Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 41
4
Examples of Tools for Costing,
Economic Modeling, and Priority Setting
A
toolkit that could support countries in their decision making related
to chronic diseases must address a complex set of tasks, including
budgeting, planning, decision making, and priority setting, said ses-
sion moderator Rachel Nugent of the University of Washington. The first
session summarized in this chapter presented several models for costing and
other economic analyses that might represent useful components of a tool-
kit, recognizing that financing is a fundamental aspect of decision making
in every country. Therefore, any tool or process to support decision mak-
ing for chronic disease control needs to provide information about costs
to support budgeting as well as, ideally, information on cost effectiveness
and potential return on investment to help convince policy makers of the
benefits of allocating resources to support intervention. Even more ambi-
tiously, Nugent added, the toolkit might also be used to address more “big
picture” types of economic questions that may also play a role in priority
setting—for example, the relationship between the prevalence of chronic
diseases and economic development.
In addition to economic analyses, decision making is also influenced by
a range of other types of information and influences. The second session
summarized in this chapter explored two examples of tools to inform pri-
ority setting that incorporate data beyond costing and economic analysis.
One is a model that focuses on estimates of impact based on the anticipated
life-saving effects of interventions, which can be a powerful policy tool. The
other is a tool that can incorporate multiple criteria that influence decision
making, including empirical data (such as effectiveness and economic data)
as well as information that reflects values and preferences. The aim of the
41
OCR for page 42
42 COUNTRY-LEVEL DECISION MAKING
tool is to provide a way to systematically and transparently establish rank-
ings or comparisons among different intervention options across a broad
range of criteria.
The following sections summarize the content of each of the presenta-
tions in this session. Chapter 6 provides a summary of the considerations
raised in this session along with the presentations and discussions through-
out the workshop.
TOOLS FOR COSTING AND ECONOMIC MODELING
Developing a Country-Validated Price Tag for Chronic Disease Prevention
Knowing how much it would cost to prevent or reduce noncommu-
nicable diseases can be important for advocacy and to promote spending
from international donors, Andrew Mirelman of Johns Hopkins Univer-
sity commented, as well as to establish priority setting at the national and
subnational levels. He noted that efforts to calculate the costs of disease
burdens and preventive interventions—for HIV/AIDS, vaccines, and child
survival, for example—have become valuable tools for advocacy and for
priority setting.
Mirelman described an effort to develop a disease prevention price
tag using a cross-validation study in which international estimates of the
costs of preventing specific diseases and reducing specific risk factors are
compared with country-level data on costs for specific population- and
individual-based interventions. The study explored prevention costs for
noncommunicable diseases in 19 resource-poor countries and was carried
out primarily by a number of Centers of Excellence set up through the Unit-
edHealth Chronic Disease Initiative and the U.S. National Heart, Lung, and
Blood Institute (NHLBI).1 The Centers of Excellence are research institu-
tions in low- and middle-income countries that collaborate with academic
institutions in high-income countries in order to address chronic diseases.
(See Box 4-1 for a list of the centers and their university partners at the
time of Mirelman’s data collection.) The goal of the Centers of Excellence
program is to build research capacity within local institutions, and for the
past year these centers have participated in developing a country validation
approach to calculating the costs of prevention.
For this project, the teams used values from international databases
(including the World Health Organization [WHO] Comparative Risk As-
sessment for Burden of Disease, WHO-CHOICE reference pricing, the
Management Sciences for Health International Drug Price Indicator Guide,
1 Formore information see http://www.nhlbi.nih.gov/about/globalhealth/centers/ (accessed
October 2011).
OCR for page 43
43
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
BOX 4-1
UnitedHealth and NHLBI Collaborating Centers of Excellencea
Argentina (Instituto de Effectividad Tulane University
Clinica y Sanitaria)
Bangladesh (ICDDR,B) Johns Hopkins University
China (George Institute) Duke University
Guatemala (The Institute of Nutrition Johns Hopkins University
of Central America and Panama)
India (Bangalore) (Population Health McMaster University
Research Institute)
India (New Delhi) (Public Health Emory University
Foundation India)
Kenya (Moi University) Duke University
Peru (Universidad Peruana Cayetano Johns Hopkins University
Heredia)
South Africa (University of Cape Town) Harvard University
Tunisia (University Hospital Farhat National Public Health Institute,
Hached) Helsinki, Finland
U.S.–Mexico Border (Pan American University of Texas El Paso;
Health Organization) University of Arizona; Whittier
Institute of Diabetes San Diego,
CA
aThis list includes those centers that were in existence at the time of Mirelman’s data
collection.
SOURCE: NHLBI (2011).
and the WHO Global InfoBase). They then validated the data at the coun-
try level using a questionnaire and interviews with technical personnel. The
key variables were risk factor prevalence, intervention coverage, and unit
prices for drugs and health staff salaries. The researchers used demographic
projections based on United Nations (UN) data to estimate, for example,
growth in the elderly populations most likely to be affected by the diseases
being studied.
The approach used in this project had several important strengths,
Mirelman explained. Tailoring the analysis of risk reduction approaches to
the individual countries was important. The teams confirmed which medical
approaches were used in each country (such as which tool was typically the
first choice for targeting hypertension), the guidelines for treatment, and
other information. They used an iterative approach to investigate confus-
ing information and to fill in gaps, and they found that doing so influenced
OCR for page 44
44 COUNTRY-LEVEL DECISION MAKING
their results. Many of the countries became interested in the benefits of the
analysis, so the validation is ongoing. The researchers used multi-variate
sensitivity analysis to address uncertainties in measurement and in the data,
such as data concerning drug prices, epidemiological information, and com-
pliance. They were also able to integrate their findings with other estimates
of the burden of diseases and with cost-effectiveness analysis.
The approach also had a number of limitations, Mirelman said. The
team needed to make assumptions that were in some cases quite optimistic.
For example, the team assumed that prevalence data could be translated
into ideal professional protocols—that patients would be identified and an
intervention of some kind would be initiated—but “that’s supremely ideal-
istic,” Mirelman said. Thus, the price tag essentially answers the question
“If we could do everything right now, how much would it cost?” The model
used in this work also required assumptions about data that were not fully
available, such as the availability of medical personnel, and it could not
fully account for barriers to implementation. In general, there was “never
enough data at the country level,” Mirelman said. The model also did not
take into account the cost offsets likely to come from the health and societal
benefits of reducing disease prevalence. Furthermore, Mirelman said, the
findings are a yearly estimate of the cost of prevention—not a projection
into the future.
The researchers hope to build on the work that has already been done,
Mirelman said, by developing more comprehensive data calculating the bur-
den of diseases used to support cost-effectiveness analysis. The researchers
hope to develop a league table to rank potential interventions based on cost
effectiveness as well as decision weights for such criteria as disease severity
and equity (multiple criteria decision analysis, an approach described in a
subsequent presentation summarized in this chapter, is one way to develop
such weights, he noted). The researchers also hope to expand the program
to more countries.
The research done through these collaborations, Mirelman concluded,
can provide valuable support for decision makers. As an example, he cited
an analysis of future projections from data from China that showed that
even though the assumption has been that population-based approaches
are the most cost-effective, in that setting “you can get an equally good
buy with individual-based approaches, even though they are expensive, be-
cause you are targeting high-risk individuals, so you realize very high-level
effects.” Furthermore, participants noted that the value of such rigorous
research goes beyond advocacy. It can reveal significant differences across
countries, which could yield insights about variation in treatment guide-
lines, prices, and other issues.
OCR for page 45
45
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
Two Types of Economic Modeling
Tom Gaziano, of Harvard University’s Brigham and Women’s Hospi-
tal, started by explaining that his work focuses not so much on a research
ideal of the best possible thing to do, but rather on “what we are able to
do” with what is available. It is unlikely that there are perfect data for any
country, he commented, and the countries with the greatest need have the
greatest data challenges. Thus, real-world modeling requires the flexibility
to adapt questions to the available data—for example, addressing a less
specific aspect of cardiovascular disease.
Gaziano described a project conducted through the UnitedHealth/
NHLBI Centers of Excellence2 that was designed to determine the potential
cost to low- and middle-income countries of adhering to current interna-
tional blood pressure guidelines, which have set a goal of bringing people
with hypertension to a target blood pressure of 140 over 90 or lower. The
researchers also examined the potential savings that might come from three
different lifestyle changes that could lessen the need for medications: a re-
duction in salt intake, an increase in physical activity, and improvements
in diet. The researchers hoped the results would be useful for determining
policy recommendations for both individual countries and regions.
High blood pressure is a significant risk factor, Gaziano noted. It con-
tributes to at least 50 percent of cardiovascular disease, particularly stroke
and ischemic heart disease. Elevated blood pressure leads to a major finan-
cial burden from both the efforts to manage the high blood pressure and
the treatment of the health problems it causes. A variety of data regarding
the economic impact of treating heart attacks and stroke are available,
but much less information is available concerning treatment of individual
risk factors at a country level. Data on the global financial burden of
hypertension are also scarce, Gaziano added, but full compliance with
drug treatment is clearly expensive. Estimates of the cost of hypertension
as a percentage of total health care costs range from 7 or 8 percent to 20
percent, depending on the region, with heart attacks and strokes being the
largest drivers of cost (Gaziano et al., 2009).
In the study carried out through the Centers of Excellence, Gaziano
said, the basic protocol was to determine the total number of people eli-
gible for blood pressure treatment by country, assess the cost of treating
this population, and determine the effect of lifestyle interventions on the
distribution of elevated blood pressure. Estimating the number of people
who would not need treatment if the lifestyle interventions were available
would make it possible to calculate the net costs and savings associated
2 Formore information, see http://www.nhlbi.nih.gov/about/globalhealth/centers/ (accessed
October 2011).
OCR for page 46
46 COUNTRY-LEVEL DECISION MAKING
with each lifestyle intervention. The researchers studied the 19 countries
represented by the 10 Centers of Excellence and 14 additional countries,
and these 33 countries contained approximately 80 percent of the popula-
tion of low- and middle-income countries worldwide.3
The researchers sought new or confirming epidemiological data on
prevalence, awareness, treatment, and control of blood pressure, using the
WHO database as well as individual country and regional data. Data on
treatment costs came from the Centers of Excellence or, when this was not
available, from WHO and other global sources. It included the costs of an-
tihypertensive medicines if prescribed according to standard of care as well
as the costs for physicians and nurses, clinic time, treatment, and laboratory
work. The associated costs of a strategy to increase physical activity were
based on a mass media campaign in Australia, and those of population-
based strategies to reduce salt intake were based on existing published
estimates (Asaria et al., 2007). The centers provided data they collected on
costs for fruit and vegetables; this was supplemented by U.S. Department
of Agriculture data. To estimate the anticipated reduction in mean blood
pressure through treatment and the three lifestyle strategies, the team used
a variety of sources, including WHO and the available published literature.
The data on estimates of prevalence, awareness, and treatment were
“somewhat shocking,” Gaziano noted. Among the population covered
in the study (as noted, roughly 80 percent of the population of low- and
middle-income countries) it was estimated that approximately 600 million
people, or about 25 to 26 percent of adults, had hypertension. Only about
40 percent of them were aware that they were hypertensive, and of that
group, only 40 percent—approximately 120 million people—were being
treated. Furthermore, of that small percentage, only about 20 percent had
their hypertension adequately controlled.
Based on the combined results across countries, the cost to treat the
people whose blood pressure is currently not being controlled and to bring
them to the point specified by the international guideline would be about
$43 billion, with human resources representing a significant portion of
the cost. The calculated net savings from the physical activity intervention
were about $1 per person. For salt reduction, the net savings were $2 per
person. By contrast, the fruit and vegetable intervention would have a net
cost of about $80 per person rather than a net savings, although Gaziano
3 The 19 countries represented by the Centers of Excellence were Argentina, Bangladesh,
Belize, Chile, China, Costa Rica, El Salvador, Guatemala, Honduras, India, Kenya, Mex-
ico, Nicaragua, Pakistan, Panama, Peru, South Africa, Tunisia, and Uruguay. The 14 addi-
tional countries included in the study were Brazil, Czech Republic, Democratic Republic of
the Congo, Egypt, Ethiopia, Indonesia, Iran, Myanmar, Nigeria, Russia, Thailand, Turkey,
Ukraine, and Vietnam.
OCR for page 47
47
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
noted that the intervention could have other benefits as well, for example
in reducing cancer rates.
When looking at individual countries, the costs varied significantly
by country, and thus the per capita net savings that would result from the
different interventions varied as well. Based on the data provided by the
countries, the variability was quite high for the cost of medications, even
those that are generically available, and there was a big range compared to
the estimates from the Management Sciences for Health International Drug
Price Indicator Guide. “This is one of the striking findings,” said Gaziano.
There was also a wide range in laboratory costs, and outpatient visits
ranged anywhere from $20 up to about $120. Similarly, the estimated per
capita costs for the lifestyle interventions considered in the study ranged,
for example, from 4 cents to 30 cents for salt reduction and from $35 to
$300 per capita to increase fruit and vegetable consumption.
Indeed, one overall finding from this work was the significant varia-
tion in results from country to country. “It’s quite a broad range,” Gaziano
explained, “depending on what they were already doing and the level of
control they had, as well as how much they were spending on health care.”
Gaziano also described a second model which was used in a study con-
ducted for the World Economic Forum on the global economic burden of
noncommunicable diseases, with a particular focus on cardiovascular dis-
ease (Bloom, 2011). In this study the researchers began with a model of the
life course of cardiovascular disease. This life course approach is important
because such lifestyle factors as excessive salt intake, consumption of trans
fats, and insufficient physical activity may start to have effects early in life,
and these and other factors become risk factors in individuals, which in turn
increase the probability of disease. Primary prevention strategies at either
the population or individual level may help control these risk factors, and
secondary prevention or acute medical treatment—both of which are more
expensive than primary prevention—come into play if primary prevention
strategies are not effective.
The researchers developed a “decisional analytic model,” Gaziano said,
which involves assessing a population in terms of age and gender distribu-
tion, blood pressure status, smoking, diabetes, and cholesterol. The model
identifies those with differing levels of risk for cardiovascular disease, and it
indicates the proportions of each in the population, which provides the op-
portunity to consider different intervention options for each subpopulation.
This type of modeling can be used to predict cardiovascular disease
events, Gaziano explained, and costs can then be attached to the various
possible interventions. The result makes it possible to predict, given a popu-
lation with a particular distribution of risk factors, the number of events
likely in a particular period of time as well as the potential treatments and
costs. Using this approach, the researchers estimated a global cost of about
OCR for page 48
48 COUNTRY-LEVEL DECISION MAKING
$860 billion annually due to various aspects of cardiovascular disease,
including the management of blood pressure and cholesterol levels and the
treatment of ischemic heart disease, stroke, heart failure, and hypertensive
heart disease. Approximately 50 percent of the cost was due to health
care costs and the rest to lost productivity. The costs differ by country and
region, he added, ranging from a low of $20 or $30 per capita in devel-
oping countries to a high of $400 to $650 per capita in North America,
Western Europe, and developed Asian countries. In high-income countries,
a considerable portion of the costs are accounted for by acute, advanced
hospital care. Although $20 to $30 per capita in developing countries may
seem low, it could be the entire health care budget in many low-income
countries, so covering all costs of cardiovascular disease would be difficult
despite the fact that the cost would be relatively low compared to high-
income countries.
Gaziano added that the global cost of cardiovascular disease is likely to
rise to as much as $20 trillion over the next 20 years, given the projected
population growth and assuming no change in risk factor estimates. “These
are probably underestimates,” he added, “because we use mostly public-
sector pricing.” A number of other factors could indicate that the estimate
is low, he said. The analysis did not include rheumatic heart disease and
other cardiovascular conditions, for example, nor did it include devices
such as pacemakers and defibrillators or some other procedures that can be
quite expensive. The researchers assumed a low level of hospital access in
low- and middle-income countries, but that hospital access could improve,
which would “vastly affect the costs over time.”
During the discussions following his presentation, Gaziano and other
participants commented on how, from the perspective of a potential toolkit,
these models could be applied at the country level by using country-specific
estimates of the costs of interventions and by adjusting the anticipated
effects of treatment and lifestyle interventions based on how they would
actually be implemented in a country and the evidence for effectiveness in
a similar population or context. There is also the potential to expand the
models to use them to explore different scenarios in a country, such as set-
ting different treatment targets or shifting treatment costs by changes in the
system’s current guidelines or standards, such as using lower-cost personnel
or changing the frequency of clinic visits for managing treatment. “When
you do these models you are forced to look at all the individual components
and say, hmm, why are we spending so much on this part?” Gaziano said.
Thus, the models can generate data that could be used to consider options
for how to make optimal use of available resources.
OCR for page 49
49
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
The World Health Organization’s Costing Methods
In the presentation following Gaziano’s, Karin Stenberg of the World
Health Organization (WHO) began by agreeing that many assumptions
go into any model. In many cases the data are somewhat shaky, she said,
so it is important to “look at the different pieces of evidence that come
together.” This means, as the previous presentations indicated, that one
should both recognize the size of the problem and explore potential solu-
tions and priorities for investment. Once priorities are established, however,
it is also important to determine the costs of implementing the chosen in-
terventions at the intended scale.
To illustrate such costing, Stenberg described a WHO study led by
Dan Chisholm which examined the costs of scaling up interventions aimed
at noncommunicable disease control (Chisholm and Mendis, 2011). The
goal of the study was to develop a financial planning tool to aid countries
in the scale-up of these health care interventions, and the cost estimates
from individual countries were then combined to produce a global “price
tag” that illustrated the total cost of scaling up noncommunicable disease
interventions worldwide. The study included analysis of data from 42 low-
and middle-income countries;4 these countries account for 90 percent of
the noncommunicable disease burden in developing countries. The scope
of the costing study was limited to the diseases and risk factors highlighted
in WHO’s Action Plan for the Global Strategy for the Prevention and
Control of Noncommunicable Diseases, specifically cardiovascular disease,
diabetes, cancers, and respiratory disorders (asthma and chronic obstructive
pulmonary disease) (WHO, 2008a). To determine which interventions to
analyze, the researchers used previous work from WHO that had identified
cost effective, feasible, low-cost interventions that were also appropriate
to implement within the constraints of the local health system where they
would be used (Alwan et al., 2011). They defined the “best buys,” or very-
cost-effective interventions, as those that could add an additional year of
healthy life for less than the country’s annual per capita income. (Table
4-1 summarizes the 14 “best buys.”) Interventions that did not meet all of
these criteria but that still offered good value for the money and had other
attributes that recommended their use were considered as “good buys” and
were also included in the costing study.
4 Included in the study were 14 low-income countries (Afghanistan, Bangladesh, Côte
d’Ivoire, DPR Korea, DR Congo, Ethiopia, Ghana, Kenya, Myanmar, Nepal, Nigeria, Su-
dan, Uganda, Tanzania), 13 lower-middle-income countries (China, Egypt, India, Indonesia,
Iraq, Morocco, Pakistan, Philippines, Sri Lanka, Ukraine, Uzbekistan, Vietnam, Yemen) and
15 upper-middle-income countries (Algeria, Argentina, Brazil, Colombia, Iran, Kazakhstan,
Malaysia, Mexico, Peru, Romania, Russian Federation, South Africa, Thailand, Turkey,
Venezuela).
OCR for page 50
50 COUNTRY-LEVEL DECISION MAKING
TABLE 4-1 “Best Buy” Interventions
Condition Interventions
Tobacco use Tax increases; smoke-free indoor workplaces
and public spaces; health information/
warnings; advertising/promotion bans
Harmful alcohol use Tax increases; restrict retail access; advertising
bans
Unhealthy diet and physical inactivity Reduced salt intake; replacement of trans fat
with polyunsaturated fat; public awareness
about diet and physical activity
Cardiovascular disease and diabetes Counseling and multi-drug therapy (including
glycemic control for diabetes) for people with
> 30 percent cardiovascular risk (including
those with cardiovascular disease); treatment
of heart attacks with aspirin
Cancer Hepatitis B immunization to prevent liver
cancer; screening and treatment of pre-
cancerous lesions to prevent cervical cancer
SOURCES: Alwan et al. (2011), Chisholm and Mendis (2011).
The WHO costing study also took into account the readiness issues
discussed earlier, by including an assessment of the current strength of
the health system in a given country as part of the scaling-up process. For
example, the researchers assumed that low-resource countries would need
more lag time than others to put infrastructure and personnel for individual
interventions into place. For population interventions, they included an
assessment of current policies and how these policies are enforced. The
model then included activities needed to strengthen policy, planning, and
implementation. Table 4-2 shows the phases of policy development and
the sorts of resources needed in each phase. The researchers developed cost
estimates for each of these elements.
The costing method was straightforward, Stenberg said, and it was
similar to those described in other previous presentations. The researchers
calculated the relevant variables: population; prevalence (percent of popula-
tion with disease or risk factor, by age and sex); current and target cover-
age (percent of population in need of intervention); resource use (resources
needed to implement an intervention); and cost per unit of resource use.
For example, in a country with a population of 1 million and a 20 percent
prevalence rate for smoking, the population in need of intervention would
be 200,000 individuals. An intervention that costs $1 per patient per year
OCR for page 51
TABLE 4-2 Resource Needs Matrix for NCD Policy Instruments
Human Supplies and
Stage of Policy Development Resources Training Meetings Mass Media Equipment Other
Planning (year 1) Program Strategy/policy Stakeholders Office Baseline survey
management; analysis equipment
administration
Development Advocacy; law Legislation Intersectoral Awareness Opinion poll
(year 2) collaboration campaigns
Partial implementation Inspection Regulation Monitoring Counter- Vehicles, fuel
(years 3-5) advertising
Full implementation Enforcement Evaluation Follow-up survey
(year 6 onward)
NOTE: NCD = noncommunicable disease.
SOURCE: Stenberg (2011a).
51
OCR for page 52
52 COUNTRY-LEVEL DECISION MAKING
would cost $200,000. If the country were to begin with a 50 percent scale-
up, the cost would then be $100,000. The researchers produced estimates
for each of the 42 countries, using country data as well as standard assump-
tions, and then totaled these numbers to come up with the global price tag
of noncommunicable disease interventions.
This particular study concentrated on the worldwide totals of certain
interventions; however, perhaps its most useful product when it comes
to actual health care planning is the tool that was developed to ana-
lyze country-specific costs of noncommunicable disease interventions. The
global estimates of the noncommunicable disease burden that the WHO
study produced are useful for advocacy, to demonstrate need and garner
additional resources, Stenberg noted, but the average costs are not very
relevant to individual countries. Similar to the experience with the other
models presented, the WHO study showed that the cost of implementation
of a given intervention varies widely among countries and thus the study
highlighted the need to tailor planning to individual countries rather than
to make assumptions based on global estimates. During the course of the
study, templates were developed that each country could use for more de-
tailed costing by plugging in more information and changing the assump-
tions as needed. These templates could be a valuable asset for country-level
planning, particularly because of the care the researchers took in providing
ways to tailor the analysis to specific circumstances.
One of the main strengths of the model, Stenberg said, is its capacity to
provide a comprehensive assessment of a broad range of both public health
and primary care interventions, considering both the “best buys” and
“good buys.” The researchers used the most current country policies and
health care systems in their analyses, which, Stenberg noted, is an approach
that could be useful in the further development of country-level tools for
the planning of noncommunicable disease control. The model also uses a
standard methodology that has been used in other WHO programs, which
makes it very easy to compare findings across diseases and interventions.
On the other hand, Stenberg said, the results will not reflect assessment
of health gains because the researchers were not able to model impact in the
available time. The model also does not include changes in epidemiology
over time, so it does not reflect the decrease in prevalence that could be
expected once interventions are implemented or the cost savings related to
such decreases. It also does not include medical personnel training costs. In
addition, as was the case with the other models, the WHO work was based
on sometimes idealized assumptions and used data inputs that could be bet-
ter validated by the countries. Stenberg also noted that countries might wish
to model other interventions that fit their needs better than those selected.
In the future, Stenberg said, it is likely that the templates will be made
available to countries, and that this noncommunicable disease work will be
OCR for page 53
53
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
integrated with another project, the OneHealth model, which is a costing
tool designed to assess public health needs in low- and middle-income coun-
tries.5 Developed by a UN interagency working group, this model is part of
an effort to standardize approaches to costing within the network of UN
agencies so that results can be compared and planning can be integrated.
The OneHealth model is also intended to address the growing awareness
of the importance of considering the health sector in national planning and
of using national health plans as a mechanism for coordination and for
ensuring that donors’ efforts are harmonized with local agendas.
Analysis has shown, Stenberg said, that a significant majority of the
new resources that will be needed in low- and middle-income countries
between 2009 and 2015 will be required for strengthening health systems
(McCoy, 2009). However, the disease programs in many countries operate
independently and develop their plans without considering the timing of
other health programs, the national health plan, or the overall development
plan for the country. This lack of synchronicity among various health plans
in different countries can be seen clearly in the WHO planning cycle data-
base, Stenberg said, which tracks the development of different health plans
across the world.6 For example, in Afghanistan, the National Health Plan
covers the years 2007-2013, the immunization plan covers 2011-2015, the
TB plan covers 2009-2013, and so on.
The OneHealth model is a tool intended to support medium-term
planning and promote integration. Its focus is on the public sector, but it
also allows for private-sector activities to be incorporated. The intended
audience is health-sector planners, disease-specific program planners, non-
governmental agencies, and donors. Six UN agencies are engaged in the
development of the tool, along with experts for each key area, who provide
technical assistance. Representatives from several countries have also been
involved in the development process to ensure that it will be useful for
individual countries.
Figure 4-1 shows the basic framework covered by OneHealth. Six
health system components form the building blocks; the bars in the center
represent the levels at which action can be taken. The model is modular, so
it can be adapted for different purposes. An additional benefit of its flex-
ibility is that it encourages the involvement of experts in particular areas
to conduct the planning for their domains, even while the model’s structure
keeps the whole system integrated.
Stenberg acknowledged that there are already many tools and models
5 For more information, see http://www.who.int/pmnch/topics/economics/costing_tools/en/
index4.html (accessed November 2011).
6 For more information, see http://www.internationalhealthpartnership.net/en/home (accessed
November 2011).
OCR for page 54
54 COUNTRY-LEVEL DECISION MAKING
HS4. Health HS5. Governance HS6. Financing
Information and Leadership Policies
Dis eas e P rograms
National
Hospital
Reproductive Health
Immunization
Child Health
…NCDs…
Nutrition
Health Center
Malaria
WASH
HIV
TB
Outreach
Community
HS1. Infrastructure HS2. Human
HS3. Logistics
and Equipment Resources
FIGURE 4-1 OneHealth framework.
NOTE: HIV = human immunodeficiency virus; NCDs = noncommunicable diseases;
TB = tuberculosis; WASH = water, sanitation, and hygiene.
Figure 4-1, color
SOURCE: Stenberg (2011b).
available for health planning. She believes, however, that this particular one
is important not only because it offers the possibility of coordination across
agencies, countries, and other units, but also because it is the first to “bring
together disease-specific planning with health systems in a unified way.”
OneHealth allows a person “to do a situation analysis, look at the capacity
of the health system, look at different strategies, do priority setting, and
look at financial implications.” OneHealth also incorporates some of the
UN’s epidemiology impact models, such as the Lives Saved Tool (LiST) and
the AIDS Impact Model (AIM), which can be used to demonstrate achiev-
able health gains and to predict reductions in disease prevalence resulting
from specific health care models.
The software is also very user-friendly, Stenberg added. The user can
adapt the model to local circumstances and chose the level of detail that
is most useful for a given purpose. For example, the model might be used
to answer such questions as “What set of interventions will have a desired
impact in my setting? What constraints in my system need to be addressed
OCR for page 55
55
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
before I can scale up a promising intervention? How feasible would a given
intervention be if I have to adapt in certain ways in order to implement
it in a given setting? What funding will I need to accomplish X?” A user
might want to cost and budget a plan that has been already developed, or
perhaps compare alternative scenarios. “It helps to make the planning more
realistic, as opposed to setting very ambitious targets that you may not be
able to achieve,” Stenberg said. For example, a user with a target of scaling
up an intervention to 90 percent coverage for a given disease or risk factor
can click on the human resources module to determine whether the health
system has a sufficient number of nurses or community health workers to
deliver the intervention. If not, the user can adjust the target and explore
other pathways for scaling up the intervention.
The OneHealth model faces the same data challenges that affect the
other models discussed, Stenberg said. Another challenge is ensuring that
each country has the capacity to take full advantage of the tool’s possi-
bilities. Despite the challenges, Stenberg concluded, the OneHealth model
provides a common platform and consistent methods for countries to use
and a way to ensure that their health systems’ capacity is what drives plan-
ning and priority setting. As funding permits, the development team will
continue to add new elements to the model, such as a health information
systems module and models for health gains for noncommunicable diseases.
Stenberg closed with her recommendations for designing and applying
a costing model as part of a toolkit:
• Be very clear about the specific policy questions to be answered,
how the tools will be used and by whom.
• Focus on broad health sector planning processes and ways to inte-
grate across programs.
• Don’t overlook the need to invest in capacity building, advocacy,
communication, and training in how to use the tools.
PRIORITY SETTING TOOLS
The Lives Saved Tool for Maternal and Child Health
The purpose of the Lives Saved Tool (LiST)7 is straightforward, ex-
plained Neff Walker of Johns Hopkins University. It is intended to estimate
the impact that increasing health coverage has on maternal and neonatal
health, child mortality, and stillbirths. It is a computer-based tool that
countries or program developers can use to estimate the relative impact of
a wide range of possible interventions and levels of coverage for purposes
7 For more information, see http://www.jhsph.edu/dept/ih/IIP/list/ (accessed November 2011).
OCR for page 56
56 COUNTRY-LEVEL DECISION MAKING
of strategic planning. It is incorporated in the OneHealth model discussed
above.
To use LiST, users begin by plugging in data for a particular country
or region, such as neonatal and maternal mortality rates, current health
coverage and interventions, and background information (e.g., vitamin A
or zinc deficiencies or exposure to P. falciparum). Data on the effectiveness
of many interventions, in terms of reducing either a cause or a risk factor
of maternal or child death, are already programmed into the model, which
currently includes more than 20 causes of death and risk factors (e.g., stunt-
ing, wasting, and intrauterine growth restriction).
The program has data for 85 low- and middle-income countries as
well as for individual states in large countries. The demographic data are
from the UN Population Division, the cause of death data are from WHO
estimates, and the mortality rates come from the Inter-Agency Group for
Mortality Estimation. The effectiveness values are from WHO’s Child
Health Epidemiology Reference Group, and the data on coverage are from
several sources: the Department of Homeland Security, the Multiple Indica-
tor Cluster Survey, the Malaria Indicator Survey, the United Nations, and
WHO/UNICEF estimates of vaccine coverage. The data also cover coun-
tries’ actions regarding HIV/AIDS and family planning, which influence
maternal and child outcomes.
As an example of how the tool can be used, Walker said that a user
focused on vaccines could assess the impact on mortality of increasing
pneumococcal and rotavirus vaccination. Such vaccines would likely have
little impact on maternal mortality but could have a significant impact in
some countries on infant mortality. The model is structured to make it pos-
sible to compare multiple scenarios—for example, comparing the impact
of 80 percent coverage of pneumococcal vaccine with 80 percent coverage
with antibiotics. The model also allows users to consider the outcome if
two or more interventions were scaled up at the same time and to generate
a variety of counter-factual scenarios—that is, asking what would happen if
an alternate course were followed. The scenarios also help users anticipate
unexpected outcomes, Walker noted. For example, an effective interven-
tion that reduces neonatal mortality might indirectly increase malaria rates
because if more infants survive, more may be exposed to malaria, unless
there is an increase in anti-malaria efforts as well.
The model is fairly simple, Walker said, but once all the factors are
combined, the result is still rather complex. As an illustration of this com-
plexity, Figure 4-2 depicts all of the factors that have an effect on pneu-
monia mortality, some of which are indirect. For example, the actions in
the upper left related to hygiene reduce the incidence of diarrhea, which in
turn lowers rates of stunting, which is beneficial because children whose
growth is stunted have a significantly increased risk of dying of pneumonia.
OCR for page 57
57
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
Hib
Pneumococcal vaccine
Improved H2O source
vaccine (0.26)
Breast
(0.18)
within 30 minutes
feeding
(0.17)
promo on
Hand washing
Breast feeding, e.g., RR 15.13
with soap (0.48)
for no breast feeding for first 6
Improved excreta
months
disposal (latrine, toilet)
(0.36)
Diarrhea
Pneumonia
Water connec on
incidence
in the home (0.63)
mortality
(RR 1.025)
Stun ng
Hygienic disposal of
children’s stools
(-2 Z score: RR 1.3,
(0.20)
-3 Z score: RR 3.2)
Complementary feeding
educa on/supplementa on
IUGR
(OR
Zinc for
Mul ple micronutrient 21.6)
Was ng
supplementa on (0.09)
preven on (-1 Z score: RR 1.6
Balanced energy
(0.15) (OR -2 Z score: RR 4.2
supplementa on (0.32) -3 Z score: RR 8.7)
1.18)
Oral an bio cs
Pregnant women protected via
IPT or sleeping under an ITN
for pneumonia
Therapeu c
(0.35) feeding
(0.7)
(0.062)
FIGURE 4-2 Factors and weights used in the LiST model that effect pneumonia
mortality.
NOTE: Hib = haemophilus influenza type B; IPT = intermittent preventive treat-
ment; ITN = insecticide-treated mosquito net; OR = odds ratio; RR = relative risk.
SOURCE: Walker (2011).
Figure 4-2, color
Similarly, interventions that affect intrauterine growth (lower left corner)
also reduce stunting. The current model is not complete, however, and a
participant pointed out a few elements that are missing from the current
model, such as tobacco use, indoor air pollution, and gestational diabetes,
all of which influence birth weight and infant death.
LiST has been used by many large organizations, such as WHO,
UNICEF, the U.S. Agency for International Development, the Global Fund,
and Save the Children, for priority setting and to support their advocacy,
Walker said. More than 40 developing countries have used it to support
their strategic planning, though only six or seven have used it as part of
their national planning processes. LiST has also been used for the evalua-
tion of programs, for example by the Global Fund and Roll Back Malaria.
There are several keys to success for LiST and other such models,
Walker said. First, it is critical to have an ongoing system for developing
and updating the assumptions that are part of the model. It is also impor-
OCR for page 58
58 COUNTRY-LEVEL DECISION MAKING
tant that the model be easy for users to learn. It should require not more
than about 2 days of training for the users to be able to easily change the
default values as they modify the model for the circumstances they are
assessing, and it should be available in multiple languages. LiST is strong
in those areas, Walker noted—it is now available in four languages, for
example—but it is also important to have some sort of organizational
backing for the model to work with individual countries, and LiST is just
now being adopted by UNICEF and WHO. Another important factor is to
have published evidence of a model’s effectiveness, Walker said, and LiST
satisfies that criterion as well.8 The most difficult key to success, Walker
said, may be to ensure that the model harmonizes with other models and
approaches. Because of the involvement of WHO and UNICEF, LiST has
been integrated with several models with broader scopes, but there are
many other disease-specific models as well, so the harmonization of LiST
with other models remains a challenge. Walker’s last word of advice was
that it is very important to “define your primary task and try to stick to
it—don’t let mission creep take over.”
A Multi-Criteria Decision Analysis Framework
The key question in setting health system priorities, said Mireille
Goetghebeur of BioMedCom and the EVIDEM collaboration, is which
interventions will contribute most to an equitable, efficient, and sustainable
health care system. To answer this question, she said, it is necessary to con-
sider both what should be done and what can be done. To tackle those two
questions, it is useful to have a mechanism to rank or compare a range of
possible interventions across a broad range of criteria. Multi-criteria deci-
sion analysis (MCDA) provides a tool for doing precisely this by assigning
weights to a range of relevant and possibly conflicting criteria. Goetghebeur
described a particular Web-based framework for applying this approach
to decision making and priority setting for health care developed by the
EVIDEM Collaboration.9 The EVIDEM Collaboration of researchers and
decision makers from a variety of countries has developed a decision-
making framework that is available on the Internet and is supported by
a Web registry of research on health care interventions and a discussion
forum. EVIDEM is intended to develop a community of MCDA practice,
8 For more information, see http://www.jhsph.edu/dept/ih/IIP/list/applications.html (accessed
February 2012).
9 EVIDEM was founded by researchers at BioMedCom, a consulting firm that specializes in
economic analysis and its application in the health sector, and its board of directors includes
policy makers, health care professionals, patients, researchers, members of the health care
industry, and other specialists. For more information, see https://www.evidem.org/ (accessed
November 2011).
OCR for page 59
59
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
Goetghebeur said, in which researchers and users develop, apply, and adapt
the tools in a continuous, open, and nonproprietary fashion.
EVIDEM began with a generic framework for assessing and ranking
interventions, based on an adaptable set of criteria. There are two modules,
the MCDA core model, which is a universal template, and the contextual
tool, which allows users to adapt it to specific circumstances. The core
model is based on four principles: that the criteria should be complete,
should have minimum overlap among them, should be mutually indepen-
dent, and should be operational (National Economic Research Associated,
2005). Those principles yielded a set of 15 universal normative criteria in
the core model, based on the assumptions that the highest value or priority
should be assigned to interventions that
• address severe diseases;
• address common diseases;
• address diseases with many unmet needs;
• are recommended by expert consensus;
• confer major improvements in efficacy/effectiveness over current
standard care;
• confer major improvement in patients’ perceived health over cur-
rent standard care;
• either confer major risk reduction or major alleviation of suffering;
• result in savings in health care intervention, medical, or non-
medical expenditures; and
• are supported by sufficient data that are fully reported, valid, and
relevant.
To address the question of what can be done in a given context, there
is a contextual tool with six criteria to help users define objectives and
priorities of the population as well as feasibility. According to this tool, the
decision-making process must address the following issues:
• Scope and mission of the health care system or plan
• Priorities for populations and access
• Opportunity costs (interventions foregone) and affordability
• System capacity (e.g., infrastructure, skills) and appropriate use of
intervention
• Political/historical context (e.g., cultural acceptance, precedents)
• Pressures/barriers from health care stakeholders
To use the framework, a user would first assign weights to the criteria,
and then, based on these weights, score and rank the potential interven-
tions. The contextual module can then be used to factor in the other
OCR for page 60
60 COUNTRY-LEVEL DECISION MAKING
elements (discussed above), and a financial tool is used to consider afford-
ability and related issues. To demonstrate the process, Goetghebeur showed
the workshop audience a prototype that is available to demonstrate the
process,10 and provided a link to additional prototypes that serve as ex-
amples of the application of the framework.11
One workshop participant noted that certain assumptions about shared
ethical preferences seem to underlie the framework and wondered how
weights are assigned to ethical or value-based criteria. The weights are
defined by the users, Goetghebeur replied, and the overall weighting would
reflect the preferences expressed by each of the stakeholders involved in the
process (who might be asked to systematically rank possible considerations
in order of priority). For each of the broad criteria, she added, there is a set
of sub-criteria designed to help users tailor their responses. The tools are
also evolving in response to user feedback.
Another participant noted that the framework appears to value inter-
ventions individually and wondered how the framework addresses interac-
tions among different approaches—that is, considering whether or how the
implementation of one worthwhile intervention might affect assessments
of how reasonable another might be. Goetghebeur responded that the as-
sessment of current interventions and what they are contributing is part of
the framework.
There are a number of applications for the framework and for the
information on the Web site, Goetghebeur said. Policy makers, physicians,
patients, researchers, and developers of new health care programs and in-
terventions all might use the framework to find information and make deci-
sions. In New Zealand and Italy, the MCDA tool is being used to assess the
reimbursement or implementation of health technology and drugs. At the
level of health care professionals, the tool has been used to develop clinical
practice guidelines, with the goal of making a link between the guidelines
and the decision making at the regulatory and reimbursement level. The
tool can also be used for identifying priority research questions and data
needs. The MCDA framework can also be used to inform the develop-
ment stage for new health care interventions or new health care programs.
Finally, the framework can be used as a tool to communicate validated
information to a range of stakeholders in a digestible format.
Goetghebeur identified some of the program’s key strengths and limita-
tions, organized into four main areas:
10 The interactive demonstration prototype is available at http://www.evidem.org/
tiki/?page=DEMO-main.
11 Additional example prototypes are available at http://www.evidem.org/evidem-
collaborative.php.
OCR for page 61
61
TOOLS FOR COSTING, ECONOMIC MODELING, AND PRIORITY SETTING
Utility to Policy Makers
• Adaptable to local context
• Systematizes decision-making process
• Combines quantitative and qualitative inputs
• Identifies applicable criteria and perspectives
• Based on a wide set of criteria
• Transparent
But
• Perceived as very complex
• May be difficult to integrate with existing processes
• There is a risk that MCDA may be used in a formulaic way rather
than as a support to priority setting.
Methodology
• Pragmatic, user-friendly and modular
• Instructions are detailed
• Open-source—so users benefit from others’ work
But
• Criteria selection and weighting process may be challenging.
Data Requirements
• Comprehensive but modular
• Open web registry—so users benefit from others’ work
But
• The Web registry is just in a beginning phase.
• Data synthesis by criteria may be challenging.
Capacity and Training Requirements
• A testing package is available in toolkit.
• There is a growing community of developers and users.
OCR for page 62
62 COUNTRY-LEVEL DECISION MAKING
But
• Expertise with MCDA is limited in the health care sector.
The EVIDEM framework, Goetghebeur concluded, provides a mecha-
nism for priority setting that is transparent and consistent and that can help
users identify the interventions that will contribute most to sustainable and
efficient disease control and that will reflect the priorities and preferences
of decision makers across a wide range of criteria.
