Mapping the Michigan Hand Questionnaire and Disabilities of the Arm, Shoulder and Hand onto the EuroQol-5 Dimension.

Abstract

Cost-utility analyses require utilities data, which are often unavailable in hand surgery, where functional scores are more commonly used. This study aims to develop predictive models to accurately map 2 dominant legacy questionnaires for the hand specialty, Michigan Hand Outcomes Questionnaire (MHQ) and Disabilities of the Arm, Shoulder, and Hand (DASH) scores onto the 5-level EuroQol-5 Dimension (EQ-5D). This will facilitate more economic evaluations of the growing number of hand procedures. Data were collected prospectively from 354 patients diagnosed with various hand conditions at the University of Michigan Hospital. Participants completed the MHQ, DASH, and EQ-5D surveys. We evaluated 6 statistical models to determine the best fit for mapping MHQ and DASH onto EQ-5D. Model performance was assessed using mean absolute error (MAE), root mean squared error (RMSE), R², and proportion of predictions within 0.1 and 1 SD of true EQ-5D values. The logarithmic response mapping model was the best fit for mapping MHQ to EQ-5D (MAE, 0.1119; RMSE, 0.1584; R² , 0.445), whereas the ordinary least squares model performed best for DASH (MAE, 0.0934; RMSE, 0.1347; R² , 0.601). Both models showed strong predictive accuracy, with 58% (MHQ) and 68% (DASH) of predictions within 0.1 of observed EQ-5D scores, and over 85% within 1 SD. Response mapping and ordinary least square models are reliable methods to map MHQ and DASH onto EQ-5D. These mapping algorithms offer a valuable alternative for obtaining utilities in the absence of direct data.