Session (CTI)

Issue: Reducing health inequities requires prioritizing the health of those who are systematically disadvantaged. Equity-informative approaches to cost-effectiveness analysis (CEA), such as distributional CEA, assess equity impacts by weighting outcomes and opportunity costs of interventions across equity-relevant characteristics. These weights have been elicited through population surveys, often fielded online, under the assumption that in a tax-funded or insurance-based system, “the public” should decide the value of reducing health inequities. Yet, this seemingly democratic solution raises questions about whether these approaches are best suited to inform real-world decision making. In this issue panel we will discuss three challenges: representativeness, heterogeneity, and context. The use of online surveys may under-represent the preferences of those most disadvantaged and requires high literacy and numeracy. Results are heterogenous and sensitive to the allocation scenario, framing effects, survey design considerations, and individual characteristics. Finally, equity is context specific, and stakeholder priorities and political dynamics can shift across countries, health conditions, and equity dimensions. As decision contexts shift, the value ascribed to improving health equity may also shift, raising concerns that societal preferences elicited via online choice experiments misalign with real-world decisions. Taken together, these challenges highlight the need to clarify whose preferences should guide equity weighting in practice, and whether population-level surveys can meaningfully represent the priorities of those who experience the greatest health disadvantages. Overview: Following an introduction to equity preference research (Cadham), panelists will briefly present on the current state of elicitations in Canada (Iragorri) and the US (Slejko) with an eye towards the challenges outlined above. Kowal will then outline future directions for gathering preferences, focusing on subgroup definitions, sampling strategies, and ways to manage data gaps. Presentations will be followed by a 15-minute moderated discussion and 15-minute audience discussion.