Key points are not available for this paper at this time.
BACKGROUND: For patients with osteoporosis, bisphosphonate therapy can reduce the risk of fractures, but its effect on reducing mortality remains unclear. Previous studies on this topic have produced conflicting results and generally have been too small to definitively answer the question of whether bisphosphonate therapy reduces mortality. Therefore, a meta-analysis may help us arrive at a more conclusive answer. QUESTIONS/PURPOSES: In a large meta-analysis of placebo-controlled randomized controlled trials (RCTs), we asked: (1) Does bisphosphonate use reduce mortality? (2) Is there a subgroup effect based on whether different bisphosphonate drugs were used (zoledronate, alendronate, risedronate, and ibandronate), different geographic regions where the study took place (Europe, the Americas, and Asia), whether the study was limited to postmenopausal female patients, or whether the trials lasted 3 years or longer? METHODS: We conducted a systematic review using multiple databases, including Embase, Web of Science, Medline (via PubMed), Cochrane Library, and ClinicalTrials.gov, with each database searched up to November 20, 2023 (which also was the date of our last search), following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We included randomized, placebo-controlled clinical trials with participants diagnosed with osteoporosis and receiving bisphosphonate treatment. We excluded papers posted to preprint servers, other unpublished work, conference abstracts, and papers that were registered on ClinicalTrials.gov but were not yet published. We collected 2263 records. After excluding records due to study type, study content not meeting the inclusion criteria, and duplicates, our meta-analysis included 47 placebo-controlled RCTs involving 59,437 participants. Data extraction, quality assessment, and statistical analyses were performed. The evaluation of randomized trials for potential bias was conducted using the revised Cochrane Risk of Bias tool. This assessment encompassed factors such as sequence generation, allocation concealment, subject blinding, outcome assessor blinding, incomplete outcome data, and reporting bias. Some studies did not provide explicit details regarding random sequence generation, leading to a high risk of selection bias. A few studies, due to their open-label nature, were unable to achieve double-blind conditions for both the subjects and the researchers, resulting in intermediate performance bias. Nevertheless, the overall study quality was high. Due to the low heterogeneity among the studies, as evidenced by the low statistical heterogeneity (that is, a low I 2 statistic), we opted for a fixed-effects model, indicating that the effect size is consistent across the studies. In such cases, the fixed-effects model can provide more precise estimates. According to the results of the funnel plot, we did not find evidence of publication bias. RESULTS: The use of bisphosphonates did not reduce the overall risk of mortality in patients with osteoporosis (risk ratio 0.95 95% CI 0.88 to 1.03). Subgroup analyses involving different bisphosphonate drugs (zoledronate, alendronate, risedronate, and ibandronate), regions (Europe, the Americas, and Asia), diverse populations (postmenopausal female patients and other patients), and trials lasting 3 years or longer revealed no associations with reduced overall mortality. CONCLUSION: Based on our comprehensive meta-analysis, there is high-quality evidence suggesting that bisphosphonate therapy for patients with osteoporosis does not reduce the overall risk of mortality despite its effectiveness in reducing the risk of fractures. The primary consideration for prescribing bisphosphonates to individuals with osteoporosis should continue to be centered on reducing fracture risk, aligning with clinical guidelines. Long-term studies are needed to investigate potential effects on mortality during extended treatment periods. LEVEL OF EVIDENCE: Level I, therapeutic study.
Lan et al. (Thu,) studied this question.