Mapping the cancer-specific FACT-B onto the generic SF-6Dv2

• DOI: 10.1007/s12282-020-01141-9
• Publication Date: 2020-07-25T18:02:23Z
• Authors: Azin Nahvijou; Hossein Safari; Mahmood Yousefi; Marziyeh Rajabi; Morteza Arab-Zozani; Hosein Ameri
• Keywords: Adult; Models, Statistical; Psychometrics; Cost-Benefit Analysis; Health Status; Breast Neoplasms; Patient Preference; Iran; Middle Aged; 3. Good health; 03 medical and health sciences; Cross-Sectional Studies; Predictive Value of Tests; Surveys and Questionnaires; Quality of Life; Humans; Female; Quality-Adjusted Life Years; Least-Squares Analysis; 0305 other medical science; Algorithms

The health-related quality of life (HRQoL) data extracted from cancer-specific questionnaires are often non-preference based, while patient preference-based utility data are required for health economic evaluation. This study aimed to map Functional Assessment of Cancer Therapy-Breast (FACT-B) subscales onto the Short Form six Dimension as an independent instrument (SF-6Dv2ind-6) using the data gathered from patients with breast cancer.Data for 420 inpatient and outpatient patients with breast cancer were gathered from the largest academic center for cancer patients in Iran. The OLS and Tobit models were used to predict the values of the SF-6Dv2ind-6 with regard to the FACT-B subscales. Prediction accuracy of the models was determined by calculating the root mean square error (RMSE) and mean absolute error (MAE). The relationship between the fitted and observed SF-6Dv2ind-6 values was examined using the Intraclass Correlation Coefficients (ICC). Goodness of fit of models was assessed using the predicted R2 (Pred R2) and adjusted R2 (Adj R2). A tenfold cross-validation method was used for validation of models.Data of 416 patients with breast cancer were entered into final analysis. The model included main effects of FACT-B subscales, and statistically significant clinical and demographic variables were the best predictor for SF-6Dv2ind-6 (Model S3 of OLS with Adj R2 = 61.02%, Pred R2 = 59.25%, MAE = 0.0465, RMSE = 0.0621, ICC = 0.678, AIC = -831.324, BIC = -815.871).The best algorithm developed for SF-6Dv2ind-6 enables researchers to convert cancer-specific instruments scores into preference-based scores when the data are gathered using cancer-specific instruments.