The Standardization of Frailty Phenotype Criteria Improves Its Predictive Ability: The Toledo Study for Healthy Aging

Several studies have assessed the performance of the original frailty phenotype criteria (FPC) and the standardized version according to the characteristics of the population. No studies exist, however, evaluating the impact of this standardization on its predictive ability.

To compare how the original FPC and the standardized-frailty phenotype criteria (S-FPC) estimate the prevalence of frailty and their ability to predict mortality, hospitalization, incident disability, and falls.

Data were taken from the Toledo Study for Healthy Aging, a population-based, community-dwelling study conducted on 1645 individuals over 65. Frailty was operationalized in two ways: FPC, using the cut-off estimated in the Cardiovascular Health Study and S-FPC, using cut-off points fitted to the phenotypic characteristics of our study sample. Frailty prevalences were compared using chi-square statistic. Cox proportional hazard models and logistic regressions evaluated the predictive ability of both tools. Lastly, survival tests were applied.

Frailty and prefrailty prevalences varied according to the tool used: 24.12% and 66.40%, respectively when we used FPC and 6.68% and 47.81% when we used S-FPC (P < .01). Regarding their predictive ability, S-FPC, but not FPC, identified consistently the prefrail persons as an intermediate risk group between robust and frail people [death 1.57 (1.15-2.16); hospitalization 1.47 (1.16-1.85); and incident disability 1.96 (1.30-2.97); P < .005]. Furthermore S-FPC predicted death and hospitalization at shorter times than FPC (P < .05).

FPC should be standardized according to the characteristics of the population in order to improve its predictive ability.