowing these changes to be
explored more fully, although the total number of dog related complaints can be used if
IMPACT 5
such categorisation is not possible. Although there are anecdotal reports of reductions in
complaints following interventions, no systematic use of this indicator was found, hence it is
presented as a suggested indicator. Measuring such an indicator requires collaboration with
local authorities and is covered in the Section ‘Secondary sources of information’. IMPACT 6
Suggested indicator – Human-dog interactions
Improvements in public perception may also be reflected in the way people interact with
roaming dogs in public places. This indicator is also covered in the Section ‘Suggested
IMPACT 7
indicator – Human-dog interactions’ under the impact of improving dog welfare, as the
behaviour of people towards dogs can impact on the level of fear, prolonged stress and
therefore welfare experienced by roaming dogs. However, this indicator could also reflect
changes in the perception of people towards roaming dogs as it measures both their IMPACT 8
positive and negative behaviours towards dogs. The Section ‘Behavioural observation
method’ describes a method for collecting data relating to 3 potential indicators:
percentage of positive human behaviours out of total of all ‘extremes’ of human-
dog interactions; percentage of relaxed dog responses out of total of all ‘extremes’
of human-dog interactions; percentage of negative human behaviours out of total
of all ‘extremes’ of human-dog interactions. These indicators have not been reported
previously in the literature and so are included here as suggested indicators.

Impact 5: Improve Public Perception
Suggested indicator – Cruelty towards dogs

The prevalence of cruelty towards dogs, whether in the form of neglect or deliberate abuse,
can be a measure of public perception of dogs, including their sentience and value. The
response to such cruelty, in the form of public outcry and prosecutions, reflects a further
dimension of public tolerance towards such cruelty and also institutional/government
perception of dogs and development of related legislation. This suggests two classes of
indicators; the prevalence of cruelty towards dogs as measured through reporting of IMPACT 1
cruelty to governments and NGOs and also the number of successful prosecutions against
cruelty towards dogs as measured through the justice system. How these indicators are
expected to change in response to an intervention will depend on the country and the stage
of the intervention; in some cases the intervention may be looking for increased reporting of
cruelty cases as an indicator of increasing awareness and sensitivity of the public towards
IMPACT 2
protecting dogs from harm, however at a later stage they may wish to see the prevalence
fall as cruelty itself becomes less widespread. Similarly with prosecutions, an intervention
may want to see an increase in enforcement of legislation against cruelty initially and then
reduction in prosecutions over time as less cruelty is reported. An increase in the proportion IMPACT 3
of reported cruelty cases that are successfully prosecuted is likely to always be desired.
The use of these indicators to assess the impact of a DPM intervention was not found and
so these are presented as suggested indicators at this time.
IMPACT 4
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Impact 6: Improve Rehoming/
Adoption Centre Performance
International
Companion Animal
Indicators related to rehoming/adoption centre performance could be argued to be a

measure of centre effectiveness and therefore related to intervention effort and not impact.
However, many parts of a DPM intervention have the potential to feed into whether a centre
IMPACT 1
is successful or not, sometimes independently of the actions of the centre itself. For example
sterilisation could reduce unwanted births which would reduce or alter the age structure of
intake and an improvement in people’s perceptions of dog could increase adoptions. Hence
a discussion of indicators relating to this impact of centre performance is included. IMPACT 2
Recommended indicator – Annual live release rate
The Asilomar Accords (Anon, 2004) is a national initiative in the US to collate data on
rehoming centre performance and hence provide an opportunity to monitor and evaluate
changes over time, potentially across large geographical areas and across several rehoming
IMPACT 3
centres. The key indicator used is annual live release rate, which is expressed as the
percentage of total outcomes for shelter animals that are live outcomes (adoptions, outgoing
transfers, and return to owner/guardian) over the year. The total outcomes include all live
outcomes plus euthanasia not including owner/guardian requested euthanasia or died/lost IMPACT 4
in shelter/care. The Asilomer Accords guidance provides definitions of the data to be used
in the calculations of these rates and also practical tools such as a data gathering form
and simple equation for the calculation of the rate itself. Annual live release rates have been
used to evaluate impact of interventions on both individual centres and whole communities
comprised of several centres (e.g. Weiss et al. 2013) and hence is presented as a
IMPACT 5
recommended indicator for situations where rehoming centres are unable to run policies of
non-euthanasia of healthy animals.
Recommended indicators – Intake, net rehoming, footfall and IMPACT 6

time in shelter
For centres that have a policy of not euthanasing healthy dogs, their annual live release rate
will always be 100% and hence they require additional indicators. These will also be useful
for centres without 100% live release rate to explore their performance in more detail. Intake
IMPACT 7
rates, split by age category are an indicator of the size of the unwanted dog population
and have been used in evaluation of intervention impact (e.g. Frank and Carlisle-Frank,
2007). Although it should be noted that if a rehoming centre is constantly running at full
capacity, their intake rate may be more a reflection of how fast they rehome and therefore
IMPACT 8
make space for new dogs than a reflection of the dog population externally. In some cases
waiting list may also be kept and the length of these lists/average waiting time scrutinised
along with intake rates. Net rehoming rates include the number of dogs rehomed and take
account of any adopted dogs that are later returned, and so are a more accurate measure
of rehoming success as opposed to gross number of rehomings. Footfall across a specified
time period is the number of visitors (families and couples count as one) to the centre.
Changes in ratio of net rehoming: footfall allows for evaluation of the success of rehoming
dogs as it takes into account the number of opportunities dogs had to be adopted. Average
time spent in the shelter can be an indicator of how long it takes for a dog to be rehomed,
the proportion of dogs over a certain length of stay (e.g. 3 or 6 months) may also be
an important indicator of shelter performance as these long-term dogs will potentially be
suffering some welfare compromise.
45
Impact 6: Improve Rehoming/Adoption Centre Performance
All these indicators require the rehoming centre to collect data on a daily basis and analyse
this data regularly. Ideally all these indicators are collected and patterns across indicators
analysed, as well as changes in individual indicators, to allow for accurate interpretation of
underlying causes (e.g. has intake increased because of a change in the external population
or as a result of increased net rehoming? Has average length of stay decreased because
we are rehoming more successfully or have more dogs returned because we have relaxed
adoption ‘rules’?) Transparency of these indicators should be encouraged; however,
rehoming centres may not feel comfortable about exposing some indicators, in particular
IMPACT 1
when their annual live release rate is below 100%, so this data may need to be treated
confidentially and protected in any public evaluation reports.
IMPACT 2
IMPACT 3
IMPACT 4
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Impact 7: Reduce Negative

Impact of Dogs on Wildlife
International
Companion Animal
The impact of dogs on wildlife can occur in several ways (see Hughes & Macdonald 2013):
(i) through predation of wildlife by dogs, as was most often reported, followed by (ii) disease
transmission to wildlife, and to a lesser extent (iii) through competition with wild carnivores,
IMPACT 1
(iv) hybridization and (v) predation of dogs by wild animals. The most common interactions,
predation and disease transmission will be the focus of this section. Note that in all cases
collaboration with wildlife stakeholders would be advised in order to collect data relating to
wildlife populations. IMPACT 2
Recommended indicator – Presence of dogs in wildlife areas
The presence of dogs in designated wildlife areas can be measured through surveys
specifically recording signs of dogs or as one of the species noted during ongoing
population surveys of wildlife. For example, Butler et al. (2004) had wildlife rangers record
IMPACT 3
sightings of dogs and dog prints along a transect within the border of a wildlife park, about
six times per month, providing a long term relative index of dog abundance in the wildlife
area. In Israel, Manor and Saltz (2004) recorded any dog sightings while surveying for
mountain gazelle at water holes, they used the proportion of observations in which dogs IMPACT 4
were sighted as a ‘dog-presence index’, allowing them to track changes in dog presence
over time with planned interventions. The increasing use of camera traps to record presence,
abundance, and population changes in wildlife, has the potential to gather opportunistic
data on the presence of dogs in designated wildlife reserves (e.g. Jenks et al., 2011). In
addition, but a rather resource heavy approach, is to radio/GPS collar a small number of
IMPACT 5
dogs. Using this method, the location and range of dogs can be quantified and mapped
(e.g. Meek, 1999), so the extent to which dogs encroach into wildlife designated areas can
6
be accurately determined (e.g. Butler et al., 2004).
IMPACT
Recommended indicator – Predation events
and predation impacts
As the presence of dogs in wildlife areas does not automatically indicate negative impacts,
additional indicators of the impact on wildlife will be needed. The number of observed
IMPACT 7
wildlife kills by dogs would appear to be an ideal direct indicator of negative impacts by
dogs. However as these are relatively rare events, data on this indicator may be difficult to
collect. Asking both community volunteers and wildlife rangers to report kills by dogs to a
central point may help with data collection (e.g. as used in Zimbabwe by Butler et al. 2004).
IMPACT 8
A relatively resource-intense approach is to radio/GPS collar a sample of dogs and follow
them as they enter the wildlife area to increase the chance of observing a predation event.
Conversely, a sample of the wildlife species can also be fitted with GPS collars containing
mortality sensors to allow for prompt necropsies and identification of the predator involved
using scat, tracks and distance between puncture wounds, although differentiating between
wild canids and domestic dogs is not possible with these signs alone (Young et al., 2011).
If the predation event is not observed, mitochondrial DNA analysis can be conducted on
the saliva left on the carcass in order to establish the species responsible (Williams and
Johnston, 2004). This approach can even be used to identify the individual responsible if
47
Impact 7: Reduce Negative Impact of Dogs on Wildlife
saliva samples can also be taken from ‘suspects’. However this must be done within a short
time period of the kill to avoid contamination of the predator’s saliva with scavenger saliva. In
some environments this may well be a matter of just a few hours, plus the tests themselves
are expensive.
The indicator of the number of observed wildlife kills by dogs alone is also not really
sufficient. As described by Hughes and Macdonald (2013); this is “unquantified in terms of
population impacts. Reporting individual instances of predation gives no indication of the IMPACT 1
impact on local prey populations and, therefore, whether it is of conservation concern”.
Additional indicators are therefore needed to reflect how the wildlife population is responding
to this predation, ideally by monitoring population numbers, distribution and structure
of wildlife prey at the same time as monitoring presence of dogs within designated wildlife
areas or number of observed wildlife kills by dogs to see if there is any correlation. For
IMPACT 2
example, the dog-presence index used by Manor and Saltz (2004) was found to correlate
with kid:female gazelle ratios; there were more kids per female (a favourable ratio in terms of
the potential for gazelle populations to grow), as dog-presence index declined.
IMPACT 3
Recommended indicator – disease incidence
in dogs and wildlife
Dog populations can act as a reservoir for diseases that are also a risk for wildlife, in
particular carnivores. Rabies and canine distemper virus (CDV) are the two mostly commonly
IMPACT 4
cited examples, but parvovirus and Ehrlichia canis have also been implicated. For diseases
with short infection cycles and high mortality (characteristics of both rabies and CDV),
transmission cannot be maintained in small endangered wild populations; as the number
of animals that succumb to infection increases, the number of new susceptible hosts IMPACT 5
diminishes, and the infection eventually fades out. New infections in wildlife populations
are usually triggered by contact with more abundant reservoir hosts, most often domestic
dogs (Cleaveland et al. 2007). Although for CDV in African wild dogs (Lycaon pictus) there
is evidence to suggest this pathogen is maintained independently of domestic dogs,
and hence management through dog vaccination against CDV may not be suitable in all IMPACT 6
situations (Woodroffe et al., 2012). Indeed, Woodroffe et al. (2012) advise careful cost-
benefit analysis before deciding whether managing disease in domestic dogs will protect
7
wildlife. Exposure to canine pathogens may actually provide some ongoing immunity and
therefore protection against large outbreaks and mortality, as well maintaining selection IMPACT
pressure for disease resistance.
Where DPM interventions have identified that it is cost effective to minimise the risks to
wildlife by reducing the incidence of infectious diseases and parasites in dogs, the key
indicator to measure success will be the disease incidence in both dogs and susceptible
IMPACT 8
wildlife species within the same area. Although a reduction in disease incidence in
both dogs and wildlife may be sufficient to attribute success of a DPM intervention, and
is likely the most affordable indicator to measure, additional evidence of the epidemiology
of the disease and how it is transmitted between dogs and wildlife is ideal. This may be
achieved by carefully mapping and monitoring location and movements of both dog and
wildlife species (e.g. using radio/GPS collars) to assess contact rates. Furthermore, detailed
serological studies of both dogs and wildlife to determine the proportion of the dog/

Impact 7: Reduce Negative Impact of Dogs on Wildlife
wildlife population with antibodies to diseases would be advantageous. Measuring such

indicators will need to be conducted in the long-term and across a range of age groups,
taking account that CDV antibodies can remain in circulation many years after exposure to
CDV and that vaccination for CDV will also lead to a positive blood result. When used in the
Serengeti this approach revealed that CDV appeared and then disappeared for many years
in spotted hyenas (Crocuta crocuta) before reappearing in juveniles, suggesting that the
virus was not persisting in this wildlife species and had been introduced by dogs acting as a
reservoir host (Cleaveland et al., 2007). The resources needed for long-term measurement of
IMPACT 1
antibody levels, and the relatively complex analysis required to explore this data, means this
indicator is usually measured as part of a long-term research programme into wildlife and
2
dog disease transmission, and may be beyond the scope of the monitoring and evaluation
plans of many DPM interventions. IMPACT
Note that due to the almost invariably fatal nature of symptomatic rabies, looking for
antibodies to rabies in serological samples may result in very few positive samples in non-
vaccinated populations, and so is not recommended for general monitoring.
All the indicators mentioned in relation to the impact of dogs on wildlife require collaboration
IMPACT 3
between those organisations working with dogs and those working with wildlife. Disease
surveillance and predation recording is likely most effective where it is run as a joint cross-
disciplinary effort between dog and wildlife stakeholders.
IMPACT 4
IMPACT 5
IMPACT 6
IMPACT 7
IMPACT 8

Impact 8: Reduce Negative Impact
Of Dogs On Livestock
International
Companion Animal
Negative impacts of dogs on livestock can occur through predation, disease transmission or
lost production due to stress resulting from close presence of dogs. DPM interventions may
be looking to evaluate their impact on livestock if they are concurrently reducing roaming
IMPACT 1
dog density and therefore the number of dogs potentially predating or harassing livestock, or
by reducing disease transmission through deworming/vaccination of dogs as reservoir hosts
for Echinococcus granulosus or rabies.
IMPACT 2
Suggested indicator – Livestock predation by dogs
The number of livestock predated upon by dogs per unit time appears the most direct
indicator for the negative impact of dogs on livestock, and will presumably be well correlated
to production losses through stress following non-lethal interactions between livestock and
dogs. Note that changes in livestock numbers would also need to be considered when
IMPACT 3
using this indicator, there may be seasonal changes in livestock numbers and vulnerability
to predation (e.g. susceptible young animals during breeding seasons), as well as potentially
changes in numbers over longer time periods. Further, changes to livestock management
practices should also be considered, such as increases or decreases in confinement of IMPACT 4
livestock that could change their susceptibility to predation.
Accessing the number of livestock predation events through secondary sources may be
challenging, as governments do not always maintain records of such predation events nor
do farmers always report them. Presumably if there is a government run compensation IMPACT 5
programme available for livestock losses to predation, there will be data recorded on the
number of predation events reported and subsequently receiving compensation, although
the level of detail included on the location of the predation event (important for determining
whether this occurred within or outside the intervention area) and which predator is culpable
(i.e. dogs or wildlife) may differ. In other countries, compensation for livestock predation is
IMPACT 6
received from insurance policies and hence data on the number of predation events may be
best accessed through insurance companies; this was the approach used by Adriani and
Bonanni (2012) to assess the impact of stray dogs on livestock in Italy.
IMPACT 7
An alternative approach is to use questionnaires of farmers regarding livestock losses. For
example, Wang and Macdonald (2006) asked farmers living around a wildlife park in Bhutan
about livestock predation events, although in this case they didn’t report losses to dogs,
only wildlife predators. The United States Department of Agriculture (USDA) uses a national IMPACT 8
survey of a random sample of producers to develop a report on cattle loss every 5 years,
including losses of cattle to predators split into species (in 2010, 11.3% of cattle predation
loss was due to dogs; NASS 2011). This indicator is presented as suggested only because
no examples of its use in evaluating dog population management interventions could be
found.
50
Impact 8: Reduce Negative Impact Of Dogs On Livestock
Suggested indicator – Livestock disease

Echinococcosis is arguably more important as a human health issue than as a livestock
health issue as often infections can be asymptomatic during the life-span of the livestock
intermediate host or at least be minimal enough to go undiagnosed in a herd environment.
However there are losses associated with E. granulosus in livestock, including direct
economic losses from condemnation of offal and indirect losses from reduced growth, milk
production and fertility in infected animals. Hence DPM interventions that aim to reduce E. IMPACT 1
granulosus in dogs can measure their impact on livestock by monitoring changes in the
number of livestock with liver or lungs infected with E. granulosus cysts at the time of
slaughter per unit time (usually per month or year) and per age group (covered in more
detail in the section on ‘Impact on echinococcosis risk’ under the impact of ‘Reduce risks
to public health’) and potentially also including a subsequent stage of economic analysis.
IMPACT 2
For example, the economic impact calculated by Benner et al. (2010) for echinococcosis in
Spain and by Budke et al. (2006) for the worldwide impact included both direct and indirect
costs related to livestock as well as humans. Note that for assessing the extent and costs of
livestock disease all species that are susceptible to E. granulosus can be included, but when IMPACT 3
assessing the risks to human health the prevalence in sheep is most relevant as the vast
majority of human cases are transmitted from the sheep-dog cycle (Eckert and Deplazes
2004).
Rabies virus can infect all mammals and in rabies endemic countries the loss of livestock to IMPACT 4
rabies is both an animal welfare and economic cost of the disease. Unvaccinated dogs are
a primary reservoir host for rabies, thus reduction of rabies in the dog population through
an intervention that includes vaccination may also lead to a reduction in rabies in livestock.
Hence a potential indicator of success of such an intervention would be the number of
laboratory-confirmed livestock rabies cases per unit time (usually per month), although
IMPACT 5
the number of clinically diagnosed livestock rabies cases per unit time is also a valid
indicator of rabies risk and can be particularly useful for increasing case detection when
laboratory infrastructure is weak. Potentially this data could also be used to estimate the
economic impact of rabies reduction following an intervention. The challenge is that rabies in IMPACT 6
livestock is rarely reported or recorded in secondary/official sources, because there is neither
a cure for rabies nor compensation for livestock, so severe underreporting will reduce the
7
reliability of this data. Setting-up key informants within the farming community or livestock
health workers/veterinarians may be an ideal first step to increase surveillance of livestock IMPACT
cases ahead of an intervention. Note that in Latin America compensation systems do exist,
however this is for rabies carried by vampire bats (which can impact large proportions of a
herd at one time) and hence not relevant to this document.
IMPACT 8

Methods of Measurement
This section provides detailed descriptions and protocols for methods of measuring the
indicators listed in the previous section. The method used will significantly influence the
data collected, for example a different population of dogs will be scored for body condition
when measured through street surveys as opposed to through clinic records. Hence the
methods and the detailed protocols used to implement them should ideally not be changed
over the lifetime of an intervention and its assessment.If a method of measurement has to be
changed, a significant period of overlap between the new and old method will expose how
the data relating to the indicator is affected by the new method and correction factors can
then be applied to permit comparison between data collected by different methods.
Methods of Measurement:
Questionnaire Surveys
International
Companion Animal
Questionnaire surveys Questionnaire

surveys
Questionnaires comprise a set of standardised questions that can be in closed (questions
with a limited set of answers to choose from, such as ‘yes’ or ‘no’) or open format
(questions that allow the respondent to answer in their own words). This is a well-tested
methodology for collecting data relating to a range of indicators. They are relatively time
Participatory
consuming to deliver and analyse and so tend to be used infrequently for DPM evaluation
Research
and more often appear as part of initial in-depth assessment of dog populations ahead
of intervention planning. However, dependent on the sampling frame and method used
to select respondents, they can provide reliable estimates relating to the wider population Street
and so may be useful for evaluation where resources allow. Annex E provides a sample Surveys
questionnaire that has been based on DPM questionnaires that have been field-tested in
several countries. It has been designed to be as short as possible and yet still to collect
data relating to indicators mentioned in this guidance document. However, adding extra
questions relevant to your intervention location is welcomed. Note the questions in Annex E Secondary
have been phrased to ask about the performance Sources
of actual behaviours, such as stating how often
a dog is fed rather than a subjective assessment
of whether the feeding is regular or not. Further, Clinic
multiple choice answers have been included to help records
with data analysis with all likely responses listed,
including a ‘don’t know’ response to avoid forcing
the respondent to guess.
Behavioural
Note that when using questionnaires you should Observation
ask respondents for their permission to use their
data, you should also explain how their name (if
asked) will be used. The questionnaire in Annex E Vaccination
starts with a statement to read to respondents and Coverage
a place where their permission can be recorded.
Relevant Impacts
Questionnaire surveys can be used to measure a wide range of indicators related to several
impacts. These include several indicators of improving dog welfare such as body condition
score, skin condition and female : male ratio (where the target dog population is visible
in households as opposed to on the street); indicators for improving care provided to
dogs, including specific dog care behaviours performed by dogs owners and knowledge
and attitudes of children following education interventions; under the impact of improving
public health the indicators of dog bites and also dog vaccination can be measured using
questionnaire surveys; and finally, indicators of an improvement in public perception
including attitudes towards dogs and the adoption of dogs can be measured through
questionnaire surveys.
53
Methods of Measurement: Questionnaire Surveys
Sampling
Questionnaires designed to measure changes in human behaviour over time ideally follow
a consistent sample (a ‘cohort’) of households, hence are longitudinal as opposed to
cross-sectional in design. This is because cross sectional questionnaires use a new sample
of people at each point in time (a ‘point sample’ of people), it is difficult if not impossible Questionnaire
to select equivalent samples, and the differences in people’s background can influence surveys
reported behaviours and attitudes, hence you may not be measuring a change in attitudes
over time but the difference in the attitudes between the two samples of people chosen.
Many factors could influence peoples’ attitudes towards dogs and the way their dogs are
Participatory
kept, including gender, religion, degree of education and previous experience of owning an
Research
animal as a child. As it is difficult to control for all these factors during analysis, or to ensure
the sample contains exactly the same representations of these different types of people, the
most robust approach is to ask the same households for their dog care behaviours in order
to assess how these have changed over time. This will require the initial sample to be larger Street
than required for final analysis, to allow for dropout but still retain a sufficient sample size; Surveys
reasons for dropout include people moving away from the intervention area, people losing
and not replacing dogs and fatigue with the study. The potential problem with a longitudinal
approach is that a change in dog care behaviour may occur in sample households as a
Secondary
result of being observed through repeated questionnaires, rather than in response to the
Sources
intervention.
Where a longitudinal approach is not possible, a cross sectional questionnaire survey is

the alternative. A larger sample size may be needed for a cross sectional approach to
Clinic
increase the chance of a representative and equivalent sample being selected each time. A
records
larger sample size is also needed detect significant change, as analysis will be on different
people (‘subjects’) at different time periods (‘between subject’ design and analysis), rather
than on the same subjects at different time periods (‘within subject’ design and analysis).
The ‘between subject’ differences at different time points may mask effects due to the Behavioural
intervention; a larger sample size of subjects at each time point can help to expose those Observation
effects. The same sampling method and frame should be used to select households at
each cross-sectional survey event. For example, every 3rd house along every street in the
intervention area, or cluster sampling where all houses in a sample of areas drawn from Vaccination
a larger sample are approached. Examples include the WHO Expanded Programme on Coverage
Immunization cluster-survey technique, used in dog-related questionnaires by Davlin and
Vonville (2012) and Kongkaew et al. (2004). Attempts should also be made to ensure the
socioeconomic background of each sample is the same; this can be achieved by ‘stratifying’
the sample for socioeconomic status in the same way at each survey event (in a ‘stratified’
sample, some dogs or people have a known greater chance of being selected than others,
this is used when your target population clearly appears as sub-populations with important
differing characteristics related to your impact, see Section ‘Making your impact assessment
robust’ for more details). This can be achieved by drawing samples from geographical areas
of known socioeconomic status (data on age, education and general socioeconomic status
by area can be available from census data).

Recruitment of respondents
Recruiting a sample of respondents can be done in a number of ways, and the best
approach to use will depend on the location and objective of the questionnaire survey. If
using a longitudinal questionnaire to assess changes in dog care behaviour in a consistent
sample of households, a face-to-face interview at the doorstep may be the best approach Questionnaire
to limit dropout. One potential limitation of the door-to-door approach of recruitment is that surveys
it may be biased towards those people that are at home most often, such as young families,
the elderly and unemployed. Administering questionnaires in the evenings and weekends
can help to avoid this bias.
Participatory
Door-to-door recruitment is also suitable when using cross-sectional approaches but it can Research
be quite time consuming, especially if looking to include a large number of dog owners in
an area where dog ownership is relatively uncommon (for example, in urban areas where
dog ownership may be <10% of households; e.g. 7% in costal urban communities of Street
Tanzania (Darryn L Knobel et al., 2008)). One approach available when using cross-sectional Surveys
designs is to use ‘convenience sampling’: accessing respondents where people are most
numerous and accessible, such as at bus stops, on buses, in parks or outside places of
worship. The limitation of this approach is that you may be accessing a biased sample
of people; however this could be seen as a benefit if an objective is to assess changes in Secondary
peoples’ perception of roaming dogs, which may be best asked of people spending time in Sources
public spaces where roaming dogs occur. One benefit of this approach is that selecting a
quota of respondents based on visible factors such as age and sex becomes possible, by
approaching people within those categories until the desired balance has been reached (e.g.
Clinic
to match the demographics of the location and therefore increase the chances of this being
records
a representative sample).
Another potentially easily-recruited and therefore

convenient sample of respondents is children at Behavioural
school (although some countries do have regulations Observation
about conducting questionnaires of children and so
appropriate approvals must be sought). Depending
on the access to education in the locality, children
Vaccination
attending a range of secondary schools could still
Coverage
represent a range of socioeconomic and religious
groups, although this approach restricts the sample
to households with secondary-school aged children.
Children of secondary-school age may both be able
to fully comprehend the questions in the questionnaire
and may well be involved in their family’s dog care, but
these assumptions would need to be tested in each
location.

Accessing respondents through landline phones has been a viable method of recruitment
in the past, however many countries are experiencing an increase in the proportion of
households with mobile phones only, leaving this method open to increasing bias. Access
through the internet is a very cost-effective method and also benefits from immediate data
entry into a database if respondents are filling in questionnaires online. However, as with
mobile phones, the use of the internet is increasing overtime and hence your sample may
also change over time, which may confound the effects you are aiming to detect. Postal Questionnaire
questionnaires can be used for both longitudinal and cross-sectional designs but this surveys
requires a fully literate population of people to avoid biases. A low return rate should be
expected for postal surveys.
Participatory
Any method of recruitment can suffer from bias resulting from only people interested in
Research
dog-related issues agreeing to respond. This can be particularly problematic when using
methods that do not include face-to-face recruitment, as the rate of responses can be very
low (although incentives can be used to increase response rates such as entry into prize
draws). As these methods are being used to assess changes in dog care and perceptions Street
over time such a bias may not be too much of a problem, so long as the demographic Surveys
of people with an interest in dog-related issues does not change over time. Recording
response rates to questionnaires can help monitor changes in people’s interest.
Interviewer bias Secondary

Sources
Interviewers can unintentionally affect responses, for example through ‘social desirability’
where the interviewee provides the responses they think the interviewer wants to hear or
attempts to create a particular perception of themselves; this may be done consciously
or unconsciously. Avoiding this completely can be difficult; basic differences between Clinic
interviewers can be relevant, including gender and age, and even without an interviewer records
present, the respondent may make assumptions about the desired responses based
on where the questionnaire has come from (e.g. the sender email address or website
hosting the questionnaire). Interviewers and question phrasing need to appear as neutral Behavioural
as possible and all interviewers should use a set script to both introduce themselves and Observation
ask the questions to avoid differences resulting from their phrasing. Consider not having
staff or anyone publicly identified with an intervention conducting the interviews to maintain
objectivity. If this is not possible, be sure to train your interviewers how not to unintentionally
Vaccination
affect the responses, they may need to wear neutral clothing (i.e. no logos) or consistently
Coverage
wear the same logo every time the questionnaire is conducted. Using a team of students
as interviewers (with the same gender and age composition at each survey event) may be
a good option, as they may appear to interviewees as less of an authority figure to appease
with appropriate responses.
Health and safety

The safety of interviewers is extremely important to consider especially when using face-
to-face interviews. The potential risks involved should be reviewed for every location and
mitigated as far as possible. Examples of mitigation include interviewers working in pairs,
stopping before sunset, having access to a mobile phone, transport and a supervisor close
by to assist when required and check everyone in and out of the survey event. Interviewers
are themselves also responsible for behaving appropriately, for example by assessing each
house before knocking (if concerned, they can miss out a household and record where,
when and why this was done), being polite but not overly friendly with respondents, politely
refusing invitations to enter households, dressing appropriately, carrying official ID and halting
an interview early if they are concerned (recording where, when and why this was done). In
some countries personal alarms may also be available for interviewers to carry with them.

Agreement with attitude statements

Monitoring changes in perceptions or attitudes towards dogs can be done through repeated
surveys that ask people for their level of agreement with dog-related statements. People are
presented with both positive and negative statements relating to dogs, such as “dogs add
happiness to people’s lives” and “street dogs pose a danger to people”, and are asked to Questionnaire
what extent they agree or disagree with the statements. Their level of agreement is indicated surveys
on a scale, called a ‘Likert scale’ (see Annex E for examples of Likert scales), and can later
be translated into a score allowing for comparison over time.
Participatory
Sample attitude statements Research
Annex E includes four lists of attitude statements that have been used in different
environments: one in Colombo, Sri Lanka where both owned and unowned dogs are
found roaming (Sankey et al., 2012); one in Tanzania where the roaming dog population is
comprised of majority owned dogs allowed to roam freely, these statements were designed
Street
specifically for dog owners (Knobel et al., 2008); one to compare attitudes between UK and
Surveys
Japanese students (Miura et al., 2000); and one that was used to compare the attitudes of
4 year old children towards dogs in three European countries (Lakestani et al., 2011). Any of
these attitude statement lists can be selected and used as presented or can be adapted to Secondary
a new location – note the attitude statements for children are for use in a classroom situation Sources
as opposed to during a household questionnaire.
The process of adaptation for the adult attitude statements could include using two or
more focus groups comprised of local people (both dog and non-dog owners) to discuss Clinic
the statements and their suitability to the location. Irrelevant statements are then removed records
and ambiguous ones made clearer using locally relevant terms/phrasing. The groups can
also be asked whether there are other important dog-related issues not covered and new
statements covering these aspects devised. This new list should then be piloted on 20+
Behavioural
people of varying backgrounds and opinions. Following the pilot test results, the list can
Observation
be updated further and retested, or accepted in its current form. Note that conducting this
process of adaptation may require accurate translation into the local language. This can be
done translating the English statements into the local language, and then back-translating
into English to check for retention of meaning. This procedure is repeated by revising the Vaccination
statement in the local language until the back-translation matches the original English Coverage
version.
For interventions that have the resources and access to relevant expertise, a novel set of
attitude statements can be developed (e.g. as described in Knobel et al. (2008)). In brief,
one potential process is to start with qualitative/open discussions about dogs within focus
groups from which a long list of attitude statements can be developed. These are then
trialled on a pilot group of 20+ people, either individually or in groups, to remove or amend
statements to improve understanding of their meaning. The improved list is then piloted on
a larger group (involving twice as many people as you have attitude statements), followed
by factor analysis of the results. Factor analysis will expose any statements that appear
redundant which can then be removed to make the final list shorter. Additional stages of
testing reliability can also be run, including testing statements with the same people but over
time.

Analysis of changes in attitudes over time

Attitude statements are commonly presented in lists of over 10 items, to allow for
measurement of a range of attitudes and several statements deliberately testing the same
issue. This can create problems at the analysis stage: if every statement is individually
analysed for change over time there is a chance of finding a falsely significant result (i.e. the
score for a statement can differ from its previous score purely by chance. The likelihood Questionnaire
of this happening for any one statement increases with the number of statements tested). surveys
Hence analysis of these statements can involve making a prior selection of key statements
or combining statements to reduce the number of statistical tests conducted. Three different
approaches to analysis are described here. The most suitable will depend on resources
Participatory
available for analysis and the level of interest in specific attitudes.
Research
One approach to analysis is to focus on individual key statements within a list of attitude
statements. Selection of statements as key to the intervention can be done in two ways.
Firstly, the results from pilot testing can indicate statements that are particularly good for
Street
capturing variability in perceptions, i.e. not everybody gave the same responses. These
Surveys
statements may be particularly good at exposing changes in attitudes over time as there
is ‘space’ for the data to show statistically significant change. Secondly, statements for
which there is a logical explanation for how your intervention will lead to a change in this
attitude plus a change in this attitude would be considered important to various intervention Secondary
stakeholders, i.e. it is a relevant attitude to your work. An example might be an intervention Sources
that aimed to reduce dog bites and improve perceptions of dogs through a combination
of rabies vaccination, reproduction control and education in safer interactions with dogs;
this intervention might in particular want to look for changes in people’s agreement with the
Clinic
statement “stray dogs pose a danger to people”.
records
The attitude statements can also be combined into a summative score for acceptance
of dogs. For example, from the attitude statements developed in Colombo, Sri Lanka,
10 statements are clearly either positive or negative with regards to acceptance of dogs
Behavioural
(statements 2, 4, 5, 7, 11, 12, 13, 14, 16, and 18). Scores of 1 to 5 can attributed for how
Observation
each interviewee responded to each statement; with a score of 1 for the response ‘strongly
disagree’ with the statement and a score of 5 when the response was ‘strongly agree’ with
the statement. The scores of negative statements are reversed (i.e. statements 4, 7, 13
and 16) so that all of the individual item scores have the same direction, which allows an
Vaccination
overall score to be calculated indicating acceptance of dogs. Within this scoring system, a
Coverage
minimum score of 10 would mean total non-acceptance and a maximum score of 50, total
acceptance of dogs. Note that this summative score does assume equal weighting to all
attitude statements, which may not be accurate, with some attitudes more important than
others. Consider using expert opinion to weight the statements as more or less important
and use these weighting to adjust the scores; for example using the ‘Delphi technique’ to
systematically achieve a degree of consensus of opinion between experts (e.g. as used
by Whay et al. (2003) to find consensus between experts on the best measures for the
welfare of dairy cattle, pigs and laying hens). Summing the scores may also mask changes
in specific attitude statements, hence this method is ideally used in addition to considering
changes in key indicator statements.
Factor analysis is a statistical method used when faced with a large number of observed
variables (in this case, responses to a large number of attitude statements); it explores the
correlation between these many variables to find a lower number of unobserved composite
variables called factors. For example, this was used to analyse the responses to 47 attitude
statements about dogs along 7-point Likert scales presented to UK and Japanese students

(Miura et al., 2000). Changes in responses to all 47 statements should not be analysed over
time, as this would lead to some significant results just by chance (even with a significance
level of 5%, this equates to one significant result out of 20 tests appearing just by chance).
So the 47 responses were analysed for how they correlated with each other using factor
analysis; this exposed clusters of responses that correlated. Each cluster is called a factor
and is given a suitable name depending on what responses it contains. For example
responses to the 4 attitude statements “I think that stray dogs are a problem in this country”, Questionnaire
“I think that stray dogs tend to bite”, “I think that stray dogs create a nuisance” and “I think surveys
that there are too many dogs in this country” were found to correlate and were given the
factor name ‘dislike of stray dogs’. Five factors were found to underlie the UK responses
and 7 in Japan, including factors described as ‘unconcern/concern for hygiene’, ‘dislike of Participatory
stray dogs’, ‘acceptance of dogs as equals’, and ‘acceptance/unacceptance of euthanasia’. Research
Each statement can be given a score depending on the average position of the responses
along the 5 or 7-point Likert scale. Scores for all statements falling within each factor can
then be combined to give a score for each factor. These factor scores can then be analysed
for changes over time by combining statements in the same way at each survey. Street
Surveys
Using questionnaires to estimate fecundity Secondary

Sources
Questionnaire surveys can be used to estimate the number of litters per female per year,
a measure of fecundity. In the sample questionnaire in Annex E, owners are asked of all
their female dogs ‘how many litters has she had in her lifetime?’; hence every female dog
will have a breeding history, although for many this will be 0 litters so far in their lifetime. Clinic
For every female dog, her average litters per year since she reached the age of 1 year records
(assumed to be the age of sexual maturity) can then be calculated (i.e. total number of
litters for that female / number of years since reaching age 1). The average fecundity for
the female population is then calculated as the average litters per year for all females of 2 Behavioural
years and above; only those females of 2 years and above are included to ensure they have Observation
experienced a full year of potential breeding after reaching the age of 1 year. The assumption
of 1 years old as the age of sexual maturity can be tested by calculating the average age at
first litter; the question ‘How old was she when she had her first litter?’ is also included in the
Vaccination
sample questionnaire.
Coverage
The average number of litters per female can also be further extended to the average
number of puppies per female per year, an alternative indicator of fecundity. The average
litter size can be calculated from responses regarding fate of puppies born in litters
produced in the previous 12 months (the sample questionnaire in Annex E includes the
following question to capture this data: ‘if she has had a litter in the previous 12 months,
please tell us about the fate of all puppies from this litter’, followed by nine outcomes). The
number of litters per female per year is then multiplied by the average litter size to calculate
the average number of puppies per female per year.

Using questionnaires to estimate survival

Questionnaire surveys can be used to estimate survival of owned dogs as part of either a
longitudinal or cross-sectional study. A simple estimate of current adult survival is available
by comparison of the total number of adult dogs owned a year ago with the number owned
currently that are at least 12+m months old, where m is the minimum age in months at Questionnaire
which a dog is considered as adult (e.g. 5 months). The respondents are asked “over the surveys
past 12 months, did any of the household’s adult dogs die or leave the household?” From
the responses, the proportion that have died or disappeared can be estimated, leaving
out those dogs that have been sold or donated to other people and which may still be
Participatory
alive. There may be cases where the fate of dogs that have left the household is unknown,
Research
however this is likely to be a relatively small number and the proportion estimated from the
dogs with known fate can be applied to the total number of dogs that left the household,
allowing the total number of dogs that have died or disappeared in the last 12 months to
be estimated. The number of adult dogs that have died or disappeared in the previous Street
12 months is then subtracted from the number owned a year ago and the remainder Surveys
(representing the number of dogs that have survived the previous 12 months) divided by
the number owned a year ago to estimate adult survival.
An alternative approach is to estimate the age structure of the current owned population Secondary
(split into equal age groups of one year) by asking the respondents the current age of Sources
their dog, or when and at what age each dog was obtained. In a population that is either
constant or that has been growing at a constant rate for some years (perhaps in line with
the growth of the human population) the age structure will have stabilised to the point where
Clinic
the number of dogs a years old divided by the number a − 1 years old equals S a λ ,
records
where S a is adult annual survival at age a and λ is annual population growth rate of dogs.
Although annual survival will be reduced in very old dogs survival of owned dogs that are
allowed to roam is probably low enough for the number of dogs that reach that advanced
age to be negligible. In that case an estimate of an age-independent adult survival Behavioural
probability S is available as the average ratio of the size of successive adult age groups Observation
multiplied by λ .
Survival of pups is certainly not independent

Vaccination
of age. Whereas respondents may not be
Coverage
able to provide reliable data on the age at
which adult dogs died, those owning dogs
that have recently had a litter of pups may
be able to remember the age in weeks
or months at which any pups died. Age-
specific survival can then be estimated using
the Kaplan-Meier method (Kaplan and Meier,
1958), which does not assume a stable
age structure (important as any seasonality
in breeding would destabilise pup age structure) and allows individuals whose survival is
unknown (because they were sold or given away) to be accounted for in the analysis. The
ages at which pups were known to die are listed sequentially. Just before each of those
ages there is a number of pups known to have reached that age (pups sold or given away
before that age are excluded from that number). The number of pups that died at that age
is then subtracted and the remainder (representing the pups surviving beyond that age) are

then divided by the total number of pups known to have reached that age. That provides an
estimate of survival over the interval since the previous age at which one or more pups died.
Multiplying those estimates up to a specific age provides an estimate of pup survival to that
age.
Questionnaire
Tools available to implement surveys
and analyse questionnaires
Participatory
Conducting questionnaires can be relatively time consuming and so tools to reduce the
Research
different processes involved are potentially very useful. Questionnaires can be designed on
survey sites such as Survey Monkey (www.surveymonkey.com) and then, if internet access
is sufficiently reliable, the responses can be filled in on a mobile phone or tablet during
face-to-face interviews (note, any drop in internet access would mean halting the interview Street
until connection is re-established). This avoids the need for printing forms and transcribing Surveys
data from forms to computer later, and it also offers some basic analysis features. If internet
access is not reliable, there are applications for phones and tablets that allow for data
entry even when offline, for example Device Magic (www.devicemagic.com), SurveyToGo
(www.dooblo.net) or Open Data Kit (www.opendatakit.org).
Secondary
Sources
Clinic
records
Behavioural
Observation
Vaccination
Coverage

Participatory Research
International
Companion Animal
The human development field has been the key innovator in devising ways of engaging
beneficiaries in the process of development, not just as passive receivers of aid but as Questionnaire
decision makers themselves who drive the development agenda in their own communities. surveys
One of the earliest examples is rapid rural appraisal (RRA) which emerged in the 1970s;
this involved communities living in poverty in rural areas being supported to come together
in groups to elucidate their problems using exercises that did not require literacy. Further,
Participatory
they prioritised those problems and explored their causes, leading to targeted development
Research
support that was driven by the desires of the beneficiaries themselves as opposed to
external experts. Central to this participation approach is the concept of ‘collective action’;
that by bringing people together through participatory exercises they can build a collective
group with the ability to work together to improve their situation, far more effectively than Street
they could as individuals. Jumping forward many decades and we see these approaches Surveys
used in many fields, including animal health, conservation and most recently in animal
welfare, perhaps best reflected through the work of the Brooke and their book ‘Sharing the
load’ (Van Dijk et al., 2011).
Secondary
The role of participatory approaches in impact assessment has been present from the Sources
outset, where the beneficiaries that initially exposed the problems they wanted to have
addressed, would also then monitor how these problems responded to the various
interventions in which they engaged. This process can be termed participatory impact Clinic
monitoring or action research, and is as much a method of intervention as it is a method of records
impact assessment. An animal welfare example can be drawn from ‘Sharing the load’ (Van
Dijk et al., 2011), where a community of working animal owners come together to assess
the welfare of their animals and the role of these animals in their own well-being, through a
set of group building exercises under the guidance of a facilitator. The community may then Behavioural
use exercises where they review the needs of their animals from the animal’s perspective Observation
(in ‘Sharing the load’ this exercise is called ‘If I were a horse’), followed by a review of how
well these needs are met (called ‘Animal welfare practice gap analysis’ in ‘Sharing the load’);
together, these exercises and others reveal the importance and suggest causes of animal Vaccination
welfare problems and hence potential solutions. The group members are encouraged to Coverage
review the welfare state of their own animals and those belonging to their group (called
an ‘animal welfare transect walk’ in ‘Sharing the load’), and as they engage in solving
these problems. The reviews are repeated so that changes can be tracked and hence
interventions evaluated for their effectiveness. This participatory approach would be possible
to replicate with dog owners, and as such, development and monitoring of indicators would
be done by the dog owners themselves as part of the process.
Even where a participatory approach to the intervention has not been used, participatory
tools can still be used to assess impact, where people within the intervention area are
encouraged to reveal whether they have experienced changes or not. This approach is
arguably extractive, with information being removed to inform evaluation of intervention
effectiveness, rather than empowering local people to make decisions and take action and
so may be a stretch of the term ‘participatory’. Regardless of this criticism, this approach
could provide insights to impact assessment that would not otherwise be found using the
other methods of measurement described in the guidance. Hence, in the rest of this section
we describe participatory tools for impact monitoring. Although we do not describe true
participatory approaches to dog management interventions that have inherent participatory
62
Methods of Measurement: Participatory Research
impact monitoring, we encourage innovation in this field and would value feedback from any
interventions that take this next step.
Most commonly, participatory approaches to assessing impact will be used alongside other
methods of measurement that produce quantitative data; a ‘mixed methods’ approach to
impact assessment. However, it should also be noted that participatory approaches do
not necessarily only result in qualitative data, participatory tools have been designed to Questionnaire
convert qualitative results into quantitative data as well, allowing for comparison over time surveys
and between communities using numerical statistics. Although participatory tools previously
described in the studies above are usually qualitative in nature, numbers can be drawn
usually from the final stages of participatory exercises to aid impact assessment
Participatory
(Chambers, 2007).
Research
Relevant impacts Street

Surveys
The participatory tools described here would be most suited to measuring indicators relating
to public perception of dogs.
Secondary
Sources
Participatory tools for impact assessment in DPM
The use of participatory tools in dog population management is still relatively rare with some Clinic
notable exceptions (Morters et al., 2014; Sankey et al., 2012). Here we describe exercises records
that can be performed with groups of people from the intervention area in order to assess
whether changes in perceptions of dogs have occurred over time.
Behavioural
Creating groups Observation
Participatory tools can be run with individuals, but they are most efficient and potentially
most powerful when run with small groups of (5 - 15) people. These people need to be
selected to represent the wider community and so should include both genders, a spread of
Vaccination
ages and economic backgrounds. Groups should also include both dog owners and non-
Coverage
owners. Dependent on local culture and the level of tension over dog issues, these groups
can mixed or kept separate. For example, dog owning groups and non-dog owning groups
could be kept separate if there is likely to be tension between these groups resulting from
recent problems with dogs (e.g. dog attacks). Another reason for keeping owners and non-
dog owners separate is if you want to ask them different questions. Your intervention may
be focused on service delivery to dog owners
and hence owner discussions will be focused
on their opinions as direct beneficiaries. While
this is not relevant to non-dog owners, you
may want to ask them about indirect impacts
on dog nuisance, welfare and density of dogs
on the streets. The groups will also work best
if they are drawn from the same community,
rather than from across a wider geographical
area, as they can more accurately confer on
the state of dog issues within a shared locality.
Recruiting groups is commonly done through

local community leaders or through snowballing, where people are asked to bring along
friends. It may also be relevant to set-up small groups that represent different stakeholders
in the issue, for example veterinary community, medical community, local government or
teachers. These groups may necessarily cover larger geographical areas but should be from
within the intervention area.
One important consideration is why people would agree to participate in these groups? They Questionnaire
may be interested enough in dogs to want to give up their time, but you may also need to surveys
provide incentives; this may be greater access to the intervention services (e.g. free parasite
control for their dogs if they usually pay) or something independent of the intervention itself,
such as a meal and drink every time they meet.
Participatory
Including a wide variety of groups increases the representativeness of the results, but Research
obviously does concurrently increase the time required. Finding the balance between
numbers of groups and time may be helped by considering that these exercises will also
increase people’s appreciation and understanding of dog related issues. Hence there
Street
may be certain groups where increasing understanding of dog issues will also help with
Surveys
implementing the intervention, and these groups can be prioritised when time is limited.
The following exercises can be run with a new group of people at each evaluation event,
however if possible, reconvening and following the same group of people over time Secondary
would provide a more reliable measure of how people’s perceptions have changed over Sources
time. A combination of both cross-sectional and longitudinal approaches may be best. If
reconvening the same group, they should be asked to bring with them their original maps
and scoring (which should have been kept by the group to increase ownership) or the
facilitator can show photos of the maps and scores taken at the original meeting. This will Clinic
remind the group of past assessments and help them make their new scores with recent records
changes in mind.
It should be noted that the composition of groups may change over time and this should be
taken into account when interpreting results. Although attempts should be made to ensure Behavioural
group membership reflects the wider community it is not always possible to achieve this and Observation
so the ability of the groups to truly reflect the wider community should be critically assessed.
Facilitation Vaccination
Facilitating participatory exercises requires skill and experience to ensure the group exposes
Coverage
its true feelings and does not reflect the opinions of a few vociferous members. There is also
a risk of social desirability if the facilitator is perceived to represent a particular perspective.
Ideally the facilitator is not a representative of the intervention itself but an independent
person seen as neutral – in many locations in the developing world there will be local experts
in participatory approaches that are used to facilitate groups for other issues, such as
poverty or health, whom may be ideal for this DPM application.
Although these participatory tools are designed to produce numerical outputs there will
be a large amount of qualitative information produced during the exercise that can be
extremely valuable for the intervention managers. Hence a representative of the intervention
can be encouraged to take part as a (passive!) recorder to take notes on key perspectives
expressed during the exercise; their affiliation to the intervention can even be left out of their
personal introduction to the group to avoid any social desirability bias. Or the entire session
can be videoed or audio recorded with permission of the participants, allowing intervention
staff to listen to the responses after the event, these recordings may also help with
later analysis.

Participatory exercises
The following exercises are described relatively
briefly. Support from local experts in participatory
approaches is advised, along with ‘how to’ texts such
as ‘Methods for Community Participation’ (Kumar,
2002) and ‘Participatory Livestock Research: A Guide’ Questionnaire
(Conroy, 2004). Exercise 1 is advised for all groups – surveys
followed by either Exercise 2a or 2b – each exercise
may take up to 2 hours to complete, so consider
asking groups to stay for up to 4 hours (with breaks)
Participatory
or using 2 sessions. Please note this is by no means
Research
an exhaustive list of possible participatory tools, but
rather just some initial ideas. We encourage innovation in this area and would value hearing
feedback on these and other tools.
Street
Exercise 1 - Dog mapping Surveys
This exercise is based on resource mapping and is used as a way of introducing the
group to each other and the concept of participation, as well as to start the group
discussing dog-related issues: Secondary
■■ The group is asked to draw a map of their local area, including any prominent local Sources
features. They can use whatever materials they choose – encourage them to be as
creative as possible.
• Creating their own map starts the group off in a participatory manner. An
Clinic
alternative that may be suitable when using stakeholder groups such as medics,
records
vets, local authorities who cover a very large geographical area, is to provide a
printed map of the extent of the intervention area and ask the group to annotate
prominent features relevant to the stakeholder group (e.g. the location of Behavioural
hospitals, vet surgeries or government service points). Observation
■■ Once the basic map is created ask the group to indicate where dogs are seen in
public places and estimate the number of roaming dogs at that location. Highlight
locations where there are both positive dog benefits and problems. The facilitator Vaccination
notes down any mention of ‘dog nuisances’ and ‘dog benefits’ on cards for later. Coverage
■■ As the discussions come to a close, the facilitator shows the group the list of dog
nuisances and dog benefits written on cards and ask if they want to change or add
to what has been written as ‘dog issues’ in their location.
■■ Output – an estimate of roaming dog population size for the local area plus a
list of dog issues for the location, including both positive and negative issues
with dogs.

Exercise 2a - Dog versus non-dog nuisance ranking

This exercise builds on the dog mapping exercise and is used to compare dog nuisances
with other public nuisances that are not related to dogs:
■■ Once dog issues have been covered on the map, ask the group to highlight the
location of any non-dog nuisances.
Questionnaire
• Non-dog nuisances are defined as attributable to the actions of neighbours surveys
(those outside the home) and something that would be considered by
almost all people as at least an annoyance, while at the same time not being
unquestionably a serious crime. For example, traffic jams or garbage/litter might Participatory
be an annoyance but not a crime. Major offences, such as burglary, are omitted Research
as these are more than a nuisance. Natural nuisances such as mosquitoes and
flooding are also not included as these are not associated with the actions of
neighbours. This definition is based on that provided by Fielding (2008).
Street
• The facilitator notes down any non-dog nuisances mentioned on cards. Surveys
■■ As the discussion come to a close, the facilitator shows the group the list of non-
dog nuisances written on cards and ask if they want to change or add to what has
been written for their location. Secondary
Sources
■■ Lay the dog and non-dog nuisances in a vertical line, with 5 columns to the right.
Creating a matrix:
• Column 1 is the severity of the nuisance. Ask the group to score the severity of Clinic
each nuisance out of a total of 10 – with 10 being the most serious nuisance. records
-- Note that if the group is not used to working with abstract numerical
proportions, proportional piling can be used instead, where the group is given
a large number of small items (e.g. 50 or 100 beans or stones) and asked to Behavioural
split the beans into piles, with the size of the pile representative of the severity Observation
of each nuisance.
• Column 2 is frequency. Ask the group to score the frequency of each nuisance
out of 10 – with 10 being several times a day and 1 being once in your life time. Vaccination
Coverage
-- It may be useful to give people an example of common activities that vary in
frequency; e.g. 10 = how often I talk to the children in a day; 5 = how often I
collect water in a week; 1 = how often I get married in a lifetime.
• Column 3 is prevalence. Ask the group to score the experience of this nuisance
for the group members out of 10 – with 10 being everybody in the group plus
their neighbours have experienced this nuisance and 1 being no one in the group
has experienced this nuisance but they have heard of it happening to others
within their community.
• Column 4 is total score for that nuisance. Total up all the scores from column
1 + 2 + 3.
• Column 5 is rank. The nuisance with the highest score gets rank 1 and lowest
score gets the lowest rank. Allow for tied ranks (nuisances with the same total
score)

• Allow the group to review the total scores and ranks and make any amendments
that they think are necessary.
• Output – a comparison between the number and rank of dog nuisances

compared to non-dog nuisances.
• Repeat this exercise with different groups and over time to see if the number
Questionnaire
and rank changes over time. If perceptions of dogs are improving some dog
surveys
nuisances will stop being mentioned at all and others will fall in rank against non-
dog nuisances, assuming non-dog nuisances do not change significantly within
the same time period.
Participatory
Research
Exercise 2b – Dog issues trend analysis
This exercise builds on the dog mapping exercise but also asks people to consider
whether both dog problems and benefits have changed over time: Street
Surveys
■■ Lay the dog problems and dog benefits in a vertical line with 3 columns to the right,
creating a matrix:
• Column 1 is the importance of this benefit or severity of the problem. Ask the Secondary
group to score the importance and severity of the issue out of a total of 10 – with Sources
10 being the most important. Note this is all about how the group feels about
this benefit or problem today, how this has changed over time is dealt with later.
(Consider the option of proportional piling described for exercise 2a).
Clinic
• Column 2 is frequency. Ask the group to score the frequency of each issue out records
of 10 - with 10 being several times a day and 1 being once in your life time.
(Consider the option of presenting common activities as described for exercise
2a)
Behavioural
• Column 3 is prevalence. Ask the group to score their current experience of this Observation
issue for the group members out of 10 - with 10 being everybody in the group
plus their neighbours are experiencing this issue and 1 being no one in the group
is experiencing this issue but they have heard of it happening to others within Vaccination
their community. Coverage
(You may want to split the exercise at this point, with a break or coming back for a
second session)
■■ Now ask the group to think about an important event that they all remember
occurring 1 or 2 years previously. This can be a public event like a sporting event
that happened in the country, a political event like an election or a natural event like
a particularly violent storm or flood, anything that places the same point in time in
everyone’s memory. This becomes the title for column 4.
■■ Dependent on the timeframe required for impact assessment, a second event

further back in time can be selected. This becomes the title for an optional column
5.
• Preferably the launch date of the intervention or intervention related events are
not used as time markers to avoid any biasing of responses in the next activity.
But the events selected should be clearly within or before the intervention

timeframe, i.e. a time that represents a pre-intervention baseline or a time that

could represent change following a known period of intervention.
• Now ask the group to state for each issue whether this issue as increased (£),
stayed the same (=) or decreased (£) since the event(s). Indicating the direction
of change with £, £ or = in column 4 and 5.
Questionnaire
Output – comparison between dog problems and dog benefits and how they
■■
surveys
have changed over time.
■■ This exercise can be conducted at several points in time after an intervention has
commenced and with different groups to see if perceptions are changing differently. Participatory
Research
Street
Surveys
Secondary
Sources
Clinic
records
Behavioural
Observation
Vaccination
Coverage

Street Surveys
International
Companion Animal
Street surveys are an efficient method of collecting data relating to a range of indicators.
Their efficiency makes them particularly suitable for monitoring and evaluation because Questionnaire
they can be repeated several times over the course of an intervention. The data collected is surveys
only related to roaming dogs seen on public property (i.e. not those behind fences or within
compounds) and the ownership status of these dogs may not be clear (although a good
state of welfare and wearing collars are potential signs of ownership): these roaming dogs
Participatory
may be owned roaming, community owned dogs or entirely unowned dogs either born
Research
unowned or abandoned/lost by their former owners. This means that confined dogs will not
be accessed through street surveys; this could be a disadvantage but as confined dogs are
commonly not a priority target this may be a minor concern for some interventions.
Street
Surveys
Relevant impacts
Secondary
Street surveys are most commonly used to measure impacts related to roaming dog density, Sources
demography and welfare. Street surveys may also be useful for assessing the coverage
of an intervention where the intervention has involved visible marking of dogs, for example
collars or paint spray during vaccination campaigns or ear notches during sterilisation.
These marks are noted and recorded for every dog observed to assess the proportion of Clinic
the population that has been reached by the intervention. See the section ‘Street surveys records
and questionnaires for measuring vaccination coverage’ for the specific use of this method
following vaccination.
Behavioural
Observation
Method
Vaccination
The method described here involves counting dogs on a route that runs along streets. It is
Coverage
similar in principle to the ‘strip transect’ method of wildlife surveys, in that observers move
relatively quickly along an extended line to avoid double counting and to collect information
on how animal density varies across the area. The difference to strip transects is that there
is no intention to extrapolate the counts to an estimate of abundance rather the concept is
to repeat these routes in a consistent manner
and compare the counts over time (see Annex
D on how estimates of abundance can be
calculated if necessary). The method thus
depends on protocol consistency; using one
or more standard routes at a consistent time
of day and year and following a standard
counting protocol, for example using a
consistent average speed and mode of
transport which will affect the ‘search effort’,
i.e. the chances you will see a roaming dog
through your observational skills. Ideal modes
of transport include motorbikes carrying a
69
Methods of Measurement: Street Surveys
2-person survey team, cars and bicycles; walking is generally too slow to allow surveying to
be completed within a reasonable timeframe and risks double counting dogs as the survey
team will be moving slower than some dogs roam.
Protocol
■■ The survey team consists of 2-3 people (a driver, navigator and observer, with navigator Questionnaire
and observer combined as one role when using a 2-person team, and no driver if using surveys
bicycles). However, all members of the survey team are responsible for spotting dogs.
■■ The team follows the predetermined route (see route selection section that follows)
travelling at maximum 15km an hour, slowing or stopping to record every dog seen
Participatory
before moving on as quickly as possible; keeping up a pace is important to avoid double
Research
counting and cover the route efficiently.
■■ Every roaming dog seen on the route is counted. Dogs confined within properties,
walking on lead or ‘close to heel’ are excluded from the survey. In some cases a dog will
be inside a fenced area but the gate will be open and the survey team needs to agree a Street
consistent rule about how these dogs will be treated. Surveys
■■ Every dog is allocated to one of 5 categories: male, female, lactating female, puppy
(under 4 months), unknown adult. This is extended to up to 10 categories when the
intervention includes visible marking, as each of the 5 categories can be marked or Secondary
unmarked. This falls to only 8 categories when the marks are ear notches applied during Sources
sterilisation as lactating females and puppies will not have been through the intervention
yet and so can only be unmarked.
■■ Every dog is also assessed for welfare status and potentially also whether they are Clinic
wearing a collar (in some countries where tethering is common, dogs on tethers but not records
confined behind a fence can be included in the survey as these dog are accessible to
roaming dogs and so are a relevant part of population in terms of breeding and disease
transmission).
Behavioural
■■ In some cases additional attributes (sex, intervention marks and welfare status) will not Observation
be possible to judge accurately due to the dog moving out of sight or lying down. The
observers must not guess these attributes but either categorise the dog as unknown or
leave welfare status unrecorded (i.e. unobserved) for that dog. Data on these attributes
Vaccination
will be obtained from the sample of dogs that can be reliably assessed.
Coverage
Route selection
Standard routes for the survey can be designed within existing administrative boundaries,
such as wards or municipalities, or routes can be drawn randomly across the entire area:
Routes within administrative boundaries (‘ward’ is used here as a generic term for
administrative boundary): One or more routes can be drawn in every ward, however
if the area is too large to cover, a sample of wards can be chosen. Routes should be
approximately 25 - 30km (15 - 18 miles) long to allow the survey to be completed within 2
hours. They should encompass different road types, excluding only those roads where dogs
are very unlikely to be found and surveying would be difficult (e.g. motorways), and also
including different environment types such as dense urban areas versus open rural settings
falling within the ward. These road and environment types should be included in the route in
approximately the same proportion as they appear in the ward. The drawing of routes can
be done in Google Maps (https://maps.google.co.uk), using the ‘draw a line along roads’
function which is found in ‘My Places’ (using the option ‘Or create with classic My Maps’).

Methods of Measurement: Street Surveys
Routes along random sample of streets across intervention area: If using wards is not
possible or not preferred, routes can instead be drawn along a random sample of streets
across the entire area of interest. This sample of streets needs to be selected without bias
to where dogs are located. One option for selecting the sample is to create a zig-zag track
across the survey area. This can be done using Google Maps ‘My Places’ (using the option
‘Or create with classic My Maps’); a zig-zag line can be drawn across the whole area using
the ‘Draw a line’ function; then the ‘Draw a line along roads’ function can be used to create Questionnaire
a route that fits as closely as possible to the original zig-zag line. This should create an surveys
unbiased route that covers both small and large roads.
Participatory
Research
Available tools for street surveys
Mobile phone applications (apps) are particularly suitable for street surveys because they can
easily be carried by the surveyors, can record Global Positioning System (GPS) information
Street
and can eliminate the need for later transcribing of data.
Surveys
If the Google account used for designing the routes on the computer is synchronized with a
smart phone, that phone’s Google Maps application can be used to display the route on the
Secondary
phone’s screen. The phone can be used first to navigate to the start of the route and then to
Sources
follow it, by taking the turns required to keep the Maps cursor moving along the displayed
route.
A second mobile phone application called ‘OSMtracker’ (free to download) can be used as
Clinic
an event recorder to record the type and welfare state of every dog seen – note this app
records
can only be used on Android smart phones. OSMtracker also records the time and GPS
coordinates of each event so that following download of the data from the phone to the
PC the spatial distribution of the dog types can be displayed on a map.
Behavioural
Observation
Vaccination
Coverage

Secondary Sources of Information
International
Companion Animal
Secondary sources of information include official or government-derived information.

These measures often do not require additional effort on the part of the DPM intervention Questionnaire
managers, other than the effort required to collate the data from the source. They also surveys
provide relatively independent data and means of validating the impacts of the intervention.
Ideally access to these data is achieved through a working collaboration between the
intervention managers and the government department responsible for collecting and
Participatory
reporting these data, which can be established at the planning and outset of the intervention.
Research
Using a freedom of information act to access data may be possible in some countries where
such legislation exists; however, a working relationship is preferable to ensure long-term
access as needed (e.g. segregated by species, geographical locations etc.).
Street
Relevant impacts Surveys
Impact indicators that can be measured by secondary sources of information include several
public health indicators such as human cases of rabies, surgical interventions for cystic
echinococcosis, dog bites and dog rabies cases, and public perception indicators relating Secondary
to public complaints about dogs to local authorities. Sources
Surveillance effort
Secondary data is particularly sensitive to changes in surveillance effort and hence close Clinic
communication with the government department will ensure any changes are known and records
incorporated into analyses. Changes in effort or effectiveness can occur at any stage of the
surveillance process from recognition of disease in the field by a member of the public/dog
owner, attendance by a veterinarian or medical practitioner in the case of a bite or suspected
human disease, diagnosis using clinical or laboratory tests, reporting of cases within the Behavioural
health centre/practice, reporting to the local authorities, through to reporting to central Observation
authorities. Ideally intervention managers work with the relevant government departments
to improve surveillance prior to starting an intervention; this may include setting up local
key informants (people in a naturally good position to collect specific data, such as school Vaccination
teachers collecting data about dog bites experienced by children, veterinary field officers Coverage
collecting data about animal rabies cases and pharmacists collecting data about dogs bites
receiving PEP), improving diagnosis by veterinary and medical professionals and improving
reporting through data management systems integrated with phone apps. Any resulting
changes in surveillance must be accounted for if using historical data as a baseline.
Often making a disease reportable (professionals including laboratories must report positive
test results to central authorities for tracking of disease trends) or notifiable (any person in
possession of an animal suspected of having a notifiable diseases must report the case to
relevant animal health authorities who will investigate the case) improves surveillance and
control. In addition, introducing the capacity for laboratory testing of suspect cases is often
promoted (e.g. by WHO) rather than reliance on diagnosis using only clinical signs. This
represents the ideal situation but often not the reality in many countries where diseases
such as rabies are endemic. Further, making a disease reportable and requiring laboratory
confirmation does not necessarily lead to good surveillance, although it should encourage
investment in systems that make reporting possible. So although the goals of making a
disease reportable/notifiable and providing accessible laboratory testing are valid, these
may not be sufficient for ensuring consistent and good surveillance.
72
Methods of Measurement: Secondary Sources of Information
Frequency versus incidence

Data relating to indicators such as cases of human or dog disease can be presented as a
frequency (i.e. number of cases per month or year) or as an incidence per 100,000 people
or dogs that were at risk of the disease during that time. Using incidence accounts for
changes in human or dog populations over time, but population estimates may be difficult
to establish reliably in some locations and so using the frequency/number of cases without Questionnaire
a denominator may be necessary. Situations when incidence is most suitable include surveys
comparisons between locations, such as treatment and control areas, or when assessing
changes over a long period of time (> 10 years) when population changes are expected to
be significant and population estimates are available for the entire period. Good practice
Participatory
when incidence appears suitable would be to present both the frequency and incidence
Research
for the indicator data and to be very clear what denominator was used and how this
denominator of human or dog population size was estimated. It may be in the future that
these population estimates are revised and hence the indicator data can be similarly revised
if original denominators were clearly stated. Street
Surveys
Geographical resolution
Access to secondary sources of information can occur at several points along the
surveillance process: direct from a health centre, hospital or veterinary station; from city, Secondary
regional or state authorities; or from central authorities. The point at which data should be Sources
gathered will depend on the geographical extent of the intervention: if it is focused on one
city, accessing data from a local hospital or veterinary station may be the most appropriate
and avoids any errors relating to communicating data up through a chain of authorities. It
also depends on the systems of reporting up from the healthcare provider to the authorities Clinic
at the central level. Where these systems are working well and the central authorities are records
transparent this may be a very efficient way of accessing data from the intervention area
and also other non-intervention areas for comparison. Ideally collect a sample of data
from both the earliest stage in the surveillance process and central authorities to assess Behavioural
whether the data matches and how any differences might have occurred. This may also Observation
reveal a difference in the level of detail for each data point, for example the exact location
of a case may be provided at the health centre/veterinary station level but reduced to only
a geographical region once data reaches central authorities. This summarising of data as
Vaccination
it moves up the chain may also indicate the best level at which to access data to inform
Coverage
impact assessment.
Dogs may be accessed by an intervention clinic one or more times. This may involve
treatment for injuries, surgical neutering or may be as simple as a single vaccination, dipping
or deworming at a field station. Whatever the nature of the intervention, this provides an
ideal opportunity to collect relatively detailed information about individual dogs. Note that the
suggested data to be collected later in this section may need to be amended dependent on
what infections/diseases are most relevant locally and what the clinic can provide in terms of
treatment.
In many locations, local vets will also be providing preventative and curative treatments.
If these vets are incorporated into the intervention the basic clinic data from these vets would
ideally also need to be collected as this can measure change in relevant indicators.
Note there are issues with dog owner/client confidentiality which may require this data
to be anonymised and/or the clients to give their permission for the data to be used.

Clinic Records
International
Companion Animal
Relevant impacts Questionnaire

Data collected through clinic records can be used to measure indicators relating to dog surveys
welfare, dog population stability and care provided to dogs, specifically reflecting owner
engagement with the intervention over time. The specific indicators include:
■■ Indicator of dog health/welfare Participatory

Research
• Reduction in the % of dog with BCS 1 (or BCS 1+2)
• Reduction in the % of dogs with a visible skin condition (can also include changes
in types of skin condition if diagnosis was possible) Street
Surveys
• Reduction in specific diseases targeted by the intervention
• Reduction in % of dogs with a TVT seen either at clinic exam or during surgery
• Female:male ratio gets closer to 1:1 – female dogs are no longer neglected/killed
Secondary
as their unwanted litters can now be prevented. Note that if the clinic catches or
Sources
invites one sex over the other this will bias the sample and the female:male ratio
will not be representative of the wider population
Clinic
■■ Indicator of population stability
records
• Adult annual survival
• Increase in the proportion of old dogs in the population (old /

puppy+juvenile+adult+old); this indicator is only possible to measure with clinic Behavioural
records where the treatment is regular (e.g. vaccination or deworming) and not a Observation
one-off, i.e. sterilisation. Where sterilisation is being used the population accessed
by the clinic may start to skew towards the younger age groups as the majority of
older dogs have already been sterilised Vaccination
Coverage
• Decrease in the % of lactating females or pregnant females and changes in
seasonal patterns of breeding.
• Indicators of the quality and value of the service and also responsible dog
ownership by owners
• Increase in the proportion of dogs that are re-treatments
• Increase in the number of dogs that are brought by owners or carers rather than
caught by staff
• Increase in the average donation/payment per dog or reduction in the size of the
difference between intervention veterinary costs and donations/payments from
owners (this takes into account changes in the costs of the intervention)
• Reduction in the number of unwanted but otherwise healthy dogs euthanised
74
Methods of Measurement: Clinic Records
Biased sample
It should be noted that the dogs coming through an intervention are very likely a biased
sample, i.e. they are ‘special’ in some way. They have come to the intervention for a
reason, either to do with their owner, carer or in the case of unowned dogs, dog catchers.
This means these dogs may, or may not, be a good representation of all dogs in the
wider area and so conclusions about general dog population welfare must be drawn with Questionnaire
caution. If the percentage of dogs passing through the intervention is high (e.g. 70% of surveys
the dog population is brought/caught for rabies vaccination), this is a very large sample
and changes in indicators measured from this population of dogs may be sufficient
for monitoring intervention impact without any attempt to extrapolate to the wider dog
Participatory
population. However, any changes to protocol that could impact on which dogs are
Research
brought by owners/carers or caught by intervention staff must be noted down and made
available during later analysis and interpretation of the data. For example, changes to the
messaging about services provided by the intervention may target different type of dogs;
for example advertising early age neutering may skew the dogs brought to the intervention Street
towards younger dogs, or advertising Surveys
free or low-cost parasite treatment
may increase the proportion of dogs
with skin condition brought to the
intervention.
Secondary
Sources
Note that for veterinary interventions
much more information may be
collected on individual dogs than
Clinic
will be necessary to ensure good
records
quality of care. The following is limited
to information that is suitable for
monitoring and evaluating population
level changes and hence is not an Behavioural
exhaustive list: Observation
Vaccination
Coverage

Data to collect for each dog

■■ If the dog is brought by an owner
• Dog identification, at least name but microchip or tattoo if present and digital
photograph if possible (note tattoo character length can be minimised by using
an alphanumeric code)
Questionnaire
• Owner name and address, including postcode, borough or ward to allow for surveys
aggregation of data to geographical areas
• Amount paid/donated by owner
■■ If dog is brought by a carer (they do not own the dog, but they have accepted Participatory
responsibility for bringing them for veterinary care) Research
• Exact address of where they collected the dog
• Amount paid/donated by owner
Street
If dog is caught
■■
Surveys
• Exact address of capture including GPS coordinates if available (e.g. by using a
mobile smart phone)
■■ Age of dog. Record this as accurately as possible, but at least whether the dog falls Secondary
into the 4 categories of; Sources
• under 4 months (puppy)
• 4 months-11 months (juvenile)
Clinic
• 1-5years (adult)
records
• over 5 years (old dog)
■■ Gender
■■ Reproductive status. Castrated/spayed, unsterilised, pregnant, lactating. Behavioural
Observation
■■ Body Condition Score. 5-point score (see Annex A)
■■ Skin condition. Presence or absence of a visible skin condition. Diagnosis of cause
if known. Vaccination
■■ Any concurrent illness, infection or injury. Coverage
■■ Transmissible venereal tumour (TVT) at clinic exam. Presence or absence of a TVT
■■ TVT at spay/castrate. Presence or absence of a TVT (it’s important to separate TVTs
that were discovered during surgery from those that were visible at clinical exam, as
the prevalence will differ)
■■ Treatment provided
• Is this the first visit for the dog or repeat visit?
• What treatment was given (e.g. sterilisation, vaccination, deworming, dipping,
wound treatment, etc)?
• If euthanasia, categorise by reason for euthanasia: physically unhealthy,
behaviour problem (as perceived by owner to be unmanageable) or unwanted by
owner for other reason (i.e. the dog is physically and behaviourally healthy)

Tools available for data recording

Clinic records can quickly become very numerous and therefore difficult to handle, hence
a data collection tool is particularly useful. Microsoft Access databases can be developed
for specific clinics, including drop down lists to reduce entry error on common terms (e.g.
municipalities in addresses, treatment type, gender, reproductive status). Ideally data is Questionnaire
entered straight into the database at the clinic; this is supported by having the database surveys
hosted on a laptop or on an online system to allow access by several devices, including
mobile phones.
There is an Access database available to download for free from Participatory

www.conservationresearch.org.uk/Home/ICAMCoalition.html. This is designed for Research
recording clinic information from interventions that access unowned dogs for sterilisation
and vaccination and was developed to suit the Animal Birth Control (ABC) approach of
DPM used in India and includes the data requirements outlined in ‘The Standard Operating Street
Procedures for implementation of the Animal Birth Control (ABC) Programme’ published by Surveys
the Animal Welfare Board of India (AWBI, 2009).
Survival estimates of roaming dogs utilising clinic data

Secondary
Interventions can provide an opportunity to apply a permanent individual mark to a dog,
Sources
such as a microchip or tattoo; most commonly this will be whilst the dog is anesthetised for
sterilisation. After these dogs have been returned to their owners (or the point of handling
in the case of unowned dogs) there can be opportunities for a sample of these dogs to
be accessed again and their individual marks read. Example opportunities include during Clinic
booster rabies vaccination; re-catching for treatment or accidentally because a mark has records
been missed; or re-catching a sample of dogs just to read their individual marks. These
dogs will provide a sample of the population that are known to have survived for a specific
length of time, from the date the marked dog was released to the date of reading. This is
Behavioural
only possible if accurate clinic records have been maintained.
Observation
A program can be downloaded and installed to estimate adult
survival from a sample of re-accessed individually-marked dogs from
www.conservationresearch.org.uk/Home/ICAMCoalition.html. This program also allows Vaccination
for the inclusion of dogs that passed through the intervention in a period of time before Coverage
individual marking was introduced; these dogs would carry a mark (such as an ear-notch) to
show they had been intervened but no individual mark (such as a tattoo). The proportion of
these dogs as compared to individually-marked dogs can also be used in the estimate so
long as the date of commencing individual marking is known. This was included because
interventions commonly do not individually mark successfully from intervention outset. Adult
survival is assumed to be independent of age but a stable age structure assumption is not
required. Instead we need to know the date of every release of a marked dog since the start
of the intervention and whether it was marked as a puppy. We also need to know the date
on which individual marking commenced and the date of re-reading the individual mark.
In addition to adult annual survival we then need two further parameters, the probability
that a surviving released dog is included in the sample (the sampling effort) and a possibly
reduced survival for those dogs neutered as puppies – in the computer program referred to
previously, these have default start values and precise values are calculated by the program
as described later. Those parameters then give the probability of survival and inclusion in

the sample for each individually marked dog and also the expected number of dogs in the
sample with an intervention mark (such as an ear notch) but without an individual mark.
That number has a Poisson distribution about its expectation whereas the appearance or
non-appearance of each individually marked dog in the sample become the outcomes of a
sequence of Bernoulli trials and hence combine to form a likelihood that can be maximised
with respect to the three parameters. The method has the potential to include covariates of
the survival probability, certainly the sex of the released dog if both males and females are Questionnaire
included in the intervention but perhaps also factors such as the type of intervention used, surveys
e.g. the method of neutering operation used.
Participatory
Research
Street
Surveys
Secondary
Sources
Clinic
records
Behavioural
Observation
Vaccination
Coverage

Behavioural Observation Method
International
Companion Animal
The behaviour of animals can be a very visible reflection of their underlying welfare state
and can present welfare challenges in its own right, for example in the case of injuries from Questionnaire
fighting. Similarly the behaviour of people towards dogs can have a very real impact on surveys
the dogs’ welfare state, both positively and negatively. By repeatedly and unobtrusively
observing dogs and people together in their ‘natural’ habitat, we can record the nature of
their interactions and how this changes over time.
Participatory
Research
Relevant impacts
Data collected through behavioural observations can be used to measure indicators relating
to dog welfare, dog population stability and public perception/tolerance of dogs:
Street
■■ Indicators of dog welfare Surveys
• The percentage of amicable interaction out of the total of all dog-dog interactions
• The percentage of aggressive interactions out of the total of all dog-dog Secondary
interactions Sources
■■ Indicators of dog population stability
• The percentage of mating interactions out of the total of all dog-dog interactions
Clinic
■■ Indicators of dog welfare and human tolerance records
• Percentage of positive human behaviours out of total of all recorded human-dog

interactions.
Behavioural
• Percentage of relaxed human-dog interactions out of total of all recorded human- Observation
dog interactions.
• Percentage of negative human behaviours out of total of all recorded human-dog

Vaccination
interactions.
Coverage
Assumptions
Time available for behavioural observation will be very limited. Ideally this will require less
than 3 days for every monitoring event (with monitoring events happening once or twice per
year). This means that video recording with later transcribing will not be possible, as this is
labour and time intensive. All data will need to be recorded in real time. A phone or tablet
event recording app would be very beneficial for this method.
Sites
At least 12 sites from within the area serviced by the dog population management
intervention are selected for their high densities of dogs and/or people, leading to a greater
chance of observing interactions between dogs, and between dogs and people. Sites
should be small enough that the entire area can be seen through a 180° scan without the
observer needing to move, and they should also be observed at peak ‘interaction time’;
when interactions are likely to be most frequent. Examples include school gates at drop-
off or picking-up time, bus stations at rush hour, squares/parks at lunchtime. As the same
79
Methods of Measurement: Behavioural Observation Method
sites observed at the same time of day should be used over the monitoring period to reduce
confounding variables, it would be wise to review a large number of sites and then select the
best on the basis of several criteria.
These criteria include:

A. High frequency of interactions between dogs or between dogs and people. Some
Questionnaire
sites may provide both, but it is likely that most sites will be better either for dog-dog
surveys
interactions or people-dog interactions. The sites should be selected to provide a
balance of both.
B. Ease of observation. This will include accessibility of the site (do you need permissions Participatory
to be there?), can you travel there easily and can you observe the site unobtrusively Research
without disturbing either dogs or people?
C. In some cases there will be sites that are locations for a high frequency of complaints.
Changes in dog-dog or dog-person behaviour at these sites may be politically Street
important and so these sites will be worth including. Surveys
D. If the number of days available to complete the observation is short, sites can be
selected with different peak interaction times to allow a number of sites to be visited in
one day, shortening the total number of days needed to complete the observations. Secondary
Sources
E. Sites should be selected that are spaced far enough apart to limit the chances of
observing the same dogs at different sites. One study in Chile found roaming male
dogs had an average home range of 22.4 hectares (Garde et al., 2012); this would
require sites to be spaced approximately 450-500 metres apart to limit the chance Clinic
of observing the same dogs. One option would be to conduct a pilot behavioural records
observation study where all dogs are photographed to assess the overlap between
sites. However, even with this level of planning these sites cannot be assumed to be
completely independent.
Behavioural
Observation
Protocol
Vaccination
Coverage
Pre-observation data collection
On arrival at the site the observer establishes the observation point (a safe position out of
peoples’ way but allowing maximum view of dogs and people; the same point should be
used for every observation in future) and then records the GPS position of the observation
point, characteristics of the site (e.g. it’s a bus stop at rush hour), time, weather, number
of dogs and number of people within view. If their arrival at the site proved interesting to
either dogs or people, the observer may also need to wait
quietly for a few minutes until their presence has become
apparently irrelevant to both dogs and people.
Observation
Behaviour sampling with continuous recording for 30
minutes of observation. If there are no dogs when the
observer arrives or if dogs leave the site resulting in the
total number of adult dogs falling below 2, the observation

should pause and restart when at least 2 adult dogs appear back on site. The observation is
considered complete once 30 minutes of observation have been recorded or if the observer
has been present at the site for 1 hour, whichever occurs first. Only the behaviour of adult
dogs should be considered, with a cut-off age of 4 months (i.e. under 4 months of age is
considered a puppy and not included). The observer should aim to keep as much of the
site in view as possible and if necessary keep their eyes moving around the site to maximise
coverage. Questionnaire
surveys
Record the outcome of every dog-dog interaction:
■■ Every interaction that is observed within the 30 minutes of active observation is
recorded. Participatory
Research
■■ An interaction is defined as when the dogs are within 3 dog lengths of each other
and are communicating with each other with body language and/or vocalisations.
This need not include looking directly at each other as direct eye contact can be
perceived as threatening, hence eyes may be averted during the interaction. Street
Surveys
■■ Record the final outcome of each interaction as either aggressive, amiable, mating
or neutral (see description of behaviours in each category that follows in table 1).
Individual behaviours do not need to be recorded, just the final outcome of each
interaction. Interactions may start with dogs appearing to behave in one way, but Secondary
they can evolve through the course of the interaction, involving communication Sources
through body language or vocalisations, into a final outcome concluding with the
dogs parting (more than 3 dog lengths apart and communication stops).
Clinic
■■ If the dogs part and then rejoin for another interaction, that subsequent interaction
records
can be scored again.
Record the frequency of the following ‘extremes’ of human-dog interactions:

■■ Positive human behaviours towards dog: Behavioural
Observation
• Person feeds a dog
• Person pets a dog

Vaccination
• Person calls a dog to come to them either using name or other friendly sounds Coverage
(e.g. kissing noise, clicks, slapping thighs)
■■ Negative human behaviour towards dog:
• Person hits or kicks a dog with any part of their body or an inanimate object like a
stick
• Person throws something at a dog to scare/hurt it (i.e. not throwing food to them),
this includes ‘sham’ throws; person pretends they are throwing something at a
dog to scare them
• Person shouts or claps hands to scare a dog

Table 1 – Categorisation of dog behaviours during interactions between dogs, categorised

into amicable, neutral, mating and aggressive (amended from ethogram used in Garde et
al. (2012); an ethogram is catalogue or inventory of behaviours or actions exhibited by an
animal and used in animal behaviour studies).
Behaviour category Descriptions of behaviours

Licks, paws, nudges with nose, or grooming between dog, Questionnaire
often with tails wagging. surveys
Amicable interaction Play behavior including charges with bouncing gait, play faces
(mouth relaxed, slightly open, teeth covered), wrestles and play
chases.
Participatory
Research
Approach and retreat, often including sniffing, limited body
language, non-demonstrative. Not aggressive but also not
Neutral interaction friendly. Includes non-reproduction related mounting, unless
this is clearly part of play or ends in aggression.
Street
Surveys
Copulation (not only mounting where one dog stands with
Mating interaction forepaws on another), usually ending in a ‘tie’
Growling, teeth visible, barking, biting, fighting. Secondary

Aggressive interaction
Sources
One dog flees with tail tucked to avoid other dog,
cowers or rolls over.
See Annex B for example recording sheet Clinic

records
Behavioural
Observation
Vaccination
Coverage

Street surveys and Questionnaires for

Measuring Vaccination Coverage International
Companion Animal
Vaccinating dogs is the most effective and humane approach to controlling rabies by
eliminating disease from the dog reservoir. This section describes how vaccination coverage Questionnaire
(the proportion or percentage of the dog population that has been vaccinated) can be surveys
measured following a vaccination campaign. As rabies vaccines that meet international
standards are very effective, vaccination coverage immediately after a vaccination campaign
has been completed can be equated to the level of herd immunity (‘herd immunity’ occurs
Participatory
when a sufficient proportion of the population has acquired immunity through vaccination,
Research
effectively creating a barrier to disease spread, such that the unvaccinated proportion of the
population is also protected and the disease outbreak dies out rather than spreads).
Relevant impacts Street

Surveys
Vaccination coverage is not an indicator of impact but an indicator of effort (note that this
may also be defined as intervention ‘effectiveness’ as it is a combination of the intervention
effort and the community response to the intervention). However, when assessing the impact
of an intervention on rabies impact indicators such as dog rabies cases, suspect bites and Secondary
human rabies cases, a measure of vaccination coverage is necessary for attribution of the Sources
intervention.
Marking
Clinic
To measure the proportion of the dog population that is vaccinated in an intervention, it records
is necessary to mark every dog that is vaccinated, for example with a collar or non-toxic
paint spray/stick (such as livestock paint/stock markers or vegetable-based dyes). Paint is
preferable for young dogs to avoid them being harmed by collars as they grow and paint is
much harder for people to remove than collars. Marking is most relevant for street surveys Behavioural
but also useful when using questionnaires to validate coverage (i.e. not only asking the Observation
owner if the dog is vaccinated but looking for the mark on the dog itself for confirmation),
and when conducting campaigns to avoid vaccinating dogs twice or missing dogs. It
is necessary to accurately record the number of dogs vaccinated and marked at each Vaccination
vaccination station/household location (some dogs will not be marked due to the owner’s Coverage
preference or dog behaviour, and the number of vaccinated but unmarked dogs should also
be recorded).
Assessment of the proportion of dogs marked provides the data necessary for evaluating
vaccination coverage, which will be the same as the proportion marked (unless a significant
proportion of vaccinated dogs were not marked). Assessment should take place as soon
as possible after the vaccination campaign to minimize mark loss, ideally no longer than 3
days. Marks can be lost very quickly, for example in Tanzania 13% of vaccinated dogs lost
their collars within one day of vaccination and 6% lost their spray paint mark (Cleaveland,
unpublished data). Owners should be encouraged to retain the mark on their dog for as long
as possible. During some interventions unowned dogs will also be sterilized at the time of
vaccination and these dogs can be ear-notched whilst anaesthetised, providing a permanent
mark with no risk of loss.
83
Street Surveys and Questionnaires for Measuring Vaccination Coverage
Sample size
It is not necessary to observe the mark status of every dog: a sample of the population can
be observed to get an estimate of coverage of known reliability.
The sample size required can be calculated using the following equation:
Questionnaire
N
n= surveys
⎜
(
⎛ e2 N − 1 )+ 1
⎞
⎟
(
⎜⎝ 1.962 p 1− p )
⎟⎠
Participatory
Research
Where,
N = the estimate of the total dog population in the area being surveyed
e = the desired level of precision, selected to be 10% in the example below, expressed as Street
a proportion when used in this equation, so 0.1 Surveys
p = the target vaccination coverage, usually set at 70%, expressed as a proportion when
used in this equation, so 0.7
Secondary
n = the sample size required to be 95% confident that when 70% vaccination coverage Sources
is estimated the true coverage is within 10% of the target, i.e. 63%-77% (using e = 0.1
and p = 0.7)
Note that this does require the total dog population to be estimated, although ‘guestimates’ Clinic
(i.e. estimates known to include some level of error) will be the reality in many cases. records
Street surveys
When using street surveys to assess vaccination coverage, the mark status of every dog
Behavioural
seen should be recorded, whether roaming or confined. When the required sample size (n) is
Observation
30% or more of the total population (N), which is the case until N is approximately 190 dogs,
it will be necessary to survey all streets to observe a sufficient sample size, based on the
experience that on average between 30-50% of the total roaming dog population is seen on
any one survey. Once n is below 30% of N the survey can be done along a sample of streets
Vaccination
to save time. This sample of streets needs to be selected without bias to where vaccinated
Coverage
dogs are located (e.g. not only the streets surrounding the vaccination station location).
One option for selecting the sample is to create

a zig-zag track across the survey area. This
can be done in Google Maps using ‘My Places’
and ‘Create with Classic Maps’. A zig-zag line
can be drawn across the whole area using the
‘Draw a line’ function, then the ‘Draw a line
along roads’ function can be used to create
a track along the roads that fits as closely as
possible to the original zig-zag line. This should
create an unbiased track that covers both small
and large roads.

If Google Maps is not available the same approach could be achieved using a compass;
starting at one side of the survey area and heading on a set course (e.g. NE) until the
opposite edge of the survey area is reached, taking whichever road seems to fit closest to
the course, then returning across the survey area this time taking the course at 90° to the
original setting (e.g. NW), repeating these transects until the survey area has been covered.
If an insufficient sample size is achieved the same approach is repeated starting at a different
point on the periphery of the survey area. Questionnaire
surveys
A similar approach has been trialled in semi-rural and rural areas of Tanzania (Hampson
and Cleaveland, pers comm); instead of starting at the periphery of the survey area the
observers are asked to start three transects at each of several common locations; a church,
Participatory
school (often at the periphery of the village) and vaccination station (often at the centre of
Research
the village). The direction of surveying is chosen by spinning a pen or bottle. This provides a
consistent but unbiased (with regards to dog marking) selection of starting points common
to many village locations and avoids the need for a compass or access to Google Maps.\
Street
Questionnaires Surveys
Where the majority of dogs are judged to be owned and confined a questionnaire will be
more suitable for measuring vaccination coverage, either using a door-to-door approach,
passer-by survey or using a telephone survey. Passer-by surveys are a form of ‘convenience Secondary
sampling’ as they make use of convenient opportunities to access large numbers of people, Sources
hence they can shorten the time needed to access a minimum sample of dog owners to
ask whether their dog has been vaccinated. It is important to try and limit any bias produced
by the sampling; for example by choosing several convenient opportunities that will provide
a sample of people from a mix of socioeconomic backgrounds. Potential opportunities to Clinic
access samples include asking children at school if their dog has been vaccinated, people records
waiting at bus stops or shoppers at grocery shops/markets. Telephone surveys will only be
suitable if landline penetration in the area is good; however with increasing exclusive use
of mobile phones the ability of telephone surveys delivered through landlines to produce Behavioural
representative samples of respondents is reducing, and as the geographical location of the Observation
respondent is essential for this survey using mobile phones or internet to deliver the survey is
not possible.
When surveying in a relatively small area (N, the estimated total dog population < 190 can be Vaccination
used as a potential cut off) it may again be most suitable to use a house-to-house approach Coverage
along every street; the frequency of stopping at houses to deliver the questionnaire will
depend on the size of n in proportion to N; e.g. where n is 50% of N stop at every other
house to deliver the questionnaire, when n is 30% of N stop at every 3rd house. In larger
areas where n is less than 30% of N, alternative sampling approaches can be used. One
approach is the WHO Expanded Programme on Immunization cluster-survey technique4.
This uses two stages of sampling: 1) selecting a sample of villages or wards and then 2)
selecting a sample of households within those villages or wards. For example in Thungsong
District, Thailand a sample of 384 households were selected from an initial random sample
of 30 clusters (villages) and then a random sample of at least 13 households (until at least 7
dog-owning households had been found) were selected per village (Kongkaew et al., 2004).
The questions included in the questionnaire can be very limited; simply asking if people have
a dog and, if so, whether it was vaccinated in the recent campaign or within the last 12
months via some other route (e.g. private veterinarians or animal health workers). Additional
4
See http://whqlibdoc.who.int/hq/2005/who_ivb_04.23.pdf for a WHO reference manual on this technique

questions can be asked, for example about whether the marks applied at the vaccination
campaign are still present (this will be particularly useful if there is a concurrent street survey
being used) and if a dog is found unvaccinated why they were not vaccinated through
the campaign. The questionnaire can, of course, be expanded further to help measure
indicators other than vaccination coverage (see Section ‘Questionnaire surveys’ for more
details).
Questionnaire
Comparison between methods surveys
When street surveys and questionnaires have been compared, the results were relatively
similar if a sufficient sample size had been achieved in each method; for example Cleaveland
et al. (2003) found 62.1% coverage estimated through street surveys and 67.8% estimated Participatory
from questionnaires in Tanzania. Potential biases are that street surveys tend to miss Research
puppies which are often poorly reached through vaccination campaigns and this leads to an
overestimation of coverage. Street surveys also tend to miss confined dogs which are more
likely to have been vaccinated leading to underestimation. One option where both confined
Street
and roaming vaccinated dogs exist is to run a combination of majority ‘resource-light’ street
Surveys
surveys and where this indicates the coverage may be marginally below or above target to
then follow-up with ‘resource-heavy’ questionnaires to provide another measure
of coverage.
Secondary
Sources
Clinic
records
Behavioural
Observation
Vaccination
Coverage

Making Your Impact
Assessment Robust
International
Companion Animal
Selecting the appropriate impacts, indicators and methods of measurement for your
intervention is an important first step in conducting an impact assessment. This section
further explores other factors to keep in mind as your plan and implement your impact
assessment, highlighting ways of increasing the robustness of your evidence.
Ethical review
In this guidance we have selected indicators and methods of measurement that minimise
animal suffering. However, it is good practice to conduct an ethical review of the indicators
and methods of measurement that you plan to use for impact assessment, in addition to
an ethical review of intervention itself, in particular veterinary and animal care activities that
can affect welfare. Such ethical reviews are also a requirement for publication in many peer-
review journals and may also be part of legislation covering research involving animals. The
overarching principle of ethical review is to ensure that the potential risks are balanced by
the likely outcome of the research and hence it encourages investigators to minimise harm
and increase potential benefit through selection of the least invasive methods possible and
good experimental design. By using non-invasive techniques such as observation (e.g. street
survey and behavioural observation) this minimises the risks to the animals and the balance
is favourably tipped. For a more thorough discussion of ethical reviews, and guidance on
how to conduct them, see RCVS and BVA (2013).
A key question of your ethical review is whether there is the potential to cause ‘pain,
suffering, distress or lasting harm’ through any of your methods of measurement, the
threshold of pain that is used is that of introducing a hypodermic needle through the
skin. In the case of taking blood samples purely for reasons of surveillance, there is
potential to cause harm and an ethical review should be conducted. In some countries,
such a procedure would also fall under animal testing and procedures legislation and an
additional licence may be needed. Where a blood test is taken in the course of veterinary
treatment and a small amount (10% or less) is used for surveillance, this may fall outside
of animal testing and procedures legislation as the primary purpose was to treat the
animal and no additional harm was caused. Where data is being collected from owned
animals (e.g. questionnaires or clinic records) the informed consent of the owner to use
the data for impact assessment should be asked and clearly recorded. This data will need
to be securely stored and not shared with agencies outside of those responsible for the
impact assessment, especially when the owner’s details need to be maintained to allow
for longitudinal study, if follow-up is not required the data can be anonymised to maintain
confidentiality.
Setting up your own ethical review board is possible but may be quite time consuming. An
alternative is to access an ethical review board through local research institutes, who usually
have a committee already formed. As this ethical review should be conducted before the
data collection starts this may also be a good opportunity to create collaboration to access
further support, such as data analysis and interpretation.
87
Making Your Impact Assessment Robust
Attribution and measuring your intervention effort

This guidance focuses on collecting indicator data relating to impact. However, evidence
that impact has occurred needs to be presented alongside evidence of intervention effort
and causality; for example, was it your intervention that caused the improvement in dog
welfare or some other change in the environment? This is commonly termed attribution:
can the change in the impact be attributed to your intervention or was this change caused
by something else?
Establishing attribution can be supported from the outset by developing impacts that can
realistically be achieved by your intervention, where a causal link can be logically argued
between your effort and the impact (this was introduced as a ‘theory of change’ in the
section on ‘Identifying Dog Population Impacts’). For example, a reduction in nuisance
complaints relating to dog reproductive behaviour can be logically linked to an intervention
that sterilised a significant proportion of the roaming dog population, especially if a
reduction in lactating females was observed over the same time period. However, the
same intervention (assuming no vaccination was involved) could not claim to have reduced
rabies prevalence and any observed reduction must have been due to some other change
in the environment, such as a dip in the natural oscillations of the disease. Conversely, an
intervention that focused on rabies vaccination of dogs, and not sterilisation, could not claim
to reduce reproductive behaviour of dogs.
Attribution can also be supported by focusing monitoring and evaluation on the intervened
population only and not spreading out beyond the immediate population where impacts may
become too diluted to be visible. Further testing of attribution requires robust experimental
design, as described further in the following section.
Elements of robust experimental design

Although most DPM interventions are not initially envisaged as ‘experiments’, rather they
are developed for social good, by utilising experimental theory they can be set-up to
provide robust evidence of intervention effectiveness. This section highlights some of the
key elements of experimental design that are applicable to DPM interventions. Although
including all elements could be considered a ‘gold standard’ and may not be achievable
for all, including any of these elements could improve the quality of any future impact
assessments.
Perhaps the most fundamental element of experimental design is using a matched control.
This is a population of dogs (and their associated people) where no intervention has taken
place but the dogs experience a similar environment and ownership style. The dogs that
have received the intervention are termed the treatment group. For example, in Jaipur the
number of human rabies cases reported by the hospital from the intervened portion of the
city was compared to those in the non-intervened outer ring of the city and found to be
lower (Reece and Chawla, 2006); note that the intervention was spread to the entire city
after a few years as the beneficial impact was so clear, hence this control no longer exists
within Jaipur. An alternative is to have a control group where only a standard intervention
is used and a treatment group where additional aspects are used; this is usually used
where the standard treatment is expected to have a beneficial impact and withholding such
a benefit from people or animals in the control group would be unethical. An example is
mass vaccination for rabies control in one group of control villages and mass vaccination
plus sterilisation in another group of treatment villages; then comparing indicators related
to rabies control between these two groups to assess whether sterilisation contributed

positively to rabies control. The approach of a standard intervention as a control and

additional aspects as a treatment may be more realistic, as a complete absence of dog
management is relatively rare, especially if the actions of individual owners to control their
own dog’s contribution to the population are defined as part of an intervention.
Note that ideally there would be several treatment groups and several control groups, termed
replicates. This is because you would expect variation in all impacts over time and by having
several groups of each type you will have a measure of this variation. You can then compare
this to the difference between the control and treatment groups and ask the question, is
this difference between the control and the treatment greater than the variation within the
treatment groups? If yes, this intervention has been effective at creating the desired impacts.
Another form of control is a temporal control. This is the period before an intervention starts
(sometimes called a baseline) and can be used for comparison with the period during or
after the intervention. Ideally both temporal and location controls are used together, in other
words the pre- and post- intervention periods are compared for both treatment and control
groups as well as comparing the different responses of treatment and controls during the
intervention. One example of replicated treatment and control groups combined with use of
temporal controls is from Cleaveland et al. (2003); all villages (>40) in the Serengeti district
were selected as treatment villages (dogs vaccinated against rabies) and 10 villages were
selected randomly from the neighbouring district as control villages. The effect of vaccination
in the treatment villages resulted in a reduction of dog rabies cases and suspect bites when
compared with the control villages. This significant reduction in rabies cases in the treatment
villages was also found when data was compared to the pre-intervention period. This finding
strengthens the evidence that vaccinating dogs against rabies, lead to a reduction of dog
rabies cases. The study also found that the number of dog rabies cases stayed the same
in the control villages across both the pre-intervention and intervention period, and the
suspect dog bites showed a slight but non-significant increase in the intervention period as
compared with the pre-intervention period. The slight increase in suspect bites was due to
the intervention providing an improved supply of human rabies vaccine to all villages during
the intervention, hence people were more likely to seek bite treatment from their local health
centre as vaccine was now available, where previously supply had been intermittent.
The most rigorous approach is the cluster randomised control trial, where the locations
that will be treated/intervened and those that will act as controls are selected at random.
This ensures any pre-existing differences between the locations do not bias whether they
are selected for intervention. There is also replication of both controls and treatments; this
is what leads to the term ‘cluster’. One example of this approach is provided by Mazloumi
Gavgani et al. (2002) who randomly selected a control and treatment village from 9 matched
pairs of villages in the provinces of Kalaybar and Meshkin-Shahr in northwest Iran. These
villages had been matched for leishmaniasis prevalence in children, so the design was a
matched-cluster randomised control trial. They used deltamethrin- impregnated dog collars
in the treatment villages and found a reduced incidence of leishmaniasis in dogs and children
in the treatment villages compared to the control villages. By using a cluster randomised
control trial the authors have removed several potential factors that could have produced
this result other than the treatment itself; creating very robust evidence for the effectiveness
of deltamethrin- impregnated dog collars on Leishmaniasis.
In reality, the use of a control group in dog population management is extremely rare,
perhaps because the resources required to monitor and evaluate both treatment and
control locations has been perceived as too large. However, we would strongly encourage
this approach wherever possible as we can be more confident of attribution from the DPM

intervention. Even if randomisation is not possible and only a subset of the indicators can
be measured in the controls at baseline and beyond, the inclusion of a matched control can
strengthen the evaluation results considerably.
As mentioned previously, the benefit of using a control group or groups is that they capture
the natural variation in impacts that would have occurred over time regardless of your
intervention. These variations can include confounding variables; those factors other than
your intervention that influence your indicators. For example, an intervention that uses
rabies vaccination of dogs to reduce human rabies cases may find that improvements in
the availability of post-exposure prophylaxis (PEP) for people bitten by dogs will also reduce
human rabies cases. These confounding variables should be identified when evaluating the
impact of an intervention. At the planning stage, all likely confounding variables should be
established so that these can be measured as well as the intervention effort. In some cases,
these confounding variables can be controlled for or even avoided with good experimental
design. For example, an intervention that includes providing low-cost surgical sterilisation
in a deprived area may be hoping to see a reduction in the percentage of roaming lactating
females. However, the percentage of lactating females may also affected by the time of year
due to one or more peaks in breeding, hence time of year can be a confounding variable for
some locations. Although the seasons cannot be controlled, their influence on indicators can
be minimised by only comparing the percentage of lactating females recorded at the same
time of year. There are many other examples of avoiding the effect of confounding variables
on indicators. These include conducting street surveys at the same time of day, avoiding
extremes of weather that will affect dog behaviour, introducing questionnaires in the same
way and avoiding unusual days like holidays when different people will be at home.
To ensure monitoring and evaluation has the best chance of exposing a real change in an
indicator, the method of measurement used needs to be reliable. The need to be reliable
extends to the observers conducting the measurement as they are an important source of
potential error and the Section ‘Increasing and testing observer reliability’ explains how this
can be tested and minimised. Furthermore, a systematic bias could result from the desire
of the observer measuring the indicators to see a change, as might be expected when
that person is involved in running the intervention. They have a vested interest in wanting
to see that the intervention has worked and so, even subconsciously, may record data
more favourably over time. One way of avoiding this is to use independent evaluators that
have no reason to want to see a change in a certain direction. Even more powerful would
be to ensure the observers recording the measurements are unable to identify which is the
intervention and which is a control areas – this is referred to as a blind experiment – this
ensures an observer could not contribute bias to the results even if they wanted to. In
reality, using blinded independent evaluators may not be possible for many interventions
due to cost (although one cost effective option could be to swap staff between different
interventions for monitoring and evaluation events) and also the ability to truly blind someone
from an intervention when the dogs themselves may carry marks indicating they have been
through the intervention. However, it is a gold standard to aim for whenever possible. Where
not possible the people measuring the indicators have to be conscious of their inherent
bias and fight against these, trying to remain objective throughout their monitoring and
evaluation work.
Measuring intervention effort

Measuring intervention effort is essential for attributing change in impact indicators to the
intervention itself, and managers need to have documented what they have done to bring
about change. This is focused on the immediate results of the intervention. However, inputs,
the time and resources to implement the activities, should also be measured as these will
be important for assessing the efficiency of the intervention. For example, an intervention
that offers low-cost sterilisation needs to record the finances required to run their clinic (the
inputs), the number of dogs that come through their clinic for sterilisation and treatment (the
intervention effort, see Clinic records for more details), and also what proportion of the dog
population this represents in their intervention area.
The intervention also needs to have described a logical chain of steps leading from the
intervention effort through to the impact (this was introduced as a ‘theory of change’ in the
section on ‘Identifying Dog Population Impacts’). By providing evidence of intervention effort
and change in impact indicators, alongside a logical explanation of how this intervention has
influenced the impacts, managers have a foundation from which to attribute the change to
their intervention. Accurate recording of inputs will also allow additional questions about cost
effectiveness to be answered.
In addition, measuring some indicators of impact will require accurate records to have
been made during intervention implementation, including when dogs were accessed by the
intervention. For example, calculating survival of unowned dogs requires knowing when the
dog was last handled by the intervention (requiring a combination of individual identification
and records showing when that dog received its individual mark). Ideally all this data should
be stored using a database that allows for later analysis rather than on spreadsheets (which
are more difficult to manipulate) or on paper. There are many database systems available, for
example ‘Access’ available from Microsoft Office, and also online databases that can allow
access from a number of computers. Whichever software is chosen the database must be
backed-up to avoiding losing data.
Sampling
Sampling methods and sample size will be a consideration for all methods of measurement.
This starts with identifying the target population: which dogs and/or people across what area
does our intervention intend to impact upon? If this target group of dogs and associated
people is small, the methods of measurement can be applied to the entire population; this
would be termed a census. However, the target group is usually too large to affordably
census on a regular basis and so just a sample of dogs and/or people are chosen to
represent the wider group; results drawn from this sample are used to infer changes
occurring in the wider group, with the acceptance that there will be a level of error in
the inference as a result.
Sampling is broadly done in two ways, simple random or stratified random. In a simple
random sample, every dog or person has an equal chance of being picked, for example
stopping at every 10th house to ask a household questionnaire. In a stratified sample,
some dogs or people have a known greater chance of being selected than others. A
stratified random sample may be used when your target population clearly appears as sub-
populations of different sizes with important differing characteristics related to your impact,
and you don’t want to run the risk of missing a sub-population when using a random
selection. For example, consider a situation where dogs living in a large rural area have
poorer welfare and greater disease risk than those living in a small urban area within your
intervention zone. You may wish to select more dogs from the rural area for your sample
and fewer from the urban area. Knowing the different chances of selection for these different
sub-populations that make up your sample will be important at the analysis stage and also
for repetition during future monitoring events.

Planning to apply the method of measurement to an appropriate sample size will also be
important; too small and there is little chance of exposing a change in an impact indicator,
too large and there will be wasted resources on unnecessary measurement. Establishing
an appropriate sample size can be done statistically using power analysis, before you
begin the bulk of your data collection; this essentially describes the ‘power’ your data will
have to expose a change in the indicator, if a change has truly occurred. The power of
your data will be increased by increasing the size of your sample, increasing the size of the
change you want to expose (e.g. a reduction in the percentage of emaciated or thin dogs
in the population by 20%, rather than 10%) and increasing how risky you are willing to be
about the accuracy of results (e.g. will you accept being 80% confident that the results are
accurate? In other words, is 20% an acceptable risk that your results are actually wrong?).
Conducting power analyses is usually done mathematically and will require the support of
a statistician who should also be able to advise you on what statistical tests will be most
suitable and most powerful for your data. They will need to know the indicator you are
interested in and how it will be measured. What is the size of the effect that you want to
measure, so what is the baseline value of your indicator and what target do you have in
mind as a definition of success, e.g. a 10% reduction from baseline of 50%? How much risk
are you willing to accept regarding the accuracy of the results? They will also need some
idea of variability or error in your measurement; for example, if you repeatedly measure the
body condition score of the same group of dogs over a very short time frame, how much
does the % of the population scored as thin or emaciated change? This last question can
be answered by pilot testing your methods on dogs or people. Note that improving your
measurement to reduce any errors will also help increase the power of your data to expose
a change.
Preparing and conducting power analysis clearly requires time and the resources of a
statistician, but is the ideal approach to ensure your monitoring has the best chance of
resulting in accurate evaluation. Where this is not possible a general approach is to increase
the sample size to the maximum you can afford whilst still being able to repeat your
measurements over time, which is the core characteristic of evaluation.
Consistency in method
Selecting an appropriate size and composition of sample will help to ensure the data
resulting from your measurement is accurate and representative of your target group of
dogs and/or people. However, perhaps even more important is maintaining a consistent and
precise protocol for how the method of measurement will be conducted, as this will reduce
error in your data resulting from variations in how the method was carried out. Even where a
sample is accidentally biased towards a group of dogs or people, if the selection criteria and
method are consistent, the data will accurately expose a change in this biased group, if such
a change has truly occurred.
Inconsistency can come from many sources, including differences between observers (see
later section on ‘Increasing and testing observer reliability’ for an example of this) and failure
to develop and consistently apply a standard method, leading to changes in the sample
selected or protocol used for measurement over time (e.g. the introduction used by an
interviewer when conducting a questionnaire changes over time because it was not written
down initially, a different introduction can influence the answers given by the interviewee).

Ensuring there are complete and accurate records of all methods used, and allowing time
to familiarize and train people to the same standard, will help to reduce error in the data.
Logistically this may be supported by assigning a lead person(s) to be responsible for
monitoring and evaluation of the intervention with consistency as a key indicator of their
performance.
Increasing and testing observer reliability

The aim of monitoring and evaluation is to explore and expose changes in targeted impacts.
In order to do this, methods need to measure changes in the indicators as precisely as
possible. One of the challenges to this will be reliability in measurement (first introduced
in the section ‘Attribution and measuring your intervention effort’). A potential source of
error that can be mitigated will be the effect of the person that conducts the method of
measurement. As far as possible, the staff involved should be consistent and make every
effort to score in the same way at each monitoring event. However, staff changes are
inevitable and people can also unintentionally drift in their approach and assessments,
especially as some changes will be slow and will require several years of monitoring to
become apparent. In order to overcome this challenge, every monitoring event (even if staff
are the same) should include a period of refresher training where the protocols are reviewed
and discussed in detail. Ideally, a bank of photos is built up of dogs in varying body and
skin condition that can be used for refresher training on scoring the body and skin
condition of dogs.
Further, key indicators can be checked for inter-observer and intra-observer reliability.
Inter-observer reliability is a measure of how much agreement, and disagreement, there
is between the different people involved in conducting the method of measurement. Intra-
observer reliability is a measure of how consistently the same person scores over time.
The following is an example of testing inter-observer and then intra-observer reliability in
scoring body condition. This should be carried out before any monitoring event and at
regular intervals:
Body condition score training and agreement test

The body condition score recommended in this guide is a 5-point scale (Annex A) requiring
observation only with no need for physical examination. The indicator produced through
body condition scoring is the percentage of the adult (non-lactating) population that is body
condition score 1 (emaciated) or 1 and 2 (emaciated and thin). Methods of measuring body
condition include street surveys and clinic records.

Before either method of measurement is conducted, all observers that will be scoring
dogs should review the protocol and category descriptions in the body condition score
tool (Annex A), and together discuss and score at least 20 dogs of varying condition to
ensure they have the same understanding of the tool. This can either be done using
photos of dogs or by directly observing dogs in the street, shelter or clinic environment. A
powerpoint presentation to help with this training titled ‘Dog body condition scoring using
visual assessment’ is available on the www.icam-coalition.org website. Ideally at least some
direct observation of live dogs should be done in the environment matching the method of
measurement they will be using; i.e. in the street if they are to score dogs during a street
survey and in a clinic if they will be scoring as dogs pass through an intervention. This
should be done both by new observers, and people who have conducted these methods
before, in a group together.
Once they have taken part in the training and agreed through open discussion, the body
condition scores for at least 20 dogs, observers can then take part in an inter-observer
test. The following is a suggestion for how this test could be run, based on the process
developed by AssureWel (www.assurewel.org):
Inter-observer test
The observers are asked to score test sets of 10 photos of dogs. You can develop your own
test sets or you can access test sets in the form of an online quiz at www.icam-coalition.org.
This online quiz is comprised of dogs falling into different categories of body condition score.
The photos are presented in a random order and the observers select the body condition
score for each dog. The quiz gives you immediate feedback on whether your assessment
was correct or incorrect.
If the observer scored less than 9/10 correct, they are advised to review the powerpoint
presentation ‘Dog body condition scoring using visual assessment’. You can also discuss
the photos in this presentation to help identify anatomical features that may have been
missed. They can then retake the quiz. If they scored 9/10 or 10/10 they also need to retake
the quiz again as observers need to score 9/10 or 10/10 on two consecutive sets to pass
the test and be considered proficient at visually assessing body condition score. If they
score below the 9/10 threshold on a set they need to start again with at least a further 2 sets
before 2 consecutive pass scores can be achieved. For an observer who scores 9/10, or
more, on two consecutive sets you can be 85% certain that they can score dogs for body
condition with at least 80% accuracy (binomial distribution, with n=10 and p=0.80).
As the aim of monitoring is to evaluate change over time, consistency in scoring over time
(intra-observer reliability) is just as important as agreement between observers (int¬er-
observer reliability). Hence observers should retake the quiz at the outset of the next
monitoring event following a period of refresher training.

Test sets can also be built-up specific to a location. Ensure that photos are at high resolution
and in sharp focus (this is especially relevant for scoring the presence of ribs) so that they
can be shown on a large computer screen or projected. The full body of the dog should
be visible, showing both one side and at least the lower part of the back, so that both hip
bones and the vertebrate can be seen if prominent, plus the extent of the waist. Ensure
that dogs from the full range of categories are present in the test set in approximately the
proportion you would expect to see in the location.
Once in the field, reliability can be informally tested by observers working pairs and asking
each other for confirmation of a score. Discrepancies can be talked through and where
agreement can’t be reached the observers can refer back to the original definitions of each
score and even take a photo of the dog(s) for discussion with a wider team after observation
is concluded.

Using your results

This guidance aims to help intervention managers to decide how to measure their impact, by
selecting the most meaningful indicators and suitable/affordable methods. However, perhaps
the hardest job falls to the intervention managers to ensure monitoring is actually done; that
time is made for analysis and interpretation; learning and improvement occurs, along with
dissemination of results to others.
This process can be helped by developing a monitoring and evaluation plan. This can
include: a detailed description of each indicator along with the impact it measures; a detailed
protocol for the method of measurement and associated budget and timeline for when this
should be conducted; name(s) of people that will be responsible for ensuring the method is
conducted and data reported; and finally a plan for regular but infrequent evaluation ‘events’.
Evaluation events are workshops where relevant project staff and representatives from
wider stakeholders, potentially including donors, come together to review indicator data and
assess the extent, or barriers, to change over time; resulting in an impact assessment of the
intervention and suggestions for improvements.
Commitment to monitoring and evaluation will also be greater if designed from the outset
as an opportunity to learn as opposed to the need to prove impact to external audiences.
This uses as a starting point the learning that intervention field workers themselves need in
order to implement the intervention more effectively, rather than the results that managers
need to demonstrate impact to senior or external people. The concept is that evidence of
impact will be an emergent property of the learning, rather than the other way around and is
termed ‘learning-based monitoring and evaluation’ rather than ‘results-based monitoring and
evaluation’.
Part of the monitoring and evaluation process includes a phase of analysis and
interpretation, requiring the support of someone with an understanding of data analysis.
We recommend, if such expertise does not exist within the intervention team, that external
scientific expertise is sought, potentially from universities, research institutes or donors,
before monitoring begins. Building such a relationship from the outset ensures that data is
collected in a way that supports later analysis to answer questions about how indicators
have changed. For example, using sufficient sample sizes and using protocols that minimise
potential confounding variables or at least concurrently collecting data on these variables so
that their effect can be tested.
Returning to the subject of evaluation events, these set a deadline for when all the relevant
data should be available for interpretation and learning. This helps to ensure those people
collecting indicator data can see that it is being valued. It also ensures that data is not simply
collected for many years, but that it is analysed and actually used for learning on a regular
basis. The evaluation event can also finish with a reporting phase, to all intervention staff,
community and government representatives, and donors. Wider dissemination to other
interventions via project evaluation reports, media releases, conference presentations and
peer-reviewed publications would allow more widespread learning. The ICAM Coalition
in particular would value receiving such project evaluation reports and any information
on performance of indicators/methods of measurement mentioned in this guidance or
innovated by the intervention team.

Appendices
Appendices
International
Companion Animal
Acknowledgements Acknowledgements
and References
Development of this guidance would not have been possible without the generosity of
expertise shown by many organisations and individuals. The ICAM Coalition would like to
thank many of those organisations and individuals by name, and we apologise in advance
for any accidental omission from this list. Firstly all contributing members of our collaborating
partners who hosted our guidance author, Elly Hiby, answered her many questions and
ANNEX A
repeatedly reviewed the document as it developed: the Boyd Orr Centre for Population and
Ecosystem Health at the University of Glasgow, the Jeanne Marchig International Centre for
Animal Welfare Education (JMICAWE) at University of Edinburgh, the Animal Welfare and
Behaviour group at University of Bristol and the Department of Veterinary Tropical Diseases ANNEX B
at University of Pretoria. In addition, the many experts and hardworking DPM implementers
in the field that gave up their time and knowledge so willingly to help others: Lex Hiby, Louisa
Tasker, Chris Baker, Harry Eckman, Jack Reece, Gad Baneth, Orin Courtney, Phil Craig,
Peter Deplazes, Amielle DeWan, Julie Bedford, Giles Webber, Suzanne Rogers, Elena Garde,
Guillermo Perez, John Boone, James Serpell, James Kirkwood, Robert Hubrecht, Steven
ANNEX C
Wickens, Liz Murchison, Alex Oppmann , Marijana Vucinic, Khageshwaar Sharma, Jim
Pearson, Tom McPhee, Andrea Strakova, Elizabeth Murchison, Melanie Conor, Francoise
D
Wemelsfelder, Stephen Blakeway, Shanis Barnard, Shuping Ho, Fiona Woodhouse, Mark
Green, Roger Lohanan and Birte Snilstveit. ANNEX
And finally the members of the ICAM Coalition themselves, for having the foresight and
dedication to invest in this guidance and consistently provide their own expertise and
experience to its development.
Through your insights we hope to help others make the world a better place for dogs and
ANNEX E
the communities amongst which they live.
98
Acknowledgements and References
References
Acosta-Jamett, G., Cleaveland, S., Cunningham, A. a, Bronsvoort, B.M. deC, Craig,
P.S., 2010. Echinococcus granulosus infection in humans and livestock in the
Coquimbo region, north-central Chile. Vet. Parasitol. 169, 102–10. doi:10.1016/j.
vetpar.2009.12.009 Acknowledgements
and References
Adriani, S., Bonanni, M., 2012. Stray dogs and damage to sheep farms in the Oristano
Province, Sardinia, Italy, in: 1st DPM Conference. York, UK. September 4-8 2012.
Accessible at https://secure.fera.defra.gov.uk/dogs2012/index.cfm, p. 1100.
Anon, 2004. The Asilomar Accords. Asilomar, Pacific Grove, California. Accessible at ANNEX A
http://www.asilomaraccords.org/.
AWBI, 2009. Standard operating procedures for sterilisation of stray dogs under
the Animal Birth Control programme. AWBI: Chennai, India. Accessible at
http://www.awbi.org/awbi-pdf/SOP.pdf. ANNEX B
Beck, A.M., 1973. The ecology of stray dogs: A study of free-ranging urban animals. Purdue
University Press, Indiana, USA.
Belsare, A. V, Gompper, M.E., 2013. Assessing demographic and epidemiologic parameters ANNEX C
of rural dog populations in India during mass vaccination campaigns. Prev. Vet. Med.
111, 139–146. doi:10.1016/j.prevetmed.2013.04.003
D
Benner, C., Carabin, H., Sánchez-Serrano, L.P., Budke, C.M., Carmena, D., 2010. Analysis
of the economic impact of cystic echinococcosis in Spain. Bull. World Health Organ. ANNEX
88, 49–57. doi:10.2471/BLT.09.066795
Broom, D.M., 1991. Animal welfare: concepts and measurement. J. Anim. Sci. 4167–4175.
Budke, C.M., Deplazes, P., Torgerson, P.R., 2006. Global Socioeconomic Impact of Cystic ANNEX E
Echinococcosis. Emerg. Infect. Dis. 12, 296–303.
Butler, J., du Toit, J., Bingham, J., 2004. Free-ranging domestic dogs (Canis familiaris) as
predators and prey in rural Zimbabwe: threats of competition and disease to large wild
carnivores. Biol. Conserv. 115, 369–378. doi:10.1016/S0006-3207(03)00152-6
Casanova, C., Costa, A.I.P., Natal, D., 2005. Dispersal pattern of the sand fly Lutzomyia
neivai (Diptera: Psychodidae) in a cutaneous leishmaniasis endemic rural area
in Southeastern Brazil. Mem. Inst. Oswaldo Cruz 100, 719–24. doi:/S0074-
02762005000700006
Chambers, R., 2007. WORKING PAPER 296: Who Counts? The Quiet Revolution of
Participation and Numbers. Institute of Developmental Studies: University of Sussex,
Brighton.
Chomel, B., Chappuis, G., Bullon, F., Cardenas, E., David de Beublain, T., Lombard,
M., Giambruno, E., 1988. Mass Vaccination Campaign Against Rabies : Are Dogs
Correctly Protected ? The Peruvian Experience. Rev. Infect. Dis. 10, S697–S702.
Cleaveland, S., Fèvre, E.M., Kaare, M., Coleman, P.G., 2002. Estimating human rabies
mortality in the United Republic of Tanzania from dog bite injuries. Bull. World Health
Organ. 80, 304–10.

Cleaveland, S., Kaare, M., Tiringa, P., Mlengeya, T., Barrat, J., 2003. A dog rabies
vaccination campaign in rural Africa: impact on the incidence of dog rabies
and human dog-bite injuries. Vaccine 21, 1965–1973. doi:10.1016/S0264-
410X(02)00778-8
Cleaveland, S., Mlengeya, T., Kaare, M., Haydon, D., Lembo, T., Laurenson, M.K., Packer,
C., 2007. The conservation relevance of epidemiological research into carnivore Acknowledgements
viral diseases in the serengeti. Conserv. Biol. 21, 612–22. doi:10.1111/j.1523- and References
1739.2007.00701.x
Conroy, C., 2004. Participatory Livestock Research: A Guide. ITDG Publishing: Bourton-on-
Dunsmore, Warwickshire, UK.
ANNEX A
Davlin, S.L., Vonville, H.M., 2012. Canine rabies vaccination and domestic dog population
characteristics in the developing world: a systematic review. Vaccine 30, 3492–502.
doi:10.1016/j.vaccine.2012.03.069
Dawkins, M.S., 2006. A user’s guide to animal welfare science. Trends Ecol. Evol. 21, ANNEX B
77–82. doi:10.1016/j.tree.2005.10.017
Eckert, J., Deplazes, P., 2004. Biological, Epidemiological, and Clinical Aspects of
Echinococcosis, a Zoonosis of Increasing Concern. Clin. Microbiol. Rev. 17, 107–135.
doi:10.1128/CMR.17.1.107 ANNEX C
Fielding, W., 2008. Dogs: A Continuing and Common Neighborhood Nuisance of New
Providence, The Bahamas. Soc. Anim. 16, 61–73. doi:10.1163/156853008X269890
Frank, J.M., Carlisle-Frank, P.L., 2007. Analysis of programs to reduce overpopulation ANNEX D
of companion animals: Do adoption and low-cost spay/neuter programs
merely cause substitution of sources? Ecol. Econ. 62, 740–746. doi:10.1016/j.
ecolecon.2006.09.011
Garde, E., Serpell, J., Pérez, G., Vanderstichel, R., 2012. Behaviour assessment of male ANNEX E
dogs pre- and post surgical and non-surgical sterilization in Puerto Natales , Chile, in:
1st DPM Conference. York, UK. September 4-8 2012. Accessible at https://secure.
fera.defra.gov.uk/dogs2012/index.cfm.
Gilchrist, J., Sacks, J.J., White, D., Kresnow, M.-J., 2008. Dog bites: still a problem? Inj.
Prev. 14, 296–301. doi:10.1136/ip.2007.016220
Hemsworth, P.H., 2003. Human-animal interactions in livestock production. Appl. Anim.

Behav. Sci. 81, 185–198. doi:10.1016/S0168-1591(02)00280-0
Hiby, L.R., Reece, J.F., Wright, R., Jaisinghani, R., Singh, B., Hiby, E.F., 2011. A mark-resight
survey method to estimate the roaming dog population in three cities in Rajasthan,
India. BMC Vet. Res. 7, 46. doi:10.1186/1746-6148-7-46
Horvitz, D.G., Thompson, D.J., 1952. A Generalization of Sampling Without Replacement

From a Finite Universe. J. Am. Stat. Assoc. 47, 663–685.
Hughes, J., Macdonald, D.W., 2013. A review of the interactions between free-
roaming domestic dogs and wildlife. Biol. Conserv. 157, 341–351. doi:10.1016/j.
biocon.2012.07.005
ICAM Coalition, 2008. Humane dog population management guidance. Accessible at

http://www.icam-coalition.org/.

Jenks, K.E., Chanteap, P., Damrongchainarony, K., Cutter, P., Cutter, P., Redford, T.,
Lynam, A.J., Howard, J., Leimgruber, P., 2011. Using relative abundance indices from
camera-trapping to test wildlife conservation hypotheses - an example from Khao Yai
National Park, Thailand. Trop. Conserv. Sci. 4, 113–131.
Kaplan, E., Meier, P., 1958. Nonparametric Estimation from Incomplete Observations. J. Am.
Stat. Assoc. 53, 457–481. Acknowledgements
and References
Kii, T., 1982. A new index for measuring demographic aging. Gerontologist 22, 438–42.
Kitala, P.M., McDermott, J.J., Kyule, M.N., Gathuma, J.M., 2000. Community-based active
surveillance for rabies in Machakos District, Kenya. Prev. Vet. Med. 44, 73–85.
Knobel, D.L., Laurenson, K.M., Kazwala, R.R., Cleaveland, S., 2008. Development of
ANNEX A
an Item Scale to Assess Attitudes towards Domestic Dogs in the United Republic
of Tanzania. Anthrozoos A Multidiscip. J. Interact. People Anim. 21, 285–295.
B
doi:10.2752/175303708X332080
ANNEX
Knobel, D.L., Laurenson, M.K., Kazwala, R.R., Boden, L. a, Cleaveland, S., 2008. A cross-
sectional study of factors associated with dog ownership in Tanzania. BMC Vet. Res.
4, 5. doi:10.1186/1746-6148-4-5
Kongkaew, W., Coleman, P., Pfeiffer, D.U., Antarasena, C., Thiptaraa, A., 2004. Vaccination ANNEX C
coverage and epidemiological parameters of the owned-dog population in Thungsong
District, Thailand. Prev. Vet. Med. 65, 105–115. doi:10.1016/j.prevetmed.2004.05.009
Kumar, S., 2002. Methods for community participation. Practical Action Publishing Ltd.
Lakestani, N., Donaldson, M.L., Verga, M., Waran, N., 2011. Attitudes of children and adults
ANNEX D
to dogs in Italy, Spain, and the United Kingdom. J. Vet. Behav. Clin. Appl. Res. 6,
121–129. doi:10.1016/j.jveb.2010.11.002
Lee, N., 2013. Expert’s best experience - Kho Tao community engagement. Banna, Italy. ANNEX E
Lembo, T., Hampson, K., Haydon, D.T., Craft, M., Dobson, A., Dushoff, J., Ernest, E.,
Hoare, R., Kaare, M., Mlengeya, T., Mentzel, C., Cleaveland, S., 2008. Exploring
reservoir dynamics : a case study of rabies in the Serengeti ecosystem. J. Appl. Ecol.
1246–1257. doi:10.1111/j.1365-2664.2008.01468.x
Lloyd, S., Walters, T.M., Craig, P.S., 1998. Use of sentinel lambs to survey the effect of
an education programme on control of transmission of Echinococcus granulosus in
South Powys, Wales. Bull. World Health Organ. 76, 469–73.
Macpherson, C.N.L., Meslin, F.-X., Wandeler, A.I., 2012. Dogs, Zoonoses and Public Health.
2nd Edition. CABI: Wallingford, Oxon, UK.
Mallewa, M., Fooks, A.R., Banda, D., Chikungwa, P., Mankhambo, L., Molyneux, E.,
Molyneux, M.E., Solomon, T., 2007. Rabies Encephalitis in Malaria-Endemic Area,
Malawi, Africa. Emerg. Infect. Dis. 13, 2–5.
Manor, R., Saltz, D., 2004. The impact of free-roaming dogs on gazelle kid/female ratio in a
fragmented area. Biol. Conserv. 119, 231–236. doi:10.1016/j.biocon.2003.11.005
Mazloumi Gavgani, A.S., Hodjati, M.H., Mohite, H., Davies, C.R., 2002. Effect of insecticide-
impregnated dog collars on incidence of zoonotic visceral leishmaniasis in Iranian
children: a matched-cluster randomised trial. Lancet 360, 374–9.

Meek, P.D., 1999. The movement, roaming behaviour and home range of free-roaming
domestic dogs, Canis lupus familiaris, in coastal New South Wales. Wildl. Res. 26,
847–855.
Miura, A., Bradshaw, J.W.S., Tanida, H., 2000. Attitudes towards dogs: A study of university
students in Japan and the UK. Anthrozoos 13, 80–88.
Acknowledgements
Morters, M.K., Bharadwaj, S., Whay, H.R., Cleaveland, S., Damriyasa, I.M., Wood, J.L.N., and References
2014. Participatory methods for the assessment of the ownership status of free-
roaming dogs in Bali, Indonesia, for disease control and animal welfare. Prev. Vet.
Med. 116, 203–208. doi:10.1016/j.prevetmed.2014.04.012
Muldoon, J., Williams, J., Lawrence, A., Lakestani, N., Currie, C., 2009. Promoting a “duty ANNEX A
of care” towards animals amoung children and young people. Defra and Child and
Adolescent Health Research Unit, University of Edinburgh, UK.
B
NASS, 2011. Cattle Death Loss. National Agriculture Statistics Service (NASS), Agriculture
Statistics Board, USDA. Accessible at http://usda01.library.cornell.edu/usda/current/ ANNEX
CattDeath/CattDeath-05-12-2011.pdf.
Nunes, C.M., Pires, M.M., da Silva, K.M., Assis, F.D., Gonçalves Filho, J., Perri, S.H.V.,
2010. Relationship between dog culling and incidence of human visceral leishmaniasis
in an endemic area. Vet. Parasitol. 170, 131–3. doi:10.1016/j.vetpar.2010.01.044 ANNEX C
Otis, D.L., Burnham, K.P., White, G.C., Anderson, D.R., 1978. Statistical inference from
capture data on closed animal populations. Wildl. Monogr. 62, 3–135.
RCVS, BVA, 2013. Ethical Review for Practice-based Research: A report of a joint RCVS / ANNEX D
BVA working party.
Reece, J.F., Chawla, S.K., 2006. Control of rabies in Jaipur, India, by the sterilisation and
E
vaccination of neighbourhood dogs. Vet. Rec. 159, 379–383.
ANNEX
Reece, J.F., Chawla, S.K., Hiby, A.R., 2013. Decline in human dog-bite cases during a
street dog sterilisation programme in Jaipur, India. Vet. Rec. 172, 473. doi:10.1136/
vr.101079
Reece, J.F., Chawla, S.K., Hiby, E.F., Hiby, L.R., 2008. Fecundity and longevity of roaming
dogs in Jaipur, India. BMC Vet. Res. 4, 6. doi:10.1186/1746-6148-4-6
Sankey, C., Häsler, B., Hiby, E., 2012. Change in public perception of roaming dogs in
Colombo City, in: 1st DPM Conference. York, UK. September 4-8 2012. Accessible at
https://secure.fera.defra.gov.uk/dogs2012/index.cfm.
Savedoff, W.D., Levine, R., Birdsall, N., 2006. When Will We Ever Learn? Improving
Lives through Impact Evaluation. Center for Global Development: London, UK and
Washington DC, USA.
Schnabel, Z.E., 1938. The estimation of the total fish population of a lake. Am. Math. Mon.
45, 348–352.
Steinberger, R., 2012. A roadmap to creating successful measurable outcomes

through high volume spay/neuter in chronic poverty on a Lakota Reservation in
the US, in: 1st DPM Conference. York, UK. September 4-8 2012. Accessible at
https://secure.fera.defra.gov.uk/dogs2012/index.cfm.

Tenzin, Dhand, N.K., Gyeltshen, T., Firestone, S., Zangmo, C., Dema, C., Gyeltshen, R.,
Ward, M.P., 2011. Dog bites in humans and estimating human rabies mortality in
rabies endemic areas of Bhutan. PLoS Negl. Trop. Dis. 5, e1391. doi:10.1371/journal.
pntd.0001391
Tenzin, Wangdi, K., Ward, M.P., 2012. Human and animal rabies prevention and control cost
in Bhutan, 2001-2008: the cost-benefit of dog rabies elimination. Vaccine 31, 260–70. Acknowledgements
doi:10.1016/j.vaccine.2012.05.023 and References
Tepsumethanon, V., Wilde, H., Meslin, F.X., 2005. Six criteria for rabies diagnosis in living
dogs. J. Med. Assoc. Thai. 88, 419–22.
Totton, S.C., Wandeler, A.I., Ribble, C.S., Rosatte, R.C., McEwen, S. a, 2011. Stray dog ANNEX A
population health in Jodhpur, India in the wake of an animal birth control (ABC)
program. Prev. Vet. Med. 98, 215–20. doi:10.1016/j.prevetmed.2010.11.011
B
Totton, S.C., Wandeler, A.I., Zinsstag, J., Bauch, C.T., Ribble, C.S., Rosatte, R.C.,
McEwen, S. a, 2010. Stray dog population demographics in Jodhpur, India ANNEX
following a population control/rabies vaccination program. Prev. Vet. Med. 97, 51–7.
doi:10.1016/j.prevetmed.2010.07.009
Townsend, S.E., Lembo, T., Cleaveland, S., Meslin, F.X., Miranda, M.E., Putra, A.A.G.,
Haydon, D.T., Hampson, K., 2013a. Surveillance guidelines for disease elimination: ANNEX C
a case study of canine rabies. Comp. Immunol. Microbiol. Infect. Dis. 36, 249–61.
doi:10.1016/j.cimid.2012.10.008
Townsend, S.E., Sumantra, I.P., Pudjiatmoko, Bagus, G.N., Brum, E., Cleaveland, S.,
Crafter, S., Dewi, A.P.M., Dharma, D.M.N., Dushoff, J., Girardi, J., Gunata, I.K., Hiby,
ANNEX D
E.F., Kalalo, C., Knobel, D.L., Mardiana, I.W., Putra, A.A.G., Schoonman, L., Scott-
Orr, H., Shand, M., Sukanadi, I.W., Suseno, P.P., Haydon, D.T., Hampson, K., 2013b.
Designing programs for eliminating canine rabies from islands: Bali, Indonesia as a
case study. PLoS Negl. Trop. Dis. 7, e2372. doi:10.1371/journal.pntd.0002372 ANNEX E
Van Dijk, L., Prtichard, J., Pradhan, S., Wells, K., 2011. Sharing the load: A guide to
improving the welfare of working animals through collective action, The Health service
journal. Practical Action Publishing Ltd: Rugby, UK.
Wang, S.W., Macdonald, D.W., 2006. Livestock predation by carnivores in Jigme Singye
Wangchuck National Park, Bhutan. Biol. Conserv. 129, 558–565. doi:10.1016/j.
biocon.2005.11.024
Weiss, E., Patronek, G., Slater, M., Garrison, L., Medicus, K., 2013. Community partnering
as a tool for improving live release rate in animal shelters in the United States. J. Appl.
Anim. Welf. Sci. 16, 221–38. doi:10.1080/10888705.2013.803816
Whay, H.R., Main, D.C.J., Green, L.E., Webster, A.J.F., 2003. Animal-based measures for
the assessment of welfare state of dairy cattle, pigs and laying hens: consensus of
expert opinion. Anim. Welf. 205–217.
WHO, 2010. Working to overcome the global impact of neglected tropical diseases.
WHO, 2011. Diagnostics evaluation series No. 4: Visceral Leishmaniasis Rapid Diagnostic
Test Performance.
WHO, 2013. WHO Expert Consultation on Rabies, 2nd report. Techincal Report Series 982.
Accessible at http://www.who.int/rabies/resources/en/.
Williams, C., Johnston, J.J., 2004. Using Genetic Analyses to Identify Predators. Sheep
Goat Res. J. 19, 85–88.
Woodroffe, R., Prager, K.C., Munson, L., Conrad, P.A., Dubovi, E.J., Mazet, J.A.K., 2012.
Contact with domestic dogs increases pathogen exposure in endangered African wild
dogs (Lycaon pictus). PLoS One 7, e30099. doi:10.1371/journal.pone.0030099
Acknowledgements
WSPA, 2007a. Surveying roaming dog populations: guidelines on methodology. Accessible and References
at http://www.icam-coalition.org/.
WSPA, 2007b. Colombo dog population survey : June / July 2007 baseline survey.
Unpublished data.
Yoak, A.J., Reece, J.F., Gehrt, S.D., Hamilton, I.M., 2014. Disease control through fertility
ANNEX A
control: Secondary benefits of animal birth control in Indian street dogs. Prev. Vet.
Med. 113, 152–6. doi:10.1016/j.prevetmed.2013.09.005
Young, J.K., Olson, K. a., Reading, R.P., Amgalanbaatar, S., Berger, J., 2011. Is Wildlife
Going to the Dogs? Impacts of Feral and Free-roaming Dogs on Wildlife Populations.
ANNEX B
Bioscience 61, 125–132. doi:10.1525/bio.2011.61.2.7
ANNEX C
ANNEX D
ANNEX E

Annex A: Body Condition Scoring
Body condition score is based on 4 main body areas; check each one in turn to assess score:
• Backbone – if clearly visible score 1, if not visible check • Abdominal tuck (area behind the rib cage where the
ribs abdomen is clearly smaller than the rib cage in score 3
• Ribs – if clearly visible score 2, if not visible check and below) – if clearly visible score 3, if just visible score
abdominal tuck 4, if not at all visible score 5, then double check by
viewing waist from above
Note, do not include the final rib before the waist in
your assessment, as this may be visible on some • Waist from above – if clearly visible score 3, if just
dogs even if they have good fat covering due to visible score 4, if no waist score 5
conformation
BCS 1 EMACIATED
Ribs, backbone, pelvic bones visible
from a distance. Obvious waist and
abdominal tuck. No body fat.
BCS 2 THIN
Ribs visible but no backbone visible.
Some body fat present. Abdominal tuck
evident. Waist visible from above.
BCS 3 IDEAL
Ribs not visible even on close
inspection. Waist visible from above.
Abdomen tucked up and in, lower line
of tummy slopes upwards from end of
ribs to back leg.
BCS 4 OVERWEIGHT
Waist barely visible from above.
Abdomen slightly rounded, flanks
concave. Lower line of tummy is
horizontal from ribs to back leg.
Moderate amount of fat – jiggle noted
when walking.
BCS 5 OBESE
Waist absent. Abdomen rounded.
Lower line of tummy bulges
downwards. Sway from side to side
when walking.

Annex A: Body Condition Scoring
Acknowledgements:
• Descriptors for 5-point body condition score amended to be observation only without
palpation from Food For Thought™ Technical Bulletin No. 77R; Innovative Research in
Dog and Cat Nutrition™ (accessed from http://www.iams.com/pet-health/cat-article/how-
to-visually-assess-cat-and-dog-body-condition#qa2 Jan 2014).
• Refinement of terms in descriptions and illustrations from Nestle Purina Body Condition Acknowledgements
and References
System.
• Photos courtesy of Professor Darryn Knobel.
ANNEX A
ANNEX B
ANNEX C
ANNEX D
ANNEX E

Annex B: Example of Recording Sheet
for Behavioural Observations
International
Companion Animal
(Ideally this would be part of a phone app to reduce time needed for transcribing data)
Acknowledgements
Observer: Date and time: and References
Location/Site name: GPS reading:

Comments (weather, events, did all dogs leave site and if so for how long did you have to
delay observations?):
ANNEX A
Behavioural observation
B
Time start: ___________ Estimated number of dogs: ___________
ANNEX
Time stop: ___________ Estimated number of people: ___________
Tally (outcome of interactions

Dog-dog interactions only, not individual behaviours) ANNEX C
Amicable: Licks, paws, nudges with nose, or
grooming between dog, often with tails wagging.
D
Play behavior including charges with bouncing
gait, play faces (mouth relaxed, slightly open, teeth ANNEX
covered), wrestles and play chases
Neutral: Approach and retreat, often including

sniffing, limited body language, non-demonstrative.
Not aggressive but also not affiliative. ANNEX E
Mating: Copulation (not only mounting where one
dog stands with forepaws on another), usually ending
in a ‘tie’
Aggressive: Growling, teeth visible, barking,

biting, fighting. One dog flees with tail tucked to avoid
other dog, cowers or rolls over
Total dog-dog interactions
107
Annex B: Example of Recording Sheet for Behavioural Observations
Important human-dog interactions Tally

Positive human behaviours: Person
feeds a dog. Person pets a dog. Person calls a dog
to come to them either using name or other affiliative
sounds Acknowledgements
and References
Extreme relaxed behaviour from
both human and dog: Person walks within
one dog length/directly over the top of a dog and
the dog doesn’t move its body to avoid them. Dog
approaches person in friendly manner, person does
ANNEX A
not avoid this approach and allows dog to touch
them or stand/sit/lie beside them
Negative human behaviours: Person

hits or kicks a dog with any part of their body or an
ANNEX B
inanimate object. Person throws something at a
dog to scare/hurt it, includes ‘sham’ throws. Person
shouts or claps hands to scare a dog
Total human-dog interactions

ANNEX C
ANNEX D
ANNEX E

Annex C: Six Criteria for Rabies
Diagnosis in Living Dogs
International
Companion Animal
(Tepsumethanon et al. (2005))

The authors analysed clinical signs from 1,170 records of rabies suspected dogs that had Acknowledgements
bitten humans between 1988 and 1996. Laboratory confirmation was carried out on those and References
dogs that died within 10 days of observation. This identified 6 criteria for rabies diagnosis
using clinical signs with 90.2% sensitivity, 96.2% specificity and 94.6% accuracy; indicating
approximately 10% false negatives and 4% false positives should be expected. The 6 criteria
are as follows: ANNEX A
1. Age of the dog?
Less than 1 month: not rabies One month or more or not known: go to 2.
2. State of health of the dog?
Normal (not sick) or sick more than 10 days: Sick less than 10 days or not known:
ANNEX B
not rabies go to 3.
3. How did the illness evolve?
Acute onset from normal health: not rabies Gradual onset or not known: go to 4. ANNEX C
4. How was the condition during the clinical course in last 3-5 days?
Stable or improving (with no treatment): Symptoms and signs progressing or not
not rabies known: go to 5.
5. D
oes the dog show the sign of “Circling”?
(It stumbles or walks in a circle and hits its head against the wall as if blind)
ANNEX D
Yes: not rabies No or not known: go to 6.
6. D
oes this dog show at least 2 of the 17 following signs or symptoms during the
last week of life?
Drooping jaw (Fig. 1) Abnormal sound in barking Dry drooping tongue
ANNEX E
Licking its own urine Stiffness upon running or walking Regurgitation
Altered behaviour Biting and eating abnormal objects Aggression
Biting with no provocation Running without apparent reason Restlessness
Abnormal licking of water Bites during quarantine (Fig. 2) Appearing sleepy
Imbalance of gait Frequent demonstration of the “Dog sitting” position (Fig. 3).
Yes: rabies No or showing only 1 sign: not rabies
109
Annex C: Six Criteria for Rabies Diagnosis in Living Dogs
Acknowledgements
and References
ANNEX A
Figure 1 Drooping jaw
ANNEX B
ANNEX C
ANNEX D
Figure 2 Bites during quarantine

ANNEX E
Figure 3 Dog sitting position
All images copied from Tepsumethanon, V., Wilde, H. and Meslin, F.X., 2005. Six criteria for
rabies diagnosis in living dogs. Journal of the Medical Association of Thailand = Chotmaihet
thangphaet, 88(3), pp.419–22. Available at: http://www.jmatonline.com/index.php/jmat.

Annex D: Calculating Dog
Population Size
International
Companion Animal
In the section on the impact of reducing dog density, the indicator of dogs per km (or mile)
of street surveyed was introduced as an accessible measure of dog density that could Acknowledgements
be monitored over time. However, in some cases an estimate of total population size will and References
be desired, for example when planning and intervention after which density alone will be
sufficient for monitoring.
The simplest approach is to multiply the average number of dogs per km of street with the
total street length for the area in question, this street length is usually accessible from local
ANNEX A
government records or from Geographical Information System (GIS) maps (exclude lengths
of motorways and trunk roads when calculating total street length, as roaming dogs are very
B
unlikely to be found on these road types). If the routes used for the street survey were within
administrative areas for which there is data available on factors potentially related to roaming ANNEX
dog density (such as the number of detached houses or socioeconomic factors relating to
housing provided by recent censuses), such factors can be tested for their relationship with
the number of dogs seen on the routes. Route length is included as a covariate first and
then factors tested for their additional predictive value using multiple regression. If significant
covariates are found these can be used along with route length to refine the estimate for
ANNEX C
areas that were not surveyed but for which data on covariate factors exist. This method will
provide the total number of dogs seen roaming at the particular time of day that was chosen
for surveying and not the total roaming dog population; as some dogs will have been missed
by the observers and others will roam at other times of the day. ANNEX D
To establish the total roaming dog population a detectability estimate will be needed - this
is the chance that a dog that roams was seen and recorded by the observers conducting
E
the street survey. The detectability estimate can then be used to ‘correct’ the estimate
calculated from dogs per km of street surveyed and total street length to a total estimate of ANNEX
population size.
Establishing detectability requires more intensive survey effort, using either questionnaires
of owners to establish the number of owned dogs that are allowed to roam or mark-
resight experiments (see explanation that follows). Because these methodologies are more
intensive, the rapid street survey method described for measuring the number of dogs per
km of street surveyed can be used widely and then, in just a sample of areas, the more
intensive method is used in addition to the rapid method to establish the total population of
dogs in that area. The estimated population size in that area from the rapid survey can be
compared to the estimate from the intensive survey, exposing the underestimate produced
by the rapid survey:
rapid street survey estimate

detectability estimate =
intensive survey estimate
111
Annex D: Calculating Dog Population Size
Which intensive survey methods to use?

The most suitable method for intensive surveying depends on the proportion of the roaming
population that are unconfined owned dogs rather than unowned dogs.
Roaming owned dogs - questionnaires Acknowledgements

and References
If almost all roaming dogs are owned, the most efficient approach is to use a questionnaire
which asks owners about the number and confinement of their dogs. This results in an
estimate of the average number of dogs per dog owning household that are not always
confined (i.e. are allowed to roam for at least part of the day/night), and the total number
of dog-owning households in the area. One reason to avoid questionnaires and use the ANNEX A
following method of mark-resight is where owners have reason to be dishonest about
confinement, for example if there is a local ordinance requiring dogs to be confined.
Unowned dogs – mark-resight ANNEX B

In some countries or regions, for example in India, it is evident that a significant proportion
of the roaming population is unowned. If tolerance of and resources for unowned dogs are
sufficient to allow some of their pups to survive to sexual maturity the unowned population
will exist at a density determined by the carrying capacity in that region. Even in areas
where resources are insufficient to allow any pups born on the street to survive to breed
ANNEX C
themselves there may be enough abandoned dogs and their surviving pups to form a
significant proportion of the roaming population, perhaps only seasonally in tourist areas.
If unowned dogs do form a significant proportion of the roaming population then the size
of that proportion will be unknown, in which case some direct estimate of the size of the ANNEX D
roaming population via intensive street surveying is required using mark-resight methods.
If the sample areas selected for the intensive surveying are small enough to deal with
E
in their entirety there will be no need to define sub-sample regions. The approach is to
conduct an initial survey in order to mark a known number n1 of randomly selected dogs, ANNEX
either with an artificial mark such as a paint spray or collar or by photographing distinctive
natural markings. If the sample areas are contained by boundaries that are meaningful to
dogs, such as the boundaries of a village, marking can be done throughout the village. If
the boundaries have no meaning to dogs and they are likely to cross them, such as the
boundaries of a municipality within a town, the streets used for marking should be far
enough within the boundaries to avoid any of the dogs in the marked sample having moved
out of the region by the time of the second survey. During a second survey of all the streets
(including those streets up to the border of the area if any were initially excluded for marking)
one or more days later n2 dogs are seen, of which m were included in the n1 sample. The
ratio m/n1 is the fraction of marked dogs that were on the streets at the time of the second
survey and is assumed to equal the fraction of all roaming dogs that were on the streets at
that time. Dividing the second survey street count n2 by that fraction gives the Peterson
estimate of the total number of dogs, which can also be calculated as follows:
n1 n2
m
If the sample areas selected for intensive surveying are too large to survey along all streets,
a sub-sample of smaller regions may be used, preferably pre-existing regions such as wards
or health post regions. Ideally this sub-sample is selected randomly but avoiding any of
the selected regions having common borders (WSPA (2007) provides a method for random
selection of regions whilst avoiding common borders). The survey method used within each

Annex D: Calculating Dog Population Size
sub-sample region is then as described above for the entire sample area. The abundance
estimates for the sub-sample regions are then used to estimate abundance over the whole
sample area, either by summing the estimated dog population for the regions and dividing
by the fraction of pre-existing sub-sample regions surveyed (Horvitz and Thompson, 1952)
or by using a regression estimator to exploit any covariates of the abundance estimates that
may be available for each of the pre-existing regions (the common practice of extrapolating
by area is an example where area is used as covariate but a poor one because region area Acknowledgements
and References
and dog abundance are usually not strongly correlated).
Another opportunity to apply mark-resight to large sample areas is to use the standard
routes established for the dog density street surveys. By marking (or photographing) dogs
along at least a sample of the standard routes, the fraction of those seen on a second
survey of the same route can be used as a detectability estimate. Just as in the Petersen
ANNEX A
method, the proportion marked is upwardly biased (leading to an underestimate of total
dogs) as a result of heterogeneity but is now also subject to a bias in the opposite direction
because in conducting the second survey only along the standard route we may have
missed some of the marked dogs that were roaming at that time but were on other streets. ANNEX B
If the intervention itself involves marking dogs, for example with collars or paint to show a
dog has been vaccinated or with an ear notch to show it has been sterilised, then these
intervention marks can be used for mark-resight (for example, as used in Hiby et al. (2011)
Animal Birth Control (ABC). This may be particularly relevant if the intervention coverage, ANNEX C
i.e. the proportion marked, is also required to evaluate the effectiveness of the intervention
(see the section on estimating rabies vaccination coverage). This will require consistent and
accurate data collection on intervention effort; specifically the number of dogs marked, on
what date and in which area they live. It may also require an estimate of mark loss, as collars
may fall off or be removed and paint wears off over time. Survival of the dogs will also be
ANNEX D
relevant in the case of a permanent ear notch that has been applied some months or years
previously. Accounting for mark loss and survival will provide an accurate estimate of the
marked population that currently exists. One option is to reduce the time between applying
the marks to resighting so that the mark loss and mortality can be assumed to be zero; i.e. ANNEX E
limiting the days between vaccination and coverage estimation. For an intervention where
survival is the relevant factor, resighting for ear notches will be best done in the first weeks
and months of the intervention.
The Petersen method suffers from the need to assume that all dogs that roam are equally
likely to be seen. That assumption is unlikely to hold because of behavioural differences
between dogs affecting when and for how long they are visible on the streets. As a result
the abundance estimates are likely to be biased downwards (underestimating the total
population), particularly if surveys are conducted at the same time of day.

Extensions of the Peterson method that use multiple subsequent surveys to reduce the
variance of the estimator include the Schnabel method (Schnabel, 1938) as used by Beck
for roaming dog abundance in Baltimore, USA (Beck, 1973) and the Schmacher-Eschmeyer
method as used by Totton in Jodhpur, India (Totton et al., 2010). As in the Peterson
method, the dogs seen during each subsequent survey are classified into those that have,
and have not, been previously marked (i.e. paint marked or photographed) but the size of
the marked sample is increased by marking some or all of the unmarked dogs seen during Acknowledgements
and References
each survey. If marking is by photography, only dogs that have distinctive natural markings
are photographed in order to facilitate the classification process. However these extensions
require the same assumption as the Peterson method (i.e. that all dogs that roam are equally
A
likely to be seen) and suffer from a similar level of bias if this assumption fails, so it would
seem more efficient to use the Peterson method in a larger sample of regions rather than ANNEX
expend the available effort on multiple subsequent surveys.
If the risk of bias is not acceptable, an alternative is to conduct a sequence of surveys and
establish the ‘encounter history’ of each dog seen (for example 0 1 0 1 1 representing
a dog first seen on the second of five surveys and then seen again on surveys four and ANNEX B
five). Encounter history can be analysed using models of type Mh or Mth (Otis et al.,
1978) to incorporate variation between dogs in the probability of being seen (referred to as
‘heterogeneity’ in the mark recapture literature) or variation in that probability both between
dogs and from survey to survey. For example Belsare and Gompper (2013) used program
CAPTURE to run the jackknife estimator model of type Mh on encounter history data from six
ANNEX C
villages in Maharashtra, India to estimate the number of roaming dogs in each village.
Such models are likely to reduce bias as compared to the Peterson method, depending
on the source of variation in the probability of being seen. Intuitively, information on that
variation is available from the numbers of dogs seen very frequently and the number seen
ANNEX D
only once in comparison to the numbers expected if all dogs were equally likely to be seen.
As expected, Belsare and Gompper (2013) report higher estimates using the jackknife
estimator than those obtained using Beck’s method, which gave point estimates below the
minimum population size (as determined by independent surveys) in each village. ANNEX E
However it is time consuming and complicated to establish the encounter histories. The
dogs have to be identified as individuals so marking via a paint spray cannot be used.
Some dogs do not have distinctive natural markings that allow them to be identified from
photographs and descriptions, for example 23% of dogs collected for spaying in Jaipur
are uniformly black or tan coloured (Reece, pers. comm.). The photographs only establish
encounter histories for the distinctively marked dogs so that the resulting abundance
estimate has to be corrected by dividing by the proportion of dogs that are distinctively
marked. During the surveys, each dog seen has to be recorded as distinctively marked
or not and be photographed only if it is distinctively marked. The criteria used to decide
if a dog is sufficiently distinctive must remain consistent between surveys. If an observer’s
ability to distinguish dogs increases over time, there is a risk that a dog will be defined as
identifiable and photographed in later surveys and recorded as not having been seen before,
whereas in fact it was seen but was considered insufficiently well marked in those early
surveys when the observer was less confident. Although encounter histories are likely to
reduce bias they do require significant time and effort, limiting the area that can be covered
by this approach and brining in other potential biases linked to small sample sizes. This
suggests simple Peterson estimates using artificial marks or photographs is the method of
choice in most cases.

Annex E – Sample Questionnaire
International
Companion Animal
Date: Interviewer: Interview no:

Acknowledgements
District: Ward: Street: House no/name: and References
Household GPS reading Latitude: Longitude:
Type of structure: Single/detached house Attached house Flat
Row house Hut/Shanty Compound

ANNEX A
‘Starter question’ (defines time commitment):
Does your household own a dog?
‘Own’ is defined as a dog living in the house or yard most of the time and being fed by the
ANNEX B
household on a regular basis.
NO YES
Ask the respondent if you can ask further questions. State: ANNEX C
“Your answers to these questions, and those of other people we ask, will be used to
help us understand the dog population better. There are no right or wrong answers
to these questions, so please answer as accurately as you can and you can skip
any question that you do not wish to answer. Your name and address will remain ANNEX D
confidential and will only be used to identify where in ___ you live and to find your
house again if we need to ask follow up questions in future”
If he/she owns a dog tell him/her that the interview will take up to 20 minutes.
ANNEX E
If he/she does not own a dog tell him/her that the interview will take up to 10 minutes.
Permission received? Yes No
YES, but please come back later

What day/time?
If YES, go to Question 1.1
SECTION 1 – HOUSEHOLD INFORMATION

1.1. First, some questions about you;
Name: Gender: Age:

(age can be recorded as ‘adult’ or ‘child’ if they don’t want to give their exact age)
What religion are you?: None Muslim Christian

Hindu Buddhist Other
1.2. How many people live in the household? _________
115
1.3. Does the household feed any dogs that are not your own on a regular basis
(regular basis defined as at least once per week)?
NO YES Don’t know
SECTION 2 – DOG BITES

2.1. Was anyone in the household bitten by a dog within the last 12 months whilst Acknowledgements
and References
physically in ______*? (*Refer to intervention area)
NO YES Don’t know
If yes, provide details of the dog bite below:

ANNEX A
What is What age Bitten by which Do you Do you know How did you treat the
the sex were they dog? know the why the dog site of the bite, if at
of person when age and bit? all? (include all relevant
bitten? bitten? sex of responses)
biting dog?
Sex Age
ANNEX B
ANNEX C
ANNEX D
1. Own dog 1. Yes, answer 1. Wash the wound with
suggests water only
E
2. O
wned dog
provocation
from the 2. Wash the wound with ANNEX
community 2. No, answer soap and water
suggests 3. Go to the bite centre/
3. Unowned
bite was hospital
dog from the
unprovoked
community 4. Nothing
3. Don’t know
4. Unknown 5. Other?
strange dog (please describe)
5. Don’t know 6. Don’t know
If asking for agreement with attitude statements include them here for non-dog owners and
at the end for dog owners.
If the household does not own a dog, please thank the respondent for their time and ask if
they have any questions before leaving.
If the household owns a dog go to Question 3.1.

SECTION 3 – HOUSEHOLD DOGS

3.1.
Adults Puppies
> 3 months < 3 months
Total M F Unk M F Unk
How many dogs are there living in this Acknowledgements
and References
household?
How many dogs were in this household
12 months ago?
Over the past 12 months, did any of the household’s adult dogs die or leave the household? ANNEX A
(only for adult dogs, puppies of under 4 months are included in female breeding history table only)
NO YES Don’t know
If yes, please record the details of all adult dogs that have left the household in the last
12 months in the table below:
ANNEX B
What sex was How old was he/she
What happened to him/her?
the dog? when this happened?
1. ANNEX C
2.
3.
4.
ANNEX D
5.
E
6.
ANNEX
7.
8.
9.
10.
1. Male 1. Sold 8. Died through

starvation
2. Female 2. Given away
9. Died other
3. Don’t know 3. Killed by owner
10. Disappeared
4. Killed by authorities
11. Abandoned/
5. K
illed by someone
Disowned
else
12. Stolen
6. Died in accident
13. Unknown
7. D
ied of disease/
parasites
Ideally, how many dogs would you like in your household? _________

3.2. Household adult dogs information (only for adult dogs currently owned by household, puppies of under 4 months are included in female breeding history table only)
Age? (be as What is the main Who looks after the

Dog Where did you get this Age when
Name? Sex? precise as Sterilised? Breed? role or function of dog?
no. dog from? acquired?
possible) this dog?
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
1. Male 1. Yes 1. Pure bred 1. Pup of own dog 1. Guard household 1. Respondent
2. Female 2. No 2. Pure-bred 2. Bought inside ____* 2. Protect livestock 2. If not respondent,
cross age & sex of
3. Pregnant 3. Don’t 3. Bought outside ____ 3. Protect crops
household
fem know 3. Mongrel 4. Gift from inside ____ 4. Pet/companion member
4. Lactating 4. Don’t know 5. Gift from outside ____ 5. Hunting 3. Don’t know
fem
6. Adopted off the street 6. Breeding
5. Don’t
know 7. Adopted from shelter 7. Other (specify)
8. Other (specify) 8. Don’t know
9. Don’t know
* ____ Refer to intervention area
International Companion Animal Management Coalition

ANNEX
ANNEX
ANNEX
ANNEX
ANNEX
and References
E
C
D
B
A
118
Acknowledgements
3.3 Household adult dog care (includes welfare scores if dog can be seen)
Dog Has dog Has dog Has dog been

no. Access treated for Was this
been been How often is Body
Is the dog How is the dog to shelter fleas/ticks? What kind of dog given Skin
(as in vaccinated? dewormed? dog fed by your condition
confined? confined? when If yes, how food? fresh water problem?
table If yes, how If yes, how household? score?
confined? long ago? yesterday?
3.2) long ago? long ago?
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
1. All day only 1. Compound/ 1. Yes, all 1. Yes, and 1. Yes, and 1. Yes, and 1. Twice a day or 1. Commercial 1. Yes 1 1. No
yard/garden the time number of number of number of more 2. Home-made 2. No 2 2. Yes
2. All night
months/ months/ months/ 2. Once per day
only 2. House 2. Yes, but 3. Leftover 3. Don’t 3 3. Dog not
years years years
not all 3. Once every 2 human food know visible
3. All day & 3. Kennel 4
the time 2. Never 2. Never 2. Never days
night 4. Static chain 4. None
3. No 3. Don’t 3. Don’t 3. Don’t know 4. Less frequently 5
4. Sometimes 5. R
unning chain 5. Don’t know
know know than once every 6. Dog not
4. Don’t
5. Never 6. O
ther (specify) visible
know 2 days
6. Don’t know
5. Don’t know
International Companion Animal Management Coalition

ANNEX
ANNEX
ANNEX
ANNEX
ANNEX
and References
E
C
D
B
A
119
Acknowledgements
3.4. Female breeding history (please include all female dogs, if a female has never had
a litter the answer to the first question will be 0 and the rest of the table can be
ignored as the follow-up questions are for breeding females only)
Female dog no. from previous table
How many litters has she had in Acknowledgements

her lifetime? (if she was owned by and References
someone else previously, please
include litters born with this previous
owner)
A
How old was she when she had her
first litter? ANNEX
Has she had any litters in the previous
12 months?
If she has had a litter in the previous
B
12 months, please tell us about the
fate of all puppies from this litter: ANNEX
# of puppies born dead
# of puppies born alive
# of puppies still in household ANNEX C

# puppies died from natural causes
(after having been born alive
# puppies killed by household member
ANNEX D
# puppies died from other causes
# puppies given away for free

ANNEX E
# puppies sold
# puppies abandoned

SECTION 4 – VACCINATION HISTORY

(only included if using questionnaire to estimate vaccination coverage; question about
roaming can be changed to a specific time window when street survey occurs)
4.1 Vaccination coverage estimation
Dog Was this dog vaccinated in the Was it marked? Is this mark still Is your dog in the Acknowledgements
no. recent campaign? present? house/yard now? and References
A
2
ANNEX
3
5 ANNEX B
6
7
ANNEX C
8
10
ANNEX D
1. Yes 1. Yes, collar 1. Yes 1. Yes
2. No, already vaccinated
3. No, didn’t know about the
2. Yes, paint
3. No
2. No, fell or wore 2. No, roaming
off 3. N
o, on walk
ANNEX E
campaign 3. Don’t know with owner
4. Don’t know
4. No, couldn’t get there 4. Don’t know
5. No, don’t want my dog
vaccinated
6. No, other (specify)
7. Don’t know
If asking dog owners for agreement with attitude statements include them here.
Please thank the respondent for their time and ask if they have any questions

Attitude statements used in Colombo, Sri Lanka

Based on questionnaire used to assess the dog population in Colombo, Sri Lanka by Blue
Paw Trust and WSPA (WSPA, 2007b).
1. Having a dog is a waste of money:

Acknowledgements
and References
1 2 3 4 5
Strongly disagree Disagree Unsure Agree Strongly agree
2. I like dogs:
ANNEX A
1 2 3 4 5
3. Dogs should always be kept outside the house: ANNEX B

1 2 3 4 5
4. I don’t like being close to dogs:

ANNEX C
1 2 3 4 5
ANNEX D
5. Dogs add happiness to people’s lives:
1 2 3 4 5
E
ANNEX
6. People who own dogs should spend time every day playing with them:
1 2 3 4 5
7. If a dog of mine got a skin disease, I would not want it around the house:
1 2 3 4 5
8. If a female dog of mine had a litter of puppies, I would not want to keep any of them:
1 2 3 4 5
9. A person should treat their dog with as much respect as they would a human member
of the family:
1 2 3 4 5

10. A dog is a valuable possession:
1 2 3 4 5
11. Street dogs should be looked after by the community:

Acknowledgements
1 2 3 4 5 and References
12. The welfare of street dogs is important to me:
1 2 3 4 5
ANNEX A
13. People should not feed street dogs: ANNEX B

1 2 3 4 5
14. I like having dogs around on my street: ANNEX C

1 2 3 4 5
15. Dogs should have the same rights and privileges as humans:
ANNEX D
1 2 3 4 5
ANNEX E
16. Street dogs pose a danger to people:
1 2 3 4 5
17. Street dogs should not be allowed to breed:
1 2 3 4 5
18. It is not acceptable to kill dogs:
1 2 3 4 5

Attitude statements used in Tanzania

Attitude statements developed and used as part of a questionnaire survey to assess the
attitude of dog owners towards their dogs in 12 sites across Tanzania (Knobel et al., 2008).
1. Our dog/s is/are a valuable possession:

Acknowledgements
and References
1 2 3 4 5
2. Our dog/s is/are an important part of the household:

ANNEX A
1 2 3 4 5
3. We like owning a dog: ANNEX B

1 2 3 4 5
4. Our dog is a member of the family:

ANNEX C
1 2 3 4 5
ANNEX D
5. You should treat your dog with as much respect as you would a human member
of your family:
E
1 2 3 4 5
Strongly disagree Disagree Unsure Agree Strongly agree ANNEX
6. Our dog/s is/are accustomed to being touched:
1 2 3 4 5
7. Our dog/s enjoy being petted:
1 2 3 4 5
8. We often play with our dogs:
1 2 3 4 5
9. We enjoy our dogs’ companionship:
1 2 3 4 5

10. It is unhealthy to touch dogs:
1 2 3 4 5
11. We never touch our dogs:

Acknowledgements
1 2 3 4 5 and References
12. Our dog will bite us if we touch it:
1 2 3 4 5
ANNEX A
ANNEX B
ANNEX C
ANNEX D
ANNEX E

Dog attitude Questions, used in Japan and UK

Based on the dog attitude statements (named the “DAQ”) developed for a study of university
students in Japan and UK (Miura et al., 2000)
Please indicate how strongly you agree or disagree with the following statements, by drawing
Acknowledgements
a circle around the appropriate number on the agreement-disagreement scale. For example, and References
if you think you disagree with the statement strongly, you might circle the 2 on the left side of
the scale. Note the circle can be drawn by interviewee or interviewer can read out statements
and ask for level of agreement and circle the correct number on behalf of the interviewee.
Disagree Agree
ANNEX A
Disagree Neutral Agree
strongly strongly
2 1 0 1 2
ANNEX B
Disagree Agree
1. I think that there are too many dogs in this country. 2 1 0 1 2
2. I think that the most important role of dogs is guarding. 2 1 0 1 2 ANNEX C

3. I think that a dog is ‘Man’s best friend’. 2 1 0 1 2
4. I wish more hotels would allow dogs to stay with their owners. 2 1 0 1 2
5. I think that dogs cannot live without human help because they ANNEX D
were domesticated by humans. 2 1 0 1 2
6. I think that keeping a dog is a waste of time and money. 2 1 0 1 2
7. I think that dogs should be allowed indoors. 2 1 0 1 2 ANNEX E

8. I feel sorry for a dog when I see the dog obey every command,
because the dog seems to be controlled by the owner. 2 1 0 1 2
9. I think that having a dog increases security. 2 1 0 1 2
10. I think that keeping a dog is annoying to its owner’s neighbours.

2 1 0 1 2
11. I think that dogs have personalities like humans. 2 1 0 1 2
12. I think that an owner should look after his/her dog until it dies
naturally, if the dog has an incurable illness. 2 1 0 1 2
13. I think that dogs are smelly. 2 1 0 1 2
14. I think that training dogs is a reflection of human arrogance. 2 1 0 1 2
15. I think that owners should keep their dogs (rather than get rid
of them) even if the dog has attacked people. 2 1 0 1 2

Disagree Agree
16. I think that dogs should be allowed to roam unless they cause
problems in human society. 2 1 0 1 2
17. I think that having a dog provides opportunities for meeting

new people. 2 1 0 1 2
Acknowledgements
and References
18. I wouldn’t want to keep a dog indoors, because they shed
their hair. 2 1 0 1 2
19. I think that only working dogs such as sheep dogs, guide dogs
and police dogs need training. 2 1 0 1 2 ANNEX A
20. I think that stray dogs should be euthanized. 2 1 0 1 2
21. I think that dogs are happier when kept outdoors rather than
indoors. 2 1 0 1 2
ANNEX B
22. I wish restaurants would allow dogs to enter with their owners. 2 1 0 1 2
23. I think that euthanizing a dog is cruel. 2 1 0 1 2
24. I think that stray dogs tend to bite. 2 1 0 1 2 ANNEX C

25. I wish shops would allow dogs to enter with their owners. 2 1 0 1 2
26. I think that dogs should be chained outdoors. 2 1 0 1 2
27. I think that an owner should let his/her dog run free in public ANNEX D
places like parks. 2 1 0 1 2
28. I think that stray dogs create a nuisance. 2 1 0 1 2
29. I think that euthanasia is justifiable if a dog is suffering. 2 1 0 1 2 ANNEX E

30. I wouldn’t want to keep a dog indoors, because they are
unhygienic. 2 1 0 1 2
31. I think that when people get a new dog, they should take the
dog to a training class. 2 1 0 1 2
32. I think that stray dogs are a problem in this country. 2 1 0 1 2
33. I think that dogs should have access to all rooms in the house. 2 1 0 1 2
34. I think that having a dog makes it difficult for its owner to travel. 2 1 0 1 2
35. I think that dogs are more loyal than people. 2 1 0 1 2
36. I get upset when I see dogs chained outdoors. 2 1 0 1 2
37. I think that training dogs is cruel. 2 1 0 1 2
38. I think that dogs are unhygienic. 2 1 0 1 2

Disagree Agree
39. I think that the most important role of dogs is giving

people companionship. 2 1 0 1 2
40. I wouldn’t want to keep a dog indoors, because they are

smelly. 2 1 0 1 2
Acknowledgements
and References
41. I think that owners should keep their dogs (rather than get rid
of them) even if the dog is aggressive to strangers. 2 1 0 1 2
42. I think that having a dog is fun. 2 1 0 1 2
43. I think that dogs should be spayed or neutered to prevent ANNEX A

unwanted puppies being born. 2 1 0 1 2
44. I think that dogs should obey their owner’s commands

all the time. 2 1 0 1 2
ANNEX B
45. I think that it is justifiable to euthanize aggressive dogs. 2 1 0 1 2
46. I am interested in TV programs and articles on dogs. 2 1 0 1 2
ANNEX C
ANNEX D
ANNEX E

Attitude statements used with 4 year old children

in UK, Italy and Spain
Attitude statements taken from Lakestani et al. 2011; original list was 12 statements long
but 3 statements relating to dog emotions (e.g. Dogs can be scared of people) were
Acknowledgements
found to reduce reliability of the test and so were removed from further analysis, only the 9 and References
statements that resulted in reasonable reliability are included here. 4 year old children were
asked to rate each of the following item as “never”, “sometimes” and “mostly”:
A
1. Dogs are dirty
2. Dogs can be friends with people
ANNEX
3. Dogs are smelly
4. I love my dog/I would like to have a dog
5. Dogs bite ANNEX B
6. Dogs are scary
7. I like hugging my dog / I would like to hug a dog
8. Dogs are fun
9. I like walking my dog and playing with him/ I’d like to walk a dog and play with it
ANNEX C
For positive items, responses were coded as follows: never = 1, sometimes = 2, mostly = 3.
This was reversed for negative items. Attitude scores were calculated by taking the mean of
these response codes for each child. Attitude scores ranged from 1 to 3, with a high score
corresponding to a positive attitude to dogs.
ANNEX D
ANNEX E

犬只数量管理监测与评估指南 英文版

Uploaded by

Copyright:

Available Formats

犬只数量管理监测与评估指南 英文版

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

犬只数量管理监测与评估指南 英文版

Uploaded by

Copyright:

Available Formats

ARE WE MAKING

International Companion Animal Management Coalition 2

■ Impact 5: Improve public perception. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

International Companion Animal Management Coalition 3

International Companion Animal Management Coalition 4

AIM OF THE GUIDANCE

PROCESS OF DEVELOPING THE GUIDANCE

HOW TO USE THIS DOCUMENT

Which impacts would you like to achieve through your intervention?

Impact 2 Dog care-giving behaviours in adults Questionnaire surveys

Impact 3 Dog density along surveyed streets Street surveys

Impact 4 Dog bites

Impact 6 Annual live release date

Impact 7 Presence of dogs in wildlife areas

Impact 8 Livestock predation by dogs

International Companion Animal Management Coalition 6

WHAT IS MONITORING AND EVALUATION?

Monitoring requires systematic and routine data collection. Monitoring an intervention

Evaluation of an intervention uses data collected through monitoring, sometimes combined

International Companion Animal Management Coalition 7

Definition DPM example 1 DPM example 2

Before an intervention is launched it is necessary to measure the baseline of indicators

International Companion Animal Management Coalition 8

WHY INVEST IN MONITORING AND EVALUATION?

Test assumptions, Develop concept

Set impacts and

International Companion Animal Management Coalition 9

The importance of monitoring and evaluation to learning cannot be overstated. Many

IDENTIFYING DOG POPULATION IMPACTS

International Companion Animal Management Coalition 10

International Companion Animal Management Coalition 11

Recommended and Suggested

1. Improve dog welfare (animal based indicators)

2. Improve care provided to dogs (resource based indicators)

3. Reduce dog density/stabilise turnover

4. Reduce risks to public health

5. Improve public perception

6. Improve rehoming centre performance

7. Reduce negative impact of dogs on wildlife

8. Reduce negative impact of dogs on livestock

Following each recommended or suggested indicator is a description of the methods of

Physical Health Indicators IMPACT 5

The recommended indicator is the percentage of adults (excluding lactating females)

International Companion Animal Management Coalition 15

International Companion Animal Management Coalition 16

International Companion Animal Management Coalition 17

Emotional well-being indicators

Suggested indicator – Dog-dog interactions

International Companion Animal Management Coalition 18

method of measurement described in section ‘Behavioural observation method’ which

International Companion Animal Management Coalition 19

Note that conducting a questionnaire face-to-face at the household also provides an

International Companion Animal Management Coalition 21

Suggested indicator – Dog care-giving behaviours in children

International Companion Animal Management Coalition 22

Suggested indicator – Owner engagement with intervention

International Companion Animal Management Coalition 23

Recommended indicator - Lactating females IMPACT 3

International Companion Animal Management Coalition 25

Suggested indicator – Pregnant females

犬只数量管理监测与评估指南英文版

犬只数量管理监测与评估指南英文版

犬只数量管理监测与评估指南英文版