Cognitive decline among adults aged 65 years and older is a substantial public health problem in terms of incidence, health burden to the individual and care givers, as well as healthcare-related costs.1 With the world's ageing population increasing, the number of older adults with dementia is estimated to rise from 26.6 million in 2007 to 106.2 million in 2050.2

The economic burden of cognitive impairment and dementia cannot be ignored. In 2000, dementia was the third most costly health condition to care for in the USA, with annual costs estimated at $100 billion (in 1997 US prices). Another study calculated that the mean annual cost for dementia care is €28 000 per patient.3 The direct costs in the UK of Alzheimer's disease were estimated at £23 billion annually.4 Clearly, the cost of care for dementia is extremely high, and any strategies that delay the onset and/or slow the progression of cognitive decline and dementia can have enormous societal return in terms of costs and consequences. To date, effective pharmacotherapy for cognitive decline remains a challenge.1 Rather, recent evidence emphasises the importance of behavioural strategies such as physical activity to promote cognitive function.5,–,10 Specifically, results from randomised controlled trials suggest that exercise has benefits for cognitive function among cognitively normal older adults5,–,7 and among older adults with mild cognitive impairment.8 9 However, more research is needed to ascertain the direct effect of exercise on cognition among those with dementia, such as Alzheimer's disease and vascular dementia. Nevertheless, as previously highlighted by Erickson and Kramer,10 physical activity provides clear benefits for cognition among seniors. These neuroscientists contend that ‘physical activity is an inexpensive treatment that could have substantial preventive and restorative properties for cognitive and brain function.’11

However, to emphasise the value in terms of costs and consequences of physical activity interventions for promoting cognitive function among older adults—such that these research findings are meaningful to government and policy makers—health economic evaluations must be incorporated in future randomised controlled trials in this field.

To date, the importance of economic evaluation applied to the field of exercise and cognitive function has largely been overlooked. Yet, under our current state of limited healthcare resources, healthcare decisions should ideally be based on efficiency. Efficiency—the cornerstone of economic theory—is a ‘guiding principle’ for decision makers as they decide among competing alternatives against a background of healthcare resource scarcity.12 Therefore, studies comparing new interventions that combat cognitive decline with existing treatments should include essential economic data in order to facilitate efficient medical decision-making.

We note that one may assume that because exercise interventions are generally more cost-effective than pharmacological treatment, all economic evaluations of effective exercise interventions are unnecessary. However, we highlight that, to date, this is only an assumption, and research studies are needed to validate this assumption. Furthermore, even if the effect size of targeted exercise training on cognitive function is smaller than pharmacological treatment, exercise as an intervention strategy may still provide good value for money because of its lower cost. Further, exercise is also likely to provide good value for money because of its established broad benefits—such as promoting cardiovascular health and optimal musculo-skeletal function.14 15

In healthcare, cost-effectiveness analysis (CEA) has emerged as one of the favoured techniques for economic evaluation. In CEA, health outcomes are quantified in terms of health benefits (ie, number of life years saved). ‘CEAs show the relationship between the net resources used (costs) and the net health benefits achieved (effects) for a specific intervention compared with a specific alternative strategy.’13 The primary outcome of a CEA is the incremental cost-effectiveness ratio (ICER). By definition, an ICER is the difference in mean costs required by the intervention compared with an alternative (eg, usual care or a ‘do nothing’ alternative) divided by the difference in mean health benefit gained from the intervention compared with an alternative.14 One major limitation of CEAs is that the units used to express the health benefits may limit comparability across disease states if an outcome specific to the intervention or disease treatment is used.13 In an attempt to mitigate this problem, CEAs can be standardised by reporting health effects of interventions by life years gained. Although this approach sounds appealing, the endpoints of clinical trials are often shorter than what would be needed for the economic evaluation. However, not all interventions have an impact on mortality, so in some cases, the endpoint of the clinical trial may be appropriate. This limitation of the CEA led to the widely recommended approach that incorporates both quantity and quality of gains or losses—the quality-adjusted life year (QALY) is an example of one such measure used in cost–utility analysis (CUA) that incorporates both quality and quantity of life.13 The prominent difference between CEA and CUA is that in CEA, the incremental cost of a programme is compared with the incremental effects of a programme where the health effects are measured in naturalistic units (ie, life years gained). In CUA, the health effects are measured in QALYs gained, a measure that includes QALYs gained.

Based on current evidence, we already know that cognitive impairment and dementia are costly to care for. However, to our knowledge there is only one published economic evaluation investigating a primary prevention strategy of resistance training on executive functions in community-dwelling older adults.15 This could be due in part to the methodological challenges of conducting economic evaluations alongside trials of individuals with dementia or mild cognitive impairment.

As we consider the methodological challenges of conducting economic evaluations alongside trials among older adults where the primary outcome is cognitive function, we highlight three key items that warrant consideration.

First, the choice of outcome measure is of particular importance for CEAs of interventions aimed at promoting cognitive function. A key limiting factor for conducting CEAs in this area is the lack of tangible units to measure cognitive function. Often for CEAs, the mean change in a tangible unit is measured, such as the drop in mean systolic blood pressure, the mean number of falls prevented or the mean number of asthma exacerbations avoided. These are all examples of tangible units that have clinically relevant meaning. The challenge with CEAs for cognitive function is that no clinically relevant tangible units have yet been defined. For example, what does a 2 second improvement on the Stroop test mean clinically? This gap in the literature highlights the need for large prospective population-based studies that determine what change on measures of cognition translates into a minimally clinically important difference (MCID). To our knowledge, there has only been one economic evaluation of exercise as primary prevention for cognitive decline.15 In this study, falls were used as the ‘effectiveness’ outcome for the economic evaluation because falls are tangible units that are clinically significant consequences of cognitive decline among senior.19 20 Further, the costs of falls are more comprehensively documented in other studies compared with cognition-related costs.20 21 One notable limitation of using falls as tangible units is that a decrease in the number of falls may be due to other factors aside from cognition that are affected by exercise, so we suggest both adjusted and unadjusted analyses be presented. One alternative way of establishing tangible units in trials aimed at promoting cognitive function is to establish a set of patient-oriented goals that would make a difference to both the individual and the carers (ie, better sleep, less wandering or fewer repetitive questions). The limitation of this approach is that these items will be more difficult to measure and quantify.

Second, the validity of participant responses among those with objective cognitive decline is of great concern. Previous studies have demonstrated that individuals who suffer the most severe cognitive impairment tend to rate their health-related quality of life (HRQL) as higher than those who are at the onset of cognitive decline.16 To our knowledge, there are no studies investigating the underlying reasons for these differences in HRQL. Rather, research has focused on the use of proxies (ie, clinicians or care givers) to assess older adults' HRQL. Thus, careful consideration needs to be given to both the validity and the mode for the collection of outcomes. Further, carers are important not only as proxies but also as beneficiaries of any improvement the individual they are caring for gains from exercise. Thus, we recommend measurement of care giver QALYs.

Third, the use of falls as an outcome measure for economic evaluations highlights other important questions to consider. First, is it appropriate to use falls as an outcome measure when the clinical trial was powered for cognitive function? Second, would it be more appropriate to consider QALYs for the economic evaluation? If so, should all trials be powered based on QALYs gained? Third, does the primary outcome measure of the clinical trial need to match the primary outcome measure of effectiveness for the economic evaluation?