OBJECTIVES: The EQ-5D is preferred by NICE to measure health-related quality of life. Versions with three (EQ-5D-3L) and five (EQ-5D-5L) response levels exist. Due to concerns about its methodology, quality and reliability, NICE does not recommend using the EQ-5D-5L value set for England published in 2018. To derive utility values from EQ-5D-5L responses, 5L data should rather be mapped onto the 3L value set. NICE previously recommended using the Van Hout crosswalk, but its 2022 Manual now states that the EEPRU mapping function developed by the Decision Support Unit (DSU) should be used. Our aim was to compare these different value sets and mapping methods.

METHODS: For each of the 3125 possible health states in the 5L system, utility values were obtained using the 2018 5L value set for England, the Van Hout crosswalk, and the DSU mapping function. In addition, utilities for the 243 possible EQ-5D-3L health states were calculated using the 3L value set for the UK, allowing comparison between corresponding 3L and 5L states. Density histograms and plots comparing utility values in comparable states were generated to visualize differences.

RESULTS: Utility values obtained with the 5L value set were generally higher than 3L value set/crosswalk estimates. Proportions of health states worse than death (utilities below zero) were 5.1% with the 5L value set, around 22% for the DSU mapping function (age and sex-dependent), 26.7% for the Van Hout crosswalk, and 34.6% for the 3L value set. Whilst the Van Hout crosswalk produces identical values for the 243 3L health states as the 3L value set, the DSU mapping function does not. Its utility values are slightly lower in the best health states, and generally higher in moderate and worst health states.

CONCLUSIONS: Utility values obtained with different UK value sets and mapping functions vary, potentially leading to differences in health-economic outcomes.