Predictive validity and responsiveness of patient-reported and performance-based measures of function in the Boston RISE study.

Abstract

Patient-reported and performance-based measures (PBMs) are commonly used to measure physical function in studies of older adults. Selection of appropriate measures to address specific research questions is complex and requires knowledge of relevant psychometric properties. The aim of this study was to examine the predictive validity for adverse outcomes and responsiveness of a widely used patient-reported measure, the Late-Life Function and Disability Instrument (LLFDI), compared with PBMs. We analyzed 2 years of follow-up data from Boston RISE, a cohort study of 430 primary care patients aged ≥65 years. Logistic and linear regression models were used to examine predictive validity for adverse outcomes and effect size and minimal detectable change scores were computed to examine responsiveness. Performance-based functional measures included the Short Physical Performance Battery, 400-m walk, gait speed, and stair-climb power test. The LLFDI and PBMs showed high predictive validity for poor self-rated health, hospitalizations, and disability. The LLFDI function scale was the only measure that predicted falls. Absolute effect size estimates ranged from 0.54 to 0.64 for the LLFDI and from 0.34 to 0.63 for the PBMs. From baseline to 2 years, the percentage of participants with a change ≥ minimal detectable change was greatest for the LLFDI scales (46-59%) followed by the Short Physical Performance Battery (44%), gait speed (35%), 400-m walk (17%), and stair-climb power test (9%). The patient-reported LLFDI showed comparable psychometric properties to PBMs. Our findings support the use of the LLFDI as a primary outcome in gerontological research.