Economic evaluations of complex services for older people
At its simplest, health economics compares the costs and benefits of an intervention. A specific treatment is said to be ‘cost-effective’ if it gives greater health gain than could be achieved by using the resources in other ways. An intervention may be better and cheaper (economically dominant), better but at a cost, or cheaper with worse outcomes (in which case costs and benefits must be quantified to guide policy makers on where to invest resources). As costs and outcomes are usually imprecise or uncertain, economists undertake sophisticated mathematical modelling to generate cost-effectiveness acceptability curves, so the output becomes the probability that an intervention will be cost effective, assuming a given level of willingness to pay.In a socialised health-care system, costs are shared across the population, and there must be limits on what any one individual can receive. This suggests the utilitarian ethic of ‘the greatest good for the greatest number’. Costing is relatively straightforward, if not without controversy (whose perspective?—the health service? health and social care? or private provision and informal care costs as well?). The big problem lies with defining and measuring outcomes. What is health care for? On the basis that it aims to improve length and quality of life, we often use a global quality of life scale, typically the Euroqol EQ5D [1]. This is a five-item scale that estimates the value or worth (utility) of a state of health given by the general population. UK research funders routinely request inclusion of an economic component in grant proposals. The data generated are influential in prioritising decisions by bodies such as the UK National Institute of Health and Clinical Excellence (NICE).
This issue of Age and Ageing includes an economic evaluation of a community hospital model of post-acute (intermediate) care compared with standard acute hospital care for older people [2]. The original publication of the trial results showed small but definite benefits (in terms of reduced disability) from community hospital rehabilitation, and broadly similar costs [3]. All other things being equal (and they may not be, if we take into account capital or opportunity costs, or accessibility), the rational policy maker would choose the community hospitals as ‘economically dominant’. But the economic study was equivocal and concluded that cost-effectiveness was similar.
How so? This is a common problem. In 2004 we reported on an ‘effective’ hospital early discharge and community rehabilitation scheme. Compared with conventional care, outcomes in terms of disability and carer psychological function were better, and overall care was cheaper [4]. It was apparently economically dominant, but the formal economic evaluation was also equivocal [5].
There are two problems: firstly defining appropriate outcomes, and secondly the sensitivity of outcome measurement scales. The EQ5D may fall short on both. It is unlikely to capture all the intended goals of rehabilitation. Improving length and quality of life are unarguable goals of health care, but are not the only goals. Regardless of effects on quality of life, it is a good thing to prevent or cure disease or complications of diseases, to relieve distressing symptoms, improve physical and social function (however measured), give explanation, information, reassurance and education, and to support families and other carers. Much of what we do as doctors is to explain, advise, support, console, teach and make choices explicit—not likely to generate any Quality Adjusted Life Years (QALYs) but valuable nonetheless. Moreover, it is not easy to take into account short-term debility, inconvenience and burden associated with interventions, if they do not lead to a monetary cost. Economic analyses risk ignoring things that are not (or cannot be) measured. The ‘MacNamara fallacy’ states that what you cannot measure does not matter.
Complex interventions can easily fall prey to outcome scale insensitivity, especially if they have a social or environmental component. As a general rule, the closer a measurement is conceptually to the intervention, the bigger the effect you will see. This alone defines ‘effectiveness’ (does this intervention have any effect?). If you want to assess physiotherapy, measure trunk control or walking speed. If you want to assess an analgesic, measure pain. One way of comparing the impact of different interventions is the Kazis effect size (change measured in units of the standard deviation of the baseline distribution). Anything over 1 is very large, anything less than 0.2 is trivial [6]. Effect size for an intervention can vary greatly depending on what you measure. In an evaluation of first eye cataract surgery, the effect size for acuity was 1.4; for a vision-specific disability scale, the VF-14, it was 1.1; for a generic disability scale, the London Handicap Scale, it was 0.5; and for the EQ5D only 0.07 (or 0.02 scale points on a 0–1 scale) [7]. An effect that is huge in terms of acuity or visual disability becomes almost unmeasurable on the EQ5D.
Methodology is against us as well. Randomised trials are designed primarily to demonstrate whether one intervention has any impact compared with another, free from bias or confounding. To keep a trial ethical, you rarely compare something with nothing, so you cannot quantify the absolute benefit of an intervention. And to avoid bias, you underestimate the size of effect (e.g. by using intention-to-treat analysis). Unlike in drug trials where a single main end-point is relatively easy to define, in complex interventions benefits may be multifaceted (e.g. effects on disability, dependency, satisfaction and psychological function). It may even be unclear as to what the precise benefits are. Why do patients ‘like’ home rehabilitation or community hospitals? It may be the location, décor or food, being ‘cared about’, or having more dignity, or choice, for example. Benefits may accrue to multiple parties. Carer outcomes are impossible to include in a conventional cost-utility analysis (an issue in the recent judicial review of NICE's appraisal of cholinesterase inhibitor drugs for dementia.)
Can we do better? In its guidance on social value judgements, NICE tentatively rejects the cost-utility approach alone (although it remains dominant in its decisions), given that we also value equity (the egalitarian ethic), and factors not well reflected in the QALY, such as reducing stigma [8]. More fundamentally, economist Joanna Coast has proposed ‘cost-consequence analysis’ as an alternative [9]. An explicit balance sheet is drawn up with all costs on one side and all benefits on the other, including carer outcomes. She points out that ‘the desire to maximise health seems to be largely the objective of economists rather than society’ and that other factors are important apart from maximising utility. She concludes that ‘use of a single outcome for cost effectiveness analysis fails to recognise that decision-making involves making judgements about a variety of important effects rather than just one.’
Economic evaluations of complex services for older people using the EQ5D may tell us something but are far from giving us the whole picture; they should be taken with a large pinch of salt. A general lack of knowledge of doctors about economic evaluations and their limitations can lead us to misinterpret these studies if we are not careful. Health economics is a discipline whose theory and sophisticated statistics have outrun its empirical practice, especially in identifying and measuring outcomes. Cost-utility alone is inadequate as an economic model. Unpicking specific costs and benefits is required if we are not to undervalue complex interventions.
Nottingham University Hospitals City Campus, Nottingham, UK.
Email: Rowan.harwood{at}nuh.nhs.uk
References
- Euroqol EQ-5D Users Guide (1996) York: Centre for Health Economics.
- O'Reilly J, Lowson K, Green J, et al. A cost-effectiveness analysis within a multi-centre randomised controlled trial of post-acute care for older people in community hospitals. Age Ageing (2008) 37:513–20.
[Abstract/Free Full Text] - Green J, Young J, Forster A, et al. Effects of locality based community hospital care on independence in older people needing rehabilitation: randomised controlled trial. BMJ (2005) 331:317–22.
[Abstract/Free Full Text] - Cunliffe AL, Gladman JR, Husbands SL, et al. Sooner and healthier: a randomised controlled trial and interview study of an early discharge rehabilitation service for older people. Age Ageing (2004) 33:246–52.
[Abstract/Free Full Text] - Miller P, Gladman J, Cunliffe A, et al. Economic analysis of an early discharge and rehabilitation service for older people. Age Ageing (2005) 34:274–80.
[Abstract/Free Full Text] - Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status. Med Care (1989) 27(3, Suppl):S178–89.[Web of Science][Medline]
- Harwood RH, Foss A, Osborn F, et al. A randomised controlled trial to assess the effect of first-eye cataract extraction on the risk of falling, and health status of women over 70 years. Br J Ophthalmol (2005) 89:53–9.
[Abstract/Free Full Text] - Social value judgements: Principles for the development of NICE guidance. Second edition draft for public consultation, http://www.nice.org.uk/media/998/50/SVJ2ForPublicConsultation.pdf (accessed 22.5.08).
- Coast J. Is health economics in touch with society's health values? BMJ (2004) 329:1233–6.
[Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||