Association of bisphosphonates with diabetes risk and glycemic control: a meta-analysis

Summary: Previous evidence suggests that bisphosphonates may improve glycemic control. The present meta-analysis, comprising seven studies with 1,233,844 participants, demonstrated that bisphosphonate use was significantly associated with a lower risk of diabetes. However, in the randomized controlled trial subgroup, a non-significant association was found. Further studies are needed to determine causality. Purpose: This study aimed to evaluate the impact of bisphosphonates on glycemic control and the risk of incident diabetes. Methods: MEDLINE, Embase, and Cochrane Library were searched from inception to February 15, 2022. Experimental or observational studies that compared fasting blood glucose (FBG) and glycated hemoglobin (HbA1c) levels and the diabetes risk with and without bisphosphonates were included. Studies without relevant outcomes, only providing crude estimates, or the absence of a control group were excluded. Two reviewers independently screened the articles, extracted data, and appraised studies. The pooled relative risk (RR) and weighted mean difference (WMD) were calculated using random effects models. Results: Seven studies (n = 1,233,844) on diabetes risk were included, including two post hoc analyses of randomized controlled trials (RCTs) and five observational studies. Compared with controls, bisphosphonates (BPs) were associated with a significant decrease in the risk of diabetes (RR = 0.77; 95% CI, 0.65 to 0.90; P = 0.002). However, in the subgroup of post hoc analyses of RCTs, the association was non-significant (RR = 0.93; 95% CI, 0.74 to 1.18; P = 0.576). Moreover, three studies (n = 4906) on FBG and one (n = 60) on HbA1c were included. We observed non-significant association between BPs and changes in FBG (WMD = − 0.61 mg/dL; 95% CI, − 2.72 to 1.49; P = 0.567) and HbA1c (WMD = − 0.11%; 95% CI, − 0.23 to 0.01; P = 0.083). Conclusion: Patients taking BPs may have a lower risk of incident diabetes than those without BPs. However, due to the high between-study heterogeneity and inconsistent findings between post hoc analyses of RCTs and observational studies, further rigorous RCTs are required to determine whether the findings are causal.