Why should economists work in the field of obesity prevention? What wisdom can they bring to bear on the issue that could make a difference to the prevalence and consequences of obesity? This chapter explores the role and content of economic analysis, summarizes the contribution which the discipline currently makes, and considers the contribution it could make to the fight against obesity. It draws on current research activity in the field to illustrate some of the specific methodological challenges that confront health economists working in the obesity field, and discusses the directions which economic research needs to take if the discipline is to make a positive and effective contribution to the search for solutions to the obesity crisis.

To assess the credentials of health economics to make a contribution, one must first appreciate the nature of economics and the roles which it performs. The fundamental problem addressed by economists is the allocation of scarce resources between competing demands—that is, how to maximize community welfare with available resources. In addressing resource scarcity, economists carry out four separate but interrelated tasks: “description”, “prediction”, “explanation” and “evaluation”.¹

Economics enables us to “describe” current activities, health status and resource use, and to “predict” future trends in the same. To date, most of health economics’ contribution to obesity prevention has centered on these two aspects. The past decade has seen a proliferation of descriptive studies which either documented the size of the disease and cost burden attributable to obesity in particular demographic groups or geographical jurisdictions, or made forecasts about the growth in the obesity problem. Whilst several writers in the 1990s^2,3 noted that such “cost of illness” (COI) studies were not particularly common in the area of obesity, more recently, authors^1,4 have referred to the growing literature documenting the economic costs of obesity. Since our earlier documentation of such COI studies of obesity,¹ there have been more studies, including for non-Western jurisdictions.^5,6

Whilst such COI studies vary in their methodology, they generally measure costs within a prevalence-based framework.³ The annual cost burden stemming from all cases of obesity-related disease (new or pre-existing) is measured, as the study purpose is generally to inform cost control or financial planning. This contrasts with incidence-based studies,⁷ which measure the lifetime costs associated with new cases only, as a baseline against which new measures can be assessed. Most studies confine the cost burden to direct health sector costs arising from the current prevalence and treatment of obesity,⁸ with few taking into consideration indirect costs arising from lost productivity⁹ or diminished social functioning and quality of life.¹⁰ The lack of consensus about obesity-related illnesses is evidenced by differences between studies in terms of the range of co-morbidities included. Studies vary in terms of the BMI cut-off points used to define obesity, as well as the perspective employed from which to measure costs. Most studies assume a national health system perspective, while some assume a narrower reference frame in terms of geographical jurisdiction¹¹ or target group.¹²

However, regardless of their choice of methods, these COI studies comprise essentially “descriptive” research focusing on the size of the obesity problem. They quantify the magnitude of the issue and estimate the associated disease burden in monetary terms. Such studies are premised on the basis that knowledge of the costs stemming from an illness will be important in informing decision making around resource allocation. They are considered a valuable tool when advocating for the deployment of additional resources towards obesity prevention, and have been employed by agencies such as the World Bank and the World Health Organization.

However, COI studies have also been the centre of active debate among economists.^13–15 While acknowledging that they may serve three purposes (to justify budgets, to help set funding priorities and to develop intervention programs), Rice¹³ argues that the methods need to be sufficiently detailed to permit transparency and to enable the reader to assess whether the results are “fact or fiction”. Byford et al¹⁴ pose three key arguments against the conduct and use of COI studies: first, high costs do not necessarily indicate inefficiency and waste; second, the supposed “cost savings” of either fully or partially preventing a disease are likely to be overstated and partly illusory; and third, the condition may not necessarily be amenable to treatment.

More recently, there has been similar questioning about the value of such studies among economists working in the obesity field.^4,16 Roux and Donaldson⁴ are highly critical of the economic credentials of such studies and conclude that they add little to the obesity debate, apart from confirming that obesity is a serious societal issue. In an earlier publication,¹ we took a more positive yet cautious approach to COI studies. While acknowledging that descriptive cost estimates can be of value to planners, we also stressed that COI estimates should not be overinterpreted. More importantly, used sensibly and carefully, COI estimates could also have a role beyond simple description and monitoring, as an input into evaluation studies and broad-based priority setting exercises.

The third task of health economics, explanation of obesity, is a relatively new and underdeveloped field. Rosin¹⁷ recently surveyed the growing economic literature on the causes of obesity epidemic, and concluded that the key economic influences on obesity prevalence are food prices, working mothers, urbanization and technological change.

Irrespective of their views about the potential contribution of COI and causal research, most economists would agree that the fight against obesity requires a “solutions-based” rather than a “problem-focused” approach, and that it is the fourth plank of economics, “economic evaluation”, which offers the potentially most valuable contribution. High quality evaluations of potential obesity interventions are required so that policy-makers know “what works” and what offers “value for money”.

A full economic evaluation of a selected intervention is characterized by the incremental assessment of both its costs and benefits measured against a comparator, usually current practice.¹⁸ This enables the analyst to answer the essential policy question of “what difference the intervention is likely to make to the disease burden and what is the net cost of doing so”. The change in costs is compared with the change in outcomes and reported as an incremental cost–effectiveness ratio (ICER). Typically, with obesity interventions, the ICER will be reported as “net cost per kg of weight lost”, “net cost per BMI (body mass index) unit saved”, “net cost per life-year gained” or “net cost per quality-adjusted life-year saved (QALY).

To date, very few obesity interventions have been subjected to rigorous economic evaluation. It has primarily been treatment options involving either surgical or pharmacological therapies that have been economically evaluated.^1,19 These two publications between them identified only one economic evaluation of a preventive intervention²⁰ and seven of lifestyle treatment interventions involving diet, exercise or behaviour therapy. Furthermore, some of these were not targeted exclusively at obese persons, but at persons for whom obesity was a serious complication or subsequent disease. In a recent discussion of the literature, Cawley²¹ listed four published cost–effectiveness studies of anti-obesity interventions,^20,22–24 of which only one was a preventive measure.

The lack of economic evaluations of potential preventive interventions for obesity has been noted in the literature.^4,21 With the growing profile of obesity as a major public health issue, governments around the world are allocating funds towards obesity programs, and public health organizations, local governments and schools, and so on, are searching for solutions to the problem. However, such efforts are severely limited by the lack of information about what works, and more specifically, how to achieve the greatest “bang for the buck”.²¹ Cost–effectiveness analysis is a tool that can help answer this question as it facilitates intervention assessment in terms of their net costs per unit of benefit.

However, before embarking on economic evaluation, there first needs to be credible evidence that an intervention actually works (or at the very least, strong program logic built on accepted theoretical foundations that facilitate defensible assumptions about effectiveness). The evidence base around the effectiveness of programs and policies remains limited. Few obesity interventions have been rigorously evaluated, and examples of funding being directed towards measures for which there is no evidence are commonplace. The need for quality evaluations of obesity prevention initiatives has been a recent topic in the literature.³¹

Sufficiency of evidence of effectiveness was a key issue in the Australian ACE-Obesity study. Unlike previous ACE studies in other disease areas where there was sufficient evidence of effectiveness based on high quality study designs (such as randomized controlled trials, cohort studies, case-control studies), the paucity of trial-based evidence around obesity interventions made it necessary to move from a traditional evidence classification based on epidemiologic study design to a new classification which incorporated other forms of evidence not usually captured, such as “parallel evidence”, “indirect evidence” and “epidemiological modeling”.

The ACE-Obesity study was criticized for its optimistic assumption about the maintenance of effect over time.³² The study acknowledged that full maintenance of benefit was highly improbable, and aimed to be transparent in its conclusion by undertaking “threshold analysis” to determine the extent to which the 100% maintenance of benefit could be reduced before an intervention would cease to be cost-effective.²⁵ This assumption and the associated sensitivity analysis was a direct product of the absence of available data on which to model an alternative trajectory and the exploratory nature of the ACE-Obesity work.

But the paucity of interventions which have sufficient demonstrated evidence of effectiveness is not the only challenge to achieving quality economic evaluation. Cawley²¹ raises additional issues in conducting cost–effectiveness analyses on obesity prevention interventions—flaws in using BMI to define different health states; the lack of consensus around the QALYs associated with such health states; the need for QALYs to vary by demographic variables, and for longer follow-up studies to ascertain the persistence of QALY savings over time. Cawley strongly advocates for the role of cost–effectiveness analyses in informing policy-makers and public health practitioners, and ponders the potential dangers of misallocated resources and unintended consequences which may arise from rushing ahead with interventions before establishing their effectiveness and cost–effectiveness credentials.

Another important challenge is that the type of interventions that offer the best prospects for preventing obesity are most likely to be in areas that lend themselves less readily to conventional economic evaluation methods. Effective health promotion interventions are often complex and multi-faceted, community-based rather than individual-based and inter-sectoral rather than restricted to the health sector. Further, they are likely to include policy initiatives, such as changes in taxes, subsidies or regulations, which offer their own unique difficulties for economic appraisal and establishing the evidence base. Such complex interventions not only create difficulties in outcome measurement, but also in attributing effects to different elements of the intervention and in apportioning costs, particularly where the intervention is of an organic community-based nature.

Richardson³³ addressed this issue when he proposed a four-fold classification of possible outcomes of health promotion based on a distinction between “disease cure”, “individual health promotion”, “community welfare” and “systemic change”. While arguing that there is justification in economic theory for including all of the benefits of a health promotion program in an economic evaluation, he acknowledges that some of the current analytical tools may be of little practical use. Besides the issue of inadequate information about outcomes, the specific problem with health promotion programs is to achieve a balance between the demand for short-run accountability and the need for programs to reach sufficient maturity to achieve long-run objectives. Potentially beneficial projects may be jeopardized by premature evaluation.

Community-based interventions pose particular challenges for economic appraisal. The study design is generally of a quasi-experimental nature and interventions may not be standardized, but more of an organic nature varying and evolving according to the needs of particular communities. While challenging to evaluate, such programs are exactly the type of intervention for which research data about cost–effectiveness is currently lacking and urgently needed. Such an example is the Pacific OPIC Project (Obesity Prevention in Communities), a large multi-country project targeting adolescent obesity in Australia, New Zealand, Fiji and Tonga.³⁴ It involves pre and post measurement of 15,000 adolescents in intervention and control sites, and the implementation of interventions in schools, churches and villages. In such a big, complex project involving large numbers of organizational entities across four countries, it was vital that the economic evaluators were included from the initial design stage, and that the economic evaluation was adequately funded and underpinned by a clear, detailed protocol. Data collection needed to be kept manageable and tractable, given the three-year intervention timeframe, and the competing demands on the project staff and others such as teachers involved in the interventions. In large community-based interventions, the issue of measurement of “current practice” can be problematic. It is difficult to recruit schools, communities as control groups, with the expectation that they meet the same anthropometric measurement and data collection requirements, but without any investment of intervention activities.

In the ACE-Obesity project,²⁵ threshold analysis was used to provide additional information to policy-makers and to public health practitioners. The interventions evaluated included several existing programs, being implemented at either a national or state level, which were shown to be cost- ineffective. Scenario analyses were used to show how such programs could realistically be made more cost-effective through cost-cutting measures, justifiable apportionment of some costs to non-obesity objectives, or measures designed to increase recruitment and participation.^29,30

Another contentious and problematic issue is the economist’s practice of discounting future benefits and costs. For most obesity prevention programs, the discounting of the value of benefits and costs occurring in the future is likely to make the “present value” of benefits obtained after 15–20 years considerably smaller. Another issue is the potential inequity of discounting the benefits of children at the rate of time preference of adults. Richardson³³ puts forward theoretically plausible arguments for employing lower than normal discount rates for distant health benefits. It is for this reason, among others, that quality economic appraisals will report their cost–effectiveness results for a range of discount rates (e.g. 0%, 3%, 5% and 7%).

Whilst the previous section discussed some of the challenges facing economists when evaluating obesity interventions, the cost–effectiveness analyses of single, stand-alone interventions are unlikely to be enough to make an effective contribution to policy decisions about strategic directions. A whole range of obesity interventions across diverse settings need to be evaluated, and given resource constraints and multiple policy objectives, decision makers need to have mechanisms for combining them into effective prevention strategies for achieving healthy weight.