Method: Utilizing the domain framework of physical Frailty from published literature, we constructed a simple five-item Patient reported (PRO)-Frailty Index using the following components- fatigue, poor balance, grip strength, anorexia and memory loss. Using Cox regressions, we analyzed the utility of this Index to predict all-cause mortality in a cohort of 3,185 (78% women, mean age 64 years) patients with rheumatoid arthritis and osteoarthritis followed for a mean 4.3 years. Receiver Operating Curve (ROC) analyses were performed to identify the optimal cutoff that can be used to dichotomize the Frailty Index.
Result: The PRO-Frailty Index (range 0-5) had a near-normal distribution in the population studied with a mean (SD) of 1.3(1.1) overall. The PRO-Frailty Index predicted mortality in age-sex adjusted analyses with a hazard ratio of 1.15(1.01-1.31) for each unit increase in PRO-frailty index. In multivariable analyses where the effect of age, gender, disease duration and education level were accounted for, each unit increase in PRO-Frailty Index was associated with 17% increased mortality risk (1.17; 1.02-1.34). ROC analyses suggested that dichotomizing the index at 0 and >0 offered the best model fit. The frailty-mortality association was observed in both the arthritis categories although the confidence interval was too wide for statistical significance individually.
Conclusion: PRO-Frailty Index can predict mortality among patients with arthritis.
Disclosure: E. Krishnan, None; J. F. Fries, None; B. Lingala, None.