Background: Physical performance measures may offer advantages over self-report in the functional assessment of older people. Estimates of the feasibility, reliability, and construct validity of these measures in large, heterogeneous samples are necessary to establish their importance relative to traditional measures of function.
Methods: Analysis of clinical data from Phase 2 of the Canadian Study of Health and Aging, a nation-wide representative survey of elderly people in Canada (N = 2,305).
Results: Both physical performance measures proved infeasible in many subjects (29.3% for the Timed Up and Go [TUG], 35.9% for the Functional Reach [FR]). Cognitive impairment was the most important determinant of inability to complete the tests. For those able to complete the tests, cognitively unimpaired subjects could reach farther (median 29 cm) and complete the TUG in less time (median 12 seconds) than those cognitively impaired (25 cm for FR, 15 seconds for the TUG). Test-retest reliability between the screening and clinical administrations of the TUG was .56 for all participants (intra-class correlations), .50 for the cognitively unimpaired, and .56 for the cognitively impaired. Construct validity was substantial, and correlations between performance measures and self-report activities of daily living (ADL) measures ranged from .40 to .70. Compared with a global clinical measure of frailty, correlations were more modest (.38 to .60).
Conclusions: The FR and the TUG were not feasible tools in this study. The TUG showed poor test-retest reliability. Our data support the observation that subsequent studies of measurement instruments typically reveal lower performance than the original reports.