Thirty RCTs (59,209 participants) were included in the review. There were some discrepancies between results reported in the text, tables and forest plots and we have used figures from the forest plots.
Non-vertebral fractures: Both classical and Bayesian analyses indicated that, compared to placebo, alendronate and risedronate showed statistically significant reductions in non-vertebral fractures. The drugs with the highest probability of being most efficacious were etidronate (0.42) and teriparatide (0.41), although etidronate was not significantly more efficacious compared to placebo. The drugs with the greatest effect sizes were risedronate (16.4) and alendronate (16.1).
Vertebral fractures: Both classical and Bayesian analyses showed that, with the exception of etidronate, all drugs significantly reduced vertebral fractures compared to placebo. The drugs with the highest probability of being most efficacious were teriparatide (0.30), zoledronic acid (0.40) and denosumab (0.20); these drugs also had the greatest effect sizes.
Hip fractures: Classical analysis indicated statistically significant reductions in hip fractures for alendronate (OR 0.62, 95% CI 0.40 to 0.96; six RCTs), and risedronate (OR 0.74, 95% CI 0.58 to 0.94; four RCTs). The drug with the highest probability of being most efficacious was teriparatide (0.44); the drug with the greatest effect size was alendronate (9.49).
Wrist fractures: Classical analysis showed that out of the drugs used, only alendronate (OR 0.44, 95% CI 0.30 to 0.67; six RCTs) and risedronate (OR 0.71, 95% CI 0.56 to 0.89; four RCTs) showed significant reductions in wrist fractures compared to placebo. The drugs with the highest probability of being most efficacious were teriparatide (0.41) and risedronate (0.22); the drugs with the greatest effect sizes were alendronate (1.80) and risedronate (1.37).
There was evidence of moderate statistical heterogeneity for the overall results for vertebral fractures (I²=65%) and wrist fractures (I²=59%). Indirect comparisons for non-vertebral fractures, hip fractures, and wrist fractures all showed no statistically significant differences in efficacy between different drugs.
Statistically significant reductions in fracture outcomes were also reported in relation to other drugs, but findings were based on only one or two trials (fully reported in the review). Other results, including meta-regression analyses and numbers needed to treat/harm, were fully reported in the review.