Sodium–Glucose Cotransporter 2 Inhibitors and the Risk of Fractures Among Patients With Type 2 Diabetes

The association between sodium–glucose cotransporter 2 (SGLT2) inhibitors and the risk of fractures is controversial. In the Canagliflozin Cardiovascular Assessment Study (CANVAS), canagliflozin was associated with a significant increased risk of fractures compared with placebo (hazard ratio [HR] 1.26, 95% CI 1.04–1.52) (1). Possible mechanisms may involve elevated serum phosphate levels or reductions in bone mineral density (2). To date, three recent observational studies investigated this association, but these did not observe an increased risk of fractures (3–5). However, these studies had some limitations, including residual confounding, a limited outcome definition, or restriction to a single SGLT2 inhibitor.

To address these limitations, we conducted a population-based cohort study using the U.K. Clinical Practice Research Datalink (CPRD). We first assembled a base cohort of all individuals, at least 40 years old, newly treated with antidiabetic drugs between 1 January 1988 and 31 December 2017. We excluded individuals with <1 year of medical history, those initially prescribed insulin in monotherapy, and women with polycystic ovary syndrome (alternate metformin indication) at the time of the first-ever prescription of an antidiabetic drug. Using this base cohort, we then assembled a study cohort of individuals who initiated a new antidiabetic drug class as of 2013 (year the first SGLT2 inhibitor was introduced in the U.K.). Cohort entry was defined by the date of this new prescription. We excluded individuals with a diagnosis of Paget disease, osteomalacia, or hyperparathyroidism, rare risk factors associated with fracture incidence, at any time before cohort entry. Patients were monitored from cohort entry until the first of an incident diagnosis of fracture, death from any cause, an incident diagnosis of Paget disease, osteomalacia, or hyperparathyroidism, end of registration with the general practice, or end of the study period (31 March 2018).

We defined exposure using a time-varying definition, where each person-day of follow-up was classified into one of the following four mutually exclusive categories: 1) current use of SGLT2 inhibitors; 2) current use of dipeptidyl peptidase 4 (DPP-4) inhibitors (with initiation after 2013; i.e., new use); 3) current use of DPP-4 inhibitors (with initiation before 2013; i.e., prevalent use); and 4) other treatments. Current and new use DPP-4 inhibitors served as the reference group. For all exposure categories, exposed person-time was defined by the prescription duration plus a 90-day grace period to allow for a potential latent exposure effect on the outcome. Time-dependent Cox proportional hazards models were used to calculate HRs, adjusted for 45 covariates (Table 1), including smoking status, BMI, medications, and duration of treated diabetes. This study was approved by the CPRD Independent Scientific Advisory Committee (protocol 18_221) and by the Jewish General Hospital Research Ethics Board.

This study included 73,178 patients, with 9,454 SGLT2 inhibitor users (1,288 canagliflozin, 5,539 dapagliflozin, 2,133 empagliflozin, and 494 users of >1 SGLT2 inhibitor) and 18,410 DPP-4 inhibitor users, monitored for a median duration of 1.9 years (patient characteristics on file). There were 1,973 fracture events, corresponding to an incidence rate of 12.88 (95% CI 12.32–13.46) per 1,000 person-years.

The results of this study are presented in Table 1 . Compared with use of DPP-4 inhibitors, use of SGLT2 inhibitors was not associated with fracture risk (HR 0.97, 95% CI 0.79–1.19). When stratifying by SGLT2 molecule, use of canagliflozin was associated with a decreased risk of fractures (HR 0.47, 95% CI 0.25–0.88), but this finding was based on few events (n = 10). There were no differences in risk according to fracture type, except for a numerically elevated HR with vertebral fractures (HR 1.76, 95% CI 0.81–3.82). Similarly, the association was not modified by history of fracture, osteoporosis, age, or sex. Finally, the results remained consistent in several sensitivity analyses investigating death as a competing risk (using the Fine and Gray method), when controlling for confounding using the high-dimensional disease risk score, when using a marginal structural model with inverse probability of treatment and censoring weighting, or when comparing to glucagon-like peptide 1 receptor agonists, as was done in a previous study (5).

In summary, the results of this large population-based cohort study indicate that use of SGLT2 inhibitors is not associated with an increased risk of fractures compared with use of DPP-4 inhibitors. Although our stratified analysis by SGLT2 molecule revealed a protective effect for canagliflozin, this finding should be interpreted with caution given that it was based on few exposed events. Overall, our real-world study provides further reassurance on the safety of this new class of drugs on bone health among patients with type 2 diabetes.