There has long been an aspiration to invest in promoting health, preventing ill health, and reducing health inequality. This aspiration can be realized through a wide variety of public health interventions, including not only screening, vaccination, and other preventive activities undertaken by healthcare professionals but also a broad range of fiscal and social programs and regulations beyond the healthcare sector with impacts on the health of the population.
Economic evaluation is increasingly used to inform decisions about which public health interventions to fund from scarce resources. However, there remains a dearth of evidence on the effectiveness and cost–effectiveness of public health interventions and the evidence that is available tends to be relatively weak – at least compared with evidence for healthcare interventions – with important methodological challenges remaining. In the UK, for example, the Wanless Report of 2004 noted the lack of evaluations undertaken in the public health field and the lack of methods development, expertise and funding available to generate the evidence.
Health economic guidelines for assessing the cost– effectiveness of healthcare technologies – such as new drugs, devices and medical procedures – have existed in many jurisdictions across the developed world for well over a decade. By contrast, methods for the economic evaluation of public health interventions are less well-established. Healthcare technology assessment guidelines use an economic evaluation framework to provide a clear and transparent approach for assessing the relative costs and benefits of alternative options with the aim of achieving an efficient allocation of resources. Typically, in relation to healthcare technology assessment, this framework focuses on the decision-making objective of maximizing health gain subject to an exogenously fixed healthcare budget constraint.
The economic evaluation of many public health interventions raises additional methodological challenges. As with the evaluation of standard ‘clinical’ healthcare technologies, it is important to determine effect estimates for use within economic evaluations of public health interventions. However, public health interventions tend to be directed at populations or communities rather than specific individuals, and can be less suited to evaluation through randomized controlled designs: the gold standard of study design for obtaining unbiased estimates of effect. In addition, public health interventions tend to generate a broad range of nonhealth benefits and opportunity costs, which may extend beyond the healthcare sector, with implications for other sectors subject to different objectives and budget constraints. Lastly, although standard health economic evaluation methods focus on maximizing health gain, a particular feature of many public health interventions is a concern to reduce health inequalities and so decision makers may be interested in information about the distribution of health levels, gains and opportunity costs within the general population as well as the average health gain for recipients of the intervention. It is, therefore, not clear how far standard methodological guidelines for healthcare technology assessment are appropriate in public health, and some public health scholars have argued that over-zealous application of standard health technology assessment (HTA) evaluation processes and criteria in public health can lead to systematically misleading conclusions.
This article briefly reviews standard methods for the economic evaluation of healthcare interventions before identifying key methodological challenges for the economic evaluation of public health interventions. To illustrate the methodological issues, it contrasts National Institute for Health and Clinical Excellence (NICE) (http://www.nice.org.uk/) methods guides for economic evaluation of healthcare technologies and public health interventions, respectively, linking this to the methodological challenges of undertaking economic evaluation of public health interventions. Finally, it explores ways forward, noting some recent methods developments in the field.
Methods For The Economic Evaluation Of Healthcare Interventions
Economic evaluation offers a clear analytical framework for assisting decision making. In the presence of limited resources and a fixed healthcare budget, economic evaluation offers a transparent approach, underpinned by explicit social value judgments, for choosing how to allocate society’s scarce healthcare resources. To do this, decision-making objectives and comparator interventions are identified and the opportunity cost of selecting a particular intervention is assessed by considering whether its’ value exceeds the value that would have been achieved if the next best alternative intervention were selected, given available resources. Costs and consequences of relevant alternative activities are compared, with the most efficient use of resources being the option that provides the best outcome.
There are two main philosophical approaches underpinning economic evaluation: the ‘welfarist’ approach and the ‘non-welfarist’ approach. Each has implications for the economic evaluation methods of choice. The key outcome in the welfarist approach is the satisfaction of individual preferences , typically measured using willingness to pay (WTP) reflecting the maximum amount an individual would pay for a particular intervention. In contrast, the nonwelfarist approach focuses on some other measure of outcome reflecting the decision-maker’s objective, such as the quality-adjusted lifeyear (QALY) which can be used as a summary measure of total population health benefit. In practice, the nonwelfarist approach and the use of cost–effectiveness analysis (CEA) based on QALYs predominates in the healthcare sector, whereas the welfarist approach and the use of cost–benefit analysis (CBA) based on WTP estimates predominates in other areas of social policy such as transport, occupational safety, the environment, employment, housing, and so on. However, each approach can be applied in different ways and both CEA and CBA could in principle be conducted using different outcome measures.
One of the most comprehensive and commonly referred to set of guidelines for health economic evaluation methods is the NICE health technology ‘References: case’ from the UK. These guidelines are periodically revised – they were first issued in 2004, updated in 2008, and a third edition is currently being prepared that may incorporate substantial revisions related to a new system of ‘value-based pricing’ due to be introduced from 2014. The 2008 version, which sets out a thoroughly ‘nonwelfarist’ approach to undertaking economic evaluations of healthcare interventions, is described in Box 1. Under this approach, the aim is to maximize health given a fixed budget constraint, whereby funding the new intervention involves displacing one or more existing interventions. A new intervention is considered cost–effective if the extra cost incurred to gain an extra one QALY, relative to the next best intervention, is less than approximately ₤20 000 to ₤30 000. This is the cost–effectiveness threshold value and represents the opportunity cost or health forgone by the displaced activity.
Methods Challenges For The Economic Evaluation Of Public Health Interventions
Unlike clinical healthcare interventions, public health interventions tend to be directed at populations or communities rather than specific individuals. One implication of this for evaluation is that public health interventions often have relatively small and hard-to-detect effects at the level of the individual, which can nevertheless sum to large effects at population level. Public health interventions also tend to generate a broader range of costs and nonhealth benefits, including costs falling on private consumption and public sector budgets beyond the healthcare sector. Finally, whereas standard economic evaluation methods focus on efficiency with the aim of maximizing health gain, typically the aim of public health interventions extends further to include a concern for reducing unfair health inequalities. Indeed, some public health professionals would go so far as to say that the primary goal of public health interventions is to generate a more equitable distribution of health in society. Based on these considerations, adjustments to standard health economic evaluation methods may be required to assess public health interventions. The formulation of social objectives, the range of outcomes and opportunity costs to be quantified, and the methods for evaluating those outcomes and opportunity costs may all need to be reconsidered, to align the methods with the broader scope and goals of public health interventions.
Box 1 Summary of the HTA and the public health References: cases (National Institute for Health and Clinical Excellence, 2008, 2009)
A number of reviews have explored the challenges of economic evaluation in public health. Four key methods challenges can be identified in applying standard health economic evaluation methods to public health interventions. These include (1) attributing outcomes to interventions; (2) measuring and valuing outcomes; (3) incorporating equity considerations; and (4) identifying inter-sectoral costs and consequences, as developed further below.
Attributing Outcomes To Interventions
Before undertaking CEA, it is essential to determine the effectiveness of relevant comparator interventions. Most healthcare interventions are directed at identified groups of individuals, and the randomized controlled trial (RCT) design is typically used as the ‘gold standard’ study design for primary data collection. Most published guidelines in the healthcare field, including the current NICE References: case, indicate a preferences for using RCT evidence to identify and measure the effectiveness of relevant comparators. Some individually focused, face-to-face public health interventions may be suitable for evaluation using an RCT. However, this might not be feasible for more community-based public health interventions, and other forms of experimental data, such as cluster randomized trials, may not be available for obtaining effect estimates. Because there are likely to be fewer controlled trials of public health programs, it will often be necessary to use other approaches for obtaining an unbiased estimate of the intervention effect. Natural experiments and the use of nonexperimental data can be used to fill some gaps in the public health evidence base. However, evidence of this kind is vulnerable to selection bias. Where study participants are not randomized to the interventions, the effects estimated for different interventions may be biased because of confounding between assignment to the intervention and the study participant characteristics. This implies that effectiveness estimates may, in part, be caused by differences in population characteristics instead of the intervention of interest. More use may be made of statistical techniques that have been developed to analyze nonexperimental data, including a range of econometric methods and simulation modeling methods. Methods are available to improve the validity of the comparator groups through study design, such as matching patients across interventions and in the analysis of results by statistically adjusting for case mix, assuming sufficient data is available to do so.
Given the dearth of RCT evidence for many public health interventions, systematic reviews of evidence which exclude all non-RCT design evidence may not yield parameter estimates that could be used in economic evaluations. Instead, a common outcome would be that there was insufficient research evidence available for assessing effectiveness. Other methods such as narrative review summaries are not particularly helpful for analysts requiring empirical estimates. Instead, broader evidence synthesis techniques are required, which enable the analyst to include all relevant evidence within an economic evaluation, including robust nonexperimental evidence from natural experiments as well as RCTs. Modeling is also required to extend the analysis to an appropriate time horizon. This may be of particular importance for public health interventions which impact health over the longer term. Modeling is also required to indicate how uncertainty in the available evidence translates to the probability that a particular decision is the correct one.
Measuring And Valuing Outcomes
Typically the main aim of a new healthcare technology is to improve health. By contrast, public health interventions tend to generate a broad range of health and nonhealth benefits, which may extend beyond the healthcare sector. Many health economic guidelines, including the NICE References: case for HTA, recommend that health outcomes are measured in QALYs. For public health interventions, however, a range of nonhealth outcomes may also be relevant – including fairly tangible crime, education and employment outcomes as well as harder-to-measure outcomes such as public reassurance, the empowerment of citizens to make informed choices, and community cohesion. Some of these outcomes may be possible to incorporate within a QALY-type framework, others not. Therefore, the use of other outcome measurement and valuation methods may be appropriate – for example, subjective well-being scores, or multidimensional indices of well-being in which health is only one component, or WTP-based methods including the possibility of using some form of ‘adjusted’ WTP after purging the influence of income, incomplete information, misperceptions of risk, protected characteristics under equalities legislation and/or other determinants of ‘raw’ WTP that social decision makers may consider inappropriate reasons for public resource allocation decisions.
Intersectoral Costs And Consequences
Public health interventions may have impacts that extend beyond the healthcare sector. Costs and benefits associated with public health interventions can fall on different sectors of the public sector. For example, a public health intervention to reduce substance abuse may reduce criminal justice costs. Interventions that are implemented in other sectors of the economy may also have public health implications. For example, improvements in housing could reduce illness and injuries, with consequent reductions in healthcare utilization and expenditure. In addition, individuals may incur out-of-pocket costs associated with accessing and using interventions. There may be ripple effects associated with an intervention that could extend across other sectors of the economy, including the private and voluntary sectors. An obvious way of addressing this challenge would be to monetize benefits. However, this still raises practical questions about how different health and nonhealth outcomes are to be valued and how to address the issue of fixed budget constraints faced by healthcare and other public sector decision makers, rather than assuming that all costs and outcomes are costlessly exchangeable between different policy sectors. It also raises deeper theoretical questions about whether and how it is possible to integrate ‘welfarist’ and ‘nonwelfarist’ approaches.
If a healthcare and personal social services (PSS) perspective is chosen, as for the NICE HTA References: case, resource use and costs falling on the healthcare and PSS sector are evaluated but impacts falling on other sectors are not. Where the healthcare sector budget is fixed, a new intervention can only be funded if other activity is displaced within the sector. There is an opportunity cost incurred in investing in the new intervention, i.e., the health forgone among the group of patients whose intervention is displaced and therefore no longer available. Under the NICE HTA References: case, this health opportunity cost is approximated by the cost– effectiveness threshold value. Using this approach, for approximately every d20 000 of expenditure the opportunity cost is one QALY lost through displacing existing interventions. However, the relevant opportunity costs and threshold values are likely to differ across sectors – with different sectors having different threshold values for both health and nonhealth opportunity costs.
The NICE HTA References: case recommends that analysts undertake CEA on the basis of benefits measured in QALYs and costs covering National Health Service (NHS) and PSSs resource use. To identify possible inter-sectoral impacts of public health interventions, the costs and benefits falling on other sectors could be considered for each comparator intervention. The cost–consequence analysis (CCA) approach, whereby a range of sector-relevant costs and consequences are measured and reported separately, could be informative. In addition, it might be appropriate to account for these impacts using real or hypothetical budgetary transfers across sectors whereby sectors that gain net benefits can ‘compensate’ sectors that lose net benefits – although the feasibility of this approach requires further investigation. For example, if a generic but sector-specific measure of outcome such as the QALY were identified for each sector, this could be used in References: to the relevant cost–effectiveness threshold value for the sector to compute net benefits for each sector.
Incorporating Health Equity Considerations
The final key methods challenge identified for the economic evaluation of public health interventions is a concern to reflect the health equity implications of public health interventions The importance of achieving health equity is recognized in many published guidelines for economic evaluation. However, in practice, health equity considerations are rarely quantified. In terms of health outcomes, it is typically assumed that the value of a QALY is the same whoever receives it. The analysis will also contain some judgment about which types of resource use to cost, and this can be influenced by equity considerations – for example, considerations of nondiscrimination may influence judgments about how far to count productivity costs, which can be much higher for highly paid workers compared with those on low pay and economically inactive groups such as children and pensioners. However, current analyses do not examine health inequality issues – in particular the distribution of QALY levels or gains between population subgroups, for example, by socioeconomic status, ethnicity and gender – which are of particular interest in public health.
To reflect health inequality considerations, data on equity relevant subgroups need to be identified, collected, and analyzed. Assuming the decision maker has the twin objectives of both reducing health inequality and improving health, if the cost-effective intervention is the option that also minimizes health inequality then the decision is clear. However, if one intervention achieves greater health outcome and the comparator intervention achieves greater health equality then under standard cost-effectiveness decision rules it is not clear which intervention to choose. Some methods have been proposed for dealing with this trade-off issue, as reviewed in the section on methods developments below.
Recent Guidance For The Economic Evaluation Of Public Health Interventions
Before reviewing recent methods developments in the field, it is useful to refer to NICE guidance that has been developed to facilitate a consistent and transparent approach to undertaking good quality economic evaluations of public health interventions in the UK. The NICE guide to methods for the development of public health guidance is periodically updated: it was first issued in 2006, and has been updated in 2009 and 2012. Described below is the 2009 version, which is the most directly comparable to the 2008 NICE ‘References: case’ for technology appraisal. The main relevant changes in the 2012 edition are a reduced discount rate for costs and benefits and an even stronger emphasis on conducting CCA and CBA as well as CEA using QALYs, following a recent shift of public health budgets in England to local government and away from the healthcare sector. As detailed in Box 1, the NICE public health References: case (right-hand column) differs in a few characteristics compared to the NICE HTA References: case (left-hand column). These differences illustrate that elements of assessment have been adapted to reflect the characteristics of public health interventions.
The NICE public health References: case reflects the fact that public health interventions may involve resources, costs, and outcomes beyond the healthcare sector. It recommends that ‘‘important health effects and resource costs are all included’’ and ‘‘effects and outcomes not related to health are included (if they are important for the public sector).’’ Therefore, NICE recommends that analysts include this information in addition to the information recommended for the NICE HTA References: case. As comparison of the two approaches shows, for the NICE HTA References: case the perspective on cost is fairly restrictive, including only NHS and PSSs costs. Also, the prescribed measure of health benefit is the QALY. Each QALY gained is assumed to have the same weight regardless of the other characteristics of the individuals receiving the health benefit (e.g., their age, socioeconomic status, or severity of their health condition). The NICE methods for developing public health guidance differ in that the perspective on costs is extended to encompass all costs falling on the public sector, recognizing the broader, cross-sector nature of most public health interventions. To facilitate comparability between NICE decisions, the QALY remains the primary measure of health outcome in the ‘References: case.’ However, for the public health References: case it can be supplemented by CCA and CBA approaches in order to take account of the broader aims and scope of public health interventions. This allows explicit consideration of multiple, nonhealth related and/or outcomes that are difficult to quantify. It also means that the impact of the interventions on the distribution in health gains can be evaluated to inform public health policy.
Methods Developments In The Economic Evaluation Of Public Health Interventions
Given increasing interest in the economic evaluation of public health interventions and current public health economic evaluation guidance, it is useful to review some of the ongoing methods requirements and developments in the area that might be used in future evaluations.
Attributing Outcomes To Interventions
In terms of primary data collection to assess the relative effectiveness of public health interventions, it is often feasible to undertake individual or cluster randomized RCTs and where possible this is recommended for measuring outcomes, although it is likely to be possible only over the short term. Where this is not feasible, nonrandomized trials may be undertaken and use of methods to restrict entry to the interventions based on those with particular characteristics or selecting controls that match the cases in terms of the confounding factor(s) may prove useful. As it is likely that outcomes of interest will extend beyond the length of trial follow-up, it is useful if outcomes measured match those available in longer term observational studies. Analytical techniques may be used to analyze nonexperimental data including econometrics. Economics has a long tradition of analyzing nonexperimental data for deriving effects attributable to a range of public health interventions. These include various matching techniques such as propensity scores, difference in differences techniques, time series analyses of natural experiments, and, where appropriate, more sophisticated econometric modeling and structural modeling. In addition, Bayesian methods may be useful in synthesizing data (modeling) including in examining the data where participants in studies do not match typical NHS patients, where intermediate outcomes are used, where relevant comparators have not been used, where long-term costs and benefits extend beyond the trial period and in quantifying the decision uncertainty and variability around the estimates. Further research might be undertaken to develop the methods for synthesizing all relevant data, experimental and nonexperimental and aggregate and individual-level data, for use in economic evaluations of public health interventions.
Measuring And Valuing Outcomes
Given that the aims and scope of public health interventions tend to be broader than for standard healthcare interventions, the measures of outcome chosen need to reflect this. As discussed above, the NICE References: case recommends CEA using QALYs as the primary form of analysis, with patients and/or carers as the source of data for measurement of health-related quality of life (HRQoL) and with values based on a representative sample of the views of the public. CCA and CBA are recommended as a secondary analysis to include other measures of outcome appropriate to decision making given the interventions evaluated.
Ongoing research includes development of sector-specific generic outcomes based on the QALY approach, for example, a social care QALY, development of a nonsector specific multidimensional measure such as a well-being index and a nonsector specific unidimensional measure such as happiness. The choice of outcome measure reflects the normative foundation underpinning the analysis and can also reflect the impact of the intervention on a particular sector or across multiple sectors of the economy.
Inter-Sectoral Costs And Consequences
Where the costs and consequences of public health interventions extend beyond the healthcare sector, the NICE HTA References: case methods would need to be extended to demonstrate this impact. The NICE public health References: case accounts for broader outcomes in the sense that the use of the cost–consequence or cost–benefit approach enables the analyst to describe other outcomes beyond the QALY and the healthcare sector. In terms of costs, besides NHS and PSSs costs, other public sector costs may be considered. Use of the cost– consequence approach could, however, mean that decision rules are not explicit as there are no standard decision rules using this approach and it is not clear the value decision makers would attach to different impacts in order to come to a decision about the cost–effectiveness of an intervention. The use of CBA may require a shift in the normative position to a more ‘welfarist’ perspective, though it may also be possible to monetize at least some nonhealth outcomes using methods other than estimating ‘raw’ WTP for those outcomes – for example, using ‘adjusted’ WTP estimates, sector-specific threshold values, and relative valuations of those outcomes in terms of other outcomes that are more readily monetized.
Research is ongoing to assess the practicalities of evaluating possible budgetary transfers across different sectors of the economy. In particular, an inter-sectoral compensation test approach to analyze the net benefit of costs, which fall on different sectors of the economy is being explored, and a stochastic mathematical programming approach is being developed to explore how to allocate resources in the context of different budgets and different budgetary policies across sectors.
Finally, research is also being undertaken to assess the use of a general equilibrium approach to simultaneously consider the impact of interventions across all sectors of the economy. The large majority of health economic evaluations undertaken to date take a partial equilibrium approach to analysis by assuming all other costs (and benefits) remain the same apart from those being evaluated. This is appropriate for evaluating the impact of most healthcare interventions. However, some health issues such as antimicrobial resistance and potentially pandemic diseases (e.g., seasonal flu, severe acute respiratory disease (SARS)) might have a macroeconomic impact that alters broader resource use and costs in the economy as a whole.
Incorporating Equity Considerations
The NICE public health References: case makes no explicit mention of equity considerations. However, the use of CCA, and assessment of subgroups in sensitivity analysis, assuming sufficient individual-level data on equity-related subgroups, may enable the analyst to include health equity issues. However, as a starting point, relevant health equity characteristics need to be identified and could include a whole range of possibilities such as socioeconomic status, degree of voluntariness or personal responsibility for health risk, and the value of treating current ill health versus preventing future health risk. If, following evaluation, the most cost–effective option is likely to be judged inequitable, either on the grounds of health inequality impact or procedural justice, it would be possible to assess the opportunity cost of not selecting that option, in terms of aggregate health gain forgone or additional resources used. Another approach that has been suggested is quantitative health impact assessment, allowing for health opportunity costs as well as health gains. Here, once a health inequality or a set of health inequalities have been determined, the distribution of net health impacts of the intervention is assessed by different equity-relevant groups. Building on this approach, it may be possible to assess the magnitude of any reduction in health inequality following adoption of the intervention and to clarify trade-offs with the objective of maximizing population health improvement.
The NICE References: cases state that an additional QALY is given the same weight regardless of other characteristics of individuals receiving health benefit. There has been some research into public and stakeholder views on equity weighting in a public health context and considerable additional research to overcome technical and practical issues is required to examine how much sacrifice in total population health is merited in order to pursue particular equity goals.
Summary
Economic evaluation provides a clear analytical framework for combining evidence and explicit social value judgments to inform decisions about how far investments of scarce resources in public health interventions are worthwhile. Given tight budgets and ballooning healthcare costs worldwide, policymakers are increasingly interested in ways of shifting the balance of effort toward preventative activity that has the potential to both improve health and reduce healthcare cost. So policymakers are likely to have an ongoing interest in assessing how far investments in public health interventions represent good value for money. Methods for the economic evaluation of healthcare interventions need to be adapted and refined in relation to the four methods challenges identified, in order to help analysts undertake good quality, relevant economic evaluations of public health interventions. Methods development in the field is still at an early stage and further research is required to improve the usefulness of methods and to pilot new methods with the aim of providing more useful information to support decisions about the investment of scarce resources into public health interventions.
References:
- National Institute for Health and Clinical Excellence (2008). Guide to the methods of technology appraisal. London: NICE. Available at: www.nice.org.uk/media/B52/ A7/TAMethodsGuideUpdatedJune2008.pdf (accessed 07.02.13).
- National Institute for Health and Clinical Excellence (2009). Methods for development of NICE public health guidance. 2nd ed. London: NICE. Available at: www.nice.org.uk/phmethods (accessed 07.02.13).
- Blundell, R. and Costa, D. M. (2000). Evaluation methods for non-experimental data. Fiscal Studies 21(4), 427–468.
- Claxton, K., Sculpher, M. and Culyer, A. (2007). Mark versus Luke? Appropriate methods for the evaluation of public health interventions. Centre for Health Economics Research Paper 31, University of York.
- Cookson, R., Drummond, M. and Weatherly, H. (2009). Explicit incorporation of equity considerations into economic evaluation of public health interventions. Health Economics Policy and Law 4, 231–245.
- Haynes, L., Service, O., Goldacre, B. and Torgerson, D. (2012). Test, learn, adapt: Developing public policy with randomised controlled trials. Cabinet Office, Behavioural Insights Team. Available at: SSRN: ssrn.com/abstract=2131581 or dx.doi.org/10.2139/ssrn.2131581 (accessed 08.05.13).
- McKenna, C., Chalabi, Z., Epstein, D. and Claxton, K. (2010). Budgetary policies and available actions: A generalisation of decision rules for allocation and research decisions. Journal of Health Economics 29(1), 170–181.
- Medical Research Council (2008). Developing and evaluating complex interventions: New guidance. London: UK Medical Resource Council. Available at: www.mrc.ac.uk/complexinterventionsguidance (accessed 08.01.13).
- Medical Research Council (2012). Using natural experiments to evaluate population health interventions: Guidance for producers and users of evidence. London: UK
- Medical Research Council. Available at: www.mrc.ac.uk/naturalexperimentsguidance (accessed 08.01.13).
- Smith, R., Yago, M., Millar, M. and Coast, J. (2005). Assessing the macroeconomic impact of a healthcare problem: The application of computable general equilibrium analysis to antimicrobial resistance. Journal of Health Economics 24, 1055–1075.
- Vining, A. and Weimer, D. L. (2010). An assessment of important issues concerning the application of benefit–cost analysis to social policy. Journal of Benefit–Cost Analysis 1(1), Article 6. Available at: www.bepress.com/jbca/vol1/ iss1/ (accessed 08.01.13).
- Wailoo, A., Tsuchiya, A. and McCabe, C. (2009). Weighting must wait: Incorporating equity concerns into cost effectiveness analysis may take longer than expected. Pharmacoeconomics 27, 983–989.
- Wanless, D. (2004). Securing good health for the whole population. Final Report. London: HM Treasury.
- Weatherly, H., Drummond, M., Claxton, K., et al. (2009). Methods for assessing the cost–effectiveness of public health interventions: Key challenges and recommendations. Health Policy 93, 85–92.