E pluribus unum: Harmonization of physical functioning across intervention studies of middle-aged and older adults

Abstract

Common scales for physical functioning are not directly comparable without harmonization techniques, complicating attempts to pool data across studies. Our aim was to provide a standardized metric for physical functioning in adults based on basic and instrumental activities of daily living scaled to NIH PROMIS norms. We provide an item bank to compare the difficulty of various physical functioning activities. We used item response theory methods to place 232 basic and instrumental activities of daily living questions, administered across eight intervention studies of middle-aged and older adults (N = 2,556), on a common metric. We compared the scale’s precision to an average z-score of items and evaluated criterion validity based on objective measures of physical functioning and Fried’s frailty criteria. Model-estimated item thresholds were widely distributed across the range of physical functioning. From test information plots, the lowest precision in each dataset was 0.80. Using power calculations, the sample size needed to detect 25% physical functional decline with 80% power based on the physical functioning factor was less than half of what would be needed using an average z-score. The physical functioning factor correlated in expected directions with objective measurements from the Timed Up and Go task, tandem balance, gait speed, chair stands, grip strength, and frailty status. Item-level harmonization enables direct comparison of physical functioning measures across existing and potentially future studies and across levels of function using a nationally representative metric. We identified key thresholds of physical functioning items in an item bank to facilitate clinical and epidemiologic decision-making.