Mapping the 25-item National Eye Institute Visual Functioning Questionnaire (NEI VFQ-25) to EQ-5D utility scores.

Objective: To develop a mapping algorithm for the estimation of EQ-5D-based utility scores from observed 25-item National Eye Institute Visual Functioning Questionnaire (NEI VFQ-25) scores, a disease-specific, patient-reported outcome measure used in several retinal disorders to evaluate vision-specific functioning.

Methods: The dataset comprised 951 paired EQ-5D/NEI VFQ-25 observations from 344 patients in RESTORE, a 12-month, randomized, double-blind trial in individuals with visual impairment due to diabetic macular edema. EQ-5D index scores (utilities) were calculated based on the UK tariff. We evaluated 11 models using predictor sets based on the NEI VFQ-25 subscales to estimate utility as a function of NEI VFQ-25 score, based on four modeling techniques. Model performance was assessed by 10-fold cross-validation comparing root mean squared error (RMSE), mean absolute error (MAE) and correlation with EQ-5D score (Pearson and Spearman correlation coefficients).

Results: Mapping results were similar across all techniques and predictor sets. The reverse two-part generalized estimating equation model used fewest predictors and had the best predictive performance (RMSE 0.200, MAE 0.140). Predicted and original EQ-5D values were not strongly correlated (squared Spearman correlation coefficient, 0.34).

Conclusions: Although mapping disease-specific instruments to EQ-5D utilities is a preferred method by some reimbursement bodies, finding an appropriate mapping equation is not straightforward. In this study, mapping NEI VFQ-25 scores to EQ-5D utilities provided low predictive power, independent of the modeling methodology applied, suggesting an inability of the EQ-5D to discriminate vision-related activities, and highlighting that mapping exercises may lead to inaccurate utility values that do not represent patients' preferences.