Fifteen RCTs were included in the review (n=17,786), of which eight double-blind RCTs were included in the primary analysis (n=2,426). Adherence ranged between 68% and 100%; seven of eight trials reported rates above 80%.
The overall pooled relative risk (RR) for vitamin D's effect on fall prevention was statistically significant at 0.87 (95% CI: 0.77 to 0.99; eight RCTs). But, statistically significant heterogeneity was detected (Q test, p = 0.05). Stratification of trials by dose of vitamin D (200 to 600 IU versus 700 to 1000 IU) resolved the heterogeneity.
There was a statistically significant benefit in trials that used higher doses of vitamin D (RR 0.81, 95% CI: 0.71 to 0.92; seven RCTs; NNT 11, 95% CI: 7 to 20), but no such effect was found in trials that used lower doses. A statistically significant benefit was found for active forms of vitamin D (RR 0.78, 95% CI: 0.64 to 0.94; two RCTs, n=624).
The ratio of the two effect sizes (high dose supplemental vitamin D: active forms of vitamin D) was 1.04 (95% CI: 0.84 to 1.31). There was a statistically significant benefit in trials in which serum 25-hydroxyvitamin D3 levels above 60n/mol were achieved (RR 0.77, 95% CI: 0.65 to 0.90); there was no statistically significant benefit in trials that achieved concentrations below this level.
There were no other statistically significant results of subgroup analyses. A sensitivity analysis that included seven trials which did not meet the criteria for inclusion in the primary analysis found a non-significant benefit in fall reduction; adding these seven to the analysis of high-dose trials gave a statistically significant result (RR 0.92, 95% CI: 0.85 to 0.99) with highly significant statistical heterogeneity (p=0.006).
There was no consistent evidence of publication bias.