The Association of SGLT2i vs DPP4i on Fracture: A Cohort Study in Veterans with Diabetes

Increased bone turnover is associated with use of SGLT2i. Patients with diabetes experience adverse effects on bone metabolism. Our aim was to evaluate if SGLT2i was associated with fractures vs DPP4i as add-on therapy to diabetes regimens.

We assembled a retrospective cohort of Veterans with diabetes combining Veterans Administration, Medicare, and National Death Index databases. Using an active comparator new user design, patients starting on SGLT2i or DPP4i were followed from prescription fill until a fracture event, death, stopping medication, loss of follow-up, or study end. Fractures included: face/skull, spine, ribs, long bones, hand/feet/digits, or hip. Fractures were identified based on a validated algorithm with positive predictive value 91.3% (86.8, 94.4). Cox models compared the association of fractures between SGLT2i and DPP4i in a propensity score-weighted cohort that balanced 70+ covariates including comorbidities, vital signs, labs, vitamin D levels, smoking, and medications.

The unweighted sample included 115,124 SGLT2i episodes (104,086 Veterans; 94% empagliflozin; 4% canagliflozin; 2% dapagliflozin) and 213,095 DPP4i episodes (173,724 Veterans; 45% saxagliptin; 15% sitagliptin; 34% alogliptin; 6% Linagliptin). After propensity score calculation and matched weighting, the cohort included 76,072 SGLT2i and 75,833 DPP4i episodes. Median age was 69.3 years and diabetes duration 9.7 (6.1, 14.0) years. In the matched weighted analyses, there were 1431 and 1564 fractures among SGLT2i and DPP4i users, respectively. There were no clinical differences in fractures per 1000 person-years: 18.2 (17.4, 19.1) vs 19.8 (19.0, 20.6). The adjusted hazard ratio (adjusted hazard ratio 0.93 [0.87, 0.99]) excluded increased risk of fractures (adjusted hazard ratio > 1) in SGLT2i users.

SGLT2i use as add-on treatment for diabetes was not associated with increased fracture outcomes compared to DPP4i.