Twelve RCTs (n=18,667) were included.
Nine of the studies were double-blind and one additional study was double-blind in its third year.
No statistically significant heterogeneity was detected (Cochran Q, p=0.998; I-squared 0.88%). The regression analysis showed no evidence of publication bias (p=0.116).
The meta-analysis showed that bisphosphonates were associated with a statistically significant reduction in the risk of hip fracture compared with control (RR 0.58, 95% CrI: 0.42, 0.80). Analyses of individual drugs showed no significant difference in hip fracture between individual bisphosphonates and control.
The absolute risk reduction for studies with a 3-year follow-up was 0.0052 (95% CrI: 0.0004, 0.011); this represented a reduction of 52 hip fractures per 10,000 women (95% CrI: 4, 110).
Sensitivity analyses.
For all types of prior distributions examined, the results were statistically significant in favour of bisphosphonates.
After the first three studies were included in the cumulative analysis, the addition of other studies did not influence the results.