6.1. Introduction

Most health systems are faced with high demand but have a limited budget with which to provide the necessary services. A fundamental objective in health systems is to determine the best use of the limited funds available to promote health and provide health care. The underlying principle in this case can be seen as maximizing value for money by selecting the optimal mix of services subject to the constraints faced by the system. The conventional approach to resource allocation is to assume that a decision-maker chooses to maximize efficiency subject to the budget constraint facing the health system. This has led to the development of an extensive suite of techniques usually referred to as cost–effectiveness analysis (CEA) in helping to set priorities, which have had widespread impact, as seen in the United Kingdom’s National Institute for Health and Care Excellence (NICE) and other institutions. However, in certain situations the conventional assumptions are too simplistic to offer meaningful information to facilitate efficient resource allocation.

This chapter discusses the potential use of CEA to achieve allocative efficiency (AE) at the health care organization (meso) and health system (macro) level. It begins with an overview of CEA as currently applied at the micro level (for example, the decisions of individual clinicians and choices between treatment options) and highlights its strengths and weaknesses. It then identifies key policy priorities that could be addressed using the tools of CEA. Given the progress in most advanced health systems in assessing and implementing AE at the micro level using CEA, this chapter looks at methodological and informational challenges to adapting these approaches to higher levels of analysis. It concludes with current applications of CEA in policymaking and potential future applications for prospective and retrospective measurement of AE.

6.2. Cost–effectiveness analysis: an overview of its strengths and weaknesses

Countries all over the world place a high priority on the health of their people. Collectively funded health systems in particular usually seek to maximize health outcomes through the provision of health service inputs. Measurement of efficiency in such systems is important in determining whether resources are being used to get the best value for money. AE involves examining the extent to which available resources are allocated across and between health services so as to maximize health outcomes. It is thus concerned with more than the relationship between health inputs and outcomes (Drummond, 2005); it is also concerned with the distribution of resources.⁸ This societal perspective on distribution distinguishes AE from other aspects of health system efficiencies (such as technical efficiency (TE)⁹). Outputs, in the case of AE, must be representative of societal utility from alternative investments.

For example, let us assume that there are two alternative uses of health care resources, one for a health care service that extends life and the other for a service that improves functional ability, both with outcomes measured in terms of utility (Figure 6.1). Following Wagstaff (1991), the utility possibilities frontier can be represented by PP’. With an assumption of diminishing utility and returns to inputs in the production of a service that extends life and one that improves functional ability, this curve is downward sloping. If society wished to maximize utility from health outcomes given a pool of resources,¹⁰ then the societal utility function can be represented by the 45-degree line SS’. The tangency between SS’ and PP’ at point A represents the Pareto-optimal point. AE through an optimal product mix occurs where the marginal rate of transformation (opportunity cost) between a service that extends life and one that improves functional ability = −1, that is, resources are reallocated between services up to the point at which the marginal (social) utilities are equal. AE thus suggests there is a unique point of the production possibilities frontier that maximizes societal values relative to all other attainable sets.

Figure 6.1

Trade-offs in utility from alternative uses of health care resources.

In a competitive market, the allocation that maximizes societal welfare will be determined by market forces of demand and supply of health technologies, and price will be indicative of the value society places on different inputs. However, health care is characterized by the absence of perfectly competitive markets (Arrow, 1963) and information asymmetry between health care providers and patients on the suitability and value of services being provided. In such a scenario, any market-driven allocation is likely to be inefficient.

This analysis of AE can be applied to different levels of the health care system, for example, at the micro (choice of treatments for specific health conditions), meso organizational (selecting the optimal mix of services for hospitals) or macro (primary versus secondary care) levels. However, as described earlier, measuring AE requires valuing health system outputs from a societal perspective. This involves an aggregation of individual utility functions and preferences to construct a societal welfare function. An allocation that maximizes this societal welfare function can be seen as allocatively efficient. In practice, however, it is not feasible to construct such a societal welfare function and the absence of perfectly competitive markets in health care necessitates the use of a tool that facilitates efficiency in allocation by determining the rationing of limited resources to unlimited demand. Methods of economic evaluation have been developed to facilitate efficient resource allocation. Economic evaluation can be defined as “the comparative analysis of alternative courses of action in terms of both their costs and consequences” (Drummond, 2005, p. 4).

CEA is one form of economic evaluation that has become a central policy tool in many health care systems (Tam & Smith, 2008). It was developed to help decision-makers with fixed resources to compare programmes that produce different outcomes. For a particular level of health care resources, the goal is to choose from among all possible combinations of programmes a set that maximizes the total health benefits produced. In keeping with the earlier discussion on AE, it uses a common unit of measure that captures utility of outcomes – QALYs. This measure allows CEA to simultaneously incorporate the increase in quantity and quality of life (Weinstein, Torrance & McGuire, 2009). Thus, in theory, it is consistent with welfare economics by allowing efficiency in production and product mix. In practice, however, the type of efficiency that can be achieved by applications of CEA is dependent on the decision rules applied. This is discussed in further detail later in this chapter.

CEA has been widely applied to health policy in Europe and other publically funded health systems such as those in Australia and Canada. In these countries, it is an important tool in informing coverage decisions. In other health systems, such as the USA, the use of CEA has played a limited role in rationing care but has influenced the use of interventions that are found to make good use of resources. These include major preventive interventions, such as HIV testing, cervical smears and influenza vaccinations. The principles of cost–effectiveness are applicable in many different contexts. Meltzer & Smith (2011) provide some examples: a private insurer can use CEA to determine the package of covered benefits that will maximize profits; or a social insurer can apply the principle to obtain the maximum health gain with a given budget.

The applications of CEA in these health systems have focused on incremental analysis. Typically, CEA describes a medical technology or health intervention in terms of the ratio of incremental costs per unit of incremental health benefit, the incremental cost–effectiveness ratio (ICER). This captures the difference in effects between the new technology under consideration and the current technology for a given population (incremental benefits), and the difference in costs between the two technologies (incremental costs).

The simplicity of the underlying principle of maximizing health gains subject to a fixed budget makes CEA applicable in many other contexts. For example, in countries looking to establish a package of treatments that could be publically funded, decision-makers would estimate the cost–effectiveness of a range of treatments, rank them in increasing order of cost–effectiveness and accept treatments until the available budget is exhausted.

In all the applications mentioned here, the focus has been on micro level resource allocation decisions. For example, in the United Kingdom, NICE identifies cost-effective technologies and makes recommendations for their use in the National Health System (NHS). NICE has developed a reference case to standardize the way economic evaluations are carried out, that “specifies the methods considered by the institute to be the most appropriate for the Appraisal Committee’s purpose … with an NHS objective of maximizing heath gain from limited resources” (NICE, 2013). NICE uses CEA to inform decisions relating to new medicines and diagnostic appraisals, as well as clinical guidelines and public health, staffing levels and service delivery guidance (NICE, 2013). More recently, NICE International under its Methods for Economic Evaluation Project developed a reference case to support health economic evaluations funded by the Bill & Melinda Gates Foundation (NICE International & Bill and Melinda Gates Foundation, 2014). NICE’s reference case provides methodological guidance to be used in the analysis, including the perspective of the analysis, the comparator on which the incremental analysis is based and the discount rate to be applied. However, each of these methodological areas has controversies, which are discussed in the following sections.

6.3. Methodological issues in the use of economic evaluations at the micro level

The objective of a health system to improve health outcomes requires a measure of population health. Within a health system, several categories of measures are available including epidemiological (mortality rates), biomedical (for example, high blood pressure), behavioural (smoking, alcohol consumption) or psychosocial (health-related quality of life) (Cookson & Culyer, 2010). In all these instances the objective is to determine the impact of interventions in improving these outcomes. QALYs were developed by economists as an overall measure of population heath that combines many of the individual categories of health outcomes (Williams, 1995). It represents one year of life, adjusted for the health-related quality of that year of life. QALYs thus enable a quantitative assessment of several aspects of health and account for improvements in length as well as quality of life (Williams, 1985). As an overall measure, it is applicable to many different kinds of interventions and is particularly useful in improving efficiency by eliminating the difficulty of comparing interventions with diverse measures of health outcomes.

6.3.2. Decision rules in cost–effectiveness analysis

The objective of the decision-maker is to maximize health benefits or QALYs generated subject to the available budget constraints. Birch & Gafni (1992) present the following combinations of incremental benefits and incremental costs, which are also represented in the cost–effectiveness plane (Figure 6.2):

Figure 6.2

Cost–effectiveness plane.

1. Incremental costs are positive and incremental benefits are negative (quadrant II in Figure 6.2): This situation is clear in terms of a decision not to adopt the new technology.
2. Incremental costs are negative and incremental benefits are positive (quadrant IV in Figure 6.2): The benefits of the new technology outweigh the costs and hence justify adoption.
3. Both incremental costs and benefits are higher/lower (quadrants I and III in Figure 6.2): Where the sign is the same, it provides a trickier situation in CEA and requires decision rules to facilitate a choice.

The decision rules of CEA allow a policymaker to make choices in situations such as statement 3 with the aim of producing maximum health gain from the available resources. Two main types of types of decision problems have been addressed using CEA at the micro level. The first is the allocation of resources on interventions across multiple diseases. The second decision problem involves choices between different medical technologies for a particular disease.

The nature of policy questions that can be addressed using CEA at the micro level go beyond these two decision problems and can be classified according to Murray, Kreuser & Whang (1994) into three categories:

1. Ground zero: given a fixed health budget and infrastructure, what allocation of non-fixed resources would result in maximizing health outcomes (or reduce disease burden)? This problem is typical of the policy question most publically funded health systems in Europe face.
2. Marginal expansion: given a current allocation that cannot be changed and a fixed infrastructure, what is the optimal allocation for a marginal increase in the budget?
3. Ground zero with political constraints: with fixed infrastructure and certain services that must be protected from budgetary reallocation for political or other reasons, given a fixed budget, how best can resources be allocated so as to maximize health outcomes without reducing resource allocation to the protected services?

The main rules used in resource allocation decisions (Weinstein & Zeckhauser, 1973) through CEA are:

1. The league table rule: programmes are selected in ascending order of their ICERs until the total available budget is exhausted.
2. ICER threshold rule: programmes with an ICER less than or equal to a defined threshold are selected.

These two decision rules can be applied to the three policy questions listed earlier. If we knew the ICERs of all the interventions, then the league table approach can be applied in all three cases. However, it is rarely the case that the ICERs of all the interventions are available and hence the threshold rule provides an alternative solution to the resource allocation problem.

The ICER threshold rule has been the most applied in policy, for example, in Australia (Commonwealth Department of Heath and Ageing, 2002), the United Kingdom (NICE, 2008) and Canada (Ontario Ministry of Health, 1994). Devlin (2002) and Birch & Gafni (2003) have suggested that the former is not easily applicable given that ICERs are not available for all programmes to create a comprehensive league table.

Both of the decision rules are based on two critical assumptions, first that programmes are perfectly divisible (that is, that a programme can be delivered to specific proportions of the population) and second, that programmes have constant returns to scale (CRS), meaning the QALYs generated do not vary by the size of the programme (Drummond, 2005). In practice, these assumptions are unlikely to hold. Birch & Gafni (1993) have suggested that this makes CEA methods unreliable. In making choices between multiple programmes, decision-makers have to decide between programmes of varying sizes. By using ICERs to compare across programmes, the decision is based on the average cost per QALY. This means that programmes of different sizes and varying opportunity costs are compared on a single statistic. On the other hand, relaxing these assumptions implies decision-making without a specified amount of resources. In contrast, Johannesson & Weinstein (1993) have argued that CEA methods provide an acceptable approximation that can guide efficient resource allocations.

Interpreting whether the ICER of a given programme makes it acceptable requires a cost–effectiveness threshold value against which it can be compared. The specification of a threshold value has generated a lot research in the field. The threshold represents the opportunity cost of the marginal programmes funded from the available resources (Weisnstein & Zeckhauser, 1973). For example, when NICE issues positive guidance that allows the adoption of a new intervention, it increases the amount of resources required to allow for this new intervention (Gafni & Birch, 1993). However, the health budget is typically fixed and the resources to deliver the new intervention must come from displacing other interventions or services (Williams, 2004). In theory, two approaches can be applied to determine the threshold value; the first involves solving the maximization problem (of allocations resulting in the highest benefits) subject to the available resources, the second is the league table approach. In either case information on the ICERs of all possible combinations of programmes must be known, making it difficult to estimate the threshold (Gafni & Birch, 2006). In a recent report Claxton et al. (2013) developed and demonstrated methods that can be applied to estimate the cost–effectiveness threshold for the NHS by relying on routine data. They presented the best estimate of the cost–effectiveness threshold given the existing data and provided recommendations for future data collections efforts that will allow more precision in threshold estimates.

6.4. Cost–effectiveness analysis as a measure of organizational and system efficiency

6.5. Methodological and informational challenges

This section reviews the potential application of the fundamentals of CEA to meso/macro level resource reallocation decisions. As in the case of micro level decisions, measurement of AE at higher levels of the health system has two aspects:

1. to estimate deviation of current allocations from an optimal allocation; and
2. to reallocate resources towards an allocatively efficient mix.

To achieve an estimate of the optimal mix must be known before any deviation can be estimated. At the macro level this means potentially that all programmes of care must be compared against one another; additionally, the opportunity costs of public funds used in the health system that can be invested in other sectors must be considered. At the meso organizational level, all services an organization can potentially provide must be compared against one another. At both these levels such an exercise is likely to be unfeasible given the variety and breadth of services and programmes that constitute a health system. Knowing all inputs and outputs of each programme and service is not possible. At the micro level, even with a more constrained set of options, it would be impossible to estimate an optimal mix from which current deviations can be compared (for example, this might require knowing all the health technologies in an Essential Medicines list). Epstein et al. (2007) highlighted this problem at the micro/individual treatment level. They proposed a more limited way of approaching this issue by focusing on the technologies currently recommended by NICE in the United Kingdom based on CEA. Their objective was to determine the optimal mix within the recommended list that maximized the gross benefit subject to the available budget constraints.

In principle, the approach of Epstein et al. (2007) can be applied to all levels of the health system. However, the primary limitation is information. To make comparisons, the value of all health system outputs at each level must be known. But even before values can be assigned to outputs, a clear definition of outputs for different organizations, services and programmes is necessary. Once these outputs are determined, their value must be ascertained. This requires a uniform measure of benefits, such as the QALYs currently used at the micro level. Such a measure must be a relevant mechanism that applies to all the services and outputs in the health system. It must also be able to capture non-health outcomes, such as patient experience and improvements in productivity both for patients and care providers. But as in the debate on QALYs used at the micro level, there are additional concerns beyond relevance that revolve around the measurement of preferences. These include whose preferences for the optimal mix should be considered and whether it is possible to aggregate these utilities. This brings into question the applicability of CEA instruments currently used for eliciting societal values of outcomes such as the EQ-5D, SF-36, visual analogue scales, and so on. Even if an appropriate instrument for measuring societal values for meso and macro level services is developed, there remains the difficulty of generating values for all services and programmes of care. This discussion has shown that informational weaknesses prevent estimation of an optimal mix from which deviations in current allocations can be estimated. In reality, the informational requirements for comprehensive monitoring of AE are enormous and policymakers must make decisions in the absence of certain information.

Applications of CEA to micro level resource allocation decisions (see Box 6.1) have often focused on achieving the maximum health for a given budget.¹² This implies that the main constraint in the decision process is the available budget. However, in reality there are several other constraints, such as transition costs, infrastructure capacity and personnel redundancies that must be included in evaluations of reforms at higher levels of the health system. For example, at the meso level, if a hospital was to consider altering its current mix of services by removing certain services currently being provided and expanding other services, it must consider the ability of its personnel to adapt to these changes and the implications for health care staff no longer being employed by this hospital because of its narrowing of speciality areas. In addition, patient outcomes for those currently in care but no longer likely to be treated at the hospital must be considered, including availability and access to suitable alternative facilities. This implies that even at the meso level a decision-maker cannot simply focus on budgetary constraints in altering the mix of services but must consider other resource and adaptability constraints.¹³ A decision-maker may also need to consider externalities generated by a package of services beyond those reflected directly in health outcomes. Conventional cost–effectiveness ratios do not reflect gains beyond direct outcomes. For example, in considering altering the current mix of services, a decision-maker might be comparing the outcomes from a service being considered for exclusion with those of a new service to be included. One possibility is similar benefits are observed when only direct outcomes are compared. However, they may vary significantly in positive externalities generated that could result in a decision that runs contrary to those implied by cost–effectiveness ratios alone. As a result of these constraints, governments may choose to focus on achieving a certain acceptable level of benefits rather than maximizing them as required by CEA, and they may choose gradual changes to service mixes using criteria that are not standard to conventional cost–effectiveness models. For example, marginal changes may prioritize those with low transition costs and minimum negative externalities. Thus, approaches to resource allocation must maximize health objectives subject to a wider set of constraints.

Box 6.1

Micro level applications: surgery versus conservative management.

Prioritizing health care services through efficient allocations is at the core of option 2 (to reallocate resources towards an allocatively efficient mix) mentioned earlier. In the absence of adequate information to estimate the optimal mix against which current allocations can be compared, option (2) provides the next best alternative. Reallocations from the current mix of services essentially involve displacing a service or sector currently being funded. To estimate whether such a reallocation is more efficient than the current mix means examining the effects of altering the balance of expenditure between programmes. This is defined as marginal analysis, where any improvement in health benefits is a result of a change in the service or programme mix rather than an increase in expenditure. “Marginal analysis takes the current expenditure allocation as the starting-point (rather than an ‘optimal allocation’) and examines the effect of small changes to that pattern” (Cohen, 1994). It thus focuses on the marginal gains from expanding a programme and marginal losses (opportunity cost) from removing or contracting a service or programme of care.

In the health sector, applications of marginal analysis have most often been combined with programme budgeting exercises and hence been termed programme budgeting and marginal analysis (PBMA) (Mitton & Donaldson, 2001). Donaldson & Mooney (1991) described how these methods can be applied by health authorities. In principle, PBMA involves first dividing the health services provided at any level, that is, health system, organization or clinical unit into a set of programmes. The divisions are based on specific objectives such as target populations or disease groups. For each, programme costs and outputs are quantified. This is followed by the marginal analysis stage where shifts of resources from one programme to another are analysed in terms of benefits generated or losses incurred. For example, if a hospital shifted £500 000 from speciality care in a disease area to more general outpatient services, what benefits would be lost from speciality care and what gains would be made in outpatient care? While this approach is in principle applicable to all levels of the health system, it also faces the same drawback of CEA in that all benefits must be measured in the same units of health gains so that comparisons can be made across programmes of care. This is evidenced by the current applications of PBMA, which have primarily focused on, for example, grouping areas of clinical activity and estimating the effects of increasing or reducing spending in some areas.

This implies that improving AE at higher levels can be a complex process when compared with the incremental comparison of competing technologies as in conventional CEA. Applications to sectoral comparisons or programmes of care would require generating optimal mixes under different configurations, with the inclusion of a range of constraints and sometimes differing objectives from health benefit maximization.

As discussed in an earlier section, a major equity assumption that underlies CEA is that the value of a QALY is the same irrespective of the beneficiary. Some of the issues relating to this assumption at the micro level were discussed earlier. In the case of reallocations at higher levels of the health system, such an assumption of equality in value of outcomes is almost impossible to justify. For example, one of the primary reasons for reallocation may be to improve equity in health outcomes or to improve access of poorer populations. Such a reason for reallocation in itself moves away from pure AE. A decision-maker must therefore first define an equity objective and an outcome measure that collectively reflect individual viewpoints on the notion of fairness. Consider a national decision-maker applying CEA to assess whether reallocation of some resources from secondary care to primary care would improve efficiency from the current allocation. Such a reallocation assumes that less access to secondary care affects all individuals in the same way. But the influence of limited secondary care could vary greatly depending on, for example, socioeconomic backgrounds. A concern with equity of outcomes implies a need to weight benefits differently depending on the population groups being considered. Alternatively, if the focus is on equity of access, then the need to adjust costs arises. For instance, costs of securing access of certain services among some groups of the population may be higher than among others. Thus, the expected costs of some services within programmes of care must we weighted differentially. CEA can incorporate these equity considerations,¹⁴ however, implementing these weights or adjustments in a cost–effectiveness model is demanding in terms of information and data requirements (for example, baseline information on the current distribution, and societal valuation of changes to the current distribution, is difficult to obtain).

The applicability of current approaches of CEA to questions of AE is also constrained by the relevance of its assumptions on scope and scale. As discussed earlier, the decision rules of CEA are based on the assumptions of indivisibility of programmes and CRS. This means that comparisons are made on long run average costs of individual treatments. Hospitals by design provide a range of services that draw on the economies of scale of providing different forms of care using an underlying infrastructure base. Decisions of reallocation for efficiency gains cannot focus on comparisons of average costs of individual services but must take into account bundles of the services being provided and the implications of shifting resources and redefining packages, and the corresponding losses or gains because of changes in scale and scope of the packages.

6.6. Cost–effectiveness analysis in policy: present and future

CEA in health care has a long-standing tradition in many high-income countries. This section begins with a discussion of current applications of CEA in health policy. The scope of CEA varies widely across regions and countries. In Europe, policymakers responded to financial pressures and growing public demand for improved quality by setting up HTA agencies, such as the Institute for Quality and Efficiency in Health Care in Germany or NICE in the United Kingdom. Across the EU, technology appraisals are used in policies for pricing, health care provider reimbursements and guiding clinical practice. For example, in the United Kingdom, NICE produces clinical guidelines (NICE, 2008) for the NHS and is required to make recommendations on the basis of both effectiveness and cost–effectiveness.

While CEA is an essential part of the evaluation process in many European countries, there is diversity in how final resource allocation decisions are made. For example, unlike the United Kingdom, which compares the cost–effectiveness ratio of a new intervention against a threshold value, Germany applies an efficiency frontier approach that compares the ICER of the new interventions with the next most cost-effective intervention, which essentially represents the prevailing efficiency level (Klingler et al., 2013). However, since the German parliament passed the Act on the Reform of the Market for Medicinal Products in 2010 there has been debate on whether the efficiency frontier approach is consistent with the law. Since then no decisions on coverage have been made using this approach.

Expenditure on pharmaceuticals is the fastest growing health care cost category in high-income countries (CIHI, 2007; Duerden et al., 2004). CEA has been used extensively in managing expenditure on pharmaceuticals in Australia, Canada and the United Kingdom (Clement, 2009). In Canada, the Common Drug Review provides recommendations for new drug listings for the 18 publically funded drug plans (Tierney & Manns, 2008). The recommendations are based on clinical efficacy as well as cost–effectiveness. In a similar manner, Australia’s Pharmaceutical Benefits Advisory Committee gives cost–effectiveness-based advice on which drugs should be funded under the Pharmaceutical Benefits Scheme.

More recently, CEA is being extended to the development of complete care pathways. Thus far, the role of cost–effectiveness in clinical guidelines has been piecemeal and selective. In some cases, CEA is applied independently at different points in the care pathway with assumptions that may not be consistent across the pathway (Lord et al., 2013). In other cases, the lack of time to build new models means cost–effectiveness estimates may not be available when resource allocation decisions are made. The risk with this selective approach is that sometimes, adequate evidence does not exist to make an informed decision. For example, in a systematic review of the United Kingdom care pathway for colorectal cancer, Tappenden et al. (2009) found no relevant United Kingdom cost–effectiveness estimates for large segments of the pathway. The Modelling Algorithm Pathways in Guidelines project (Lord et al., 2013) was developed to evaluate the feasibility and relevance of modelling complete care pathways for the NICE clinical guidelines. The rationale for such an approach (Tappenden et al. (2012) refer to this approach as whole disease modelling) is that a model that captures a full guideline should allow CEA of a range of different scenarios. By using a common framework and similar assumptions throughout the care pathway, the accuracy and consistency of the estimates would be improved (Box 6.2).

Provider reimbursement schemes play a critical role in improving productivity and efficiency in health systems. In England, GP payments include a pay-for-performance (P4P) scheme known as the Quality and Outcomes Framework (QOF). The scheme rewards performance in four areas: clinical; organizational; patient experience; and other services. Walker et al. (2010) evaluated the cost– effectiveness of a subset of nine QOF indicators with direct clinical impact. The authors found that QOF incentive payments are likely to be cost-effective even if the actual improvement in care outcomes is modest. However, the study did not include the costs of administering the QOF scheme in the analysis. We now consider the potential to develop a framework for extending CEA to studying AE at the health system level (see Box 6.3).

The discussion of AE can be considered from either an ex ante perspective or an ex post perspective. In the ex ante case, the focus is on prospective assessments of the health system. The role of AE in this case is to guide decisions on the purchase of health care. For example, this could mean whether a new intervention should be adopted or whether a reallocation of resources to a new service mix achieves greater AE. The ex ante case is of particular importance to countries that are looking to establish publically funded health systems. Such an exercise is key to the WHO’s move to promote universal health coverage in developing countries (WHO, 2010). The objective of AE in this case is to establish a package of health care services that maximizes the health outcomes of the population in the country and is made available free to all individuals at the point of access.

Box 6.2

Meso level applications: redesigning care pathways.

Box 6.3

Macro level applications: a research agenda.

On the other hand, there is a strong case for also measuring AE in ex post analyses. For example, in low to middle income countries looking to improve AE in their health systems, ex post analyses can establish a baseline to identify the potential for efficiency savings. The retrospective approach to measuring value for money is particularly applicable in established publically funded systems such as the ones in Europe, Australia or Canada. In this case the focus is more on improving efficiency through changes in the current mix of services provided by different levels of the health system. Take, for example, the case of a publically funded health system, where the government allocates resources to services based on perceived societal valuations of those services. It may then be relevant to ensure that providers are making available services that are consistent with the societal valuations rather than diverting resources to other service areas. For example, at the hospital level retrospective analysis might identify excessive capacity or investment in specialities or services not consistent with the needs of society. It may also mean divergence from prescribed clinical guidelines that reflect cost–effectiveness. Deviations from the preferred options can reflect inefficiency and therefore a reduction in the value for money.

Measuring allocative inefficiency at the organizational level can offer insights into the performance of the different organizations that comprise a health system. There have been several metrics developed to achieve this objective (Hollingsworth, 2003). However, health care organizations, particularly hospitals, are complex structures providing a range of services to heterogeneous populations. Thus, any such metrics of deviation and inefficiency must be scrutinized to ensure that they indeed reflect allocative inefficiency rather than being caused by constraints faced by the organization in servicing its population. For example, observed variations in AE must be conditional on adjustments for the case mix in hospitals (see Chapter 2).¹⁵ They must also reflect the policy constraints, environmental factors and determinants of demand for services that are likely to influence performance. For example, the use and take-up of services offered by a hospital depend on the demand for its services and the elasticity of this demand with respect to substitute and complementary services. The elasticity of demand for any of its services varies to different extents by price, distance and convenience.

Scrutinizing the accuracy of metrics of inefficiency becomes even more important at the health system level which by design might preclude flexibilities that allow adherence to an allocatively efficient mix. Such constraints might include the structure of physical capital and administrative arrangements, financial constraints including long-term commitments to certain groups of patients, and at least in the short-term, workforce constraints. In addition, there may be governance constraints including the absence of effective accountability mechanisms that prevent health systems from maximizing performance.

Retrospective analyses of AE can also be applied by international donors to evaluate the value for money received on their external aid to developing countries. Teerawattananon et al. (2013) described the potential applications of CEA for monitoring and evaluating the performance of the Global Fund to Fight AIDS, Tuberculosis and Malaria. They emphasized the importance of retrospective CEA being included as part of final reports submitted by grant recipients. Such analyses not only facilitate an understanding of the value for money achieved by Global Fund grants, but equally they provide important information to recipient countries on the interventions that are cost-effective in their settings and therefore worthy of long-term finance. Thus, ex post cost–effectiveness evaluations provide opportunities to inform decision-makers in developing countries of the implications of potentially sustaining or rolling out programmes initially funded by external donors. An example of ex post CEA in the case of HIV prevention is presented in Tosanguang et al. (2012). The programme aimed to expand HIV preventive services among high-risk populations in Thailand. Concern over the long-term sustainability of the programme beyond the initial five-year international funding led to a retrospective CEA. The analysis indicated that the programme had a much higher cost per person in Thailand than similar programmes in other countries, such as India and Bangladesh, and did not perform as well. These examples illustrate the importance of incorporating CEA in decision-making at international organizations and global health initiatives to ensure allocatively efficient resource allocation.

6.7. Conclusion

The final part of this chapter highlights potential areas for future research that will enable the application of CEA to meso and macro level efficiency analysis, both in retrospective assessment of past performance, and as a tool for guiding future allocation decisions.

There is growing awareness that the design and performance of national health systems has large implications for other sectors within a country. This focus has been particularly accentuated with the stagnation of health budgets in most European countries and with the recognition of the importance of maximizing health outcomes given the constraints. Health system level analyses of AE offer a powerful tool for identifying potential gains that can be made with the given resources. The currently widely applied tool of CEA provides an important framework within which assessments of AE can be made and offers a framework for both prospective purchasing decisions and retrospective evaluations. In the absence of information on all possible health care interventions and services, marginal analysis offers an opportunity to improve AE. However, several challenges must be addressed before efficiency is evaluated at the meso or macro levels of the health system.

First, research is required into the appropriateness of existing measures of preference valuation for higher levels of evaluation. The second challenge relates to finding and allowing for one or more equity criteria that influences policymaker decisions in allocating resources across and within health sectors. Current applications of equity-adjusted CEA are limited. Some examples from the emerging literature that allow for broader objectives in the CEA objective function include the evaluation of the impact of policies across multiple domains using extended CEA (Verguet, Laxminarayan & Jamison, 2015); allowing for distributional concerns (Asaria et al., 2013); and financial concerns (Smith, 2013). But at higher levels of the health system these criteria can be major determinants of allocations. Third, factors beyond the budget constraints present challenges to adopting an optimal mix of services. These might include human resources constraints, transition costs of either eliminating or incorporating new services or even the dynamic aspect of the health system. Thus, current methodologies for incorporating equity considerations and other non-financial constraints in conventional CEA must be explored. It is important to acknowledge that many health outcomes currently observed are the results of decisions taken in earlier time periods and that often changes may not reflect in the outcomes for a long time. Finally, decision-making tools such as CEA must also allow for externalities and system-wide effects in estimating the gains and losses from changing the current mix of services or in moving resources between sectors.

References

• Arrow KJ. Uncertainty and the welfare economics of medical care. American Economic Review. 1963;53(5):941–973.
• Asaria M, et al. York: Centre for Health Economics; 2013. Distributional cost–effectiveness analysis of health care programmes. CHE Research Paper 91. https://www​.york.ac.uk​/media/che/documents​/papers/researchpapers​/CHERP91_distributional​_CEA_healthcare.pdf, accessed 22 July 2016.
• Birch S, Gafni A. Cost effectiveness/utility analyses: do current decision rules lead us to where we want to be? Journal of Health Economics. 1992;11(3):279–296. [PubMed: 10122540]
• Birch S, Gafni A. Changing the problem to fit the solution: Johannesson and Weinstein’s (mis) application of economics to real world problems. Journal of Health Economics. 1993;12(4):469–476. [PubMed: 10131757]
• Birch S, Gafni A. Economics and the evaluation of health care programmes: generalisability of methods and implications for generalisability of results. Health Policy. 2003;64(2):207–219. [PubMed: 12694956]
• Canadian Institute for Health Information (CIHI). Ottawa, ON: CIHI; 2007. Drug expenditure in Canada, 1985 to 2006.
• Claxton K, et al. Discounting and decision making in the economic evaluation of health-care technologies. Health Economics. 2011;20(1):2–15. [PubMed: 21154521]
• Claxton K, et al. York: University of York; 2013. Methods for the estimation of the NICE cost effectiveness threshold.
• Clement FM. Using effectiveness and cost–effectiveness to make drug coverage decisions: a comparison of Britain, Australia, and Canada. JAMA. 2009;302(13):1437–1443. [PubMed: 19809025]
• Cohen D. Marginal analysis in practice: an alternative to needs assessment for contracting health care. BMJ. 1994;309(6957):781–784. [PMC free article: PMC2541009] [PubMed: 7950568]
• Commonwealth Department of Health and Ageing. Canberra, Australia: Commonwealth of Australia; 2002. Guidelines for the pharmaceutical industry on preparation of submissions to the Pharmaceutical Benefits Advisory Committee. Including major submissions involving economic analyses.
• Cookson R, Culyer A. Measuring overall population health: the use and abuse of QALYs. In: Killoran A, Kelly MP, editors. Evidence-based public health: effectiveness and efficiency. Oxford: Oxford University Press; 2010.
• Cookson R, et al. York: Centre for Health Economics; 2016. Years of good life based on income and health: Re-engineering cost-benefit analysis to examine policy impacts on wellbeing and distributive justice. CHE Research paper 132. https://www​.york.ac.uk​/media/che/documents​/papers/researchpapers​/CHERP132_income​_health_CBA_wellbeing_justice.pdf, accessed 19 September 2016.
• Culyer AJ, Wagstaff A. Equity and equality in health and health care. Journal of Health Economics. 1993;12(4):431–457. [PubMed: 10131755]
• Devlin N. An introduction to the use of cost–effectiveness thresholds in decision-making: what are the issues? In: Towse A, Pritchard C, Devlin N, editors. Cost effectiveness thresholds: economic and ethical issues. London: King’s Fund and Office of Health Economics; 2002.
• Dodd S. Designing improved healthcare processes using discrete event simulation. British Journal of Healthcare Computing & Information Management. 2005;22(5):14–16.
• Dolan P, Tsuchiya A. The elicitation of distributional judgements in the context of economic evaluation. In: Jones AM, editor. The Elgar companion to health economics. Cheltenham: Edward Elgar Publishing; 2006.
• Donaldson C, Mooney G. Needs assessment, priority setting, and contracts for health care: an economic view. BMJ. 1991;303(6816):1529–1530. [PMC free article: PMC1671822] [PubMed: 1782499]
• Drummond M. Methods for the economic evaluation of health care programmes. 3rd edn. Oxford: Oxford University Press; 2005.
• Duerden M, et al. Current national initiatives and policies to control drug costs in Europe: UK perspective. Journal of Ambulatory Care Management. 2004;27(2):132–138. [PubMed: 15069991]
• Epstein DM, et al. Efficiency, equity, and budgetary policies: informing decisions using mathematical programming. Medical Decision Making. 2007;27(2):128–137. [PubMed: 17409363]
• Fetter RB. Diagnosis related groups: understanding hospital performance. Interfaces. 1991;21(1):6–26.
• Gafni A, Birch S. Guidelines for the adoption of new technologies: a prescription for uncontrolled growth in expenditures and how to avoid the problem. CMAJ. 1993;148(6):913–917. [PMC free article: PMC1490730] [PubMed: 8448705]
• Gafni A, Birch S. Incremental cost–effectiveness ratios (ICERs): the silence of the lambda. Social Science & Medicine. 2006;62(9):2091–2100. [PubMed: 16325975]
• Gravelle H, et al. Discounting in economic evaluations: stepping forward towards optimal decision rules. Health Economics. 2007;16(3):307–317. [PubMed: 17006970]
• Ham C, et al. London: The King’s Fund; 2011. Where next for the NHS reforms? The case for integrated care. http://www​.kingsfund​.org.uk/sites/files/kf​/where-next-nhs-reforms-case-for-integrated-care-ham-imison-goodwin-dixon-south-kings-fund-may-2011.pdf, accessed 22 July 2016.
• Hollingsworth B. Non-parametric and parametric applications measuring efficiency in health care. Health Care Management Science. 2003;6(4):203–218. [PubMed: 14686627]
• Johannesson M, Weinstein MC. On the decision rules of cost–effectiveness analysis. Journal of Health Economics. 1993;12(4):459–467. [PubMed: 10131756]
• Klingler C, et al. Regulatory space and the contextual mediation of common functional pressures: analyzing the factors that led to the German Efficiency Frontier approach. Health Policy. 2013;109(3):270–280. [PubMed: 23380191]
• Lord J, et al. Economic modelling of diagnostic and treatment pathways in National Institute for Health and Care Excellence clinical guidelines: the Modelling Algorithm Pathways in Guidelines (MAPGuide) project. Health Technology Assessment. 2013;17(58):v–vi. 1–192. [PMC free article: PMC4781470] [PubMed: 24325843]
• Lowe FC. Economic modeling to assess the costs of treatment with finasteride, terazosin, and transurethral resection of the prostate for men with moderate to severe symptoms of benign prostatic hyperplasia. Urology. 1995;46(4):477–483. [PubMed: 7571214]
• Meltzer DO, Smith PC. Theoretical issues relevant to the economic evaluation of health technologies. In: Pauly MV, McGuire TG, Barros PP, editors. Handbook of health economics. Vol. 2. Amsterdam: Elsevier; 2011.
• Mitton C, Donaldson C. Twenty-five years of programme budgeting and marginal analysis in the health sector, 1974–1999. Journal of Health Services Research & Policy. 2001;6(4):239–248. [PubMed: 11685789]
• Murray CJ, Kreuser J, Whang W. Cost–effectiveness analysis and policy choices: investing in health systems. Bulletin of the World Health Organization. 1994;72(4):663–674. [PMC free article: PMC2486597] [PubMed: 7923545]
• Naylor C, et al. Clinical commissioning groups: supporting improvement in general practice? London: The Kings Fund; 2013. http://www​.kingsfund​.org.uk/sites/files/kf​/field/field_publication_file​/clinical-commissioning-groups-report-ings-fund-nuffield-jul13.pdf, accessed 22 July 2016.
• National Institute for Health and Care Excellence (NICE). London: NICE; 2008. Social value judgements: principles for the development of NICE guidance. [PubMed: 27905706]
• NICE. London: NICE; 2013. Guide to the methods of technology appraisal 2013. https://www​.nice.org​.uk/process/pmg9/chapter/1-foreword, accessed 22 July 2016.
• NICE International, Bill and Melinda Gates Foundation. NICE International; 2014. The Gates Reference Case: what it is, why it’s important and how to use it. https://www​.nice.org​.uk/Media/Default/About​/what-we-do/NICE-International​/projects​/Gates-Reference-case-what-it-is-how-to-use-it.pdf, accessed 22 July 2016.
• Ontario Ministry of Health. Ontario guidelines for economic analysis of pharmaceutical products. Toronto: Ministry of Health, Ontario; 1994.
• Smith PC. Incorporating financial protection into decision rules for publicly financed healthcare treatments. Health Economics. 2013;22(2):180–193. [PubMed: 22241688]
• Tam TY, Smith MD. Pharmacoeconomic guidelines around the world. ISPOR CONNECTIONS. 2008;10:4–5.
• Tappenden P, et al. Systematic review of economic evidence for the detection, diagnosis, treatment, and follow-up of colorectal cancer in the United Kingdom. International Journal of Technology Assessment in Health Care. 2009;25(4):470–478. [PubMed: 19845977]
• Tappenden P, et al. Whole disease modeling to inform resource allocation decisions in cancer: a methodological framework. Value in Health. 2012;15(8):1127–1136. [PubMed: 23244816]
• Teerawattananon Y, et al. Health technology assessments as a mechanism for increased value for money: recommendations to the Global Fund. Global Health. 2013;9:35. [PMC free article: PMC3848836] [PubMed: 23965222]
• Tierney M, Manns B. Optimizing the use of prescription drugs in Canada through the Common Drug Review. CMAJ. 2008;178(4):432–435. [PMC free article: PMC2228339] [PubMed: 18268271]
• Tosanguang K, et al. Nonthaburi: Health Intervention and Technology Assessment Program; 2012. Economic evaluation of comprehensive HIV prevention interventions targeting those most at risk of HIV/AIDs in Thailand (CHAMPION)
• Verguet S, Laxminarayan R, Jamison DT. Universal public finance of tuberculosis treatment in India: an extended cost–effectiveness analysis. Health Economics. 2015;24(3):318–332. [PubMed: 24497185]
• Wagstaff A. QALYs and the equity-efficiency trade-off. Journal of Health Economics. 1991;10(1):21–41. [PubMed: 10113661]
• Walker S, et al. Value for money and the Quality and Outcomes Framework in primary care in the UK NHS. British Journal of General Practice. 2010;60(574):e213–e220. [PMC free article: PMC2858553] [PubMed: 20423576]
• Weinstein MC, Zeckhauser R. Critical ratios and efficient allocation. Journal of Public Economics. 1973;2:147–157.
• Weinstein MC, Torrance G, McGuire A. QALYs: the basics. Value in Health. 2009;12 Suppl. 1:S5–S9. [PubMed: 19250132]
• Williams A. Economics of coronary artery bypass grafting. British Medical Journal. 1985;291(6491):326–329. [PMC free article: PMC1416615] [PubMed: 3160430]
• Williams A. York: University of York; 1995. The measurement and valuation of health: a chronicle. (Centre for Health Economics Discussion Paper). http://www​.york.ac.uk/che/pdf/DP136.pdf, accessed 22 July 2016.
• Williams A. Intergenerational equity: an exploration of the ‘fair innings’ argument. Health Economics. 1997;6(2):117–132. [PubMed: 9158965]
• Williams A. What could be nicer than NICE? Office of Health Economics annual lecture. 2004. https://www​.ohe.org/publications​/what-could-be-nicer-nice, accessed 22 July 2016.
• WHO. Health systems financing: the path to universal coverage. Geneva: WHO; 2010. The World Health Report. http://apps​.who.int/iris​/bitstream/10665​/44371/1/9789241564021_eng.pdf, accessed 22 July 2016.