EDITORIAL The inevitable pursuit of efficiency

Why health economics?

We cannot escape the fact that the resources available for health care are scarce relative to needs and demands, necessitating tough decisions about how to allocate them across people, disease areas, treatments and a whole host of other potentially competing factors. In the context of such resource constraints, some form of priority-setting is inevitable to ensure the most optimal mix of healthcare provision. This is an enormous and complex task and a range of criteria can guide such decisions, for example, the pursuit of equity (in the distribution of health and the distribution of resources) and consideration of patient and public preferences. One important criterion is cost-effectiveness or efficiency. Concerns about the relevance of economics to health care can often stem from a misconception that it concerns a focus on costs alone, hence distaste for terms such as ‘rationing’. However, pursuing efficiency is really about ensuring value for money and it must necessarily consider goals of the healthcare system, whether they be the maximisation of health and well-being or other societal values. This special issue is dedicated to the application of economics to the area of mental health care. It illustrates some of the particular issues that apply to this area and also the distance we still need to go to measure and account for efficiency reliably and acceptably.

Priority-setting

Explicitly examining the relative efficiency of alternative treatment options is a somewhat recent preoccupation across the world and has taken many forms with varying degrees of success or acceptance (Sabik & Lie, 2008). A current notable example is the National Institute for Health and Clinical Excellence (NICE) in England and Wales, which produces guidelines for the management of specific health conditions and assesses the cost-effectiveness of specific treatments. Another early initiated and well-known example is the priority-setting system implemented in Oregon in the USA. To determine the best use of its limited Medicaid budget and to try to extend health insurance coverage to a larger proportion of its population, Oregon created an explicit ranking of condition–treatment pairs based on cost-effectiveness (according to the Quality of Well-Being Scale). The approach proved unpopular both in terms of the methodology used to generate the list and the treatment decisions it implied. While it remains in place, there is now more emphasis on broader criteria alongside ongoing methodological refinement. In this issue, Mihalopoulos et al. (2013, this issue) usefully describe some key healthcare prioritisation attempts specifically in the area of mental health care. Also, Cruz et al. (2013, this issue) document Brazil's evolving priority-setting process. While most economic evidence is generated in high-income countries, issues of resource allocation are arguably even more important in low- and middle-income countries, where there is often less resourcing for health care, significant socio-economic inequality and larger gaps between the number of people with mental health problems and the number receiving treatment for such problems. A decade ago, Brazil embarked on a process of defining its prioritisation criteria, developing methodology for assessing healthcare interventions, commissioning new research and producing mechanisms for dissemination of their assessment findings. There is of course still a long way to go: much of their cost-effectiveness evidence continues to be generated through economic modelling based on international evidence, rather than through primary research, but there is nevertheless an explicit consideration of efficiency; and whilst there have been few evaluations in the mental health area in the past, the government has recently encouraged and committed funding to more studies in this area.

Assessing outcomes, costs and cost-effectiveness – a long way to go

A key criticism of Oregon's early attempts at incorporating efficiency considerations into priority-setting centred on the ability of the Quality of Well-Being Scale to capture public preferences (Nord, 1993). This clearly illustrated that the palatability of efficiency considerations is very much dependent upon the robustness through which efficiency is assessed and the extent to which broader contextual factors are considered. The context issue is particularly relevant to the area of mental health because it carries particular challenges for assessing both outcomes and costs.

Aside from logistical issues such as difficulties in recruitment and retention to research due to the physical, psychological and social impacts of mental health problems, there are more conceptual hurdles such as the appropriateness of focusing on a single-outcome measure given the vast array of effects attributable to mental health conditions. NICE currently focuses on a single-outcome measure for considering the cost-effectiveness of drugs across all clinical areas (it takes a broader assessment of outcomes for other assessments, e.g. medical technologies): quality-adjusted life years (QALYs) (Williams, 1985) which combine two important outcome dimensions, length and quality of life. QALYs are useful for broader levels of decision-making because they (potentially) allow comparisons between different disease areas. However, there has been continued concern about the appropriateness of using generic health state description systems in the mental health area (Brazier, 2010; Chisholm et al., 1997) due to the construct of the instruments themselves and the challenges in obtaining credible data from people with cognitive impairment and communication difficulties (Awad et al., 1997; Neumann, 1999). It is thus unsurprising that such concerns have spawned a welcome spate of investigations of the validity of such measures in particular patient groups. In this issue, Byford (2013, this issue) explores the relevance of a commonly used (but also heavily criticised) health-related quality of life measure, the EQ-5D, among adolescents with major persistent depression. This was done through an examination of construct validity (by exploring the ability of the EQ-5D to discriminate between characteristics such as age, gender and severity of illness), convergent validity (correlation of the EQ-5D with other outcome measures) and responsiveness (the ability of the EQ-5D to detect changes over time). The findings generally support the use of the measure in this patient group (wider contextual issues, such as culture, aside) but with a caveat that it may be more appropriate where the focus in on average, rather than individual-level effects.

On the cost side, Bendeck et al. (2013, this issue) highlight the lack of consensus on methods to estimate the costs of an illness and the effect of this on the usefulness of such estimates. A similar point is also raised by Mihalopoulos et al. (2013, this issue), who call for a focus on true differences in efficiency, rather than differences in methodology. While pharmaceutical treatments are commonplace in mental health care, care is still very much characterised by a range of non-pharmacological treatments which are complex to deliver, access and evaluate. An array of agencies and services (including outside of the healthcare sector) are involved in caring for people with mental health problems (hence, for example, the recent focus on developing effective employment support interventions; see Heffernan & Pilkington, 2011). Furthermore, inputs can continue over long periods of time and high rates of physical and psychiatric comorbidities create complex patterns of service use. Measuring costs comprehensively to enable decision-making that is relevant to all stakeholders is challenging in terms of data collection and for interpretation as it may involve trading off costs and impacts across different stakeholders. ‘Cost-of-illness’ or ‘burden of illness’ studies commonly attempt to describe the full range of costs associated with a health condition. Although they do not actually assess efficiency because they do not consider the resulting outcomes, there is a role for such studies in resource allocation decisions. They transform a diverse range of economic impacts into a single, cost-based measure of the consequences of a condition which can in turn indicate the relative magnitude of the burden of one condition compared with another. Also, whether we like it or not, such studies are more likely to draw widespread attention to a condition compared with more formal economic evaluations (except where the latter are implicated in unpopular policy decisions about whether a treatment will be provided via the public purse). However, methodological variations in which prevalence/incidence rates are used, the range of services considered, the nature of unit costs and inclusion/exclusion criteria can limit the comparability of one cost of illness study against another (Smith & Wright, 1996). Bendeck et al. (2013, this issue) demonstrate an approach for producing cost of illness estimates which importantly represents such variations.

More broadly, it is clear that while methodological approaches to economic evaluation are broadly agreed upon and standardised, with widely accepted guidelines for conduct and reporting, some methodological aspects continue to evolve and be debated. An example of this is provided by Sabes-Figuera et al. (2013, this issue), who re-visit an evaluation they previously reported on the cost-effectiveness of selective serotonin reuptake inhibitors (SSRIs) plus supportive care compared with supportive care alone for people with depression. This had employed conventional approaches to economic evaluation but they now sought to (a) identify relevant co-variates to include in the estimation of cost-effectiveness and (b) explore whether the findings hold ground in particular sub-groups within their study sample. Their findings are interesting – the probability that SSRIs plus supported care is a cost-effective approach is reduced by the inclusion of co-variates and is higher among those without a previous history of depression than for those with – and important because they highlight the risk of erroneous decisions if nuances within broad methodological frameworks are ignored.

The importance of stakeholder involvement

Both Mihalopoulos et al. (2013, this issue) and Bendeck et al. (2013, this issue) highlight the importance of involving stakeholders in research. Prioritisation, design, conduct and dissemination of academic-led research are influenced by agendas which may or may not be directly relevant for policy-making. Failing to involve relevant stakeholders and account for the agendas important to them can thus limit the usefulness of such research for real-world decisions. Bendeck et al.'s (2013, this issue) findings have now been translated into regional priority-setting decisions and this is likely a result of a formal cooperation plan with relevant public agencies throughout the study process. Importantly, they also utilised a range of dissemination approaches, from leaflets through to technical reports, and utilised existing templates to facilitate comparability with other international estimates. This represents an important conceptual shift away from cost of illness studies being attention-seeking tools towards a more serious role in healthcare planning.

Conclusion

No matter how good the economic times are, there will never be enough resourcing, monetary or other, to meet all of modern society's needs and wants. The need for appropriate allocation comes into sharper focus in economically lean times such as now and it is important to remind ourselves that investing in mental health, and ensuring this is done cost-effectively, takes on greater, not less, importance (Knapp, 2012).

Declaration of Interest: The authors report no conflict of interest.