Thirty-four RCTs (73,464 women) were included in the review. Five RCTs scored 5 on the Jadad scale, nine scored 4, 11 scored 3, four scored 2 and five scored 1.
Direct comparisons: All osteoporosis treatments except etidronate showed statistically significant reductions in risk of new vertebral fractures compared to placebo (RR ranged from 0.30 to 0.72). Denosumab, risedronate and zoledronic acid also statistically significantly reduced risk of non-vertebral and hip fractures compared to placebo. Alendronate, strontium ranelate and teriparatide showed statistically significant reductions in risk compared to placebo for non-vertebral fractures. There were no statistically significant differences in risk of wrist fractures for any treatment compared to placebo.
Mixed treatment comparisons: The findings for each comparator versus placebo were consistent with direct comparisons for new vertebral fractures. Compared to placebo, risk of non-vertebral fracture was only statistically significantly reduced with teriparatide (RR 0.47, 95% CrI 0.22 to 0.90) and risedronate (RR 0.80, 95% CrI 0.65 to 0.95).
Mixed treatment comparisons showed that denosumab was statistically significantly more effective in preventing new vertebral fractures compared to strontium ranelate, raloxifene, alendronate and risedronate (RR ranged from 0.45 to 0.56). These reflected the adjusted indirect comparison results. No other statistically significant differences between active treatments were identified. Mixed treatment comparisons and indirect comparisons that compared osteoporosis treatments were generally consistent for risk of non-vertebral fracture.
Sensitivity analyses did not significantly alter the findings. Findings from meta-regression were reported in the review.