Background and importance The Canagliflozin Cardiovascular Assessment Study (CANVAS) indicated that use of sodium glucose cotransporter 2 inhibitors (SGLT2i) may increase the risk of fracture compared with placebo. However, the association has remained uncertain in previous real world studies, and many factors (eg, severity of diabetes) were not considered in the analyses.
Aim and objectives To evaluate the risk of fracture associated with the use of SGLT2i in type 2 diabetes patients.
Material and methods We conducted a retrospective cohort study by analysing the electronic medical record (EMR) data from the Chang Gung Research Database (CGRD), covering 1.3 million people in Taiwan (6% of the population). We selected type 2 diabetes patients newly receiving SGLT2i from 2016 to 2018. Dipeptidyl peptidase 4 inhibitor (DPP4i) was considered the active comparator as no association has been found between DPP4i and fractures. Using the propensity score derived from the patient’s age, sex, baseline body mass index (BMI), systolic blood pressure, HbA1c, renal function, comorbidities and co-medications, we matched SGLT2i new users 1:1 with DPP4i new users. The primary outcome was hospitalisation due to fracture at any site. We followed patients from initiation of SGLT2i or DPP4i to fracture hospitalisation, death, last clinical visit or 31 December 2019. We used the Cox proportional hazard to estimate the fracture risk associated with SGLT2i and DPP4i use.
Results We identified 10 736 patients for each group receiving SGLT2i or DPP4i. Mean age was 59.3 (SD 12.5) years and 39.6% were women. Mean BMI, HbA1c and eGFR were 28.0 (SD 4.7) kg/m2, 8.8 (SD 1.9)% and 94.7 (SD 31.8) mL/min/1.73 m2, respectively. The incidence rates of fracture were 3.1 and 4.2 per 1000 person years for SGLT2i and DPP4i users, respectively. The risk of fracture was similar for SGLT2i users (HR 0.79, 95% CI 0.51 to 1.23) and DPP4i users.
Conclusion and relevance Different from the findings of CANVAS, our analysis of real world EMR data in Taiwan did not reveal any positive association between SGLT2i and fracture risk.
Conflict of interest No conflict of interest