Abstract
OBJECTIVES: To perform an economic evaluation, mapping from a disease specific measure onto a generic preference-based measure is essential when utility values of health states are not available in the same sample. In fibromyalgia (FM), no mapping studies have estimated the EQ-5D-3L from the Revised Fibromyalgia Impact Questionnaire (FIQR). This study aimed to predict an EQ-5D-3L utility score from the FIQR. METHODS: Data for 160 Spanish adults with confirmed FM according to the American College of Rheumatology 1990 criteria were used. Econometric models investigated were ordinary least square (OLS), censored least absolute deviations (CLAD) and multinomial logistic (MNL) models in ten alternative specifications. Choice of model was based on the 'Best' performance which was defined as: the lowest absolute difference (AD), mean absolute error (MAE), and root mean squared error (RMSE), and the highest R2statistics. RESULTS: The majority of the sample were females (n=156, 97.5%), with mean (SD) age of 57.3 (8.8) years. Patients had a mean (SD) FIQR total score of 68.9 (18.87) and a mean (SD) EQ-5D-3L utility value of 0.47 (0.22). The predicted mean utilities in all OLS models were identical (up to four decimals) to the observed means. The OLS model performed best compared with CLAD and MNL with the lowest error statistics (AD = 0.0000; MAE = 0.1279; MSE = 0.1576; R2= 0.4675). The best performing mapping function, defined by the OLS, is able to predict the EQ-5D-3L tariff score using the 21 FIQR collapsed items in a discrete form, age and educational level. CONCLUSIONS: It was feasible to map from the FIQR to the EQ-5D-3L. Over the ten model approaches tested, the OLS was the best performing model which predicted the tariff score of the EQ-5D-3L. Further research is required to verify the mapping function between these two measures.