Preference laundering in health economics

Measuring the value of health with preferences

Healthcare resources are limited. This means that difficult decisions need to be made on how to allocate these limited resources efficiently and ethically. In order to approach these decisions in a systemic and transparent manner, a measure of the benefits of alternative allocations of healthcare resources is needed. A commonly used measure is quality-adjusted life-years (QALYs), which can be seen to capture the improvements in well-being of different health interventions. This is done by asking people, usually from the general public, to make choices, or state their preferences, between different health states. For example, they may be asked: ‘Would you prefer seven years in full health to ten years with moderate pain and moderate problems with mobility and self-care?’ With enough responses to this type of questions, a complete measure of the public’s preferences over health states can be derived, which is taken to indicate the value of health and can be used to make judgements about how to best allocate healthcare resources.

Why we shouldn’t trust people’s preferences

However, a quick pause to consider the question above should make it clear that it is a question that is incredibly difficult to answer. Answering that question adequately would require deep knowledge of how the diminished state of health (with moderate pain and moderate problems with mobility and self-care) would affect my life and the projects I hold as important. There is no reason to believe that typical respondents will have that type of knowledge since they will likely never have contemplated this type of trade-offs before. The point is that theorists and planners should be cautious with using people’s preferences over health states as they are likely based on a poor understanding of the consequences of those health states and will likely be affected by biases and framing effects, since people do not have pre-settled judgements about these issues. In short, people’s preferences over health states cannot be trusted, unless there is evidence to the contrary.

How to launder preferences

What would this evidence to the contrary look like? Answering this question essentially amounts to answering the question of ‘How to launder people’s preferences?’—the topic of my PhD project. The basic idea behind ‘laundering preferences’ is that, while preferences will sometimes be problematic and cannot be ‘trusted’ (as discussed above), if certain conditions are met, preferences will be reliable indicators of what makes people’s lives go well—that is, their well-being. The central questions are then (i) what those conditions are, (ii) how to measure whether those conditions are satisfied, and (iii) how to make people’s preferences satisfy those conditions (if they are not already satisfied). Key conditions of laundered preferences are commonly taken to be that they are self-interested, based on complete and correct information, and based on error-free and exhaustive reasoning around the alternative choices. As we have discussed, preference over health states will likely not satisfy these conditions, but can we make them do that?

In my PhD project, I explorer ways to launder people’s preferences. In brief, this can be done either by—what I call—ex ante or ex post laundering, which amounts to approximating a situation where the conditions are satisfied before eliciting people’s preferences (for example, by reframing the questions to make sure that elicited answers are self-interested, by inviting respondents to deliberative discussions over the alternative choices, by making respondents repeat their choices to learn from previous decisions, by confronting respondents with their seemingly problematic preferences, and so on) or substituting problematic (aspects of) people’s preferences after they have been elicited (for example, by estimating structural models of people’s choices and substituting ideal for problematic parameters, or removing inconsistent choices, and so on), respectively. While these efforts will likely not solve all ‘trust-issues’ with people’s preferences over health states, they can significantly mitigate these issues and enable more reliable measures of the value of different allocations of healthcare resources.