Presentation (CTI)

OBJECTIVES: The National Institute for Health and Care Excellence (NICE) has launched a public consultation on adopting the EQ-5D-5L value set (Rowen, 2026) to replace the current 3L value set. An impact assessment commissioned by NICE and conducted by SCHARR (Wailoo et al., 2026), based on 37 published technology appraisals, suggests that this change could significantly affect cost-effectiveness results and future reimbursement decision-making. While NICE’s Decision Support Unit (DSU) provides algorithms to map EQ-5D-3L data to EQ-5D-5L at the individual level, the non-linear relationship between 3L and 5L utilities introduces uncertainty when these methods are applied directly to population-level mean values. This study aimed to develop and validate a tool to address this limitation for analyses without access to individual patient data (IPD).
METHODS: An R-based simulation tool was developed to generate pseudo IPD from aggregate inputs, including population mean EQ-5D-3L utility, age, sex distribution, and their correlations using a truncated multivariate normal distribution. The DSU mapping algorithm is applied to each pseudo-observation, generating individual EQ-5D-5L utilities, which are then averaged to produce a population-level estimate. Performance was validated using IPD from a publicly available dataset (Health Survey for England) and Takeda clinical trials data.
RESULTS: Tool-derived EQ-5D-5L estimates were consistently closer to directly observed 5L values than estimates obtained by applying the algorithm to aggregate means. However, validation is limited by the datasets used and assumptions underpinning the simulated distributions.
CONCLUSIONS: This tool provides a practical solution for converting EQ-5D-3L aggregate data to EQ-5D-5L estimates without IPD. Key limitations include the assumption of a truncated normal distribution, which may bias results at high utility levels. Further validation across diverse datasets is ongoing.