Diabetic ketoacidosis (DKA) has been reported with sodium–glucose cotransporter 2 inhibitors, but comparative risk estimates in routine clinical practice remain limited. A retrospective new-user cohort study published in the Diabetic Medicine evaluated DKA risk among patients with type 2 diabetes initiating ertugliflozin compared with sulfonylureas (SU), thiazolidinediones (TZD), or incretin-based therapies using Medicare and Medicaid fee-for-service adjudicated claims within the Innovation in Medical Evidence and Development Surveillance (IMEDS) network.
After 1:1 propensity score matching, 42,907 matched pairs were analyzed for ertugliflozin versus SU/TZD and 42,247 matched pairs for ertugliflozin versus incretin-based drugs. Baseline characteristics were balanced after matching. DKA incidence rates per 1000 person-years were 2.95 for ertugliflozin and 1.49 for SU/TZD, and 2.76 for ertugliflozin and 1.06 for incretin-based drugs.
For ertugliflozin versus SU/TZD, the overall hazard ratio (HR) for DKA was 1.88 (95% CI 1.17–3.02). Among patients not using insulin at baseline, the HR was 2.34 (95% CI 1.27–4.31). In contrast, among insulin users, the HR was 1.17 (95% CI 0.54–2.52). For ertugliflozin versus incretin-based drugs, the overall HR was 2.40 (95% CI 1.40–4.11). Among non-insulin users, the HR was 2.84 (95% CI 1.42–5.66). Among insulin users, the HR was 1.87 (95% CI 0.79–4.46).
As an observational claims-based study, findings demonstrate association rather than causation. The findings may be influenced by residual confounding or misclassification.