Experimental Tests of the Robustness of Health Decisions Using the EQ-5D

Three experiments systematically explore the robustness of stated preference estimates of health-related quality of life (hereafter HR-QoL), a core component of the QALY calculation. The presentation of health information is manipulated to discover how the judged severities of health states respond to these presentation changes. Specifically, splitting up an EQ-5D dimension into two increases its weight in severity judgments, and combining two dimensions reduces their weight, even though the information provided is unchanged. This result implies that preferences are malleable. Telling participants the name of the health state significantly alters the judged severities of some health states, which suggests that the standard information presented in an HR-QoL estimation study may not adequately capture all relevant aspects of health states as perceived by members of the public. This effect is shown to be driven by fear of certain health states, and the implications for allocative policymaking are discussed. Behaviour across all three experiments demonstrates systematic deviations from normatively ideal behaviour, but the conclusions are optimistic: understanding where standard approaches fall short is necessary for the design of behaviourally realistic yet tractable models and appropriate survey designs to better understand the impacts of different health states on people’s anticipated quality of life.