A total of 25 diagnostic cohort studies (n>5,920) were included in the review. Methodological quality was variable. For example, most studies (21 out of 25) reported details of screening test procedures, but few (eight out of 25) reported results for all the screening tests performed. Ten of the 25 studies had errors in their reported analyses. Loss to follow-up ranged from 0% to 33.5%.
Eight studies evaluated the Tinetti gait, balance and mobility scales. Sensitivity among five studies that evaluated the overall mobility scale ranged from 0.27 to 0.76 and specificity ranged from 0.52 to 0.83. Seven studies that evaluated the balance scale alone reported sensitivity values of 0.08 to 0.80 and specificity values of 0.66 to 0.97; one study reported an AUC of 0.56. Four studies evaluating the gait scale alone and reported sensitivity values of 0.20 to 0.68 and specificity values of 0.63 to 0.95.
Three studies of the TUG test reported sensitivity values of 0.10 to 0.54 and specificity values of 0.73 to 0.95; one study reported an AUC of 0.61. Tests and cut-offs varied between these studies.
Results of studies that evaluated Mobility Interaction Fall Chart, functional reach, tandem stance and walking tests were inconsistent (two studies each).
All other tests were evaluated in single studies.