University of South Florida
Scholar Commons
Graduate Theses and Dissertations
Graduate School
10-30-2007
A Comparative Study of Healthcare Procurement
Models
Arka Bhattacharya
University of South Florida
Follow this and additional works at: https://scholarcommons.usf.edu/etd
Part of the American Studies Commons
Scholar Commons Citation
Bhattacharya, Arka, "A Comparative Study of Healthcare Procurement Models" (2007). Graduate Theses and Dissertations.
https://scholarcommons.usf.edu/etd/630
This Thesis is brought to you for free and open access by the Graduate School at Scholar Commons. It has been accepted for inclusion in Graduate
Theses and Dissertations by an authorized administrator of Scholar Commons. For more information, please contact scholarcommons@usf.edu.
A Comparative Study of Healthcare Procurement Models
by
Arka P. Bhattacharya
A thesis submitted in partial fulfillment
of the requirements for the degree of
Master of Science in Industrial Engineering
Department of Industrial and Management Systems Engineering
College of Engineering
University of South Florida
Major Professor: Kingsley A. Reeves, Ph.D.
Grisselle Centeno, Ph.D.
José Zayas-Castro, Ph.D.
Date of Approval:
October 30, 2007
Keywords: GPO, healthcare organizations, Wilcoxon, comparative study, DEA, Delphi
method
© Copyright 2007, Arka P. Bhattacharya
DEDICATION
To my parents and my lovely wife for their unconditional support and love. To
my advisor Dr. Kingsley Reeves for his constant guidance, encouragement and for his
excellent mentorship. To Dr. Bob Sullins, Dr. Janet Moore, Ms. Margaret Martinroe and
Ms. Mia Fluitt from the Dept. of Undergraduate Studies for their unstinted support,
encouragement and inspiration.
ACKNOWLEDGEMENTS
I would like to thank Dr. Kingsley Reeves for his guidance, support, belief,
encouragement and patience. I also would like to thank the committee members and
faculty members of the Industrial Engineering Department at the University of South
Florida for their teaching and support. I also would like to thank my colleagues and the
Department of Undergraduate Studies for their assistance in fulfilling my research goals.
TABLE OF CONTENTS
LIST OF TABLES
iii
LIST OF FIGURES
iv
ABSTRACT
v
CHAPTER 1 INTRODUCTION
1
CHAPTER 2
2.1
2.2
2.3
4
5
7
8
OBJECTIVES AND SIGNIFICANCE
Hypothesis
Effects of GPO Sourcing
Broader Impact of the Research
CHAPTER 3 RELATIONSHIP TO CURRENT LITERATURE
3.1
Evolution of GPOs
9
11
CHAPTER 4
4.1
4.2
4.3
4.4
4.5
13
13
14
16
19
20
INSIGHT INTO GPOS
Functions and Services of GPOs
Importance of GPOs in Healthcare Industry
Importance of Innovation/Innovative Products
Impact of Purchasing Groups
Classifications of GPOs
CHAPTER 5 COST COMPARISON ANALYSIS
5.1
Cost Comparison Methodology
5.1.1 Building the Cost Model
5.1.2 Comparison of Unit Overall Cost (Wilcoxon Paired Test)
5.2
Results of Cost Comparison
5.2.1 Procurement Model A versus Procurement Model B
5.2.2 Procurement Model A versus Procurement Model C
5.2.3 Procurement Model B versus Procurement Model C
5.2.4 Summarization of Results of Cost Comparison Study
5.3
Analysis of Cost Comparison
5.3.1 Overall Comparison
5.3.2 Medical Devices Comparison
5.3.3 Surgical Devices Comparison
i
24
24
25
29
34
34
36
38
39
41
41
44
46
CHAPTER 6 MEASUREMENT & COMPARISON OF INNOVATION (DEA)
6.1
Methodology of Innovation Measurement & Cost Comparison
6.1.1 Identifying Innovation Metric
6.1.2 Analyzing Innovation Metric Using Delphi Method
6.1.3 Innovation Metric Scale and Innovation Score
6.1.4 Theoretical Analysis versus Empirical Analysis
6.1.5 Comparison and Ranking Using DEA
6.1.6 Data Envelopment Analysis (DEA)
6.1.7 Selection of Decision Making Units (DMUs)
6.1.8 Simulated Costs of DMUs (Input)
6.1.9 Simulated Innovation Score (Input)
6.1.10 Simulated “No. of Beds” (Output) of DMUs
6.1.11 Selection of (DEA) Model
6.2
Results of Comparison of Access to Innovation with Cost
48
48
49
51
53
54
56
57
59
62
65
68
69
71
CHAPTER 7 CONCLUSION AND DISCUSSIONS
78
REFERENCES
81
APPENDICES
Appendix A Process Map of Self Sourcing Model
Appendix B Process Map of GPO Model
Appendix C Process Map of Hybrid Model
83
84
85
87
ii
LIST OF TABLES
Table 5.1
Wilcoxon Results of Comparison of HC A and HC B
35
Table 5.2
Wilcoxon Results of Comparison of HC A and HC C
36
Table 5.3
Wilcoxon Results of Comparison of HC B and HC C
38
Table 5.4
Cost Efficiency of Procurement Models
41
Table 6.1
Simulated DMUs
59
Table 6.2
Simulated Costs of DMUs (Input)
63
Table 6.3
Simulated Innovation Scores of DMUs (Input)
66
Table 6.4
Simulated Values of Outputs (No. of beds)
68
Table 6.5
Ranking and Efficiency Scores of DMUs
72
Table 6.6
Statistics on Input/Output Data
75
Table 6.7
Projection of DMUs
76
iii
LIST OF FIGURES
Figure 1.1
Healthcare Value Chain
3
Figure 2.1
Procurement Models Used in the Research
5
Figure 4.1
Rationale for Group Purchasing
14
Figure 4.2
Importance of Innovative Capabilities
17
Figure 4.3
Classification of GPOs
21
Figure 4.4
Ranking of GPOs by Contract Purchases and Memberships
22
Figure 4.5
Market-share of GPOs
23
Figure 5.1
Screen Shot of Cost Model Template
29
Figure 5.2
Distribution of Data Obtained from Healthcare Organization C
30
Figure 5.3(a) Overall Comparison Based on Total Price Difference
41
Figure 5.3(b) Overall Comparison Based on Mean Price Difference
42
Figure 5.3(c) Overall Comparison Based on Mean Percentage Difference
42
Figure 5.4(a) Medical Device Comparison Based on Total Price Difference
44
Figure 5.4(b) Medical Device Comparison Based on Mean Price Difference
45
Figure 5.5(a) Surgical Device Comparison Based on Total Price Difference
46
Figure 5.5(b) Surgical Device Comparison Based on Mean Price Difference
46
Figure 6.1
Flowchart Showing Processes of Delphi Method
53
Figure 6.2
Graph Showing Efficiency Scores of DMUs
74
iv
A Comparative Study of Healthcare Procurement Models
Arka P. Bhattacharya
ABSTRACT
Group Purchasing Organizations (GPOs) play a significant role in the healthcare
industry. The presence of GPOs helps the healthcare centers to offload their
responsibilities so that they can focus on more critical areas which require attention like
providing quality care.
This thesis involves the comparison of three models of procurement operations in
terms of cost efficiency. This cost comparison model features a healthcare organization
associated with a National GPO, a healthcare organization which procures by Self
sourcing (not associated with a GPO), and a Hybrid procurement model involving a
National GPO and a regional GPO. The comparison model highlighted the cost
effectiveness of these three different ways of procurement, which threw significant light
on the purchasing operations of healthcare organizations.
In the second part of this research study, we formulated a method to measure the
degree of access to innovative products across the above mentioned procurement models
either involving on-contract (from a GPO) purchasing, or off-contract purchasing (from
v
individual manufacturers not affiliated to GPO) or both. We also identified the metrics
for innovation and measure the innovativeness of products. Based on the literature study,
it was found that purchasing groups may also be an entry barrier to new suppliers (Zweig
1998), with big National GPOs dominating the market and dictating the pricing of
commodities.
The first hypothesis H1 of this research study was stated as “National GPOs
(Group Purchasing Organizations) enable the healthcare establishments to lower the cost
of medical services and operations.”
The second hypothesis H2 of this research study was acknowledged as “National
GPOs a barrier to the entry of innovative product manufacturers in the healthcare
industry.”
This thesis will identify the advantages and disadvantages of each type of
procurement operation and address the economic issues which affect the relationship
between a healthcare center and a GPO. The proposed research would indirectly help to
identify whether cost savings are being shared by the links in the downstream supply
chain and if savings are being percolated to patients for the added welfare of the society.
It will also identify the importance of innovative products in the society and will raise the
bar of specialty treatments without compromising on the level of service being offered to
the patients. This thesis will also highlight positive aspects of niche manufacturers of
innovative products with smaller volumes, currently marginalized in the market by the
big National players.
vi
To the best of the author’s knowledge, the research objective of measuring
innovation of products has not been addressed yet in academic literature and will have the
benefit of comparing three different purchasing models used in healthcare industry.
vii
CHAPTER 1 - INTRODUCTION
Group Purchasing Organizations have become very significant in the healthcare
industry. The GPOs (also called purchasing groups) have mostly become popular in
healthcare, education and government organizations. The healthcare industry is faced
with the constant pressure to cut down costs and stiff competition among healthcare
centers which have led to mergers and acquisitions resulting in suppliers of larger size.
The most frequent reason given by a healthcare center to be affiliated with a GPO is
advantageous contractual conditions. The modern GPOs have changed the conservative
method of procurement. The huge pressure of lowering the prices has mostly been
beneficial to the end user which in our case is a customer to a healthcare center.
A GPO is a formal and virtual structure that facilitates the consolidation of
purchases for many organizations (Nollet 2005). The outsourcing of purchasing to GPOs
has facilitated healthcare centers to focus on their critical areas like providing healthcare.
This has taken off the burden of purchasing operations which many healthcare centers
used to face previously. It has been estimated that more than 70 percent of the healthcare
purchases are done through group purchasing (Nollet 2002). These purchasing groups
have a stronger negotiating capacity in dealing with their suppliers and have the
necessary volume to support, which lowers the cost of commodities (standardized
objects). Purchasing groups empower their members in negotiation and create favorable
1
conditions for their members. However, the price advantages are greater for larger GPOs
as they have more negotiating capacity. There is also a general agreement that GPOs
generate savings between 10 and 15 percent amounting to $12.8 billion to $19.2 billion
(Hendrick 1997) and (Schneller 2000). Thus, it is quite evident that in the healthcare
industry, the existence of GPOs cannot be ignored.
According to a recent Health Industry Group Purchasing report, goods and
purchased services accounted for the second largest dollar expenditure (55% labor and
45% non labor supplies, services and capital equipment) in the hospital organization
(Schneller 2000). The main rationale for group purchasing is to achieve lower prices,
ensure price protection, implementing improved quality programs, reduced contracting
costs and monitoring market conditions (Schneller 2000). Estimates place the GPO
market for healthcare organizations and nursing homes at between $148 and $165 billion
dollars and growing to $257 and $287 billion per year by 2009 (Hewitt 1995). It is also
noteworthy that 72 to 80 percent of every healthcare (acute care organization) supply
dollar is acquired through group purchasing (Schneller 2000).
In addition to purchasing options, GPOs offer information sharing, clinical and
operational benchmarking and value assistance benchmarking that could strategically
differentiate GPO members in their market (Schneller 2000). Schneller also stated from a
report that product standardization and entering into GPO contracts were the most
effective cost reduction strategies (Schneller 2000). In choosing to contract with a GPO, a
company must evaluate the performance of its suppliers with that of the GPO’s
performance in terms of purchasing power (Schneller 2000).
2
The purchasing groups facilitate their members to get more favorable conditions
than they would have obtained individually (Rozemeijer 2000). The administrative costs
also get lowered due to the fact that a single organization performs the negotiations
instead of many. There are two types of structures among GPOs. The first type is
cooperative structure where the purchases to be performed by the group are distributed
among members. Second type of structure is the third party structure where a distinct
organization negotiates and writes contracts according to a mandate given by the
members (Hendrick 1997). The healthcare value chain is shown below in figure 1.1.
Figure 1.1 Healthcare Value Chain
3
CHAPTER 2 - OBJECTIVES AND SIGNIFICANCE
In this thesis our main objective is to compare the three models of procurement
operations in terms of cost effectiveness by capturing the purchasing costs per unit of the
commodities/items procured by different healthcare organizations. This comparative
study focuses on three scenarios featuring a healthcare organization which procured
through a National level GPO, a healthcare organization which procured by Self sourcing
and a Hybrid procurement model (comprising a National GPO and a regional GPO). In
this study, items procured by the healthcare organizations are classified into 2 main
categories, medical devices and surgical devices. These items are common items required
in daily operations and procured in bulk quantities by the healthcare organizations. The
procurement costs of these items are highlighted through this comparative study. Apart
from supplying commodities and surgical instruments to healthcare organization, many
GPOs are now focusing on diversifying their product range and providing additional
support services like maintaining medical records and training hospital employees in new
technologies. This has led to an overall growth of technology and made the daily
operations more efficient. Our comparison model evaluates and highlights the economic
benefits of these three different ways of procurement, which will throw significant light
on the purchasing operations of healthcare organizations.
4
Figure 2.1 Procurement Models Used in the Research
We have also captured the degree of access and compared the cost associated with
the procurement of innovative products through a National GPO, a Hybrid model, and
Self sourcing. This is done by firstly identifying metrics for innovation and formulating a
technique to measure the degree of innovation of products sourced from healthcare
organizations involving a National GPO model, a Self sourcing and a Hybrid one either
involving on-contract (from a GPO) purchasing, or off-contract purchasing (from
individual manufacturers not affiliated to a GPO) or both. The measure of innovation is
tied to the cost of the products and both the factors are used in the comparison of models.
This may highlight the fact that when it comes to innovative products (for example
pacemakers), whether a procurement model is rated higher in terms of innovation and
also whether the advantages of low cost outweigh the advantages of innovation.
2.1 Hypothesis
Based on our literature review, it was found that purchasing groups may also be
an entry barrier to new suppliers (Zweig 1998). Big National GPOs provide commodities
at a much lower price due to large volumes, which gives an advantage to existing
5
suppliers, since suppliers with innovative products do not have sufficient sales volume to
allow them to take advantage of economies of scale and to offer competitive prices
(Elhauge 2002).
Formally, the hypothesis can be stated as follows:
H1 - National GPOs (Group Purchasing Organizations) enable the healthcare
establishments to lower the cost of medical services and operations.
H2 - National GPOs a barrier to entry of Innovative product manufacturers in the
healthcare industry.
This proposed thesis will throw light on the pros and cons of each type of
procurement operation. There is a strong need to address these economic issues as they
will affect the relationship between a healthcare center and a GPO. These factors can
affect the consistency of healthcare delivery quality which will have a social impact. Our
thesis will highlight positive aspects of niche manufacturers of innovative products with
smaller volumes which are currently marginalized in the market by the big National
players. Most of the big players in the GPO market are driven by costs and big volumes
and they sideline the smaller players which manufacture innovative products (Zweig
1998; Everard 2005). One added advantage of our project will be that it will help the
specialty healthcare organizations realize the importance of innovative products and help
them choose the procurement model for innovative products which will be most cost
effective. This will help to reduce the cost of innovative products in the market and help
the patients in accessing high end products at reasonable price. To these healthcare
organizations, technology of the products will be of higher priority which will help to
raise the quality of specialty healthcare services. To the best of the author’s knowledge,
6
the research objective of measuring innovation of products has not been addressed yet in
academic literature and will have the benefit of comparing three different purchasing
models used in healthcare industry.
2.2 Effects of GPO Sourcing
GPOs have a large effect on the healthcare industry. Based on our study, we have
found that there can be positive as well as negative effects when a healthcare organization
affiliates with a GPO. However, since the minimization of cost is a top priority for many
healthcare organizations, these negative effects are sometimes overshadowed. With
affiliation to a GPO, healthcare organizations enjoy lower prices, protected pricing,
improved quality control programs, reduced contracting costs and GPOs also monitor
market conditions (Schneller 2000). This price protected market gives healthcare
organizations some form of security against price fluctuations. However, along with these
favorable conditions there are quite a number of cons associated. Affiliation to a GPO,
reduces the autonomy of an individual healthcare organization and often it gets bound by
a contract and cannot come out of it (Nollet 2005). This may lead to dissatisfaction among
some physicians who wish to maintain their autonomy in choosing products, and may
sometimes circumvent the contract terms of the GPO to access those products (Burns
2002). This reduces the overall cost savings of the GPO in the long run. GPOs also create
an entry barrier to small innovative product suppliers, who cannot compete with large
volume existing suppliers due to small volumes (Zweig 1998). This may affect quality of
specialized commodities in the long run and result in dissatisfied customers. Thus, the
cost associated with loss of business has to be considered. Due to the large size of GPOs
7
and long term contracts with the suppliers, few big players dominate the market and many
small players have complained about the lack of competitive access (Burns 2002). This
creates an oligopolistic market scenario, and big players end up dominating the market,
while the smaller ones are marginalized.
Close cooperation among healthcare competitors sourcing from the same GPO
also gives rise to anti-trust issues(Nollet 2005). Sometimes members do not want to share
sensitive information with their rivals.
Sometimes National players are not able to deliver their products on time due to
logistic problems and during calamities. This causes unscheduled delays to patients and
the service quality of the healthcare organization suffers. In this aspect, local or regional
suppliers are sometimes better off as their logistic operations prove to be better. Many
local players also share their warehouses with their clients which may help in product
delivery and reduce overhead. This is our added research objective and will determine
innovation metrics to analyze whether regional sourcing improves quality of products and
results in a superior distribution model.
2.3 Broader Impact of the Research
The proposed research would indirectly help to identify whether cost savings are
being shared by the links in the downstream supply chain and if the savings are being
percolated to patients for the added welfare of the society. It will also identify the
importance of innovative products in the society and will raise the bar of specialty
treatments without compromising on the level of service being offered to the patients.
8
CHAPTER 3 - RELATIONSHIP TO CURRENT LITERATURE
One of the most common issues dealt in the past and current literature is about the
optimal size of GPOs and the benefits which healthcare organizations gain by affiliating
with a GPO. There is an overall consensus that affiliation with a GPO indeed results in
cost savings. The past literature has dealt with issues like size of purchasing group and
the types of benefits which can be extracted with affiliation to a GPO (Nollet 2005). This
study was based mostly on the interviews with health managers. Jean Nollet and Martin
Beaulieau identified the different aspects of a relationship with a GPO. The paper
evaluated the impacts of a GPO on a supply market. The issue related to the size of a
GPO and its effects on the buyers and the suppliers were also discussed. They further
went on to discuss the member characteristics and the issues faced by them.
M. Essig described the concept of group purchasing as “Purchasing Consortium”
and has introduced it as a supply management concept combining symbiotic horizontal
relationships and strategic understanding to gain competitive advantage (Essig 2000).
This paper focused on the symbiotic relationship among the members in a similar
hierarchy level. This literature further described and classified sourcing options available
to a GPO and illustrated the benefits associate with each type.
Member commitment has a huge role to play in the success of a GPO and also the
growth of member enrollment depends on it (Nollet 2002). W. R. Doucette pointed out
9
that the transparency in sharing of information between the members and the trust issues
shape the success of GPOs by creating a strong member commitment (Doucette 1997).
Member commitment is influenced by other members to a great extent.
The major chunk of literature has dealt with the identification of costs and the cost
saving benefits enjoyed by the healthcare organizations (Schneller 2000), (Rozemeijer
2000) and (McFadden 2000). Any healthcare center affiliated to a GPO benefits from
three types of cost reductions: price, administrative costs and utilization costs (Anderson
1998). As has been discussed earlier that affiliation to a GPO can generate savings up to
10 to 15 percent which is a direct cost savings (Hendrick 1997) and (Schneller 2000). The
healthcare organizations can utilize the savings generated in more vital areas which relate
directly to technical quality (quality of healthcare delivery).
Chapman mentioned that the real savings in the healthcare savings come from
product standardization (Chapman 1998). However certain types of purchases like
commodities are suited for larger savings and standardization may be enforced by certain
purchasing groups by forcing healthcare centers to use all the products in the package
(Nollet 2005). A significant amount of study has been carried out previously about the
role of GPOs and identifying the economic costs and its impact on the entire supply chain
(Schneller 2000).
However, based on our literature research, it is found that there has been a dearth
of work related to the comparison of procurement models through Self sourcing, National
GPO sourcing and regional GPO sourcing. Most of the earlier or present literatures have
identified economic and non economic costs associated with a GPO (Dobler 1996;
Anderson 1998; Chapman 1998; Schneller 2000), but there has been no direct
10
comparison between three different procurement models. Also most of the earlier
literatures have just mentioned non economic costs like loss of autonomy of physicians
and barrier to entry of innovative products (Zweig 1998) and (Elhauge 2002). Our current
research focus will be to address this issue of access to innovative products while
sourcing from a GPO. This thesis will identify innovation metrics to evaluate the degree
of innovation in products and will also illustrate which procurement method gives access
to the highest level of innovative products and at the same time keeping the cost to a
reasonable level.
Based on our literature research, it can be said that this proposed research topic is
unique because it identifies innovation metrics to analyze the degree of innovation in a
particular item which again reflects the procurement model. To the best of our knowledge
their has been no scholarly work which deals with the comparison of procurement
processes associated with a GPO and measures a product’s innovativeness by analyzing
particular metrics.
3.1 Evolution of GPOs
The concept of GPO took birth way back in 1910 in New York with the formation
of Hospital Bureau of Standard Supplies of New York (Barlow 2005).
However, with the advent of 1970s, the formation and growth of GPOs really
took shape and the regional groups gave in to National level organizations. Almost close
to 37% of the purchasing groups were set up during this period (Nollet 2002). This
sudden surge in the number of purchasing groups was due to the increased government
pressure on cost reduction.
11
This augmentation in the number of GPOs resulted in stiff competition among the
members in a price sensitive market. GPOs could not enlarge by adding members as most
of the healthcare organizations were already serviced by one of them (Nollet 2002). To
stay afloat with the competition, GPOs started extending additional services which
facilitate the operational efficiency of a healthcare organization. These services include
consulting, contact management, human resource management, computing services, etc.
The consolidation of purchasing groups began in the 1990’s. This age was the age
of mergers and acquisitions. For example, Novation resulted from the merger of VHA
and HealthSystem Consortium (Doucette 1997).
12
CHAPTER 4 - INSIGHT INTO GPOS
4.1 Functions and Services of GPOs
A GPO is a group of organizations which consolidate their resources to have more
leverage on their suppliers. Based on past literature, it is seen that a GPO’s procurement
strategy creates better operational links with the suppliers, shorter lead times, and creates
more competition in the market. This favors the end user in terms of lower costs and has
a socio economic benefit too. GPOs also provide joint purchasing programs to clinicians,
and other healthcare entities.
The areas where GPOs use their influence in negotiating the prices are pharmacy,
laboratory, diagnostic imaging, office facilities, dietary, maintenance, IT, and insurance
(Burns, 2002). Apart from negotiating prices, a GPO serves as an instrument for price
protection for its members as it functions like a link between the large number of vendors
and the healthcare organization.
In past, GPOs offered their members the same standardized pricing irrespective of
the volume they purchased. This proved to be advantageous to the smaller members and
the bigger players felt that they had to bear the burden of subsidizing the smaller ones
(Burns 2002). However, the concept of “tiered pricing” (Burns 2002) has come into
effect recently which relates the pricing of products to the volumes members purchase.
Affiliation to a GPO is a “win-win” situation for both the healthcare organization
as well as the GPO. The healthcare organization gets the benefit of cash incentives, cost
savings and outsourcing of purchasing function to a third party which in turn helps them
13
to streamline their operations. GPOs gain by the added negotiating capacity in controlling
the price of the products and also the contract administration fees paid by the vendors.
Apart from controlling and lowering of costs, GPOs offer additional services to
their members like materials management, contract management, operations consulting,
programs to improve product standardization, insurance services, technology
management programs, disease management, human resource management, education,
and marketing (Burns 2002). Figure 4.1 briefly summarizes the justifications as
mentioned in the past literature for a healthcare organization to be affiliated to a GPO.
Figure 4.1 Rationale for Group Purchasing
4.2 Importance of GPOs in Healthcare Industry
The process of sourcing through GPO in healthcare industry is mostly seen for
non critical items where the level of customization is almost non existent. Most
commonly, commodities which are required in bulk quantities and other pharmaceutical
14
products along with office supplies are sourced through GPOs. Over the years it’s been
seen that this model of procurement through GPO has seemed to be more cost effective
and more and more healthcare organizations are getting affiliated to big National GPOs.
The most distinctive factors which have contributed to the rise of GPOs are high volume
of commodities being sourced which has helped to lower the price of commodities and
the standardization of products being sourced. The levels of customization of products
sourced through GPOs have been minimal. According to Dobler and Burt (Dobler 1996),
evidence is plentiful that simplification or standardization can result in big savings.
Standardization and simplification is also the focus of major efforts in healthcare
providers as a tactic for reducing costs (McFadden 2000). This is where the GPOs have a
substantial advantage over smaller players in the industry. This leads them to dominate
the market.
In healthcare industries, since GPOs are mostly concerned with non critical items,
issues like loss of confidentiality are not important. Innovation and technology are not
given high importance in the business of non critical items.
Unlike other verticals, healthcare industry has socio-economic obligations.
Providing quality healthcare at a low cost has been one of the challenges of the modern
healthcare industry. Since it affects the medical services to the common man, controlling
the cost becomes a crucial factor. GPOs have played a big role in this by providing
commodities at very low costs. This has enabled the healthcare organizations to offer
medical services to the common man at reasonable prices and has raised the standards of
healthcare quality in the country. Moreover, GPOs also offer other services like data
warehousing, information technology services, and training of staff in the latest
15
technologies to the healthcare establishments and healthcare organization, which have
taken off the workload out of these healthcare organizations and helped them focus on
more critical issues like operational issues.
4.3 Importance of Innovation/Innovative Products
Based on our research, it can be said that most of the items procured through a
GPO are commodities which are pretty simple items and require little or no innovation.
These items are sourced in bulk quantities and are supplied at a very low cost. However,
it is also seen that healthcare organizations source few type of items like pacemakers and
other surgical quantities which are sophisticated and require a high degree of innovation.
Though the volumes of these advanced items are quite low compared to the bulk
quantities, they have a huge impact on the quality of specialty care. These specialized
items require sustained innovation in their lifecycle which is essential for their existence.
Many of these state of the art items are manufactured and supplied by niche
manufacturers who do not have a strong influence on the National GPOs and cannot
deliver at a rock bottom price to the market because of their low volumes. Sometimes
these manufacturers are sidelined and the market dominated by the big National GPOs
creates a barrier for their entry in the business. According to Muller, “companies must
exploit their innovative capabilities to develop new businesses if they are to successfully
confront the disruptive effects of emerging technologies, empowered customers, new
market entrants, shorter product life cycles, geopolitical instability, and market
globalization” (Muller 2005).
16
Geopolitical Instability
Counters
Innovative
Capabilities/ Sustained
Innovation
Globalization threat
Shorter product Life
cycles
New market entrant threats
Emerging Capabilities
Figure 4.2 Importance of Innovative Capabilities
Source: Adapted from Strategos et (Muller 2005)
Innovation management can be also defined as coping with rapidly changing
environment or in turbulent environments. Calantone, Garcia and Droge define turbulent
environments as those in which market needs or technology are uncertain and have
impact on new product development processes (Buganza 2006). Further this paper
discusses that to manage turbulent environments, companies have to reduce the
development time and increase the ability to react to changes. That is, a product must
have high degree of “Life Cycle Flexibility”. Life Cycle Flexibility of a product is the
ability to introduce innovations during life cycle processes at a low cost and shortest
time.
This allows the product to adapt and be redesigned according to contextual
changes and opportunities, i.e., flexibility after the product has been released. Such
examples are quite common in the industry and can be clearly seen in the automotive
industry. In the automotive industry, cosmetic changes to a product happen often within a
17
product life cycle based on changing customer views and perceptions. Some cars even
undergo major changes in the form of engine capacity and technology to cater to the
customer demands.
According to Tommaso Buganza and Roberto Verganti, the metrics for LCF (Life
Cycle Flexibility) are (Buganza 2006) :
1. Frequency of adaptation: Number of new features per unit time.
2. Rapidity of adaptation: Inverse of the time needed to adapt to the service/product
as a reaction to the launch of the new feature by the competitor = (1/Time needed
for reaction).
3. Quality of adaptation: Ability to be consistent with quality through different
service package adaptations such as robustness as a dimension of quality.
In our proposed thesis we have identified certain metrics for innovation which
were proposed in a more generic way by Amy Muller, Liisa Valikangas and Paul Merlyn
(Muller 2005) to cater closely to the healthcare industry and which will make it more
feasible for data collection. These innovation metrics are essentially product based and
the ones which are the most appropriate and will make data collection feasible will be
considered. The innovation metrics we have identified for data collection are (adapted
from (Muller 2005)) :
1. Measure R&D budget as a % of annual sales of a particular product.
2. No. of patents/new ideas filed by the company in the last year/last month.
3. Measure % of capital that is invested in radical projects.
4. Average time required from idea generation to product/service launch.
5. Ratio of revenue from innovative/new products to commodities.
18
6. Measure % of employees that are involved in developing an innovative product.
7. Measure innovation revenue per employee from the new product/service
developed.
8. % of management that is accountable to development of a new product in terms of
time (man hour).
9. No. of incentive schemes to support innovation.
Out of these metrics only few will be considered based on accessibility of data from the
three different sources.
4.4 Impact of Purchasing Groups
The purchasing groups play a very important role in manipulation of the
commodity prices. The bigger players in the market exploit their leverage with the
supplier resulting in the wiping out of the smaller players who lack the negotiating
capability. Due to the concentrated market share by the big dominant players, entry of
small players in the market becomes very difficult (Sethi 2006). It is prohibitively
expensive for a new entrant to gain significant market share because most current and
potential customers are already locked in to existing GPOs through various contractual
arrangements (Sethi 2006). The extermination of smaller players from the market creates
an oligopolistic market scenario where the bigger ones sometimes dictate terms to the
buyers as well as the suppliers. This sometimes results in poor service quality by the
healthcare organizations. The growth of bigger purchasing groups may result as an
advantage to the existing suppliers as smaller volume suppliers may lose out in a price
19
conscious market even though their product may be technologically superior. This affects
the quality of products in the long run.
4.5 Classifications of GPOs
GPOs can be classified based on their ownership, membership, geographical
scope, and size (Burns 2002). When GPOs are classified based on ownership, they are
distinguished as for-profit, non-profit and public GPOs (Burns 2002). The two largest
for-profit GPOs are divisions of the two largest investor owned hospital systems: HCA (
Health Trust Purchasing Group) and Tenet (BuyPower) (Burns 2002). The three largest
non profit GPOs are hospital cooperatives like Novation, which is a group purchasing
arm for VHA/UHC (Burns 2002). The largest public GPO is the VA. Healthcare
organizations which are a part of for-profit and the VA systems are more committed to
their group purchasing contracts. Healthcare organizations within the non profit alliances
join their GPOs voluntarily (Burns 2002).
GPOs also differ in the type of membership. Some GPOs are committed to the
larger healthcare organizations where as some of them focus on smaller buyers like
ambulatory centers and physicians offices. Many GPOs try to focus on two types of
market in order to have a stronger presence (Burns 2002).
Many GPOs differ in their reach to cater to different markets. Some or rather
smaller players focus on regional healthcare organizations. This helps them to consolidate
their resources and sometimes perform better in logistical operations than National
players. Large GPOs generally focus on a National level. They have better reach which is
facilitated by their financial muscle and volumes of purchases. These GPOs sometimes
20
result in extermination of regional players which has been discussed earlier. Figure 4.3
briefly summarizes the classification of GPOs.
Figure 4.3 Classification of GPOs
21
Figure 4.4 Ranking of GPOs by Contract Purchases and Memberships
22
Figure 4.5 Market-share of GPOs
23
CHAPTER 5 - COST COMPARISON ANALYSIS
In this project, we would like to focus on two aspects of healthcare industry. The
first aspect would be comparing the prices of the bulk items across three different
procurement models. The second aspect would be to compare the degree of access to
innovative items across different procurement model using a technique called data
envelopment analysis (DEA). The second part of the research is explained in chapter 6.
5.1 Cost Comparison Methodology
Firstly, we would like to capture the total cost of procuring items through three
different procurement models. This involves comparison of procurement models of
commodities through a Self sourcing unit, a National GPO and a Hybrid model. A Self
sourcing healthcare organization procures items through individual contracts with
vendors and manufacturers. A healthcare organization affiliated to a National GPO
procures most of its items through the GPO and is bounded by GPO contracts. These
organizations also have to take into account the mandatory compliance rate sometimes
being enforced by some GPOs. It is important to note here that during the course of
interaction with the staff of healthcare organizations under consideration in this study, the
compliance rate was found to vary among GPOs ranging from 60 percent to 90 percent.
The Hybrid model in the study features healthcare organizations which are affiliated to a
National GPO as well as to a regional GPO. In this model, the healthcare organization
24
procures items from a National GPO as well as a regional GPO and has the flexibility to
choose products from either of them depending on the lower prices. This chapter will
include a comparison of the costs associated with these procurement models. In order to
achieve this, a clear understanding of the series of processes and operations undertaken in
each of these procurement models is required. This involved mapping out the entire
process/operation in the form of a flow diagram (Process Map) with each operation
described briefly and the resources associated with it for each of the procurement model
(Please refer appendix A, B and C).
5.1.1 Building the Cost Model
The cost model was designed to capture an estimated overall price/cost of the
items which includes the overhead. Please refer figure 5.1 on page 29 which displays the
screenshot of the MS Excel based model.
Primary Overhead includes the human resource cost only which means the salary
of individuals involved in the purchasing operation or part of their function relates to
purchasing. Secondary overhead comprises of the administrative fees paid by the
healthcare organization to the GPO (applicable only for GPO members) and the rebates
gained by the hospital from the GPO due to various reasons like compliance/loyalty etc.
Primary Overhead has been classified into seven types:
1. Legal staff - negotiates the contracts with the supplier.
2. Follow up staff - checks the price and the quality of the suppliers (gets the price
quotes) and chooses the suppliers.
25
3. Administration staff - works on the purchase orders and sends the orders to the
suppliers.
4. Inventory staff - maintains the inventory and works on them and notifies when
there is some short fall of some items.
5. Finance staff - processes the funding associated with funding and releases the
money.
6. Stocking staff - manages the stocking of the products in the warehouse after they
are obtained from the suppliers.
7. Transportation Staff - manages the transportation of products/items within the
campus.
This is a general classification and only those teams/buckets which are applicable
to a particular hospital or healthcare organization are taken into consideration. For
example: There may be legal staff involved in the procurement of items for hospital A
whereas it might be absent for hospital B. In calculating the overhead, the average annual
salaries of the individuals or the titles they represent are taken. Once the final annual
overhead is calculated (sum of the annual average salaries of all the individuals
involved), it is then calculated per day by taking the number of work days in a year as
260.
Also, it should be noted here that the proportionate salary of the average annual
salary should be taken into comparison. If a staff has a fraction of the responsibility
involved in procuring operations, then that fraction should be multiplied with the annual
average salary and then that amount should be filled in as the annual average salary. For
26
example, if staff A earns $50,000 per year and 50% of his job responsibilities fall in the
procurement operations, then .50* 50,000=25,000 will be his annual average salary.
Secondary Overhead has been classified into:
1. Administrative fees - The fee paid by the GPO members to the GPO on an annual
basis. This fee is only applicable to the healthcare organization associated with
the GPO. This amount calculated on a yearly basis is added to the Total Overhead
in a year.
2. Rebates - The money paid to the GPO members by the GPO for a variety of
reasons. This could be loyalty of the member to the GPO or for maintaining good
compliance rates or sometimes to clinch deals with the members in a very
competitive market. This amount taken annually is subtracted from the Total
Overhead calculated annually.
Spending per category is defined as the total amount spent on each category like
drugs, office supplies, medical devices, etc. We have coined a term “ICV” which is the
“Total average cash value of the items in the inventory per day (ICV)”. The ICV of the
four categories in consideration are finally added to get the total amount spent or total
ICV per day.
The cost model is then built on the “average daily inventory” of the items (for
example: 1, 3, 6, 10, 200, etc.) which are procured and their “standard price” in the next
column in the spreadsheet. The product of these two will give the “Total $$ amount/day”
of products in the column next to the standard price in the spreadsheet.
“Spending % Value” is the percentage of “Total $$ amount/day” spent on each
item to the total amount spent or “total ICV per day”.
27
“% Overhead” is the product of the “Spending % Value” with “total
overhead/day” to get the % of overhead added in every product's cost.
The sum of “% Overhead” and the “Total $$ amount/day” will give an estimated
“Overall Cost” of each item. This “Overall Cost” is then divided by the “average daily
inventory” to finally calculate the estimated “Unit Overall Cost” for each of the items.
The estimated “Unit Overall Cost” for each item will be taken for comparative study. It is
this cost which will be used for the comparison of each item across the different
healthcare organizations.
Maintaining the confidentiality of data has been given the utmost importance in
this thesis. This cost model based on the MS Excel was created to get an estimated
pricing of the products/items, since information regarding actual pricing of the products
could not be accessed by us. The Excel sheet has been designed in a way that the items
which are commonly procured are divided into two categories (Please refer figure 5.1 on
page 29). They are medical devices and surgical supplies. The excel model would be
populated with 50 to 100 items for each of the above mentioned categories for the
procurement models (with the help of information accessed from each of these three
healthcare centers respectively) being in operation in the healthcare organizations
considered in the study. The items in each of those two categories must be common to all
the three healthcare centers having different procurement models.
To maintain the confidentiality, the data concerning overhead, spending per
category, standard price and daily inventory are entered by the staff from the hospital.
After the data was filled in these shaded cells (refer to the screenshot), the spreadsheet
would automatically calculate the overhead, overhead% and finally the “unit overall
28
cost”. Also, to maintain error proofing, the cells other than those shaded ones (where the
data is entered by the hospital staff) were formulated and locked. This is the column we
were interested in and this column “unit overall cost” and the “products/items” column
were then copied and pasted in a different excel sheet and sent to us. We had no access to
the actual pricing information and actual salary figures which are confidential.
Figure 5.1 Screen Shot of Cost Model Template
5.1.2 Comparison of Unit Overall Cost (Wilcoxon Paired Test)
The “unit overall cost” of each of the items in the two categories, i.e., medical
devices and surgical devices were used for the comparison model to compare the prices
of the bulk items.
The categories medical devices and surgical devices were taken into
consideration, because these were the two categories of products which were common
across the three healthcare organizations in this comparison study. During the course of
29
this research, we were assisted by the staff from the materials management department of
all the three healthcare organizations. Also, there was no data available for the
procurement of office supply equipment. Among the inventory of bulk items which were
procured by the three healthcare organizations, we could get 222 items which were
common across all the three healthcare organizations. This number could be further
broken down into 156 medical device items and 66 surgical devices items. Utmost effort
was made to match the products having similar generic names and features. The data
obtained from the three healthcare organizations were non-parametric in nature which is
shown in the figure 5.2 below.
5
Frequency
4
3
2
1
0
46.041068
28.225177
24.256975
21.218927
20.097927
16.514731
14.082562
13.572106
13.211785
12.521169
11.11778
9.888821
8.894224
8.137258
8.007142
7.396598
6.796062
6.365678
6.035384
5.164607
4.80086
4.1439
3.913491
3.234263
3.032122
2.962643
2.792491
2.502232
2.201964
1.991093
1.77158
1.40125
1.161036
1.000893
0.920821
0.725647
0.680607
0.570509
0.47042
0.400357
0.323426
0.230205
0.161713
0.050045
HC Setting C
Figure 5.2 Distribution of Data Obtained from Healthcare Organization C
Since prices obtained from the three healthcare organizations do not follow a
normal distribution, we considered implementing Wilcoxon signed-rank test.
30
Wilcoxon signed-rank test is a non parametric alternative to the paired Student’s
t-test for comparison of two related samples of data. In this study, we are comparing the
prices of same bulk items (paired) across two healthcare organizations at a time, i.e., first
we will be comparing the prices of bulk items between healthcare organization A and
healthcare organization B, followed by comparison between healthcare organizations A
and C, and between B and C. Wilcoxon test is used to compare differences between
measurements at interval levels. This enables to compare differences between arbitrary
pairs of data.
1. Wilcoxon test assumes that the difference between two samples of data,
i.e. di = Ai – Bi for i = 1 to…..n. are simulated to be independent.
Where A and B are two related samples of data and d is the difference between
these two samples at each measurement.
2. Each difference di is drawn from a continuous population.
Testing the null hypothesis (for a paired t-test):
Ho: μd = 0
H1: μd ≠ 0
Where μd is the mean difference between the measurements. In a paired t-test, the
null hypothesis Ho: μd = 0 will be rejected if the mean difference between the sample
measurements is not equal to zero. However, the null hypothesis in Wilcoxon test is that
the median difference between pairs of observations is zero. By testing and rejecting the
null hypothesis, it can be shown that the data samples do not have the same median and
are drawn from different populations. This is done by ranking the absolute value of the
differences between observations from the smallest to the largest, with the smallest
31
difference getting a rank of 1, followed by the next larger difference getting the 2nd rank,
etc. Ties are given average ranks. The ranks of all differences in the (positive) direction
are summed, and the ranks of all differences in the negative direction are summed. In this
study, only after the null hypothesis in the Wilcoxon test has been rejected, the cost
efficiency of one procurement model can be compared with the other by measuring the
mean difference of prices and the mean percentage differences of prices (difference
between the prices of commodities and expressed as a percentage of the price of the
commodity from which it is subtracted). If the null hypothesis is not rejected, then the
mean difference of the prices and the mean percentage difference between the samples
will be zero and the cost efficiency of one procurement model versus the other cannot be
determined.
For example: If we compare procurement models of healthcare (now onwards we
would call healthcare as HC) organization A versus HC organization B, then the mean
difference of prices of all the bulk items is calculated, i.e., we are subtracting the prices of
the bulk items of HC organization B from HC organization A. If the mean difference is a
positive value, then we can conclude that HC organization B is more cost efficient than
HC organization A or vice versa. This will enable us to rank the different procurement
models in terms of cost efficiency. In this study, comparison model will comprise of HC
organization A versus HC organization B, HC organization A versus HC organization C
and HC organization B versus HC organization C, with the latter models’ procurement
prices subtracted from the former ones.
Similarly, the percentage difference between the prices of all the commodities are
calculated, and if the mean percentage difference of all the commodities in comparison is
32
zero, then the two procurement models cannot be compared. If it’s not zero but a positive
or a negative value then there exists a difference and it can be determined which
procurement model is more cost efficient. When HC organization A and B are compared
the percentage difference is calculated by
% difference = ((Cost A – Cost B)/Cost A)*100.
The mean percentage difference is given by = (∑in % difference)/n (where n = 222).
The null hypothesis is tested by the relations below.
(Adapted from http://www.nist.gov/speech/tests/sigtests/wilcoxon.htm):
The mean is given by
μ=
n(n + 1)
4
The variance is given by σ 2 =
Eq. 5.1
n(n + 1)(2n + 1)
24
Sum of the positive and negative ranks Z + + Z − =
Eq. 5.2
n(n + 1)
2
Eq. 5.3
Where Z+ and Z- are sum of positive and negative ranks respectively.
The test statistic W is
W=
Z+ − μ
Eq. 5.4
σ
The null hypothesis in the Wilcoxon test will be rejected if
P(W) ≤ α value
Eq. 5.5
Where α = P(type I error) = P(reject Ho|Ho is true), is the significance level.
The assumption here is that α, the significance level of the test is simulated to be
0.05. Therefore the percent confidence interval of the test is 100(1 – α) = 95.
33
As discussed earlier, only if the null hypothesis is rejected, we can compare the
cost efficiency between the procurements models and rank them accordingly from the
most to the least.
The procurement model with the lowest price in the comparison study will
emerge as the most economical leader in procurement operations and also help to
determine whether GPOs are the most economical ways of procuring items.
5.2 Results of Cost Comparison
As mentioned in the previous section, the cost comparison study was performed
by matching almost exact bulk items across three different procurement models, i.e., Self
sourcing, GPO model and Hybrid model. All in all, 222 bulk items were found common
across these three procurement models which can be further classified as 156 medical
devices and 66 surgical devices.
5.2.1. Procurement Model A versus Procurement Model B
In this cost comparison model we compared the procurement model of HC
organization A versus HC organization B, i.e. the comparison of “Self source” model
with that of “National GPO” model. The prices of each of the bulk items (totaling 222
items) of the HC organization B are subtracted from those of HC organization A to get
the differences at each sample point. Using the assumptions and equations mentioned in
section 5.1.2, we get
n = 222.
The mean is given by Eq 5.1
μ=
n(n + 1)
= 12,376.5
4
34
The variance is given by Eq. 6.2
σ2 =
n(n + 1)(2n + 1)
= 917,923.75
24
After running Wilcoxon test in the MS excel solver for the sum of ranks, the results are
displayed in table 5.1
Table 5.1 Wilcoxon Results of Comparison of HC A and HC B
Differences
N
Rank-Sum
Negative
Positive
Zero
59
163
0
6991
17762
P
1.89725E08
From table 5.1 it can be said the number of positive ranks are higher than negative
ranks. Thus, sum of positive ranks and negative ranks Z + + Z − =
n(n + 1)
= 24,753. The
2
Z+ and Z- values are 17762 and 6991 respectively. Thus test statistic given by Eq. 6.4
W=
Z+ − μ
σ
= 5.621.
Now, P(5.621), i.e. the significance of the difference from the table 7.1 as given
by the solver is 1.89725E-08. Thus using Eq. 5.5, the null hypothesis in the Wilcoxon test
will be rejected as: P(W) ≤ α value = 1.89725E-08 < 0.05.
Since the null hypothesis is rejected, two procurement models can be compared
against each other based on their cost efficiency. The mean difference when the prices of
commodities of HC organization B (GPO model) are subtracted from those of HC
organization A is $2.77, which is a positive value. The mean percentage difference of the
prices of the commodities is 6.17 percent, again a positive value. Thus, we can say that
35
the average prices of commodities procured by the Self source model are more than the
GPO model as shown by a positive value of mean difference and mean percentage
difference. Thus, in this comparison model the GPO procurement model of HC
organization B is more cost efficient than that of Self sourcing model of HC organization
A.
5.2.2. Procurement Model A versus Procurement Model C
In this cost comparison model we compared the procurement model of HC
organization A versus HC organization C, i.e. the comparison of “Self source” model
with that of “Hybrid” model which encompasses procurement through a National GPO as
well as a regional GPO. The prices of each of the bulk items (totaling 222 items) of the
HC organization C are subtracted from those of HC organization A to get the differences
at each sample point. Using the assumptions and equations mentioned in section 5.1.2,
and the values of mean, variance and n= 222, the sum of positive ranks is displayed in
table 5.2.
Table 5.2 Wilcoxon Results of Comparison of HC A and HC C
Differences
N
Rank-Sum
Negative
Positive
Zero
51
171
0
4967
19786
P
1.04361E14
From the table 5.2 it can be said the number of positive ranks are higher than the
negative ranks. Thus, sum of positive ranks and negative ranks Z + + Z − =
36
n(n + 1)
2
=
24,753. The Z+ and Z- values are 17762 and 6991 respectively. Thus, test statistic given
by Eq. 5.4 W=
Z+ − μ
σ
= 7.73
Now, P(7.73), i.e. the significance of the difference from the table 5.2 as given by
the solver is 1.04361E-14. Thus, using Eq. 5.5, the null hypothesis in the Wilcoxon test
will be rejected as: P(W) ≤ α value = 1.04361E-14 < 0.05.
Since the null hypothesis is rejected, the cost efficiency of one model versus the
other can be determined. The mean difference when the prices of commodities of HC
organization C (Hybrid model) are subtracted from those of HC organization A is $3.96,
which is a positive value and the mean percentage difference is 14.87 %, again a positive
value. Thus, we can say that the average prices of commodities procured by the Self
source model are more than the Hybrid model as shown by these positive values. Thus, in
this comparison model the Hybrid procurement model of HC organization C is more cost
efficient than that of Self sourcing model of HC organization A. One interesting
observation can be made here, the mean difference of prices between HC organization A
and HC organization C (when subtracted) is more than the mean difference of prices
between HC organization A and HC organization B. Thus, we can say that HC
organization C is not only more cost efficient than HC organization A but also HC
organization B. This can be illustrated further in the following section which shows the
comparison between HC organization B and HC organization C.
37
5.2.3. Procurement Model B versus Procurement Model C
In this cost comparison model we compared the procurement model of HC
organization B versus HC organization C, i.e. the comparison of “National GPO” model
with that of “Hybrid” model. The prices of each of the bulk items (totaling 222 items) of
the HC organization C are subtracted from those of HC organization B to get the
differences at each sample point. Again repeating the steps mentioned in the preceding
sections using the assumptions and equations mentioned in section 5.1.2, and the values
of mean, variance and n= 222 , the sum of positive ranks is displayed in table 5.3.
Table 5.3 Wilcoxon Results of Comparison of HC B and HC C
Differences
N
Rank-Sum
Negative
Positive
Zero
97
125
0
10243
14510
P
0.025957856
From the table 5.3 it can be said the number of positive ranks are higher than
negative ranks. Thus, sum of positive ranks and negative ranks Z + + Z − =
n(n + 1)
2
=
24,753. The Z+ and Z- values are 17762 and 6991 respectively. Thus test statistic given by
Eq. 5.4 W=
Z+ − μ
σ
= 2.226. Now, P(7.73), i.e. the significance of the difference from
the table 5.3 as given by the solver is 0.025957856. Thus using Eq. 5.5, the null
hypothesis in the Wilcoxon test will be rejected as: P(W) ≤ α value = 0.025957856 <
0.05.
Again while comparing the cost efficiency of the GPO model versus the Hybrid
model, the mean difference when the prices of commodities of HC organization C
38
(Hybrid model) are subtracted from those of HC organization B (GPO model) is $1.18,
and the mean percentage difference is 11.61 percent. Thus, we can say that the average
prices of commodities procured by the GPO model are more than the Hybrid model. Thus
in this comparison model the Hybrid procurement model of HC organization C is more
cost efficient than that of GPO model of HC organization B. However, it should be noted
that mean difference of prices between these two models is the least, which goes on to
show that these two models are quite close in terms of cost efficiency with Hybrid model
being the most.
5.2.4. Summarization of Results of Cost Comparison Study
The results obtained by the comparison study shows the comparative cost
efficiency of each of the three procurement models in consideration. From the results
obtained above, it can be concluded that GPOs overall deliver products to healthcare
organizations at a much reduced price, or in other words they are more cost efficient
compared to Self sourcing models. This can be attributed to the volume of bulk products
the GPOs carry in their inventory and their negotiating skills with the manufacturers.
However in this study, two healthcare organizations B and C are affiliated to GPOs with
C being further associated with a regional GPO. HC organization C fared the best with
being the most cost efficient among the three models due to their flexibility of
procurement contracts with a National GPO as well as a regional GPO. During the course
of interaction, staff from HC organization C acknowledged that affiliation to both
National GPO and regional GPO is important to drive prices low. Moreover, this gives
the model more leverage to procure items through two different sources depending on
lower prices. Apart from the benefit of choosing the lowest priced products being offered
39
by the competing National and regional GPO, another factor which might be responsible
for the Hybrid model to achieve the highest cost efficiency would probably be larger
volume of items. High compliance rate with the National as well as regional GPO
contracts may be another reason why the prices of the items in the Hybrid model are low
as compared to others. During the course of interaction with the staff from HC
organization C, it was brought to our knowledge that the compliance rate is very high and
that helps them to drive costs low. The compliance rate varies from one healthcare
organization to another and has a significant effect on the pricing of the items being
procured. By staying within the contract with the GPOs, the healthcare organizations are
making use of the power enjoyed by the GPOs with manufacturers in reducing costs. As
suggested in the past literature, GPOs have the capacity to supply the varied items at
large volumes as compared to small scale manufacturers. This might be the same reason
why HC organization A is the least cost efficient as all the items are Self sourced from
individual manufacturers locally. By increasing the volume of items, the cost per item
reduces, and this can be exploited by many healthcare organizations which have high
volumes of procurement to get affiliated to a GPO and contract for a variety of items.
Thus, in this study we can rank the Hybrid model of HC organization C as the
most cost efficient, followed by GPO model of HC organization B and the least being
Self source model of HC A. The table 5.4 in the following page would summarize the
results.
40
Table 5.4 Cost Efficiency of Procurement Models
Healthcare
Procurement
Ranking based on Cost Efficiency
Organization Model
(Most to Least)
C
Hybrid
1
B
GPO
2
A
Self
3
5.3 Analysis of Cost Comparison
The analysis of the results for each of the three comparison studies of
procurement models has been classified into further three sections, i.e, overall
comparison, medical devices comparison and surgical devices comparison.
5.3.1. Overall Comparison
Overall comparison involves the comparison of the total price difference and the
mean price difference of all the bulk items (totaling 222 in number) between HC
organizations A and B, A and C and B and C.
HC Organizations
Overall Comparison
B-C
B - C, $262.93
A- C
A - C, $878.71
A- B
$0.00
Total Price Difference
A - B, $615.79
$200.00
$400.00
$600.00
$800.00 $1,000.0
0
Total Price Difference
Figure 5.3 (a) Overall Comparison Based on Total Price Difference
41
HC Organizations
Overall Comparison
Mean Price Difference
B - C, $1.18
B-C
A - C, $3.96
A-C
A - B, $2.77
A-B
$0.00
$1.00
$2.00
$3.00
$4.00
Mean Price Difference
Figure 5.3 (b) Overall Comparison Based on Mean Price Difference
HC Organizations
Overall Comparison
B-C
B - C, 11.61%
A-C
A - C, 14.87%
A-B
0.00%
Mean Percentage
Difference
A - B, 6.17%
5.00%
10.00%
15.00%
20.00%
Mean Percentage Difference
Figure 5.3 (c) Overall Comparison Based on Mean Percentage Difference
From the figures 5.3(a) and 5.3(b), it can be concluded that the Hybrid model of
HC organization C is the most cost efficient, followed by the GPO model of HC
organization B and the Self sourcing model of HC A being the least efficient. This is
because the difference of both the total price as well as the mean price is maximum with
a positive value of $878.71 and $3.96 respectively when the HC organization C is
compared with HC organization A (with prices of HC organization C subtracted from HC
42
organization A) with respect to other comparison between HC organization B and HC
organization A. From figure 5.3 (c), the mean percentage difference between HC
organization A and HC organization C (14.87 percent) is also higher than HC
organization A and HC organization B (6.17 percent). This again proves that the Hybrid
model is the most cost efficient as the percentage difference is the largest when compared
with HC organization A as compared to when HC organization B with HC organization
A. However, the difference in the total price as well as the mean price of items between
HC organization B and HC organization C (with prices of HC organization C subtracted
from HC organization B) are smaller positive numbers of $262.93 and $1.18 (HC
organization C being more cost efficient) with respect to the other two comparisons. This
can be due to the fact that both the healthcare organizations are affiliated to a GPO which
helps to bring down the cost of procurement. The Hybrid model of HC organization C
has more leverage to choose between commodities based on lower prices as it is affiliated
to a National GPO as well as a regional GPO. As mentioned earlier and acknowledged
during the course of this study by the staff of HC organization C, in a Hybrid model the
healthcare organization has more freedom to negotiate the prices of items with the
National GPO and many times they procure items from the regional GPO at costs lower
than those offered by a National GPO. In fact, as acknowledged by the HC organization
C staff, that by negotiating contracts with regional GPOs, the healthcare organization can
get rebates on a yearly or quarterly basis which can result in huge savings in the long run.
This may not be possible for National GPOs, as they are bounded by much standardized
pricing across the country and may have many healthcare organizations affiliated to
them. Based on the past literature and recognized by the staff of HC organization C, the
43
regional GPOs have the added advantages of better knowledge of regional market
dynamics and may provide better logistics and warehousing facilities to the regional
healthcare organizations as compared to the National GPOs. At the same time, the
regional GPOs have fairly large inventory and sufficient number of healthcare
organizations in that particular region to keep costs low. Thus, a healthcare organization
following a Hybrid model of procurement involving a National GPO and a regional GPO
gets the best of both worlds and has more flexibility in choice of products as compared to
a healthcare organization following only the GPO model.
5.3.2. Medical Devices Comparison
This gives more in depth analysis of the cost comparison of procured medical
devices by comparing the total price difference and the mean price difference of only the
medical devices (totaling 156 in number) between HC organizations A and B, A and C
and B and C.
HC Organizations
Medical Devices Comparison
B-C
B - C, $73.28
A-C
A - C, $646.09
A-B
$0.00
Total Price Difference
A - B, $572.81
$200.00
$400.00
$600.00
$800.00
Total Price Difference
Figure 5.4 (a) Medical Device Comparison Based on Total Price Difference
44
HC Organizations
Medical Devices Comparison
B-C
B - C, $0.47
A - C, $4.14
A-C
A - B, $3.67
A-B
$0.00
Mean Price Difference
$1.00
$2.00
$3.00
$4.00
$5.00
Mean Price Difference
Figure 5.4 (b) Medical Device Comparison Based on Mean Price Difference
From figures 5.4(a) and 5.4(b), it can be again concluded that Hybrid model is the
most cost efficient followed by GPO model and the least being Self sourcing model
concerning the procurement of medical devices. However, it should be noted here that the
differences in total price as well as the mean price of medical devices when healthcare
organization B and C (prices of medical devices of HC organization C being subtracted
from those of HC organization B) are compared are very small positive values. This
suggests that though Hybrid model followed by HC organization C is more efficient than
the GPO model followed by HC organization B, but it’s by a very narrow margin. With
the difference in total price and mean price being small, positive numbers of $73.28 and
$0.47, respectively, it can be concluded that the cost efficiency of GPO model of HC
organization B comes very close to that of Hybrid model of HC organization C (almost as
cost efficient as C) when the procurement of medical device items is concerned.
45
5.3.3. Surgical Devices Comparison
This section involves the comparison of the total price difference and the mean
price difference of only the procured surgical device items (totaling 66 in number)
between HC organizations A and B, A and C and B and C.
HC Organizations
Surgical Devices Comparison
Total Price Difference
B - C, $189.65
B-C
A-C
A - C, $232.63
A-B
A - B, $42.98
$0.00
$50.00
$100.00
$150.00
$200.00
$250.00
Total Price Difference
Figure 5.5(a) Surgical Device Comparison Based on Total Price Difference
HC Organizations
Surgical Devices Comparison
B - C, $2.87
B-C
A - C, $3.52
A-C
A-B
$0.00
Mean Price Difference
A - B, $0.65
$1.00
$2.00
$3.00
$4.00
Mean Price Difference
Figure 5.5 (b) Surgical Device Comparison Based on Mean Price Difference
From figures 5.5(a) and 5.5(b), the analysis of the comparative cost efficiency of
the procurement models leads to the same inference of Hybrid model being the most cost
efficient, followed by the Hybrid model and the Self sourcing one being the least when
46
surgical devices are concerned. It is important to note that the GPO model fairs poorly in
terms of cost efficiency compared to the Hybrid model. It’s just the reversal of GPO
models performance in the medical devices comparison. In fact, the GPO affiliated HC
organization B is slightly better than the Self sourcing model of HC organization A as
displayed by the small positive values of differences of total price and mean price when
HC organizations A and B are compared. Here, the Hybrid model of HC organization C
outperforms the GPO model of HC organization B by a big margin in terms of cost
efficiency. The differences in the total price and the mean price of the surgical devices
procured by the HC organization B and HC organization C are quite large positive values
of $189.65 and $2.87 respectively which suggests that the GPO model is trailing behind
the Hybrid model in terms of cost efficiency significantly.
47
CHAPTER 6 - MEASUREMENT & COMPARISON OF INNOVATION (DEA)
6.1 Methodology of Innovation Measurement & Cost Comparison
This is the second aspect of our research. Here we will be identifying certain
innovation metrics in measuring the degree of access to innovative products procured
through off- contract as well as on-contract negotiations in different procurement models.
Innovative products in this study can be classified as those products which have a fairly
higher degree of sophistication and advanced technologies as compared to bulk items.
Common items which falls under this category are temporary pacemakers, splines,
ventilators, beds, etc, which might be procured in very few quantities or in low numbers
for specialized medical cases. This aspect of the project would enable us to compare the
different procurement models based on the off-contract and on-contract price of
innovative products and degree of access to innovative technologies. This is independent
of the cost comparison study and would involve more than three healthcare organizations.
During the course of the study it was found that there exists lot of variation in the models
employed by the healthcare organizations in procurement of innovative items. Many
healthcare organizations procure innovative items strictly on an on-contract basis either
through a GPO or a regional GPO whereas some healthcare organizations purchase these
items through off-contract means and some procure similar type of items through both
on-contract as well as off-contract means. This would throw light on the concern as
mentioned in the past literature (Zweig 1998) that sometimes National GPOs (mostly on48
contract purchases) are a hindrance to the entry of niche manufacturers of innovative
products (mostly off-contract purchases) and as a result the healthcare organizations
affiliated to only GPOs may sometime lose out on the more advanced technologies
available in the market. The items which are procured through GPOs are termed as “on
contract” purchased items and the ones which are procured from local manufacturers or
vendor which are not affiliated to any GPO are called “off-contract” items.
In this study three types of models of procurement of innovative items are studied:
1. Healthcare organizations which procure innovative items only through oncontract purchases, i.e. through contracts with GPOs.
2. Healthcare organizations which procure innovative items only through offcontract purchases, i.e. through contracts with individual manufacturers and
vendors not affiliated to any National or regional GPO.
3. Healthcare organizations which procure innovative items through on-contract
purchases as well as off-contract purchases, i.e., again different models of a
particular generic item like pacemaker, purchased though both the sources (dual
sourcing).
6.1.1 Identifying Innovation Metric
Based on the literature research, we have identified certain innovation metrics
which will help us in measuring access to innovative technologies.
The innovation metrics we have identified for data collection can be classified into two
types:
Product Innovative Metrics (metrics specific to the product):
1. Product features and specifications of the product.
49
2. Life cycle of the product.
3. Warranty details of the product.
4. Support from the manufacturer of the product in terms of training and technical
expertise.
5. Ease of operation of the product.
6. Reliability and quality of the product.
Corporate Innovative Metrics (adapted from (Muller 2005)):
These are the metrics applicable at the corporate level for the manufacturers of
innovative products which are listed below:
1. Measure R&D budget as a % of annual sales of a particular product manufacturer.
2. No. of patents/new ideas filed by the manufacturing company of that particular
item in the last year/last month.
3. Measure % of capital that is invested in radical projects by the company
manufacturing the innovative product.
4. Ratio of revenue from innovative/new products to commodities.
5. Measure % of employees that are involved in developing an innovative product.
6. Measure innovation revenue per employee from the new product/service
developed.
7. % of management that is accountable to development of a new product in terms of
time (man hour).
8. No. of incentive schemes to support innovation.
The most substantial metrics will be considered which is totally dependent on
accessibility and feasibility of data from the different sources, i.e., the data obtained from
50
the healthcare organizations as well as the product manufacturers. Data concerning
innovation metrics will be accessed from these sources and analyzed using the Delphi
Method.
6.1.2 Analyzing Innovation Metric Using Delphi Method
The data available for the innovative products from different sources will be
utilized to measure the degree of access to innovative technologies of the healthcare
organizations which procures that particular item. To measure the degree of access of
innovation, the Delphi method will be used which will enable us to get an “innovation
score” which is discussed in the latter part of this study.
Delphi method is extremely useful in cases where there is lack of scientific
knowledge. Delphi method becomes handy in forecasting and making judgments. This
involves expert opinion, intuition and experience. Most of the Delphi applications are
used for generating information for decision making
Delphi method involves a panel of experts from the related disciplines who are
given questionnaires concerning the particular subject. The experts chosen are
knowledgeable individuals who can draw from their extensive experience to assist in
forecasting results. In this study, the panel of experts will involve physicians who use the
products on a daily basis and material management staff who procure these items when
they are on-contract. During the initial contact, the nominated persons are told about the
Delphi and invited to participate. They are assured of anonymity in the sense that none of
their statements will be attributed to them by name (Gordon 1994). Each expert is
provided with a feedback on the preceding round of replies before the beginning of the
next round of questionnaire. In the first round the participants are asked to provide their
51
views on the subject in discussion. Then an analysis of the first round will throw light on
the range of opinions. In the second round, the range would be presented to the group,
and experts holding opinions at the extremes of the range would be asked to reassess their
opinions in view of the group's range and provide reasons for their positions (adapted
from Theodore (Gordon 1994)). These reasons would be synthesized by the researchers
at the end of round two; the synthesized reasons would form the basis for the third
questionnaire (Gordon 1994). In the third round, the questionnaire comprises of the new
opinions of the panel members. The opinions along with the reasons are then presented to
the participants. Each member of the group would be asked to reassess his or her position
in view of the reasons presented. They might also be asked to refute, if appropriate, the
extreme reasons with facts at their disposal (Gordon 1994).
In a fourth and final round, these arguments would be presented, along with the
evolving group consensus, and a reassessment requested (Gordon 1994). In a sense,
Delphi method is a controlled debate. The reasons for extreme opinions are made
explicit, fed back coolly and without anger or rancor (Gordon 1994).
The idea is that the consensus will lead to the best response. Statistically, the
midpoint of responses is identified by the median score. With every round, the range of
responses by experts is supposed to reduce which will help the median move closer to the
best response. A flowchart of the Delphi method processes is shown in figure 6.1 on the
next page.
Some of the advantages of the Delphi method include that the panelists need not
be physically present at the same location to give their responses and the process does not
require agreement by all the members as consensus is sought to arrive at the median.
52
Start
Define and identify a problem
Selection of panel of members
based on expertise.
Preparation and distribution of
questionnaire
Feedback and analysis of responses
No
Provide additional/requested info.
Has
consensus
been
reached?
Yes
Final consensus/report
Figure 6.1 Flowchart Showing Processes of Delphi Method
(Adapted from http://www.ryerson.ca/)
6.1.3 Innovation Metric Scale and Innovation Score
After the opinions and the feedback expressed by the panel of experts, each
innovative product is rated on a scale of 1 to 5 by these experts which has been termed as
“innovation metric scale” (IMS). The IMS will be used to assign each product a
particular rating termed in this study as “innovation score” from 1 to 5 with 1 being the
53
most innovative and 5 being the least. Thus, every product will have a specific innovation
score from a range of 1 to 5 which will be used in the data envelopment analysis (DEA)
later.
6.1.4 Theoretical Analysis versus Empirical Analysis
During the course of this research, getting access to data from the product
manufacturers as well as the healthcare organizations has been a very thorny task since
the start. Data concerning the pricing of innovative products, warranty details, product
features manufacturer’s details and model numbers are crucial to have a realistic analysis.
As mentioned earlier in the past literature (Zweig 1998; Everard 2005; Sethi 2006), the
big National GPOs quite often create a hindrance to the entry of manufacturers of high
technology products in to the market. Due to this reason, as was mentioned in the past
literature (Zweig 1998), physicians sometimes circumvent the hospital contracts with a
GPO and procure these items through off-contract means, which becomes an added cost
to the healthcare organizations. Due to less volume of these items coupled with high level
of sophistication, healthcare (HC) organizations may not be in a position to negotiate and
reduce costs. Also, the compliance rate of the HC organizations reduces as more and
more physicians prefer the off-contract route. However, contrary to the past concerns
about the HC organization not maintaining the compliance rate as discussed earlier, we
found that based on the interaction and our correspondence with the materials
management departments of almost seven HC organizations, that contract compliance is
more or less enforced by these healthcare organization managements on the physicians. If
a particular product is being preferred by the physicians, the physician has to provide
suitable justifications for choosing the particular product. The chances for the physician
54
preferred items to be approved by the material management department for purchase
would depend on strict evaluation of the products capabilities, long term cost savings and
relevance. Only after the product has satisfied the entire requirements specific to a
particular HC organization, would the products be approved for purchase. Once the
product has been approved for purchase it becomes the new standard and would be added
in the list of on-contract items. Thus, by ensuring large volumes of purchase, the HC
organizations tend to drive costs low. This trend was observed across all the seven HC
organizations we had interacted. This made the accessibility of data even more difficult
and so the comparison between on-contract items and off-contract ones could not be
undertaken as all the HC organizations seem to have products sourced through oncontract means only. Secondly, information regarding product manufacturers and
warranty details were considered confidential by both the HC organizations and the
manufacturers and could not be accessed by us. In the first aspect of this research project
involving the comparison of cost of bulk items, we could get hold of only an estimated
price for each of those products using the MS Excel model and thus we could compare
the estimated cost of those products. The information regarding manufacturers details,
warranty, technical expertise (considered confidential) were not required unlike in this
second part. Without specific information of a product like features, model numbers,
pricing and warranty details, it is impossible to carry out our research into measuring the
degree of access of innovative technologies across different HC organizations following
different procurement models. As a result, the main idea about this section is to propose a
research methodology which can be implemented if real data from the industry is
available. This is more of a theoretical analysis and if empirical data is available, this
55
methodology can be used to realistically compare the HC organizations. In order to
simulate real world scenarios, suitable and realistic data has been taken into
consideration, which would give us simulated results. However, this research idea can be
replicated not only in healthcare vertical but also other sectors where similar procurement
models are followed and GPOs have been known to exist. The fundamental question
which has been addressed in this section is to determine whether items classified as
innovative have the same levels of innovation (compare the degree of innovation and cost
of procurement) when they are procured by different means like on-contracting in the
form of a GPO or off-contracting through niche manufacturers. This also underlines the
concerns in the past literature that Nationalized GPOs cause a barrier to the entry of niche
manufacturers and slow down the entry of new advanced products (Everard 2005) than
currently available in the market. Data which is vital for undertaking Delphi method has
been simulated and so is the outcome of Delphi method which is the “innovation score”.
6.1.5 Comparison and Ranking Using DEA
In actual scenario with available real data, after the products are assigned a
particular innovation score by the panel of experts, a data envelopment analysis (DEA) is
carried out to compare and rank every procurement model followed by HC organizations
considered in this study for a particular product. For example, say healthcare organization
(HC) 1 uses both off-contract and on-contract models to procure an innovative product
like pacemaker. Thus it will have two different models of pacemakers with different
pricing and different innovation score (every model of a particular generic item will have
a unique pricing and innovation score), i.e., on-contract and off-contract pricing. So the
pacemaker model procured by HC 1 by on-contract means with a unique pricing and
56
innovation score will be compared with the pacemaker model procured by HC 1 with offcontract means with a different pricing and innovation score, as well as with other
pacemaker models procured by other healthcare organizations using either of the three
procurement models like only off-contract, only on-contract and both.
6.1.6 Data Envelopment Analysis (DEA)
Data envelopment analysis is a performance measurement approach used to
measure the relative performance of a number of entities called decision making units
(DMUs) by evaluating their efficiencies. DMUs can be various entities like departments,
HC organization, manufacturers etc. This is mostly used where there are multiple inputs
and multiple outputs where the DMUs are rated based on their efficiency as there are
limitations when multiple inputs and outputs are involved to evaluate efficiency of units
in conventional statistical approaches. The statistical approaches reflect “average” or
“central tendency” behavior of the observations while the DEA deals with the best
performance and evaluates all performances by deviations from the “efficient frontier
line” (Cooper 2006). The “efficient frontier line” connects the most efficient DMUs and
all the lesser efficient DMUs are either above or below this line in a output versus input
graph. This approach helps to identify the performance leaders which have the most
efficiency in a particular group and compares every DMU with these leaders. DEA had
been used to benchmark particular organizations as most efficient ones to highlight
inherent inefficiencies of the poor performing ones in that industry vertical.
DEA utilizes mathematical programming techniques which can handle large
number of variables and relations (constraints) in terms of inputs and outputs and this
relaxes the requirements that are often encountered when one is limited to choosing only
57
a few inputs and outputs because the techniques employed will otherwise encounter
difficulties (Cooper 2006). For every DMU a fractional programming problem is
formulated where the relative efficiency of the DMU is obtained by maximizing the
objective function which is a ratio of output DMU weights to input DMU weights. The
Fractional Programming (FP) solution for every DMU produces weights which are most
favorable to that particular DMU for maximizing efficiency. Since the objective function
is a ratio of the DMU output weights to DMU input weights, the optimal efficiency is at
most 1.
The mathematical model is as follows (Cooper 2006)
max θ =
u1 y1o + u 2 y 2 o + ............................ + u s y so
v1 x1o + v2 x2 o + ......................... + vm xmo
ST KK
u1 y1 j + ......................... + u s y sj
v1 x1 j + ............................ + vm xmj
Eq 6.1
≤ 1( j = 1,......n)
v1 , v2 ,.........................vm ≥ 0
u1 , u 2 ,..........................u s ≥ 0
The constraints signify that the ratio of the output weights to input weights should not
exceed 1, i.e., the optimal objective value can be at most 1.
After replacing the above fractional programming model to linear programming model
the basic DEA algebraic model becomes
∑
max θ =
∑
s
r =1
m
u r y ro
v xio
i =1 i
ST
∑
∑
s
r =1
m
u r y ro
≤ 1, (o = 1,.....n)
Eq 6.2
vx
i =1 i io
u r ≥ 0; r = 1,...........s
vi ≥ 0; i = 1,............m
58
Where vi is the optimal weight for the input item i and it’s magnitude expresses how
highly the item is evaluated. Similarly, ur does the same for the output item r. (Cooper
2006).
6.1.7 Selection of Decision Making Units (DMUs)
As mentioned earlier, non accessibility of real data led us to consider realistic
simulated data. In this section, the decision making units will be “model type of
product/name of hospital” for a particular generic product. For example, for temporary
pacemakers, the different DMUs will be model1/hospital 1 (Hybrid on-contract),
followed by other different permutations. Thus, there will be a unique product, i.e.,
unique model of say pacemaker which will be compared with other models of
pacemakers procured through different sources. Even within the same HC organization
two different types of pacemakers may be procured if they follow both on-contract as
well as off-contract purchasing. The table 6.1 below displays all the DMUs which will be
considered in this analysis.
Table 6.1 Simulated DMUs
DMUs
Hospital Type
Model 1/Hosp A
Model 2/Hosp B
Model 3/Hosp C
Model 4/Hosp D (oncontract model of
Hosp D)
General
General
General
General
Model 5/Hosp D
(off-contract model General
of Hosp D)
Procurement model considered by the
hospital.
Self sourcing off-contract
GPO model with on-contract
GPO model with off-contract
GPO model with both off and oncontract purchasing
GPO model with both off and oncontract purchasing
59
Table 6.1 (continued)
DMUs
Hospital Type
Model 6/Hosp E
Model 7/Hosp F
Model 8/Hosp G (oncontract model of
Hosp G)
Model 9/Hosp G
(off-contract model
of Hosp G)
Model 10/Hosp H
General
General
General
Model 11/Hosp I
Specialty
Model 12/Hosp J
Specialty
Model 13/Hosp K
Specialty
Model 14/Hosp L
Specialty
Model 15/Hosp M
(on-contract model
of Hosp M)
Model 16/Hosp M
(off-contract model
of Hosp M)
Model 17/Hosp N
(on-contract model
of Hosp N)
Model 18/Hosp N
(off-contract model
of Hosp N)
Specialty
Procurement model considered by the
hospital.
Hybrid model with on-contract
Hybrid model with off-contract
Hybrid model with both off and oncontract purchasing
General
Hybrid model with both off and oncontract purchasing
Specialty
Specialty Hospital with Self sourcing
(off-contract)
Specialty Hospital with GPO model
(on-contract)
Specialty Hospital with GPO model
(off-contract)
Specialty Hospital with Hybrid model
(on-contract)
Specialty Hospital with Hybrid model
(off-contract)
Specialty Hospital-GPO model with
both off and on-contract purchasing
Specialty
Specialty Hospital-GPO model with
both off and on-contract purchasing
Specialty
Specialty Hospital-Hybrid model with
both off and on-contract purchasing
Specialty
Specialty Hospital-Hybrid model with
both off and on-contract purchasing
The DMUs under consideration in this research are different models of a
particular generic product like pacemakers sourced from diverse healthcare organizations.
There are two types of healthcare organizations under consideration (2nd column in table
6.2 titled “hospital type”) i.e., general and special. General type represents HC
organizations which cater to all sorts of medical cases from orthopedics to pediatrics.
60
These may have a wide range of departments like cardiology, neurosurgery, orthopedics
etc.
The specialty type represents the HC organizations which cater to special
treatments and research like cancer, cardiac ailments, geriatric disorders, neuro-surgery
etc. These are more focused in treatment of special cases and research and tend to be
more advanced over general HC organization. However they cater to a smaller population
sample and it is simulated that the features of the items they procure are more
sophisticated.
The next sub classification of DMUs is the procurement model they follow like
Self sourcing, GPO or Hybrid. Again these procurement models can be followed by
either general or specialty HC organization. These HC organizations following a
particular procurement model for example Hybrid model can procure innovative items by
only on-contract means, or only off-contract means or both (Model 8 and model 9).
Sometimes the same hospital can procure two different models of an item with one being
procured through on-contract way and the other through off-contract. It should be noted
the models of products are unique which are the DMUs in this study. For example: the
model 2 is different from model 3 of a particular item as they are procured by two
different HC organizations, with model 2 being procured by on-contract means which is
being followed by hospital B whereas model 3 is being procured by on-contract means
which is being followed by hospital C though hospital B and C are affiliated to a GPO.
Similarly in a more complicated case where the same hospital follows two
different sources of contracting, the models of the items procured will be different and
unique for that particular source of contracting. For example: model 4 and model 5 are
61
two different models of an item procured by the same hospital using two different sources
of contracting as shown in table 6.1. Same pattern is repeated in specialty HC
organizations. However, Self sourcing HC organizations are unique as they do only offcontracting as they are not affiliated with any GPO or bounded by any contract. They
procure their bulk items as well as innovative items the same way.
DEA models vary from having single output and single input to multiple inputs
and outputs. In this research methodology, all the data concerning outputs and inputs
have been realistically simulated due to non accessibility of data from the industry. Two
inputs are considered in this study with one output. These two inputs are in the form of:
1. Cost of innovative products.
2. Innovation Score (1 being the most innovative and 5 being the least).
The output considered is the “the number of hospital beds”. This would give an
idea about the maximum number of patients a hospital can treat. It is simulated and
highlighted in the past literature that there is an underlying link with the cost of the
product as larger volume of products will drive costs low. Thus, it is simulated that if a
hospital has the capacity to treat a large number of patients, its cost for procuring a
particular item will be lower than another hospital which has lesser capacity of treating
patients.
6.1.8 Simulated Costs of DMUs (Input)
Cost of innovative products has been adjusted based on the healthcare
organizations. That is, the underlying assumption is that the cost of a particular DMU of a
general HC organization is lower than a specialty one under both off-contract and oncontract means. Similarly as evident from the first aspect of the thesis, it has been
62
simulated that the DMUs under the Hybrid models are the most cost efficient, followed
by those under GPO models for both general and specialty HC organization only when
the items are sourced through on-contract means. Off-contract purchase costs for all HC
organizations having different procurement models, are simulated to be quite close but
not same as it depends on the negotiating power of the respective HC organization with
the vendors when they off-contract. But it differs in a case where a particular HC
organization has two sources of contracting as mentioned in this study. For the same HC
organization, cost of a DMU by off-contract means is simulated to be higher than an oncontract one for both general and specialty type organizations having dual sources of
contracting. Another assumption is that the cost of procurement for Self sourcing will be
higher than those of GPO and Hybrid models by on-contract means for both general and
specialty HC organizations. Table 6.2 displays the simulated costs of innovative items for
the DMUs. The simulated cost of generic item pacemaker is taken into consideration with
a price range between $4000 and $5000 for a general HC organization and between
$5100 and $6000 for the specialty ones.
Table 6.2 Simulated Costs of DMUs (Input)
DMUs
Hospital
Type
Procurement model
Simulated Costs in US
considered by the hospital. Dollars (input)
Model 1/Hosp A
Model 2/Hosp B
Gen
Gen
Model 3/Hosp C
Gen
Model 4/Hosp D
(on-contract model
of Hosp D)
Gen
Self sourcing off-contract
GPO model with oncontract
GPO model with offcontract
GPO model with both off
and on-contract
purchasing
63
5000
4400
4900
4500
Table 6.2 (Continued)
DMUs
Hospital
Type
Procurement model
Simulated Costs in US
considered by the hospital. Dollars (input)
Model 5/Hosp D
(off-contract model
of Hosp D)
Model 6/Hosp E
Gen
Model 7/Hosp F
Gen
Model 8/Hosp G
(on-contract model
of Hosp G)
Model 9/Hosp G
(off-contract model
of Hosp G)
Model 10/Hosp H
Gen
Model 11/Hosp I
Sp
Model 12/Hosp J
Sp
Model 13/Hosp K
Sp
Model 14/Hosp L
Sp
Model 15/Hosp M
(on-contract model
of Hosp M)
Model 16/Hosp M
(off-contract model
of Hosp M)
Model 17/Hosp N
(on-contract model
of Hosp N)
Model 18/Hosp N
(off-contract model
of Hosp N)
Sp
GPO model with both off
and on-contract
purchasing
Hybrid model with oncontract
Hybrid model with offcontract
Hybrid model with both
off and on-contract
purchasing
Hybrid model with both
off and on-contract
purchasing
Specialty Hospital with
Self sourcing (offcontract)
Specialty Hospital with
GPO model (on-contract)
Specialty Hospital with
GPO model (off-contract)
Specialty Hospital with
Hybrid model
(on-contract)
Specialty Hospital with
Hybrid model
(off-contract)
Specialty Hospital-GPO
model with both off and
on-contract purchasing
Specialty Hospital-GPO
model with both off and
on-contract purchasing
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
Gen
Gen
Sp
Sp
Sp
Sp
64
4850
4200
4900
4300
4950
5800
5400
5900
5150
5900
5300
6000
5250
5850
In the above table “Gen” represents general category of HC organizations and “Sp”
represents specialty units.
6.1.9 Simulated Innovation Score (Input)
When assuming innovation scores of DMUs under consideration, similar pattern
is seen. As discussed earlier, the innovation score ranges from 1 to 5 with 1 being the
highest or most innovative and 5 being the least innovative. For example, when the
innovation score of a particular model A is 3 and another model B is 5, it can be said that
model A will have higher innovation score than the model B. The innovation score of a
particular DMU of a specialty hospital having a certain procurement model and
contracting source is simulated to be higher than that of a corresponding DMU of a
general hospital. Again the innovation score of the DMUs of Hybrid models is simulated
to be higher than those of GPO models only for on-contract means for both general and
specialty HC organizations. This is because of the assumption that Hybrid models have
more flexibility than just GPO models in the choice of products and have generally wider
range. However, for off-contract purchases, innovation score is simulated to be constant
for all the three models of procurement for comparisons within general and specialty HC
organizations. It is also simulated that the innovation score for DMUs of HC
organizations having two sources of contracting for both GPO as well as Hybrid models
of procurement under both general and specialty categories will be higher for the ones
through off-contracting than those procured through on-contracting. Again, it has been
simulated that the innovation score of the Self sourcing DMUs for both general and
specialty categories will be higher than their respective Hybrid and GPO on-contract
DMUs, whereas remaining the same as that of their respective off-contract DMUs.
65
It should be noted here that under a category of HC organization like general,
certain DMUs will have the same innovation score whereas their cost will vary slightly.
For example, models 1, 3, 5, 7, and 9 under general HC organization type are simulated
to have same innovation score whereas they may not have same cost prices as cost is
dependent on the negotiating power and the volume of the items a HC organization
purchases when making off-contract purchases. However, their costs are simulated to be
quite close if not the same.
The DEA will be unique for each kind of generic item, i.e. say for pacemakers,
there will be a DEA model with different models of pacemakers numbered from 1 to 18
forming the DMUs. Similarly, other generic items like implants etc. will have their own
respective DEA. Thus in this study, innovation score and the output “the number of
hospital beds” is simulated to remain constant for a particular DMU under different DEA
models. For example, model 3 will have same simulated values for “innovation score”
and “no. of beds” constant for all generic item DEA models. Only the cost of innovative
items (simulated input) will change across the DEA models because every generic
innovative item costs differently. Table 6.3 listed in the following page displays the
simulated innovation scores for the DMUs.
Table 6.3 Simulated Innovation Scores of DMUs (Input)
DMUs
Model 1/Hosp A
Model 2/Hosp B
Hospital
Type
Gen
Gen
Model 3/Hosp C
Gen
Model 4/Hosp D
(on-contract model
of Hosp D)
Gen
Procurement model
considered by the hospital.
Self sourcing off-contract
GPO model with oncontract
GPO model with offcontract
GPO model with both off
and on-contract
purchasing
66
Innovation
Score
(input)
3
5 (same for all general
GPO on-contracts)
3 (same for all general
off- contracts)
5
Table 6.3 (Continued)
DMUs
Hospital
Type
Gen
Model 5/Hosp D
(off-contract
model of Hosp D)
Model 6/Hosp E
Gen
Model 7/Hosp F
Gen
Gen
Model 8/Hosp G
(on-contract
model of Hosp G)
Gen
Model 9/Hosp G
(off-contract
model of Hosp G)
Model 10/Hosp H Sp
Model 11/Hosp I
Sp
Model 12/Hosp J
Sp
Model 13/Hosp K
Sp
Model 14/Hosp L
Sp
Model 15/Hosp
M (on-contract
model of Hosp
M)
Model 16/Hosp
M (off-contract
model of Hosp
M)
Model 17/Hosp N
(on-contract
model of Hosp N)
Model 18/Hosp N
(off-contract
model of Hosp N)
Sp
Procurement model
considered by the hospital.
GPO model with both off
and on-contract
purchasing
Hybrid model with oncontract
Hybrid model with offcontract
Hybrid model with both
off and on-contract
purchasing
Hybrid model with both
off and on-contract
purchasing
Specialty Hospital with
Self sourcing (offcontract)
Specialty Hospital with
GPO model (on-contract)
Specialty Hospital with
GPO model (off-contract)
Specialty Hospital with
Hybrid model
(on-contract)
Specialty Hospital with
Hybrid model
(off-contract)
Specialty Hospital-GPO
model with both off and
on-contract purchasing
Innovation
(input)
3
4 (same for all general
Hybrid on-contracts)
3
4
3
1
(same for all specialty
off-contracts)
3
(same for all specialty
GPO on-contracts)
1
2
(same for all specialty
Hybrid on-contracts)
1
3
Sp
Specialty Hospital-GPO
model with both off and
on-contract purchasing
1
Sp
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
2
Sp
67
Score
1
6.1.10 Simulated “No. of Beds” (Output) of DMUs
The simulated values of the output of the DMUs in the DEA the specialty HC
organization has will have lesser number of beds compared to general ones as the former
ones are more focused to a particular type of treatment whereas the latter ones cater to a
wider range of treatments and population. However, the number of beds (output) for
DMUs which fall under the category of HC organizations which have dual contract
sources is simulated to be same as they are the same HC organization. Table 6.4 listed
below shows the simulated values of outputs (No. of beds) for the DMUs
Table 6.4 Simulated Values of Outputs (No. of beds)
DMUs
Model 1/Hosp A
Model 2/Hosp B
Hospital
Type
Gen
Gen
Model 3/Hosp C
Gen
Model 4/Hosp D
(on-contract model
of Hosp D)
Model 5/Hosp D
(off-contract model
of Hosp D)
Model 6/Hosp E
Gen
Model 7/Hosp F
Gen
Model 8/Hosp G
(on-contract model
of Hosp G)
Model 9/Hosp G
(off-contract model
of Hosp G)
Model 10/Hosp H
Gen
Model 11/Hosp I
Sp
Gen
Gen
Gen
Sp
Procurement model
considered by the hospital.
Self sourcing off-contract
GPO model with oncontract
GPO model with offcontract
GPO model with both off
and on-contract
purchasing
GPO model with both off
and on-contract
purchasing
Hybrid model with oncontract
Hybrid model with offcontract
Hybrid model with both
off and on-contract
purchasing
Hybrid model with both
off and on-contract
purchasing
Specialty Hospital with
Self sourcing (offcontract)
Specialty Hospital with
GPO model (on-contract)
68
No. of Beds (output)
300
350
325
250
250
400
350
325
325
150
175
Table 6.4 (Continued)
DMUs
Model 12/Hosp J
Hospital
Type
Sp
Model 13/Hosp K
Sp
Model 14/Hosp L
Sp
Model 15/Hosp M
(on-contract model
of Hosp M)
Model 16/Hosp M
(off-contract model
of Hosp M)
Model 17/Hosp N
(on-contract model
of Hosp N)
Model 18/Hosp N
(off-contract model
of Hosp N)
Sp
Sp
Sp
Sp
Procurement model
considered by the hospital.
Specialty Hospital with
GPO model (off-contract)
Specialty Hospital with
Hybrid model
(on-contract)
Specialty Hospital with
Hybrid model
(off-contract)
Specialty Hospital-GPO
model with both off and
on-contract purchasing
Specialty Hospital-GPO
model with both off and
on-contract purchasing
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
Specialty Hospital-Hybrid
model with both off and
on-contract purchasing
No. of Beds (output)
160
180
145
170
170
165
165
6.1.11 Selection of (DEA) Model
The DEA model chosen in this study will be “CCR input-oriented bounded”
model. The CCR model was proposed by Charnes, Cooper and Rhodes in 1978 (Cooper
2006). The main assumptions of the CCR model are (Cooper 2006)
1. Constant returns to scale which assumes that a proportional change in the inputs
also increases the output by the same proportion.
2. Since all the data (inputs and outputs) are simulated to be positive, translation
invariant capability is not required. Translation invariance converts negative data
to positive values, which are not a concern in this study.
69
Expressing the linear programming model of DEA (Eq. 6.2) from section 6.2.5 in the
form of vector matrix notation (Cooper 2006),
(LPo)
maxv,u = uyo
subject to
vxo = 1
Eq. 6.3
-vX + uY ≤ 0
v ≥ 0, u ≥ 0.
Where matrix(X,Y) comprises of row vector v as input multipliers and u for as output
multipliers.
Input-oriented CCR models minimize inputs to satisfy the desired output levels.
In this research study, the main objective would be to minimize the values of inputs, i.e.
the cost and innovation score. The DMUs with the relative minimum innovation score
and relative minimum cost would be the optimal DMU against which the other DMUs
will be measured. It was decided to minimize input because the output which is the
number of beds, cannot be varied as that is constant and specific to a hospital.
Minimization of input is the sole reason for choosing a inverted innovation scale with 1
being the most innovative and 5 being the least.
The dual problem of the (LPo) in equation 6.3 expressed with a real variable θ
and the transpose of non negative vector λ = (λ1,….. λn) (Cooper 2006)
(DLPo)
Subject to
minθλ θ
θxo – Xλ ≥ 0
Eq. 6.4
Yλ ≥ yo
λ ≥ 0.
70
Since the innovation score is bounded by an inverted scale of innovation score of
a maximum value of 1 and minimum value of 5. The innovation score cannot go out of
this range and this is the sole reason why bounded input-oriented CCR model is applied
as the solution will try to minimize the innovation score and give optimal efficiency for
each DMU.
The bounded equations are:
lxo ≤Xλ ≤uxo
Eq. 6.5
lyo ≤Yλ ≤uyo
Eq. 6.6
where ( lxo, uxo) are lower and upper bound vectors to inputs and ( lyo, uyo) to outputs
respectively.
6.2 Results of Comparison of Access to Innovation with Cost
This section of the study deals with the comparison of degree of access to
innovative products across various HC organizations following different procurement
models and contract sources. As discussed earlier, accessibility of data for off-contract
items has not been possible due to the recent phenomenon of HC organizations to lay
stress on on-contract purchasing. Also, information regarding product features,
manufacturer’s details was not available to us due to confidentiality concerns shared by
the hospital staff and the manufacturing units. This comparison model is totally based on
simulated data which has been tailored to suit real world scenarios as closely as possible.
The conditions and the justifications for the assumptions have been explained in detail in
the methodology section. The main idea here is to bind the two factors’ cost and
innovation score and rate the models of pacemakers which are the DMUs in terms of
71
efficiency and ultimately rank them. There is a link between the cost and the degree of
access of innovation rated by innovation score and based on these two factors, the DEA
model tries to find out the optimal efficient DMU which would be rated as the most
efficient and all other DMUs will be measured against it in terms of efficiency.
The DEA model is run for the generic item pacemaker using DEA solver and the
results are shown below.
Table 6.5 Ranking and Efficiency Scores of DMUs
Rank
1
2
3
4
5
6
6
8
9
10
11
12
13
14
15
16
16
18
DMU
6
8
2
4
5
3
7
9
1
13
17
15
11
10
18
12
14
16
Score
1
0.976734
0.954536
0.933324
0.865971
0.857134
0.857134
0.848476
0.839992
0.815526
0.799992
0.792445
0.77777
0.724131
0.717942
0.711857
0.711857
0.699993
Reference
set
(lambda)
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
Table 6.5 above displays the efficiency score of each DMU and the rankings
based on efficiency score. The efficiency score is the efficiency of each DMU evaluated
against the most efficient DMU which in this study is model 6. The reason for model 6 to
be ranked most efficient is due to lower cost as Hybrid model in this study is simulated to
have lowest cost for on-contract purchases (assumption is taken from the cost comparison
72
analysis where the Hybrid model had the lowest cost) as compared to GPO and Self
sourcing and at the same time fair better on the innovation scale than the GPO model.
Understandably, the second ranking DMU is model 8 which again is from Hybrid
model and has on-contract purchase sources. It can be seen that there is a huge difference
in ranking between second ranking model 8 and eighth ranking model 9. Both the models
are procured by the same hospital G, however the extremely low prices of on-contract
Hybrid models (low even compared to GPO on contract) as compared to off-contact
prices which are quite similar across the Hybrid, GPO and Self sourcing model, drive the
difference in rankings. Because the difference between the simulated off-contract pricing
and on-contract pricing for the GPO model of the same hospital is less as compared to the
Hybrid model, model 4 and model 5 trail closely in ranking at 4 and 5, with model 4
being the more efficient one. However, it is closely followed by GPO on-contract DMUs
models 2 and 4 at ranks 3 and 4. Since DMU model 6 is the most efficient, it is taken as a
reference set against which other DMUs will be rated. Quite expectedly, model 1, which
is procured by off-contract Self sourcing is the least efficient among general items as it is
generally simulated to have a highest price. From the table it can be said that the general
HC organization are more efficient than specialty ones. The major factors behind this
might be the fact that the average cost of items in specialty units are much higher than
those of general HC organization (higher inputs) and at the same time lower outputs in
terms of “number of bed”. When rated on innovation score, specialty ones will
outperform the general ones (as they have better innovation score), but when costs are
tied with innovation they seem to be less efficient overall. Similar trends as in the general
73
hospital can also be observed in the specialty HC organization. The figure 6.2 shows the
efficiency of DMUs in a graphical form below.
Score
17
15
13
DMU
11
9
7
5
3
1
0
0.1
0.2
0.3
0.4
0.5
0.6
Efficiency
Figure 6.2 Graph Showing Efficiency Scores of DMUs
74
0.7
0.8
0.9
1
Table 6.6 Statistics on Input/Output Data
(IB)InScore
Max
Min
Average
SD
5
1
2.666666667
1.290994449
(I)Cost
6000
4200
5141.667
564.764
(O)Beds
400
145
241.9444
84.19783
(I)Cost
-0.92579
(O)Beds
0.741941
1
-0.84902
-0.84902
1
Correlation
(IB)InScore
(IB)InScore
(I)Cost
(O)Beds
1
0.925789863
0.741940575
DMUs with inappropriate Data with respect to the chosen Model
No.
DMU
None
No. of
18
DMUs
Average
0.826934118
SD
0.09234159
Maximum
1
Minimum
0.699993
No. of DMUs in Data =
No. of DMUs with inappropriate Data =
No. of evaluated DMUs =
18
0
18
Average of scores =
No. of efficient DMUs =
No. of inefficient DMUs =
0.826934
1
17
Table 6.6 above shows the statistics like maximum, minimum and average values
of outputs and input. It also displays the correlation between the inputs to the outputs,
standard deviation, and average efficiency score along with the maximum and minimum
values. The correlation is of particular importance here. It describes the strength and the
nature of relationship between the inputs and between the inputs and the output. As seen
from the table, the bounded input “InScore” (Innovation Score) has inverse correlation
with output variable “Cost” and is equal to -0.92579. The relationship between “InScore
75
and Cost” is however stronger than the relationship between “Cost and Beds” and
“InScore and Beds”. The inverse correlation shows that as “InScore” increases the cost
would decrease. This is because of the usage of inverted Innovation Scale where the most
efficient is rated as 1 and the least 5. The model also has been based on assumptions that
on-contract purchases are low on innovation as compared to off-contract ones, i.e. oncontract purchases have an innovation score range between and 3 and 5 whereas offcontract ones range between 1and 3. At the same time on-contract items have a much
lower price as compared to off-contract and this would be the reason for inverse
correlation. Similarly, the correlation between input “cost” and output “bed” is also
shown as inverse with a value of -0.84902, which can be justified as the cost decreases
with the increase of capacity of the hospital to treat patients.
Table 6.7 Projection of DMUs
No.
DMU
I/O
4
Score
Data
0.933324
(IB)InScore
(I)Cost
(O)Beds
5
(IB)InScore
5
4500
250
0.865971
3
3.99996
4199.958
399.996
-1.00004
-300.042
149.996
3.99996
0.99996
(I)Cost
(O)Beds
4199.958
399.996
-650.042
149.996
6
(IB)InScore
(I)Cost
(O)Beds
7
(IB)InScore
4850
250
1
4
4200
400
0.857134
3
33.33%
13.40%
60.00%
4
4200
400
0
0
0
0.00%
0.00%
0.00%
3.99996
0.99996
(I)Cost
(O)Beds
4900
350
4199.958
399.996
-700.042
49.996
33.33%
14.29%
14.28%
4
5
6
7
Projection
Difference
76
%
20.00%
-6.67%
60.00%
Table 6.7 (Continued)
8
9
8
(IB)InScore
0.976734
4
(I)Cost
(O)Beds
4199.958
399.996
9
(IB)InScore
4300
325
0.848476
3
(I)Cost
(O)Beds
4950
325
4199.958
399.996
3.99996
3.99996
0
100.042
74.996
0.00%
-2.33%
23.08%
0.99996
750.042
74.996
33.33%
15.15%
23.08%
The above table 6.7 shows the projection of DMUs to the efficient frontier. The
projection of model 4 to model 9 has been chosen to be displayed here. Model 6 is the
reference set and is the most efficient DMU. In order to achieve optimal efficiency, the
DMUs are projected to the efficient frontier, and their difference and percentage changes
are also highlighted. As can be seen, since DMU model 6 is the most efficient, the
percentage change and difference in input and output weights for it to be projected to the
efficient frontier are both zero. The other DMUs either have positive or negative changes
to their input and output values for them to be projected to the efficient frontier, as they
are less efficient than model 6.
77
CHAPTER 7 - CONCLUSION AND DISCUSSIONS
As discussed earlier, this research study has two contributions. In the first section,
the goal was to determine whether National GPOs help the healthcare organizations
affiliated to them to drive their costs low as compared to the healthcare organization
which Self contract. Based on the available data from three healthcare organizations
involving a Self sourcing model, a GPO model and a Hybrid model, our results clearly
prove that healthcare organizations affiliated to National GPO are indeed more cost
efficient than the Self sourcing ones. A Hybrid model was also used in the comparison
and it was the clear winner in terms of cost efficiency in the comparison test. The Self
sourcing model has significantly higher overall costs compared to the GPO model and the
Hybrid model. The Hybrid model achieves the efficiency by having the flexibility of
wider range of products and the ability to choose between the best prices offered by a
National GPO and a regional GPO. Thus, based on the results of this cost comparison
study, the first hypothesis “H1 - National GPOs (Group Purchasing Organizations) enable
the healthcare establishments to lower the cost of medical services and an operations” is
valid. It should be mentioned that during the course of research, it was found that no two
healthcare organizations affiliated to the same GPO have the same price figures. Thus,
two different healthcare organizations under the same GPO will have different cost
efficiencies. This is dependent on the negotiating capacity of each healthcare
organization, volume of purchase, and the compliance rate of the healthcare organization.
78
For example, a healthcare organization having a high compliance rate and high volume of
purchase will have lower prices of products from the GPO as compared to another which
has lesser compliance rate and volume under the same GPO. Also, many GPOs have
mandatory compliance rates.
The second aspect of this research was to measure and compare the degree of
access to innovative products across HC organization with different procurement models
and modes of contracting. In this study, since data was not available, realistic data were
simulated. The results achieved with the simulated data, models of items procured
through contracts by a GPO and Hybrid model driven HC organization faired most
efficient as compared to Self sourcing and off-contracted models. In spite of oncontracting models having lesser innovative features as compared to off-contracting ones,
the fact that they are more cost efficient, improves their efficiency. Thus, many HC
organizations in their attempts to drive costs lower might go for on-contracting source of
procurement and might compromise on the quality of products as they are more cost
efficient. This might create a barrier to the entry of niche manufacturers of high end items
whose products are more advanced than the ones offered by the GPOs, but do not have
the necessary volumes to drive the cost low. They might be beaten out in the race and
since they are generally not affiliated to GPOs, they may not find the support from the
HC organization to sustain in the competitive marker. Thus, if this research study is
performed with real world data and if it is quite similar to the simulated data used in this
project, the hypothesis “H2 - National GPOs a barrier to entry of innovative product
manufacturers in the healthcare industry” can be proved, which again reflects the
concerns shared by the past literature.
79
In this research project, the most common procurement models like Self sourcing,
GPO and Hybrid are compared and discussed in terms of cost efficiency. But, during the
course of the study, it came to our knowledge that one more type of procurement model
is gaining acceptance in the healthcare industry. This model is the recent phenomenon of
formation of “regional cooperatives”. In this model, multiple healthcare organizations
which are in the close proximity geographically create a purchasing and logistics
subsidiary, which is solely responsible for procurement operations to those healthcare
organizations. Based on our interaction with the healthcare professionals, we could
interpret that regional cooperatives generally drive high compliance rate and have
contracts with local manufacturers. Contracting with local vendors and maintaining high
compliance rate for the items might result in low costs and better distribution facilities
and supply lines. They might also have better access to latest technologies in the industry
through contracting with niche manufactures, and since multiple hospitals have shares in
a regional cooperative, a large volume would help to drive costs low. It would be
interesting to compare this procurement model with the three compared in this study as a
future research study.
Future research can also involve actually comparing off-contract pricing with oncontract pricing for innovative items, as there would be several healthcare organizations
where the physicians circumvent and procure their preferred products. Unfortunately, in
this study, almost four healthcare organizations we worked with, had no way of
accounting for off-contract purchasing, as they strictly enforce the compliance rate to
drive costs low.
80
REFERENCES
Anderson, M. G., and Katz, P.B. (1998). "Strategic sourcing." International Journal of
Logistics Management 9(1): 1-13.
Barlow, R. D. (2005). "Glancing back." Healthcare Purchasing News.
Buganza, T. a. V., Roberto (2006). "Life-Cycle flexibility: How to measure and improve
innovative capability in turbulent environments." The Journal of Product Innovation
Management 23: 393-407.
Burns, L. R. (2002). "The Health Care Value Chain."
Chapman, T. L., Gupta, A., Mango, P.D. (1998). "Group Purchasing is not a panacea for
US Hospitals." McKinsey Quarterly 1: 160-165.
Cooper, W. W., Seiford, Lawrence M., Tone, Kaoru (2006). "Introduction to Data
Envelopment Analysis and Its Uses."
Dobler, D. W., and Burt, David N. (1996). "Test and Cases." Purchasing and Supply
Management.
Doucette, W. R. (1997). "Influences on Member Commitment Group Purchasing
Organizations." Journal of Business Research 40: 183-189.
Elhauge, E. (2002). "The exclusion of competition for hospital sales through group
purchasing organizations." unpublished.
Essig, M. (2000). "Purchasing consortia as symbiotic relationships: developing the
concept of ‘consortium sourcing'." European Journal of Purchasing & Supply
Management 6: 13-22.
Everard, J. L. (2005). "Defining and Measuring Hospital Product-Based Cost Savings."
Gordon, T. J. (1994). "THE DELPHI METHOD." AC/UNU Millennium Project.
Hendrick, T. E. (1997). "Purchasing Consortiums: Horizontal Alliances Among Firms
Buying Common Goods and Services: What? Who? Why? How?"
Hewitt, D. (1995). "The Consortium Option." Purchasing and Supply Management: 32.
81
McFadden, C. D., and Leahy, Timothy M., (2000). US Healthcare Distribution.
Muller, A., Liisa Valinkangas, and Merlyn, Paul (2005). "Metrics for Innovation:
Guidelines for developing a customized suite of innovation metrics." Strategos, An
International Strategic Management Consultancy.
Nollet, J., and Beaulieu, Martin (2002). "The development of group purchasing: an
empirical study in the healthcare sector." Journal of Purchasing & Supply Management.
Nollet, J., and Beaulieu, Martin (2005). "Should an organization join a purchasing
group? " Supply Chain Management: An International Journal 10(1): 11-17.
Rozemeijer, F. (2000). "How to manage corporate purchasing synergy in a decentralized
company? towards design rules for managing and organizing purchasing synergy in
decentralised companies." European Journal of Purchasing & Supply Management 6(1):
5-12.
Schneller, E. S. (2000). "The value of group purchasing in Health Care Supply chain."
Sethi, P. (2006). "Group Purchasing Organizations: An Evaluation of their Effectiveness
in Providing Services to Hospitals and Their Patients." HGPII Report, 07-20-06.
Zweig, P. L., Zellner, W. (1998). "Locked out of the Hospital." Business Week 3569: 7576.
82
APPENDICES
83
APPENDIX A
Process Map of Self Sourcing Model
Orders
Medical
clinic
<=$500
Is PO
value <
=$500
AP Clerks
create PO
from invoices
No
>$500 &
<$5000
Yes
Supervisor
approves PO
online
Director
approves PO
online
Is PO
value >$500
& <$5000
Yes
No
>$5000
CFO
approves PO
online
After Approval, POs sent back to AP
Copies are sent
to UMSA
Financial
analysts
complete the
receipts
Copies are
made and
paperwork
completed by
financial
analysts.
Checks are
cut at UMSA.
84
Vendors
receive
checks.
APPENDIX B
Process Map of GPO Model
There are 4 kinds of items sourced by this GPO affiliated
1. Inventory Items (Frequently ordered and officially booked inventory)
2. Non-Inventory Items:
a. Non Stocks----Not officially booked inventory (Not in the ledger,
frequently ordered)
b. Special Items-------- Not officially booked inventory (infrequently
accessed).
3. Services.
Process- Maps:
1. Inventory Items
Inventory
items ( max
stock—week
supply)
Vendors receive the
computer generated
orders to supply
Computer
system
decrements
inventory
Min. inventory
level--- (buffer
stock level)
EDI (Electronic Data
Interchange) system orders
vendors to supply the
inventory.
85
Re-order
requisition
generated by
the computer
system within
Requisition
converted to
POs
APPENDIX B (CONTINUED)
2. Non Inventory Items
Depts. Needs
non inventory
items.
Depts.
Generate
requisition in
the computer
system
Requisition
goes
through
approval
Requisition
s converted
to POs.
Dept.
Chairs
or
directors approve the
requisition
EDI (Electronic Data
Interchange) system
orders vendors to supply
the inventory.
Vendors receive the
computer generated orders
to supply
86
APPENDIX C
Process Map of Hybrid Model
Rebates
Department
(Requestor)
Accounts
Payable/GL
Delivery
Lawson
(Requisition)
VAT
Procurement
Sourcing
Purchasing
Price
Confirmation
87
Receiving
&
Distribution
Rebates