Background & Scientific Rationale
Competing healthcare interventions are increasingly prioritised on the basis of relative cost-effectiveness, or the additional cost per unit of health benefit. Health benefits are most commonly measured in terms of quality-adjusted life years (QALYs) gained, which combine changes in quality of life and years of life into a single index measure. Any improvement in quality or survival is associated with a proportional increase in QALYs. Interventions that generate a greater number of QALYs for a given cost, or equivalently, have a lower cost per QALY gained, have greater priority for funding. This evaluative framework has become known as ‘QALY maximisation’.
Advocates of palliative care, however, argue that this framework unfairly favours curative interventions over end-of-life care. They note that improvements in quality at the end of life are inevitably of a very limited duration, and therefore the overall QALY gains from such improvements will always be smaller than similar improvements to patients with a longer life expectancy. Indeed, as the primary objective of end-of-life care is not to extend survival, some suggest that it is inappropriate to even include a time element in its evaluation. They also suggest that the QALY as currently measured neglects less quantifiable non-health benefits of end-of-life care. In these regards, they argue, the QALY maximisation approach discriminates against patients nearer to the end of life.
Collectively these arguments, sometimes referred to as the ‘QALY problem’, suggest that the relatively low priority assigned to end-of-life care may reflect a failure of the evaluation framework rather than the true societal value of such care. This implies that there may be a subset of interventions, including but not limited to end-of-life care, that are highly valued by the public and by decision makers but that are unable to demonstrate their value within the conventional QALY maximisation framework.
A first step in testing for a bias in the evaluation of end-of-life care is to understand the relationship between societal value and QALY gains. A weak or even negative relationship would suggest that QALY gains do not adequately reflect the societal value of end-of-life care and that such care may be systematically disadvantaged relative to curative interventions that are more likely to demonstrate their value within a QALY maximisation framework. To this end we propose using stated preference methods to explore the relationship between societal value and QALY gains for hypothetical end-of-life care. For comparison, we will also estimate this relationship for curative care interventions.
The project will use stated preference (SP) methods to estimate a societal value function over QALYs gained for hypothetical end-of-life and curative interventions. SP methods present two or more alternatives, each described by a common set of attributes, and ask respondents to indicate their preferred option. There are a number of potential techniques, but for this research we will consider discrete choice experiments (DCE) and constant-sum paired comparisons (CSPC). DCEs present respondents with two (or more) hypothetical scenarios that differ in the levels of relevant attributes and ask respondents to identify a single preferred option. The likelihood of choosing a particular alternative is taken to be proportional to the relative value of the alternatives. CPSCs are similar to DCEs in terms of their presentation but ask respondents to allocate hypothetical budget shares between the alternatives. The difference in budget shares is taken to be proportional to the difference in value between the alternatives. The two approaches have different practical and theoretical properties and we will identify a preferred technique with the assistance of a project Steering Group and through pre-pilot work with small groups of the public. The preferred technique will be comprehensible to respondents and allow for consideration of the trade-offs and opportunity costs associated with different choices.
A critical phase of the research will be the identification of the non-health benefits associated with end-of-life care. A literature review has suggested that these benefits may include, among others, a sense of control and dignity, avoiding an inappropriate prolongation of dying, relieving the burden on family and carers, and strengthening the relationship with loved ones. To refine this list to a set that can be feasibly incorporated into an SP task we will consult with our Steering Group and conduct focus groups with end-of-life care advocates, including hospice and palliative care professionals, as well as general practitioners. Potential focus group participants will be identified by members of the Steering Group. We will also collaborate with health care professionals with relevant experience to ensure that the hypothetical scenarios developed are plausible and policy relevant.
Based on feedback from this pre-pilot phase we will further refine the language and presentation of the tasks and conduct a pilot with a larger sample of the public, drawn from the Department of Economics database of volunteers. In this piloting phase particular attention will be paid to respondents’ comprehension of the tasks in terms of the language used and the instructions given. On the basis of these results the language and instructions of the questionnaires will be further refined and the experimental design will be updated to maximise the statistical efficiency of the survey.
Once the experimental design and the presentation of the SP questionnaire is finalised, a survey sampling company will be commissioned to host the survey and identify a representative sample of the UK population. Each respondent will answer some basic demographic questions followed by approximately 12 DCE and/or CSPC tasks. Respondents will be asked to indicate their preferences over paired end-of-life or curative care scenarios that differ in terms of their attribute levels, including QALY gains. By systematically changing attribute levels over a series of tasks it will be possible to statistically model the change in societal value associated with a unit change in each attribute. Depending on the task format, responses will be modelled using nonlinear logit (DCE) or linear tobit (CSPC) models. Latent class techniques may be used where appropriate to understand and classify respondent heterogeneity.
Expected Output of Research / Impact and added value
The increasing proportion of elderly citizens in the population means that the relative share of healthcare resources to end-of-life care is likely to grow over the short to medium-term. To ensure that this share is fair and efficient it is important to identify any bias against end-of-life care that may exist in the current evaluation framework.
It is anticipated that the statistical models will show a significant and positive relationship between QALY gains and the societal value of curative interventions but the anticipated relationship for end-of-life care is less clear. It may be positive as well, but an insignificant (flat) relationship would suggest that the value of end-of-life care is independent of any QALY gains, whilst a negative relationship would suggest that value is higher when QALY gains are smaller (i.e. when patients are closer to death). Both cases would suggest that strict QALY gains are not an appropriate measure of the societal value of end-of-life care. Such information would inform changes to the current evaluation framework and ensure a fairer and more efficient allocation of health care resources to end-of-life care.