Discrete choice experiments in health care

In many publicly provided healthcare systems, limited resources coupled with unlimited demand result in decisions having to be made about the efficient allocation of scarce resources. This raises questions of how services should be provided (for example, how should patients with cancer be treated? should central clinics, which reduce waiting time but increase travel time for patients, be introduced?) through to the optimal provision and the financing of health care (for example, how should we pay doctors to encourage them to work in remote and rural areas? what would encourage nurses to return to the labour market?). Given the lack of a market for health care, economics techniques inform such decisions.¹ One approach adopted by and further developed in health economics over the past decade is discrete choice experiments.²,³ In this issue Sculpher et al use this approach to consider patients' preferences in the treatment of prostate cancer (p 382).⁴

Discrete choice experiments are an attribute based measure of benefit² that is based on the assumptions that firstly, healthcare interventions, services, or policies can be described by their characteristics (or attributes) and secondly, an individual's valuation depends on the levels of these characteristics. For a review of the stages involved in conducting discrete choice experiments and the information they provide, see Ryan and Farrar³ as well as the paper by Sculpher et al in this issue.⁴

Discrete choice experiments were introduced into health economics as a technique to go beyond the quality adjusted life year (QALY) paradigm.⁵ Users were concerned with many aspects of health care beyond health outcomes. Such factors included waiting time, location of treatment, type of care (for example, surgical or medical), and staff providing care (consultant or specialist nurse) and were referred to as process attributes. Discrete choice experiments allow investigation of the trade-offs between such process and health outcomes attributes.^2-5 Applications of discrete choice experiments have been extended to consider provider preferences⁶,⁷ such as strength of hospital consultants' preferences for various aspects of their work.⁶

More recently the technique has been used to value health outcomes in the provision of care (often beyond those valued within the QALY). For example, Sculpher et al use the technique to establish which health attributes of conservative treatments for prostate cancer are most important to men.⁴ They included eight attributes and found that men were willing to contemplate trading off some life expectancy in order to be relieved of the burden of troublesome side effects, such as limitations in physical energy.

At the methodological level, studies find that respondents will complete discrete choice experiments in an internally valid and consistent manner.²,⁸,⁹ An important question in the use of any survey technique is that of external validity—that is, do individuals behave in reality as they state in a hypothetical context?

Although limited research has been conducted in this area and future research is clearly important (which is the case for all economic evaluation techniques, including those used in the QALY framework), experience from other areas such as the valuation of environmental goods and services implies that we can be optimistic.

Given the role of the National Institute for Clinical Excellence (NICE) in making recommendations concerning optimal treatments, can it make use of discrete choice experiments? The institute is under increasing pressure to take account of patients' preferences. To date systematic consideration of such preferences has been limited. Typically public preferences are required to elicit quality weights in the QALY paradigm.¹⁰ This is not enough since patients may value outcomes differently to the public and have preferences over aspects of care beyond QALYs.⁵,¹¹

NICE plans to have a patient centred evaluation of technologies in addition to the current assessments of clinical and cost effectiveness. Using the approach of discrete choice experiments allows the integration of patients' values on all aspects of care in one measure. We will be able to see how patients trade different health outcomes as well as process type attributes, alongside each other. Valuation of process and health outcomes from the patients' perspective may well lead to conclusions that conflict with the recommendations of the cost per QALY approach. This is more likely to be the case in comparisons of technologies that differ with respect to outcomes beyond those measured in a QALY, as well as process attributes. Recent examples include the reviews commissioned by NICE on the effectiveness and cost effectiveness of metal on metal hip resurfacing arthroplasty for treatment of hip disease where treatment options differed with respect to process (surgical versus conservative management)¹² and of haemodialysis at home versus hospital or satellite unit for people with end stage renal failure.¹³ For such technologies, the crucial question then becomes: what are the implications of patient centred care for the institute's guidelines? This is an important area for future research.