RESUMO
OBJECTIVE: The aim of this study was to derive a function that can map the Nottingham Health Profile (NHP) questionnaire onto a utility measure, the EuroQol five-dimensional (EQ-5D) questionnaire index, for diabetic patients. METHODS: A cross-sectional study was performed on diabetic patients in Hungary with different complications in which quality of life was measured by using both the NHP questionnaire and the EQ-5D questionnaire. Ordinary stepwise-backward least-squares regression was used to develop a mapping function. Adjusted R2, Akaike's information criterion, and root mean square error were used to assess the performance of the model. The robustness of the models was tested using 10-fold cross-validation and bootstrapping. RESULTS: The best-fitting models were those that contained all the NHP statements as predictors and a stepwise reduced version that contained only 19 statements. The latter model, however, showed considerable variability in the selection of predictors. The adjusted R2 of the former model was 0.68, the root mean square error was 174, and the Akaike's information criterion was -559.9. CONCLUSIONS: The expected value of the EQ-5D questionnaire can be reasonably predicted on the basis of results of the NHP in patients with diabetes mellitus. The mapping function of the NHP onto the EQ-5D questionnaire is capable of estimating the expected EQ-5D questionnaire utility values in a group of patients with diabetes. The function's applicability for individual-level predictions, however, is limited. Further research is needed to find out whether mapping functions developed in Central-Eastern European countries are transferable to Western European countries.
