Thirteen studies (n=1,473, range 39 to 374) were included in the review. Four of the included studies used the same population, but reported different periods of follow-up. Validity scores ranged from 4 to 10 points. The authors reported the main biases as lack of blinding of therapists and participants, lack of allocation concealment and lack of intention-to-treat analysis.
There were no significant differences between exercise and usual care on both fall rates (rate ratio 1.22, 95% CI 0.76 to 1.98; two RCTs) and proportion of fallers (risk ratio 0.77, 95% CI 0.24 to 2.43; two RCTs). There were no significant differences between bisphosphonate and placebo for proportion of fallers (risk ratio 0.95, 95% CI 0.73 to 1.22; two studies). There were not enough data to enable pooling of other outcomes.
For the analysis of individual studies, only one RCT reported a significant reduction in fall rate (p=0.0002) and proportion of fallers (p=0.004) for vitamin D compared with placebo for female stroke survivors. There were no other significant differences between interventions and control in the remaining RCTs.