Fifty-three studies (12,832 people, range 12 to 1,200 people) were included in the review; 25 had prospective designs. Follow-up ranged from four weeks to five years. Nine studies used representative samples and 14 studies followed the recommendations for falls data collection. Eight studies were considered high quality trials; four of these included representative samples.
The mean difference in Timed Up and Go performance was small but statistically significant. Performance was slower in healthy higher-functioning fallers compared to non-fallers (MD 0.63 seconds, 95% CI 0.14 to 1.12; seven studies; moderate heterogeneity Ι²=52%).
Timed Up and Go times were statistically significantly slower in fallers compared to non-fallers in studies of higher- and lower-functioning people living independently (MD 2.05 seconds, 95% CI 1.47 to 2.62; 21 studies; Ι²=61%) and slower in fallers compared to non-fallers in institutional settings (MD 3.59 seconds, 95% CI 2.18 to 4.99; nine studies; Ι²=0%).
The diagnostic accuracy of Timed Up and Go tests in correctly identifying fallers was poor to moderate across studies (results were reported for individual studies). Other results were reported in the review.