The Methods of Cost-Effectiveness Analysis to Inform Decisions about the

Use of Health Care Interventions and Programs

Oxford Handbooks Online

The Methods of Cost-Effectiveness Analysis to Inform
Decisions about the Use of Health Care Interventions
and Programs
Simon Walker, Mark Sculpher, and Michael Drummond
The Oxford Handbook of Health Economics
Edited by Sherry Glied and Peter C. Smith

Print Publication Date: Apr 2011

Subject: Economics and Finance, Health, Education, and Welfare, Econometrics, Experimental
and Quantitative Methods
Online Publication Date: Sep 2012 DOI: 10.1093/oxfordhb/9780199238828.013.0031

Abstract and Keywords

Health care systems exhibit their features and nuances that impose constraints on the
appropriate way to analyze interventions and make decisions about their implementation.
This article seeks to explain the theoretical foundations of these methods and their
implementation in practice. It focuses on a budget constrained health care sector, where
implementing a new technology with additional costs will result in other health care
services being displaced hence forgoing health improvement elsewhere. It is intended to
provide grounding in the policy motivation for economic evaluation and the underlying
normative issues in undertaking studies. The article further develops the key elements of
undertaking economic evaluation in practice. It outlines a simple scenario regarding the
comparison of chemotherapy treatments for breast cancer. This example helps to
illustrate some of the conceptual issues raised in this article. The approaches that have
been taken to quantify outcomes and costs from health care interventions are also
discussed.

Keywords: health care systems, costs, economic evaluation, decisions, normative issues

Page 1 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs

31.1 Introduction
THE developed world is faced with rising real health care costs. An important driver of
increased cost pressure is the arrival in the market of new medical technologies such as
pharmaceuticals and devices. The question facing health care systems is whether these
new technologies should be funded by collectively funded health care systems and made
available to patients. A range of other resource allocation decisions face health systems,
including whether to invest in new equipment and buildings and the appropriate
organization of services.

In standard economic theory, the supply and demand for new technologies and health
services in perfectly competitive markets would result in efficient outcomes (so called
“market clearing” prices and quantities). However, health and healthcare have many
characteristics which, taken together, limit the functioning of markets (Arrow 1963).
These include significant uncertainty (about both the occurrence and recovery from
illness); limits to effective individual choice arising from asymmetries of information
between healthcare (p. 734) professionals and patients and individuals not facing the full
cost of their care at the point of delivery; and the presence of significant ineradicable
monopoly characteristics among providers. As a result health care markets, without
government intervention, are unlikely to produce efficient solutions. Furthermore,
unregulated markets for health care raise a number of profound distributional issues as
individuals' use of services is necessarily governed by their willingness and ability to pay
rather than medical need. Private insurance is a partial market-based response to this but
insurance itself is characterized by potential inefficiencies (such as moral hazard; Nyman
2006) and fails to address equity concerns if cover is unavailable or prohibitively
expensive for those with pre-existing or chronic health conditions.

The basic economic problem is often described as a problem of unlimited wants but with
limited resources to satisfy them. This applies as much to health care as to any other
market, but the shortcomings of free markets in health care, on the basis of efficiency
and equity, means that alternative ways of allocating resources need to be established—
that is, choices must be made about which health care interventions to provide and which
not to provide within a health care system. The methods of economic evaluation have
been developed to help to inform these choices.

Economic evaluation can be defined as “the comparative analysis of alternative courses of

action in terms of both their costs and consequences” (Drummond et al. 2005). The
economic evaluation of health care technologies has become a central policy tool in many
health care systems (Tam and Smith 2004). These methods have most notably been
adopted to inform decisions about the funding of expensive new medicines—for example,
by the National Institute for Health and Clinical Excellence (NICE 2008) in the UK.
However, they have also been applied to inform decisions as disparate as investment in

Page 2 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
workplace health (Tompa et al. 2008) and safety programs and incentivizing family
doctors to change practice (Mason et al. 2008). This chapter seeks to explain the
theoretical foundations of these methods and their implementation in practice.

Health care systems around the world are not homogenous. Each exhibit their own
features and nuances which imposes constraints on the appropriate way to analyze
interventions and make decisions about their implementation. This chapter will, in the
main, focus on a budget constrained health care sector, where implementing a new
technology with additional costs will result in other health care services being displaced
hence forgoing health improvement elsewhere. While not all health care systems operate
under such a “hard” budget constraint in the long run, in the short run they must make
trade-offs within the health care sector. This simplification also makes easier the
exposition and interpretation of what is presented in this chapter.

31.2 Principles of Choice in Health Care

Although all economic evaluation methods have formal links to the principles of
microeconomic analysis, there are several different normative viewpoints present in the
field—that is, alternative frameworks for judging whether one allocation of resources
(p. 735) should be judged socially preferable to another. The choice of normative

viewpoint has an impact on analytic methods and the types of recommendation likely to
flow from the analysis. These normative viewpoints can be classified broadly as either
“welfarist” or “non-welfarist” as described briefly below.

31.2.1 Welfarism

Welfarism is based on orthodox neoclassical welfare economics, the central principles of

which are: (i) the welfare of society is based on the utility (well-being) of the individuals
in the society; and (ii) individuals are the best judges of their own well-being (Broadway
and Bruce 1984). In principle, interpersonal comparisons of utility are not permitted.
Instead improvements in societal welfare can only come about when a change (for
example, a change in policy) results in at least one individual being better off and no
individual being made worse off (a so-called “Pareto” improvement). However, policy
decisions which produce winners but no losers are likely to be few and far between

For this practical reason the welfarist framework has been extended using the concept of
“potential Pareto improvement.” Developed by Kaldor 1939) and Hicks 1939) the idea is
that if the “gainers” from a policy change, those individuals made better off by a change,
can potentially compensate the “losers,” those made worse off, such that no individual
would be worse off and therefore such a change would result in an increase in social
welfare. Note the emphasis on “potential compensation”: if compensation was actually
paid this would be equivalent to a Pareto improvement. The Kaldor-Hicks compensation

Page 3 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
test forms the normative basis for a widely used approach to economic evaluation (known
as cost-benefit analysis) as used, for example, in most appraisals of new transport
investments (Mishan 1971).

The welfarist approach is open to criticism (Sen 1979). For many people the assumptions
underpinning it are too restrictive and the notion that social welfare is only a function of
individual utilities too narrow (Sen 1979; Culyer 1991). Welfarism has also been criticized
because of the mechanisms needed to employ it as a method for economic evaluation. In
general, utilities are expressed through individuals' willingness to pay; that is, the largest
amount an individual would be willing to pay for a change in policy. Willingness to pay is
often calculated from market prices (if these are prices arising in competitive markets) or
imputed using other methods if market prices are not available (“shadow prices”), but
these methods raise major issues in healthcare. First, an individual' willingness to pay is
likely to be closely related to their ability to pay, but when individuals' incomes are widely
dispersed the sum of willingness to pay values may not actually reflect aggregate utility.
This is because those with higher income are prepared to pay higher prices but they are
likely to derive less utility from their last unit of income (termed a lower marginal utility
of money). One response is to give greater weight to a pound spent by a poor person than
by a rich person—but all attempts to reweight these “pound votes” are likely to prove
controversial (Lerner 1944).

Second, as a result of problems in the functioning of the market for healthcare (e.g.
asymmetries of information) consumer sovereignty may be compromised, and this raises
(p. 736) still further issues about whether the willingness to pay of individuals actually

reflects the utility they derive. Applications of welfarism are also usually based on the
assumption that we are in a “first best” world, one in which markets are perfectly
competitive and result in efficient outcomes. In such a world, at the market clearing price
the marginal social benefit is equal to the marginal social cost. Thus the price represents
the opportunity cost in terms of both demand and supply and as such the market price
can be used to value both the outcomes and the resources used (where the opportunity
cost represents the value of the next best alternative forgone). Several authors (Sculpher
et al. 2005; Claxton et al. 2007) have argued that this first best world does not exist, for
example as a consequence of the existence of externalities whereby individuals do not
incur the full cost of their actions, and as such making decisions based on prices that are
assumed to be derived in a first best world may actually reduce social welfare (Diamond
and Mirrlees 1971a and 1971b).

31.2.2 Non-welfarist approaches

Non-welfarist approaches, which can take several forms such as extra-welfarism or social
decision-making, exchange the notion of the objective being the maximization of some
aggregation of individual utilities for another set of objectives. When considering
decisions in health care this generally involves maximizing some target measure of
health, though this does not necessarily have to be the case. Some would argue that the

Page 4 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
non-welfarist approach has been influenced by Sen's capability theory (Sen 1993). The
theory emphasizes functional capabilities, i.e. what a person can do or can be, instead of
the more standard welfarist emphasis on utilities. However, in its general use in the
economic evaluation of health care, adopting a non-welfarist approach has led to a focus
on a much narrower objective, i.e. health alone, rather than the wider ranging measures
of well-being implicit in Sen's theory.

When making decisions in health care, the use of an objective function under which
health is maximized may be controversial given that this would appear to negate
individuals' preferences and can thus appear paternalistic. However, if such objectives
arise from some form of legitimate process, for example from the decisions of a
democratically elected government, or through the voluntary purchase by a consumer of
an insurance plan which makes decisions on the basis of such an objective function, then
it is more difficult to interpret the function as paternalistic. Although such functions
clearly involve strong assumptions and social value judgments, Claxton et al. (2007)
argue that these value judgments can be seen as explicit.

31.3 Making Choices

In selecting a welfarist or a non-welfarist approach to making resource allocation
decisions in health care, a series of value judgments are effectively being made about
what should matter in terms of both outcomes and opportunity costs. In either approach
an (p. 737) objective function is being maximized, whether that be the summation of
individual utilities or, for example, the health of the population, subject to one or more
constraints. Different choices may be made under each approach as both the outcome
and the opportunity cost could be valued very differently by the methods associated with
each of the approaches.

Although rarely used in health care, cost-benefit analysis (CBA) is an approach to

economic evaluation which has its roots in the welfarist normative framework (Sugden
and Williams 1979). Although there are many variants, in general CBA quantifies the
costs of interventions in terms of their impact of resources and takes market prices where
possible to value these. In principle, all outcomes would be included with valuation based
on the willingness to pay of individuals who stand to gain or lose from a new policy. These
methods rarely explicitly consider the constraints imposed by finite budgets.

Page 5 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
The non-welfarist
paradigm predominates in
the economic evaluation
used by most health care
systems to inform resource
allocation. NICE in the UK,
for example, seeks to make
decisions which are
consistent with maximizing
health subject to a fixed
Click to view larger
NHS budget using a set of
figure 31.1 Assessing the cost-effectiveness of an
tools called cost-
intervention given an objective of maximizing health
subject to a fixed budget effectiveness analysis
Source: Claxton et al. 2008. (CEA) (NICE 2008). This is
illustrated in Figure 31.1
below which shows the comparison of a new intervention (which could be generalized to
any particular new policy) compared with existing practice. The y-axis shows the
additional cost of the new intervention when it is priced at 3 alternative levels (P⋆, less
than P⋆ or greater than P⋆). The x-axis shows the gain in health (p. 738) generated by the
new technology in terms of individuals' survival duration adjusted for the quality of life
they experience—so called quality-adjusted life-years (QALYs) which are discussed in
more detail below.

Consider first the situation where the new intervention generates two QALYs per patient
but at an additional cost of £20,000 per patient (based on a price 〈 P⋆). Given the fixed
budget, to fund this new intervention another existing service must be displaced
elsewhere in the health system (possibly in a quite different clinical area). This
displacement generates an “opportunity cost”: forgone health for another group of
patients who no longer receive the service. The cost-effectiveness “threshold” reflects the
estimate that, for every £20,000 generated through displacement of one or more existing
service, the opportunity cost is 1 QALY. In this first scenario, then, the additional cost of
the new intervention has an opportunity cost of 1 QALY but itself generated two QALYs,
hence its introduction would be cost-effective as there is a net gain in health of one QALY
from the same fixed budget. In contrast, Figure 31.1 shows that, if the additional cost of
the new intervention is £40,000 (due to a higher price = P⋆), then the benefit of the new
treatment is exactly offset by that lost elsewhere; the health system gains nothing from
funding the new intervention. If the additional cost per patient is £60,000 (with a price 〉
P⋆) then the opportunity cost, resulting from the displacement of existing services to fund
the new intervention, is greater than the additional health it generates (3 QALYs versus
two QALYs); this could not be considered cost-effective as there is a net loss in health.

CEA (or its variant, cost-utility analysis) is the most widely used form of economic
evaluation in health care. The NHS Economic Evaluation Database (See 〈http://
www.crd.york.ac.uk/crdweb/〉) shows that, of 7212 published full economic evaluation
studies in the database, over 98 percent used some form of cost-effectiveness analysis.
Page 6 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
The way these studies are used to inform decisions varies. For example, some systems do
not have a “hard” budget constraint so the opportunity cost of funding new (more
expensive) interventions may fall, in part at least, outside the health system (e.g.
additional insurance premiums or taxation, or reduced expenditure elsewhere). The
methods also raise a number of issues including which existing services new
interventions are compared with, how health is measured and what costs are included.
These are discussed in more detail below.

31.4 Defining the Decision Problem

Healthcare revolves around choices. These choices occur at all levels of the healthcare
system: for example, a physician must decide which treatments to give to a specific
patient with a particular condition; a hospital formulary committee must decide the set of
pharmaceutical therapies from which physicians are permitted to choose for a particular
condition; managers must decide whether it is worth investing in expensive equipment
with large upfront cost, for example magnetic imaging machines—the list of points of
choice is almost endless.

Any explicit approach to making these types of choices needs to specify the
(p. 739)

decision problem being addressed which involves two key questions. These are described
below in the context of choosing between alternative medical interventions, but this can
be readily generalized to the other types of choice summarized above.

31.4.1 Who is the Choice Directed At?

This typically takes the form of the patient group which is the intended recipient of, for
example, a new therapeutic or diagnostic intervention. This will often need to be quite
specific—for example, in making decisions about whether to give patients access to new
pharmaceuticals, the recipient group of patients will usually be tightly defined by the
product's license. There may be a series of sub-groups of recipients which are
distinguished in the decision problem—for example, the cost-effectiveness of a new
treatment for ischemic heart disease may be considered separately for sub-groups of
patient defined in terms of the severity of their current condition.

31.4.2 What is the Choice Between?

It is necessary to define the options being compared. Economic evaluation is often used
to assess new medical technologies, so these need to be specified together with the
options against which the new intervention will be compared. In principle, comparators
should reflect all the existing ways in which the recipient group is currently (or could be)
managed. This may include a “do nothing” option or a sequence of possible interventions.

Page 7 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
More generally, the aim of a study might be to identify the most cost-effective way of
managing a particular group of patients from a number of existing interventions. Again,
the principle is to include all possible options in the analysis.

31.5 A Clinical Scenario

The chapter so far is intended to provide a grounding in the policy motivation for
economic evaluation and the underlying normative issues in undertaking studies. We now
further develop the key elements of undertaking economic evaluation in practice.

Box 31.1 A Clinical Scenario to Illustrate the Methods of Economic Evaluation:

Introduction

You are in control of a central budget from which the healthcare services for a
specific population must be paid for. For example, you may be an insurance
company or an institute with the remit to assessing health care interventions and
programs for possible reimbursement or coverage within a national health care
system.

You must decide which chemotherapy (if any) to make available to patients
presenting with early breast cancer, while being mindful of your obligation to
provide healthcare to patients with other diseases from your finite budget.

The regulatory body with the remit for licensing drugs lists three approved
regimens for use in patients with early breast cancer:

• First generation chemotherapy (FGO);

• Second generation chemotherapy (SGO);
• FGO in combination with SGO (COMB).

Box 31.1 outlines a simple scenario regarding the comparison of chemotherapy

treatments for breast cancer. This example will be used throughout the remainder of the
chapter to illustrate some of the conceptual issues raised in this chapter. This clinical
scenario is also used as part of Chapter 32 by Griffin and Claxton on uncertainty in
economic evaluation. (p. 740)

31.6 Measuring Outcomes

Page 8 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
The assumed objective, in terms of the outcomes of health care programs and
interventions, is central to economic evaluation. As shown in Figure 31.1 above the focus
of outcome measurement in most evaluations is on health. Let us consider the treatment
of a single disease. If there are a set of mutually exclusive treatment options, where the
set of treatment options includes all relevant treatments for the disease (perhaps
including a “no treatment” option), how should we decide which treatment is the “best”?
If we are treating a single disease, and making decisions only within that disease area, a
key source of evidence on the comparative effect of different treatments will be
randomized controlled trials (RCTs), if available. The main results of such trials are
typically clinical measures of outcome—for example, in a transitory illness, time with
symptoms or, in cancer, overall survival. If the objective of the decision-maker is simply to
choose the treatment from the mutually exclusive interventions which the trials suggest
gives the best result for the clinical measure of outcome then it is generally a
straightforward choice.

However, things can become rather more complicated if we have two different clinical
measures within a disease area, which are not perfectly correlated—for example, severity
of symptoms and duration with symptoms. Or, similarly, we may be trying to make
decisions across disease areas which have different clinical measures of outcome. For
example, a choice may be faced between funding a new drug for a cancer which gives an
extra eight weeks of progression free survival or a new drug for the treatment of flu
symptoms which reduces the time with symptoms by two days. In these circumstances we
need to be able to establish a trade-off between the two measures, but how would this be
addressed?

It would be helpful to have some “generic” (non-disease specific) unit of health

measurement which could allow us to make choices between treatments across disease
areas. (p. 741) The reason for this should be clear from Figure 31.1: it is necessary to
compare the health gained from a new intervention, relative to an existing comparator,
with the health decrement associated with displacing existing services (typically from
quite different clinical areas) to fund the additional cost of the new technology. One unit
of health outcome that has a degree of comparability across disease areas is overall
survival duration, the estimated mean time until death. One issue that arises from using
overall mean survival duration is that, while it is a clinical measure of outcome, results
from trials will generally not provide this information. This is because, for many diseases,
trials do not have a long enough follow-up to estimate mean survival with any degree of
accuracy and are often not large enough to estimate differences between interventions
with any degree of precision. This raises an issue of how to link other “intermediate”
endpoints, such as time to progression in cancer, which clinical randomized controlled
trials are powered to find differences in between treatments, and the overall survival. We
will return to this later in the extrapolation and modeling section of this chapter.

Overall survival clearly has benefits over disease specific units of health by facilitating
comparison across treatments in different disease groups where the change in survival is
considered one of the key impacts of health care interventions on health. However, few

Page 9 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
would argue that health improvement is purely about extending survival. If, for example,
we have a choice between two cancer drugs, one of which has minimal side effects and
the other has significant side effects, is it inevitable we would select the latter just
because it had very marginally higher overall survival? There are also disease areas
where treatments have no effect on overall survival but instead lead to improvements in
health-related quality of life (HRQoL), for example some treatments for mental illnesses
or most interventions in dermatology. By only considering overall survival we would
clearly ignore the health gain generated by interventions in such clinical areas.

There are arguments, therefore, for measuring health in such a way that it captures both
changes in the quantity of life (overall survival) as well as the HRQoL (morbidity). The
most commonly used and well-known measure, which jointly encompasses both the
quantity and quality of life, is the quality adjusted life year (QALY), although other
measures have been proposed (Murray et al. 2002). QALYs are generated by the
summation across all health states of the length of time in a particular health state
multiplied by a weight representing the HRQoL attached to that health state (see Box
31.2). These weights are often referred to as “utility” although this has a quite different
meaning to the same term used by the neoclassical economist. The HRQoL weights are
based on a scale where 1 represents perfect health and 0 represents death. Negative
values are also permitted (i.e. health states considered worse than death).

Box 31.2 Quality-adjusted life-years from an Intervention

With treatment 1 the

patient's HRQoL
would deteriorate
along the lower path
and they would die at
time X. However,
with treatment 2 the
patient would
maintain a higher
quality of life and
would live longer and
would instead die at time Y. The area below treatment path 1 represents the QALYs
associated with treatment 1 and the area below treatment path 2 represents the
QALYs associated with treatment 2. The area between the two paths is the QALYs
gained by treating the patient with treatment 2 instead of treatment 1. This can be
separated into 2 parts: area A represents the QALY gain as a result of quality
improvement and area B the QALY gain as a result of quantity improvement (figure
based on Gold et al. 1996).

Page 10 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
There are various approaches for deriving the HRQoL weights (See Box 31.3).
Importantly, these weights represent preferences and an important issue is whose
preferences should be used in estimating the weights. The main contending sources of
preferences are the general public (who are typically the ultimate payers) and patients.
There are arguments for and against both groups, and the choice is essentially a political
rather than scientific one (Brazier et al. 2005). The HRQoL weights should satisfy three
requirements. First, more desirable health states should receive higher weights. Second,
they should be anchored on perfect health and death, as these two states are relevant if
we are (p. 742) to make comparisons between all possible interventions. Finally, they
should be measured on an interval scale, such that an increase from 0 to 0.2 is
considered equivalent to an increase from 0.8 to 1.

There are now a number of measurement instruments that facilitate the measurement of
individuals' health states in a way that can be expressed in terms of HRQoL weights. One
of these measures is the EuroQol (EQ)-5D (Kind 1996) which is used widely in RCTs and
other clinical studies (See Box 31.4).

Box 31.3 Deriving Health-related Quality of Life Weights

Two most common

widely used methods
for establishing the
health related quality
of life weights are
time trade-off (TTO)
Click to view larger and standard gamble
(SG). Time trade-off
involves asking an individual to make a choice between living the rest of his/her life,
for time t, in a given health state, which has an as yet unknown quality of life value
of hi, and living a shorter period, for time x, in perfect health, which has a quality of
life value of 1. Time x is varied until the individual is indifferent between the two
alternatives. At this point the quality of life weight attached to the state is equal to
x/t.

Page 11 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
With standard
gamble, an individual
is asked to make the
choice between the
certainty of
remaining in a given
health state, with an
as yet unknown
quality of life x, or an
alternative with two
possible outcomes,
Source: Drummond et al. 2005
perfect health and
death. The
probability of the outcome perfect health in the alternative, p, is then varied until
the individual is indifferent between remaining in the state with certainty or taking
the gamble between perfect health and death. At this point the health related
quality of life weight for the state x is equal to p.

Box 31.4 Euroqol (EQ)-5D

Getting all patients in a clinical trial to value their health states using the TTO or SG
would be very time consuming. Instead values using these techniques have been
placed on health states through the use of questionnaires. An example of such a
questionnaire is the EQ-5D which is a standardized instrument for use as a measure
of health outcome. Applicable to a wide range of health conditions and treatments,
it provides a simple descriptive profile and a single index value for health status.
Individuals are asked 5 questions about various aspects of their health (mobility,
self-care, usual activities, pain/discomfort, and anxiety/depression) on which there
are 3 levels to choose from (no problem, moderate problems, severe problems). The
states described by the instrument are linked to existing preference weights which
are available for several countries (see 〈www.euroqol.org〉).

Tick one box for each group of statements:

Mobility
I have no problems in walking about □

Page 12 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
I have some problems in walking about □

I am confined to bed □

Self-Care
I have no problems with self-care □

I have some problems washing or dressing myself □

I am unable to wash or dress myself □

Usual Activities
I have no problems with performing my usual activities (e.g. work, study,
housework, family, or leisure activities) □

I have some problems with performing my usual activities □

I am unable to perform my usual activities □

Pain/Discomfort
I have no pain or discomfort □

I have moderate pain or discomfort □

I have extreme pain or discomfort □

Anxiety/Depression
I am not anxious or depressed □

I am moderately anxious or depressed □

I am extremely anxious or depressed □

Page 13 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
QALYs have been widely criticized (Nord and Kamlet 2009). For many they lack sensitivity
to actual changes in health, in part due to their generic nature. Another criticism is that
QALYs provide a crude representation of how individuals would make the trade-off
between length and quality of life—for example, the assumption of constant (p. 743) (p. 744)
proportional trade-off assumes that if someone is indifferent between, say, ten years in a
particular (dysfunctional) health state and five years in good health, they will also be
indifferent between one year in that health state and six months in good health. There
has also been concern that while QALYs focus on individuals' health, they can be a poor
reflection of individuals' preferences regarding the distribution of health improvement
between different types of patient. Alternatives to the QALY such as the healthy-years
equivalent and the patient trade off (Drummond et al. 2005) have been suggested
(p. 745) to address these problems, but there are very few examples of their use in

applied economic evaluation studies.

Box 31.5 A Clinical Scenario to Illustrate the Methods of Economic Evaluation:

Measuring and Valuing Outcomes

Published studies indicate FGO offers an improvement in survival when compared

to no treatment, while the second generation regimen SGO offers an improvement
in survival when compared to FGO, and the combination regimen of FGO and SGO
offers an improvement in survival when compared to SGO alone. However, the side
effect profile of COMB is considered to be worse than the side effect profile of FGO
alone and SGO alone.

Patients will receive chemotherapy for three months. Patients on COMB experience
a health-related quality of life (HRQoL) score (in terms of a 0 to 1 “utility”) of 0.5
while on treatment, those on SGO a score of 0.6 and those on FGO 0.7.

Following treatment and until progression all patients have a HRQoL score of 0.75.
Patients who receive FGO are expected to have a cancer recurrence four years after
the end of treatment, those who receive SGO have a recurrence five years after the
end of treatment and those who receive COMB have a cancer recurrence six years
after the end of treatment.

Following progression, all patients have the same expected survival of two years
with a HRQoL score health-related quality of life of 0.6.

Those patients who receive no chemotherapy treatment are expected to receive four
QALYs.

Page 14 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs

Treatment QALYs Expected QALYs Total expected

generated accumulated post- QALYs
during treatment during accumulated per
treatment progression free patient
survival

No 4 QALYs
treatment

FGO 0.7 × 0.25 = 4 × 0.75 = 3 3.175 + (2 × 0.6)

0.175 = 4.375

SGO 0.6 × 0.25 = 5 × 0.75 = 3.75 3.9 + 1.2 = 5.1

0.15

COMB 0.5 × 0.25 = 6 × 0.75 = 4.5 4.625 + 1.2 =

0.125 5.825

Despite these criticisms, QALYs are now widely used in economic evaluations supporting
decision-making in health care. For example, NICE requires their use in studies
undertaking to inform its decisions about whether to recommend new technologies for
use in the NHS (NICE 2008); similar agencies internationally recommend QALYs, at least
as an element in studies submitted to them (Tam and Smith 2004). The QALY, then, is an
extensively used and understood measure of health in cost-effectiveness analyses. For
example, they were recommended as the appropriate unit for health measurement by the
Panel on Cost-Effectiveness in Health and Medicine, a US panel with expertise in cost-
effectiveness analysis, clinical medicine, ethics, and health outcomes measurement,
which was convened by the US Public Health Service (Gold et al. 1996).

Box 31.6 Examples of Revealed Preference and Contingent Valuation

Revealed preference: the value of a

statistical life
“Suppose jobs A and B are identical except that workers in job A have higher annual
fatal injury risks such that, on average, there is one more job-related death per year
for every 10,000 workers in job A than in job B, and workers in job A earn $500

Page 15 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
more per year than those in job B. The implied value of statistical life is then $5
million for workers in job B who are each willing to forgo $500 per year for a 1-
in-10,000 lower annual risk” (Fisher et al. 1989).

Contingent valuation: road safety

“Suppose that you are buying a particular make of car. You can choose to have a
new kind of safety feature fitted to the car at an extra cost. The risk of a car driver
being killed in an accident is 10-in-100,000. You could choose to have a safety
feature fitted to your car which would halve the risk of the car driver being killed,
down to 5-in-100,000. Taking into account how much you can personally afford,
what is the most that you would be prepared to pay to have this safety feature fitted
to the car?” (Jones-Lee et al. 1985).

Hypothetical example
Current risk of death without safety feature: 10-in-100,000

New risk with safety feature: 5-in-100,000

Reduction in risk (dR): 5-in-100,000

Example maximum premium willing to pay (dV): $100

Implied value of life = dV/dR = $(100/5) × (105) = £2,000,000

How would the effects of alternative options on outcomes be captured in CBA

(p. 746)

within a welfarist paradigm? Given the general absence of unregulated markets in health,
the focus would be on estimating the shadow price of the health outcomes— that is, the
price at which the health outcome would be valued if a market for them existed and all
market distortions were removed. There are two approaches which have been used to
find this shadow price: revealed preference and contingent valuation (see Box 31.6).
Revealed preference involves finding real life situations where individuals trade health
and wealth, and then estimating an implicit value based on the individuals' behavior in
these circumstances. An example of this method is the use of wage premiums for more
dangerous jobs as a measure of how much the individual is willing to accept in return for
increased risk of loss of life or injury (Johanesson 1995). Contingent valuation attempts to
calculate the shadow price of health by creating hypothetical markets where trade-offs
between health and wealth are investigated in a sample of individuals (Drummond et al.
2005). Discrete choice experiments have also been used to value different treatments,

Page 16 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
based on the premise that any treatment can be described based on its characteristics
and the extent to (p. 747) which individuals value the treatment depends on its level of
these characteristics (Ryan and Gerard 2003).

The use of contingent valuation and discrete choice methods is not confined to valuing
health outcomes. Indeed, one of the suggested advantage of these methods is that they
enable a wider set of outcomes to be incorporated into studies (e.g. effect of interventions
on other areas of individuals' well-being such as convenience and the value they attach to
information). These allow the comparison of benefits across different fields although it is
unclear how decisions can be made, particularly when transfers between different sectors
are not possible.

31.7 Costs
It is not the money costs of resources that we are directly interested in, but instead the
opportunity costs, i.e. what is the next best thing we could do with the resources we use
to provide a course of treatment. However, the most direct way to estimate the
opportunity cost is, first, to estimate the resource costs associated with an allocation and
then transform this into the opportunity costs (whether they be in terms of health or
utilities). Which resources will be included depends on the perspective taken, e.g.
whether we should include productivity costs if the focus is on health and there is a fixed
budget, where productivity costs are any gain in output as a result of the health care
intervention. The transformation into opportunity costs also depends on the view we take
on prices, i.e. do we accept that prices reflect opportunity cost or do we instead need to
calculate shadow prices?

Costs associated with different health care programs and treatments can be split into
several groups: costs arising from the use of resources within the health care sector;
costs relating to resource use by patients and their families; costs arising from resource
use in other sectors of the economy; and costs stemming from productivity changes.

31.7.1 Which Costs Should Be Included?

The range of the above costs considered in an economic evaluation is likely to be based
upon the consideration of four points (Drummond et al. 2005). First, what is the
perspective of the analysis? This is perhaps the most essential, as it is determined by the
budget constraint imposed on the decision-maker and also on specific judgments that we
are making. If the perspective is that of the healthcare sector, we are unlikely to be
interested in the costs imposed on patients or their families. However, these costs would
clearly be of importance if we took a broader societal perspective, i.e. an all-
encompassing perspective whereby changes in costs and productivity are all included.
Many health care systems use economic evaluation methods to inform (p. 748) resource

Page 17 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
allocation decisions within that sector—as a result, they take a perspective reflecting
their own budget constraint (e.g. NICE's technology appraisal program in the UK). Other
agencies take a wider cost perspective because they are responsible for a wider set of
budgets or have no direct budget responsibility at all (e.g. the TLV, which makes decisions
about the use of new pharmaceutical products in Sweden).

Second, is the comparison restricted to two or more programs immediately under study?
If the comparison relates to set of specific options with specific costs being common to all
the interventions being compared, then these costs need not be considered as they would
not affect the choice between interventions. However, if it is thought that, at some later
stage, a wider comparison of interventions may be considered, including some
interventions not considered in the original evaluation, it would be wise to include costs
which are common. Third, are some costs merely likely to confirm a result that would be
obtained by consideration of a narrower range of costs? If it is felt that the costs are
unlikely to affect the balance of costs between the various interventions it may not be
worth the extra resource use required to collect them. Finally, what is the relative order
of magnitude of costs? Similarly to the last point, if it is considered that some costs will
be very small and unlikely to change the balance of costs between different interventions
then it may not be worth considering them given the cost of collecting information on
them.

31.7.2 Measuring and Valuing Costs

Once the Costs To Be Included In the Evaluation Have Been Decided Upon, Then the
Method For Valuing the Costs is Based On the Measurement of Two Components. First,
the Physical Quantities of the Resources Consumed (E.g. Days In Hospital, Doctors' Visits,
Etc.) Must Be Estimated. Second, the Value of These Resources In Money Terms Must Be
Estimated (I.e. the Prices).

The measurement of the physical quantities of the resources consumed will often depend
upon the context for the economic evaluation. If the economic evaluation is being
undertaken as part of a randomized controlled trial, data on the resource use may be
collected as part of the study's case record forms. This could be supplemented by patient
questionnaires for resource use not easily captured from routine clinical notes (e.g.
patients' use of community care services). If there is no trial (or similar clinical study)
providing a vehicle for the collection of resource use data, these could be collected from
patients' clinical notes, hospital data systems, or through patient questionnaires.

For many resources, market prices may be available. Theoretically the proper price for a
resource is its opportunity cost (the value of the forgone benefits given the resources use
in its next best alternative). However, the pragmatic approach to costing is to take
existing market prices which would reflect the opportunity cost in a “first best” world
(Drummond et al. 2005). However, there are many reasons why the market price may not
reflect the true cost of the resource—for example, some prices may be subsidized or set

Page 18 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
under government issued monopoly patents. If a health system perspective is taken,
(p. 749) then the financial costs falling on the system is the sole focus. Here the

opportunity cost is in terms of health (see Figure 31.1).

If we take a societal perspective we may also need to value productivity costs if there are
productivity gains or losses associated with the health care interventions being assessed,
for example due to the patient or one of their relatives being unable to work. These are
commonly valued using one of two approaches: the human capital approach or the
friction cost method, with the latter only being used to assess productivity losses. More
discussion of these approaches and their strengths and weaknesses can be found in
Sculpher 2001).

Once resource use and prices have been collected, the total resource cost of a patient can
then be calculated by the summation of the resources included multiplied by their price.
Box 31.7 considers the measurement and aggregation of costs in our clinical scenario.

31.8 How Do We Make Decisions Based on

Costs and Outcomes?
Box 31.7 A Clinical Scenario to Illustrate the Methods of Economic Evaluation:
Dealing With Costs

The published pricing lists indicate that SGO is the most expensive chemotherapy
regimen simply in terms of purchasing the drugs, followed by COMB, with FGO
being the cheapest. Due to the more serious side effects associated with COMB
patients will require more treatment than those on SGO, and similarly those on FGO
will require less treatment than those on SGO. Each year of progression free
survival costs $250 in medical care resources. Following progression, care costs
$5000 until death.

Page 19 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the Use of Health Care Interventions and
Programs

Chemotherapy Cost of treating Cost of care Cost of care Total cost of

cost side-effects during following care
progression free progression
survival

No treatment $7500

FGO $5000 $500 $1000 $5000 $11,500

SGO $10,000 $1000 $1250 $5000 $17,250

COMB $7500 $5000 $1500 $5000 $19,000

Page 20 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an
individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs

Earlier we briefly touched upon how costs and outcomes can be combined for the purpose
of decision-making (see Figure 31.1). We will now examine this further. Let us consider a
decision-maker allocating resources across the health system. Ideally, with (p. 750) a
fixed budget, the decision-maker would select from all available treatments, which
includes all possible treatments in all disease areas, the subset that maximizes health
benefits for the entire patient population subject to the given budget constraint. This
approach could be achieved using mathematical programming (Epstein et al. 2007).
However, in practice such an approach is not possible: there is no single decision-maker
deciding on all treatments, nor is there enough information about the effectiveness and
cost of all services to be able to formulate such a problem. Instead, when making
decisions using economic evaluation, decision rules are used which attempt to proxy this
ideal solution.

Page 21 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs

31.8.1 Decision Rules for Cea

Let us first consider the decision between a set of mutually exclusive interventions/
programs before extending our analysis to consider independent interventions/programs.
We consider two interventions for a disease, A and B, from a health system perspective
where the relevant unit of output is a QALY. Figure 31.2 shows the cost-effectiveness
plane which represents the incremental costs and effects of one of the treatments
compared to the other. Note that the north east quadrant is what we illustrated in Figure
31.1. The origin represents the comparator treatment, in this case let it be B. If treatment
A lies to the right of the origin it is more effective (has more QALYs than B), or if is to the
left it is less effective. If treatment A lies above the x-axis then it is more costly than B,
and vice versa.

If intervention A is less
costly and more effective,
then A is said to dominate
B. Similarly if A is more
costly and less effective
than B, then it is
dominated by B. In both
cases the choice between
A and B is obvious: the
dominant option is clearly
Click to view larger
preferable. In practice,
figure 31.2 The incremental cost-effectiveness plane however, it is rare that
costs and outcomes lend
themselves to the dominance rule. Usually A will be both more costly and more effective
than B or vice versa, i.e. in the (p. 751) north-east or south-west quadrant. The critical
issue in these cases is whether the additional (incremental) cost of the intervention is
worth paying for its incremental benefits, or similarly if it is worth making the cost saving
given the lower outcomes which are associated with it. The decision rules developed to
address this issue generally focus on the incremental cost-effectiveness ratio (ICER)
which is defined:

The ICER defines the amount that must be paid, per extra QALY, for the more expensive,
but more effective, intervention. Whether to implement A given an ICER of a particular
value will depend on the perspective the analysis is taking, with the key being what is
displaced at the expense of the extra expenditure, i.e. the opportunity cost. If we define
the threshold that we are willing to pay per extra QALY as λ, then if the ICER is greater
than this we would not fund A, if it is less than λ we would fund A; if it is equal to λ we are
indifferent between funding A and not funding A. This takes us back to the principles we
covered earlier as illustrated in Figure 31.1. But where does λ come from? In general the

Page 22 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
choice is between a λ which represents the opportunity cost under a fixed budget (i.e. the
ICER of the displaced treatment) or some measure of societal willingness to pay under an
adjustable budget (also see McCabe et al. 2008).

An ICER compares two alternatives. However, when there are a number of mutually
exclusive alternatives, then a number of comparisons can be made which will generate a
series of possible ICERs. Table 31.1 looks at a number of hypothetical alternative
interventions. Treatment X is dominated by all other treatments, it is both more expensive
and less effective. Therefore, as discussed previously, it can be ignored. The choice is
then between W, Y, and Z. These should be ranked in order of cost. The ICERs for each of
these non-dominated options is then calculated relative to the next cheapest as shown in
Table 31.2.

Figure 31.3 represents the cost-effectiveness plane for these three treatments where the
origin is Z. As can be seen from the figure, there is a combination of Z and Y (e.g. if we
give 50% of the patients Z and 50% of the patients Y), whereby the combination
dominates W. Therefore, W is said to be subject to “extended dominance,” and can be
(p. 752) removed from consideration. Extended dominance can also be found from the

ICERs. Once all dominated treatments are removed and the treatments ranked in order
from least to most costly and the ICERs calculated, then if the ICER of the next most
effective and expensive treatment is less than the ICER for the treatment compared to
the next less effective and expensive treatment, then the treatment is subject to extended
dominance.

Table 31.1 Costs and QALYs of Mutually Exclusive Alternatives

Treatment Cost QALYs

W $8000 7

X $10,000 5

Y $9000 8

Z $4000 6

Page 23 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the Use of Health Care Interventions and
Programs

Table 31.2 Incremental Costs and QALYs of Mutually Exclusive Alternatives

Treatment Cost QALYs Incremental cost Incremental ICER

QALYs

Z $4,000 6

W $8,000 7 $4,000 1 $4,000

Y $9,000 8 $1,000 1 $1,000

Page 24 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an
individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
All dominated and
extended dominated
strategies should be
removed from the option
decision set before making
decisions about which
treatment should be used.
A failure to do so will lead
to comparisons not with
Click to view larger the next best treatment
figure 31.3 Top right quadrant of the cost- but with clearly inferior
effectiveness plane for options Z, W, and Y as defined alternatives. Consequently,
in Tables 31.1 and 31.2
the ICERs will be under-
estimated and we will not
ensure we select the optimum treatment. Box 31.8 considers decision rules in the context
of the hypothetical clinical scenario.

(p. 753)It should be noted that decision rules outlined above are based on several
assumptions, notably constant returns to scale, perfect divisibility and non repeatability,
which may not always hold. A critique can be found in Birch and Gafni 1992). While
extended dominance may not hold, in that a combination of two programs may not be
possible in practice, its implications for decision-making still holds as long as λ is constant
no matter how expensive a new treatment is. This is based on the argument that if you
were willing to pay for the extendedly dominated treatment you must also be willing to
pay for the next more expensive and more effective treatment. The largest criticism of
ICER decision rules is related to the argument above in that ICERs only give us
information about the cost per unit, not the total cost of the program. With perfect
divisibility and constant returns to scale this does not matter. However, when this is not
the case the size of the project does matter as it can alter the cost per QALY on the
marginal project which will be displaced, or the societal WTP (Birch and Gafni 1992).

31.8.2 Decision Rules for CBA

Box 31.8 A Clinical Scenario to Illustrate the Methods of Economic Evaluation:

Decision Rules

You must now make a decision about which treatment, if any, to make available to
patients with early stage breast cancer. The incremental costs and QALYs are
presented in the table below:

Page 25 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs

Incremental Incremental Incremental cost per

cost QALYs additional QALY

FGO vs. No $4000 0.375 $10,667

treatment

SGO vs. FGO $5750 0.725 $7931

COMB vs. $1750 0.725 $2414

SGO

COMB vs. $11,500 1.825 $6301

No
treatment

As can be seen from the ICERs, both FGO and SGO are subject to extended
dominance by a combination of No treatment and COMB. Assuming we accept the
exclusion of treatments based on extended dominance, this leaves us with a choice
between two options: No treatment or COMB.

Consider how you would now choose between No treatment and COMB, and what
additional information you would need to know.

Decision rules for CBA are rather different to those of CEA. Both costs and benefits of
health care interventions are expressed in monetary terms using appropriate prices (or
shadow prices), and the “net present value” of an intervention (the difference (p. 754)
between its aggregate benefit and its aggregate cost) is assessed. Any project with a
positive net present value should be considered as a potential Pareto improvement using
the welfarist principle discussed above. When selecting between mutually exclusive
alternatives then the alternative with the highest net present value should be selected.
The simplicity of this rule relies upon the assumption that budget constraints can
typically be ignored. If budget constraints are recognized under CBA by health system
decision-makers then, similarly to ICER decision rules, it would be necessary to compare
the net present value of the treatment with the net present value of the treatments which
would be displaced to fund it. If a broader view of public sector decisions is taken, CBA
methods can be used to compare policies across sectors, although such decisions would
typically be outside the remit of health system decision-makers.

31.9 Vehicles for Economic Evaluation

Page 26 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
A large proportion of economic evaluations are conducted based on the information of a
single randomized controlled trial (Sculpher et al. 2006). However, relying on a
randomized trial as a single vehicle for economic evaluation has a number of key
limitations (Sculpher et al. 2006). As a result of these limitations, economic evaluation for
decision-making would normally need to draw upon a range of sources for evidence,
including RCTs, but also cohort studies, quality of life surveys, etc.

Economic evaluation is not only concerned with the measurement of costs and effects but
also with the appropriate synthesis of relevant evidence to inform resource allocation
decisions within healthcare. It should be noted, therefore, that RCTs and decision analytic
models are not competing alternatives (Drummond et al. 2005). Instead, RCTs are
focused on different types of measurement whilst decision models are concerned with
informing specific decisions.

All relevant comparators must be considered for the correct calculation of ICERs.
However, many RCTs will fail to compare all the relevant options that might be
considered by decision-makers, for example in RCTs drugs will typically just be compared
to one other treatment which may not even be an active intervention but instead a
placebo. To compare all the relevant options it is probable that evidence from a number of
RCTs will be required to be synthesized through a meta-analysis (Egger et al. 1993). A
decision analytic model will allow the results of the meta-analysis to be combined with
other types of evidence, such as costs and utilities.

As was described in the outcomes section of this chapter, clinical trials often compare
interventions in terms of intermediate endpoints rather than final outcomes, for example
in cancer trials, time in remission. However, as discussed previously, the economic
evaluation may require outcomes to be in QALYs. Decision models can be (p. 755) used to
link intermediate outcomes with the outcomes required for decision-making.

Decision analytic modeling provides a framework for the synthesis of the data from all of
the relevant sources. It also allows for the consideration of decision-making under
uncertainty (see Chapter 32).

Decision analytic modeling provides a framework for decision-making which can satisfy
the important objectives of economic evaluation (Drummond et al. 2005), namely:

1. Structure. It can provide a structure which appropriately reflects the prognoses of

patients and how the treatments will impact on these prognoses.
2. Evidence. It can provide a framework within which all relevant evidence can be
synthesized.
3. Assumptions and scientific value judgments. It provides an explicit way of bringing
to bear the assumptions and judgments that are inevitably needed in making
decisions based on uncertainty evidence.

Page 27 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
4. Evaluation. It provides a structure which allows the translation of the relevant
evidence into the outcomes of interest, namely costs and effectiveness, so that
decisions can be made.

There are many types of decision models that can be used in economic evaluation, with
perhaps the most common being decision trees and Markov models. However, two
features that they all have in common are probabilities and expected values (Drummond
et al. 2005). Probabilities represent the strength of belief that an event will happen; this
may be based on prior evidence when available, but this may not always be the case and
such values could be elicited from expert opinion if required. The other key feature is
expected values; given a treatment it is still uncertain what the patient's outcome will be,
for example, with a cancer drug there will be probabilities that they will progress or not
progress. The expected costs and outcomes are a weighted sum of the possible costs and
outcomes dependent on what occurs, with the weight based on the probability that those
events will occur. The expected values represent the best estimate of the endpoints of
interest for decision-making. More detailed discussion of decision modeling can be found
in Drummond et al. (2005).

31.10 Summary
Resources are finite but wants are infinite. The scarcity of resources will always lead to
decisions about which competing treatments to choose in health care. Methods for the
economic evaluation of health care interventions facilitate a formal joint consideration of
costs and outcomes for the purpose of decision-making. They allow for a decision over
whether one allocation of resources is more preferable to another, not just within disease
areas but also across them.

Two broad principles of choice have been proposed for economic evaluation,
(p. 756)

welfarist, based on orthodox welfare economics, and non-welfarist, which changes the
objective from utility maximization to another set of objectives. The latter is more
prominent in policy, but the choice between the two has key implications for some aspects
of methodology including the types of outcomes and costs that should be included in any
analyses.

This chapter has discussed the approaches that have been taken to quantify outcomes
and costs from health care interventions. Perhaps the most predominant measure of
outcome used in economic evaluations in health care is the QALY. The choice of costs to
use depends on the decision-making context and the perspective taken. Incremental cost-
effectiveness ratios can then be compared to cost-effectiveness thresholds to allow for
efficient decision-making. It should be clear that using solely the results of individual
trials to conduct such analyses has significant drawbacks, and that it is more appropriate

Page 28 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
to synthesize all the relevant evidence to allow economic evaluations to meet the needs of
decision-makers. A useful tool for doing this is decision analytic models.

References
ARROW, K. (1963) Uncertainty and the welfare economics of medical care. American
Economic Review, 53, 941–73.

BIRCH, S. AND GAFNI, A. (1992) Cost effectiveness/utility analyses: do current decision rules
lead us to where we want to be? Journal of Health Economics, 11, 279–96.

BRAZIER, J., AKEHURST, R., BRENNAN, A., DOLAN, P., CLAXTON, K., MCCABE, C., SCULPHER, M., AND TSUCHYIA,

A. (2005) Should patients have a greater role in valuing health states? Applied Health
Economics and Health Policy, 4, 201–8.

BROADWAY, R. AND BRUCE, N. (1984) Welfare Economics. Oxford, Blackwell.

CLAXTON, K. AND AKEHURST, R. (2005) It's just evaluation for decision making: recent
developments in, and challenges, for cost-effectiveness research. In Smith, P., Ginnelly, L.,
and Sculpher, M. (eds.) Health Policy and Economics: Opportunities and Challenges.
Oxford, Oxford University Press.

—— SCULPHER, M., AND CULYER, A. (2007) Mark vs Luke? Appropriate methods for the
evaluation of public health interventions. Centre for Health Economics Paper 31,
University of York.

—— BRIGGS, A., BUXTON, M., CULYER, A., MCCABE, C., WALKER, S., AND SCULPHER, M. (2008) Value
Based Pricing for NHS drugs: an opportunity not to be missed? British Medical Journal,
336, 251–4.

CULYER, A. (1991) The normative economics of health care finance and provision. In
McGuire, A., Fenn, P., and Mayhew, K. (eds.) Providing Health Care: The Economics of
Alternative Systems of Finance and Delivery. Oxford, Oxford University Press.

DIAMOND, P. AND MIRRLEES, J. (1971a) Optial taxation and public production: I. Production
efficiency. American Economic Review, 61, 8–27.

—— —— (1971b) Optimal taxation and public production: II. Tax rules. American
Economic Review, 61, 261–78.

(p. 757) DRUMMOND, M. F., SCULPHER, M. J., O'BRIEN, B. J., STODDART, G. L., AND TORRANCE, G. W. (2005)
Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford
Medical Publications.

EGGER, M., DAVEY-SMITH, G., SONG, F., AND SHELDON, T. (1993) Making sense of meta-analysis.
Pharmacoepidemiology and Drug Safety, 2, 65–72.

Page 29 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
EPSTEIN, D., CHALABI, Z., CLAXTON, K., AND SCULPHER, M. (2007) Effciency, equity and budgetary
policies: informing decisions using mathematical programming. Medical Decision Making,
27, 128–37.

FISHER, A., CHESTNUT, L., AND VIOLETTE, D. (1989) The value of reducing risks of death: A note
on new evidence. Journal of Policy and Management, 8, 88.

GOLD, M. R., SIEGEL, J. E., RUSSELL, L. B., AND WEINSTEIN, M. C. (1996) Cost-Effectiveness in Health
and Medicine. New York, Oxford University Press.

HICKS, J. (1939) Welfare propositions in economics and interpersonal comparisons of

utility. Economic Journal, 49, 696–712.

Johanesson, P. (1995) Evaluating Health Risks. Cambridge, Cambridge University Press.

JONES-LEE, M., HAMMERTON, M., AND PHILIPS, P. (1985) The Value of Safety: Results of a National
Sample Survey. Economic Journal, March, 95/377, 49–72.

KALDOR, N. (1939) Welfare propositions and interpersonal comparisons of utility. Economic

Journal, 49, 549–52.

KIND, P. (1996) The EuroQoL instrument: an index of health-related quality of life. In

Spilker, B. (ed.) Quality of Life and Pharmacoeconomics in Clinical Trials. 2nd edn.
Philadelphia, Lippincott-Raven.

LERNER, A. (1944) The Economics of Control. New York, Macmillan.

MASON, A., WALKER, S., CLAXTON, K., COOKSON, R., FENWICK, E., AND SCULPHER, M. (2008) The GMS
quality and outcomes framework: Are the quality and outcomes framework (QOF)
indicators a cost-effective use of NHS resources? Quality and Outcomes Framework. Joint
executive summary. Reports to the Department of Health from the University of East
Anglia and the University of York.

MCCABE, C., CLAXTON, K., AND CULYER, A. (2008) The NICE cost-effectiveness threshold: what it
is and what that means. Pharmacoeconomics, 26, 733–44.

Mishan, E. (1971) Cost-Benefit Analysis. London, Allen and Unwin.

MURRAY, C., SALOMON, J., MATHERS, C., AND LOPEZ, A. (2002) Summary Measures of Population
Health: Concepts, Ethics, Measurement and Applications. Geneva, WHO.

NICE (NATIONAL INSTITUTE FOR HEALTH AND CLINICAL EXCELLENCE) (2008) Guide to the Methods
of Technology Appraisal. London, NICE.

NORD, E. AND KAMLET, M. (2009) QALYs: some challenges. Value in Health, 12 (Suppl. 1), S10–
S15.

Page 30 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
NYMAN, J. (2006) The value of health insurance. In Jones, A. (ed.) The Elgar Companion to
Health Economics. Cheltenham, UK, Edward Elgar.

RYAN, M. AND GERARD, K. (2003) Using discrete choice experiments to value health care
programmes: current practise and future research reflections. Applied Health Economics
and Health Policy, 7, 373–8.

SCULPHER, M. J. (2001) The role and estimation of productivity costs in economic

evaluation. In Drummond, M. F. and McGuire, A. E. (eds.) Theory and Practice of
Economic Evaluation in Health. Oxford, Oxford University Press.

—— CLAXTON, K., DRUMMOND, M., AND MCCABE, C. (2006) Whither trial-based economic
evaluation for health care decision making. Health Economics, 15, 677–87.

(p. 758) SEN, A. (1979) Personal utilities and public judgments: or what's wrong with
welfare economics. Economic Journal, 89/355, 537–58.

—— (1993) Capability and well-being. In Nussbaum, M. and Sen, A. (eds.) The Quality of
Life. Oxford, Clarendon Press.

SUGDEN, R. AND WILLIAMS, A. H. (1979) The Principles of Practical Cost-Benefit Analysis.

Oxford, Oxford University Press.

TAM, T. AND SMITH, M. (2004) Pharmacoeconomic guidelines around the world. ISPOR
Connections, 10, 4–5.

TOMPA, E., CULYER, A., AND DOLINSCHI, R. (2008) Economic Evaluation of Interventions for
Occupational Health and Safety: Developing Good Practice. Oxford, Oxford University
Press.

Simon Walker

Simon Walker is a member of the Team for Economic Evaluation and Health
Technology Assessment in the Centre for Health Economics, University of York. He
joined in October 2006 after completing an M.Sc. in Health Economics at York. He
had previously graduated from Clare College, Cambridge, with a B.A. in Economics.

Mark Sculpher

Mark Sculpher is Professor of Health Economics at the Centre for Health Economics,
University of York, where he is Director of the Programme on Economic Evaluation
and Health Technology Assessment. He has over 160 peer-reviewed publications and
is a co-author of two major text books in the area. Sculpher has been a member of
the National Institute for Health and Clinical Excellence (NICE) Technology
Appraisal Committee and currently sits on the NICE Public Health Interventions
Advisory Committee. He chaired NICE's 2004 Task Group on methods guidance for

Page 31 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018

The Methods of Cost-Effectiveness Analysis to Inform Decisions about the
Use of Health Care Interventions and Programs
economic evaluation. He is also a member of the Commissioning Board for the UK
NHS Health Technology Assessment program and the UK Medical Research
Council's Methodology Research Panel.

Michael Drummond

Michael Drummond is Professor of Health Economics and former Director

(December 1995-September 2005) of the Centre for Health Economics at the
University of York. His particular area of interest is the economic evaluation of health
care treatments and programs. He has undertaken evaluations in a wide range of
medical fields including care of the elderly, neonatal intensive care, immunization
programs, services for people with AIDS, eye health care, and pharmaceuticals. He is
the author of two major textbooks and more than 600 scientific papers, has acted as
a consultant to the World Health Organization, and was Project Leader of a European
Union Project on the Methodology of Economic Appraisal of Health Technology. He
has also served on the boards of directors of the International Society of Technology
Assessment in Health Care and the International Society for Pharmacoeconomics and
Outcomes Research. He has been president of the International Society for
Pharmacoeconomics and Outcomes Research. He is currently coeditor-in-chief of
Value in Health. His recent projects relate to the conduct of economic evaluations
and their use in health care decision making.

Access is brought to you by

Page 32 of 32

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2015. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use.

Subscriber: Gothenburg University Library; date: 01 May 2018