Recurrent Measurement of Frailty Is Important for Mortality Prediction: Findings from the North West Adelaide Health Study

OBJECTIVES Frailty places individuals at greater risk of adverse health outcomes. However, it is a dynamic condition and may not always lead to decline. Our objective was to determine the relationship between frailty status (at baseline and follow‐up) and mortality using both the frailty phenotype (FP) and frailty index (FI). DESIGN Population‐based cohort. SETTING Community‐dwelling older adults. PARTICIPANTS A total of 909 individuals aged 65 years or older (55% female), mean age 74.4 (SD 6.2) years, had frailty measurement at baseline. Overall, 549 participants had frailty measurement at two time points. MEASUREMENTS Frailty was measured using the FP and FI, with a mean 4.5 years between baseline and follow‐up. Mortality was matched to official death records with a minimum of 10 years of follow‐up. RESULTS For both measures, baseline frailty was a significant predictor of mortality up to 10 years, with initially good predictive ability (area under the curve [AUC] = .8‐.9) decreasing over time. Repeated measurement at follow‐up resulted in good prediction compared with lower (AUC = .6‐.7) discrimination of equivalent baseline frailty status. In a multivariable model, frailty measurement at follow‐up was a stronger predictor of mortality compared with baseline. Frailty change for the Continuous FI was a significant predictor of decreased or increased mortality risk based on corresponding improvement or worsening of score (hazard ratio = 1.04; 95% confidence interval = 1.02‐1.07; P = .001). CONCLUSIONS Frailty measurement is a good predictor of mortality up to 10 years; however, recency of frailty measurement is important for improved prediction. A regular review of frailty status is required in older adults. J Am Geriatr Soc 67:2311–2317, 2019