Physical Function and Frailty for Predicting Adverse Outcomes in Older Primary Care Patients

Objective: To explore the predictive ability of the Short Physical Performance Battery (SPPB), Late Life Function and Disability Instrument-Function component (LLFDI-function) and frailty phenotype, for falls, hospitalizations, emergency department (ED) visits, and low self-rated health (SRH) over 1 and 2 years in older adults.

Design: Secondary analysis of data from a longitudinal study, the Boston Rehabilitative Impairment Study of the Elderly.

Setting: Primary care.

Participants: Adults 65 years and older at risk for disability who completed ≥1 follow-up call (N=391).

Interventions: None.

Main outcome measures: We computed separate logistic regression models using the SPPB, LLFDI-function, and frailty phenotype as independent variables and falls, hospitalizations, ED visits, and SRH over 1 and 2 years as dependent variables. Receiver operating characteristic curves were constructed and the areas under the curves calculated.

Results: Participants had a mean age of 76.5±7.1 years. The SPPB, LLFDI-function, and frailty phenotype all predicted hospitalizations and low SRH over a 1- and 2-year timeframe (odds ratio [OR] min-max, 1.35-1.51 and 1.67-3.07, respectively). Over 2 years, the SPPB predicted ED visits (OR, 1.28), and the LLFDI-function predicted falls (OR, 1.31). The LLFDI-function predicted low SRH better than the frailty phenotype over 1 year. There were no differences between the measures for any of the other outcomes.

Conclusions: The SPPB, LLFDI-function, and frailty phenotype had similar accuracy for predicting falls, hospitalizations, ED visits, and low SRH over 1 and 2 years among older primary care patients at risk for disability. As a result, when considering the optimal screening tool for older adults, the choice between a measure of function and frailty may ultimately depend on clinical preference and context.