Abstract
BACKGROUND: To determine the clinically meaningful changes and responsiveness of widely used frailty measures. METHODS: We analyzed data from a prospective cohort study of 1,135 community-dwelling older adults who underwent assessments of frailty and health-related quality of life using the EuroQol-5D at baseline and 1 year later. Frailty measures included deficit-accumulation frailty index (FI); frailty phenotype; Fatigue, Resistance, Ambulation, Illness, and Loss of Weight scale; and the Study of Osteoporotic Fracture (SOF) index. We determined the clinically meaningful changes by the distribution-based method and the anchor-based method using the EuroQol-5D score and responsiveness indices. RESULTS: Frailty measures were available in 925 participants at 1 year (81.5%). Based on the distribution-based method, small and large clinically meaningful changes were 0.019 and 0.057 for FI, 0.249 and 0.623 for frailty phenotype, 0.235 and 0.587 for FRAIL scale, and 0.116 and 0.289 for SOF index, respectively. The anchor-based estimates of small and large changes were 0.028 and 0.076 for FI, 0.097 and 0.607 for frailty phenotype, 0.269 and 0.368 for FRAIL scale, and 0.023 and 0.287 for SOF index, respectively. Based on the responsiveness index, per-group sample sizes to achieve 80% power in clinical trials, ranged from 51 (FI) to 7,272 (SOF index) for a small change and 9 (FI) to 133 (FRAIL scale) for a large change. CONCLUSIONS: The estimates of clinically meaningful change of frailty measures can inform the choice of frailty measures to track longitudinal changes of frailty in clinical trials and clinical care of community-dwelling older adults.