OBJECTIVES: In the absence of an accepted local EQ-5D-5L tariff, NICE requires EQ-5D-5L responses to be mapped onto the long-standing EQ-5D-3L value set for the UK. Their 2022 Manual recommends using the mapping function based on the EEPRU dataset recently developed by the NICE Decision Support Unit (DSU), instead of the van Hout crosswalk that was previously endorsed. Our aim was to compare utility values derived using these two mapping methods.

METHODS: For all 3,125 unique health states existing in the EQ-5D-5L system, utility estimates were calculated using both mapping functions. Absolute differences in resulting values were calculated and a scatterplot was generated to identify outliers and allow visual comparison. Differences in mean utility values were evaluated in subsets of health states, grouped by the level of severity of reported health issues. Simulations in hypothetical patient populations were conducted to evaluate the potential impact of the mapping functions on utility values derived from EQ-5D-5L clinical trial data.

RESULTS: For 1,898 (61%) health states spread across the severity spectrum, the DSU EEPRU mapping function generates a higher utility value than the van Hout Crosswalk. The mean difference was 0.062 (± 0.139 standard deviation), ranging from +0.619 to -0.198 for individual health states. Our analyses and simulations suggest that mean utility values for a patient population estimated with the DSU mapping function would typically lie 0.050 to 0.100 higher than those obtained with the van Hout Crosswalk.

CONCLUSIONS: Our study shows that not only the choice of the EQ-5D instrument, but also the mapping approach used can have a substantial impact on utility values and, hence, estimated health gains and cost-effectiveness ratios. The DSU EEPRU mapping function tends to shift UK utility values further upwards, potentially making it harder for new technologies to demonstrate utility gains and obtain NICE recommendation.