Abstract
PURPOSE: To investigate the comparative efficacies of the five most commonly used bisphosphonates for the secondary prevention of osteoporotic fractures in a Bayesian network meta-analysis. METHODS: Five databases and the reference lists of all acquired articles from inception to July 2017 were searched. A Bayesian random-effects model was employed, and vertebral, hip and nonvertebral nonhip fractures were assessed by odds ratios (ORs) and 95%credible intervals. Furthermore, with respect to each endpoint, rank probabilities for each bisphosphonate were evaluated using the surface under the cumulative ranking curve (SUCRA) value. RESULTS: Thirteen eligible studies were identified involving 11,822 patients with osteoporotic fractures. Overall in the pairwise meta-analyses, bisphosphonate use significantly reduced the risk of new vertebral, hip, and nonvertebral nonhip fractures, with ORs and 95% confidence intervals of 0.56 (0.49-0.64), 0.69 (0.48-0.98), and 0.82 (0.70-0.97), respectively. In network meta-analyses, significant differences were found between placebo and any one of the five bisphosphonates for new vertebral fractures. The rank probability plot and the SUCRA calculation results suggested that alendronate was the best intervention (14.6%) for secondary prevention of vertebral fractures, followed by zoledronate (15.3%) and etidronate (22.1%). In terms of the incidence of new hip fractures, alendronate was associated with the lowest incidence (18.5%), followed by zoledronate (43.1%) and risedronate (52.5%). However, zoledronate ranked lowest (16.6%) regarding the incidence of new nonvertebral nonhip fractures, followed by risedronate (23.8%) and alendronate (44.1%). CONCLUSIONS: Bisphosphonates show significant efficacy for secondary prevention of new vertebral fractures, and alendronate is most likely to be successful at secondary prevention of vertebral and hip fractures compared with the other four bisphosphonates.