A comparison of frailty indexes for prediction of adverse health outcomes in an elderly cohort

This study aimed to compare the predictive accuracy for several frailty-related adverse health outcomes of a cumulative index derived from the Italian population-based elderly cohort of the Conselice Study of Brain Aging (CSBA), which takes into account multiple different domains (demographic, clinical, functional, and nutritional parameters), with that of an index derived from the Study of Osteoporotic Fractures (SOF), modified for application to the CSBA database and henceforth called mSOF, which is exclusively focused on muscular fitness. Data are for 1007 CSBA participants aged >= 65 years. Investigated adverse outcomes included 4- and 7-year risk of death and 4-year risk of fractures, falls, disability, hospitalization, and nursing home placement. Accuracy for prediction of these outcomes was investigated using area under the curve (AUC) statistics. CSBA index performed better than mSOF index for prediction of mortality (p < 0.001), hospitalization (p = 0.002), and nursing home placement (p = 0.049). For all outcomes excluding falls, frailty defined by CSBA index had a slightly lower specificity but a much higher sensitivity than frailty defined by mSOF Index. In conclusion, in this elderly cohort, the multidimensional CSBA index is a better predictor of frailty-related adverse health outcomes than the unidimensional mSOF index