Keeping Pace with Assessing Cost-Effectiveness: Economic Efficiency and Priority-Setting in Mental Health

The last two decades have seen an increasing interest in, and demand for, economic analysis of mental health care and policy, fuelled to a large extent by government concerns in high-income countries about escalating total health-care expenditures. Considerations of cost and costeffectiveness have legitimately appeared on the health agenda in a number of different guises, including regulatory decisions relating to drug approval or pricing, priority-setting exercises across as well as within health programs and broader health-care reform processes. In short, the application of economic analysis and evaluation to mental health care has come a long way. Over the last 5 years or so and nowhere is this more apparent than in Australia, where a concentrated period of research effort – by the Assessing Cost-effectiveness (ACE) project and its forebears – has produced what can already be claimed to represent the most robust economic evidence base worldwide underpinning national mental health policy and resource (re)allocation.

Remarkable as it may sound, no country to date has been able to clearly link mental health strategic policy or investment decisions to a credible, consistent and evidence-based assessment of what interventions actually work best and at what cost. That has now changed with this accumulation of cost and effectiveness data across the spectrum of psychiatric disorders, from attention deficit hyperactivity disorder (ADHD) and depression in children and adolescents [1], [2], through common mental disorders in adulthood [3], [4], to more severe mental illnesses such as schizophrenia [5], [6]. The remaining challenge in ACE's home country, needless to say, is how to translate the accumulated evidence into policy and the policy into practice [7], while for other countries in the region and elsewhere the gauntlet is whether they can pick up the pace in order to catch up with ACE.

Mental health economics is a relatively recent, and to some an unseemly, addition to ways of thinking about the prevalence, treatment or outcome of psychiatric disorders (see [8] for an overview of its role in mental health). In an (imaginary) world with unlimited resources available to meet identified mental health needs, there would indeed be little place for analysis of by-now familiar issues in the health lexicon – efficiency, cost containment and cost shifting, reimbursement rates and so on. The reality, of course, is that resources are not endlessly on tap, meaning that rationing is inevitable (despite the protestations of some politicians) and rather difficult allocation choices have to be made. Such choices may be informed by an evaluation of what intervention represents the most cost-effective strategy to reduce the burden of a particular mental disorder such as schizophrenia (technical efficiency analysis), or a broader assessment of the relative costs and consequences of investment into mental health care vis-a-vis alternative uses of public resources within or beyond the health sector (allocative efficiency analysis). At its best, such analyses can provide objective information to policy makers about how to optimize the return on a society's resources, including good and bad buys in health. Many methodological as well as empirical limitations stand in the way of this goal, however, including non-standardized measurement of costs and benefits, divergent analytical perspectives and incomplete coverage of key policy areas.

The question is, how does ACE fare in this regard? Put more specifically, do the results of these analyses provide a valid and reliable mechanism for (re)allocation of mental health resources at the national level that all relevant stakeholders can sign up to? Before addressing this question, it is worth dwelling for a moment on the antecedents to the application of the ACE approach to mental health care, in particular work relating to disease burden, which for the first time offered a consistent basis for ascertaining the epidemiological profile of mental disorders and the burden it produces relative to other causes of ill-health [9]. From the resource allocation point of view, a tempting subsequent question to ask is how much of the burden attributed to a particular disease is actually avertable, and at what cost, either by current or scaled-up implementation of effective intervention strategies [10], [11]. Importantly, analyses of both these concepts, the attributable and avertable burden of disease, rely on the disability adjusted life year (DALY) as a summary measure of population health. Important because there are a number of specific challenges associated with using DALYs as the primary measure of health outcome in the economic evaluation of mental health care, as discovered at some cost by the ACE team as well as by others attempting to use this metric.

With respect to mental health care, a key problem is that local or regional data on the effects ofmental health treatment have to be squeezed into a summary measure called years lived with a disability (YLD), the basic formula for which is incidence multiplied by average duration (in years) and disability level (measured on a 0–1 scale, where 0=death, 1=full health). Unfortunately, there remains a notable shortage of data from published trials of mental health interventions upon which to compute expected levels of disability before or after treatment, resulting in a reliance on symptom or quality of life (QoL) measures instead. Then there is the issue of how to map these symptom or QoL scores onto a 0–1 disability index. ACE-Mental Health uses a two-stage procedure, first extracting (mainly symptom) effect sizes from the literature and then either applying a ‘conversion factor’ to estimate the equivalent impact of this effect size on a 0–1 disability index or via the ‘survey severity’ method, using existing disability weight values for different categories of illness severity to derive population-level weighted averages with or without treatment [12]. The ingenuity behind this approach, however, should neither disguise the many assumptions that underpin this crucial component of the final analysis, nor the consequent need to apply the approach to other populations. In particular, and despite the standardized approach used, there remains some variation with respect to the robustness and generalizability of efficacy and cost findings across the ACE-Mental Health papers [6], [13], which to the purist would suggest that comparisons of allocative efficiency between the various mental disorders might really be a bridge too far.

Beyond technical issues underlying the DALY and its use in economic evaluation, what about ACE's engagement with other, more pragmatic considerations? Here, a set of ‘second-stage’ criteria – strength of evidence, equity, feasibility and acceptability – are defined and used to qualify the preceding efficiency findings in a qualitative way, for example, stating that while cognitive behavioural therapy (CBT) for depression may be a good thing from a cost-effective point of view, there are obvious challenges in making this widely available in the population [4]. Few would argue with the inclusion of these criteria, although many would perhaps want to add others, while still others would surely take issue with seeing these other criteria considered as somehow secondary to efficiency! If I were a family member and/or informal carer of an individual with a psychotic illness, for example, I might not give a fig about cost-utility results, but would push hard for more explicit recognition of carer or family burden, user rehabilitation and disability rights, which are not so well catered for by the selected ACE criteria. This brings us into the complex arena of priority-setting in health, including whose values or preferences should be taken into account when determining and measuring these criteria. Clearly, a steering committee such as that used in ACE cannot represent all possible viewpoints, and indeed did not include a service user, but on the positive side encourages engagement with other key stakeholders from the outset.

Perhaps wisely, no attempt was made to integrate these (often competing) criteria into a composite prioritization score, but that means there is no sense of the relative importance or weight associated with these diverse considerations. To illustrate how this can be done, overall program benefits in an earlier mental health prioritysetting exercise carried out in South Australia [14] were couched quantitatively in terms of individual health gain (contributing an importance weight of 28%), community health gain (27%) and equity (45%). From the rational, explicit perspective on priority-setting, such a uniformly quantitative approach has some appeal, although others argue that this may just be adding uncertainty on top of uncertainty.

In summary, ACE-Mental Health breaks new ground in its standardized approach to and broad sweep across the comparative efficiency ofmental health interventions. Although there remain a number of limitations and challenges for improved analyses, clear policy recommendations can be elicited. In this regard, some of the more striking findings from this work include the much more favourable ratio of cost to effect for neuroleptics compared to atypical antipsychotics (and to a lesser extent, tricyclic antidepressants [TCAs] vs selective serotonin reuptake inhibitors [SSRIs]), the economic worth of psychosocial approaches to psychiatric disorders (whether alone or in combination with pharmacological treatment) and the strong economic efficiency argument in favour of long-term or maintenance strategies for depression [15]. It is of some note that, quite independent of ACE although similar in overall method, exactly the same findings emerged from a more global analysis of these interventions as part of the WHO's sectoral costeffectiveness analysis CHOICE project [11], [16]. The conundrum for high-income countries like Australia and New Zealand is that a number of these findings fly in the face of prevailing clinical practice, leaving a (largely political) choice to either maintain the status quo and accept the inefficiency of currently preferred (or industryinduced) treatment strategies, or orchestrate whole-scale change in the mental health system in the interests of increased efficiency. Not an easy one.