A comparison of the discriminative and evaluative properties of the SF-36 and the SF-6D index

Abstract

To examine whether the move from the multidimensional SF-36 patient-reported outcome measure to the single-index preference-based SF-6D entails a loss in discriminative and evaluative properties, the magnitude of that loss and whether it matters. Retrospective analysis of data from studies that used the SF-36 and the derived SF-6D. The discriminative and evaluative properties of the two measures were compared by calculating effect size (ES), standardized response mean (SRM), and relative validity (RV) statistics using the SF-6D as the reference. Data were available from seven studies and 8,522 subjects. At least one SF-36 scale was always more sensitive than the index. Cross-sectional pooled results showed that physical functioning (RV=0.19 and ES=0.13) and PCS (RV=0.18 and ES=0.13) were generally most sensitive compared to the index (RV=0.16 and ES=0.12). Longitudinal pooled results showed that PCS (RV=0.20), MCS (RV=0.17), general health (RV=0.18), and social functioning (RV=0.17) were generally more sensitive than the index (RV=0.14) based on RVs. Longitudinal pooled SRMs were all very small in magnitude. No scale/dimension consistently had the largest RV, ES, or SRM across all conditions studied. Moving from the SF-36 to SF-6D entails losses of a small magnitude in discriminative and evaluative properties.