Answers are grounded in this paper and cite their sources.
Background: Sodium-glucose cotransporter-2 (SGLT2) inhibitors reduce cardiovascular and renal events in type 2 diabetes mellitus but are associated with a rare form of diabetic ketoacidosis (DKA), which may occur with only modest hyperglycaemia. This review evaluated the relative and absolute risk of DKA in large placebo-controlled cardiovascular and renal outcome trials and explored whether the signal differed across major trial populations. Methods: Eight large, placebo-controlled phase 3 cardiovascular or renal outcome trials of empagliflozin, dapagliflozin, canagliflozin, or ertugliflozin that reported arm-level DKA event counts were included. Rare-event data were pooled using the Peto one-step odds ratio, with a Mantel-Haenszel odds ratio as a sensitivity analysis. A prespecified subgroup analysis compared heart-failure trials with the remaining trial populations. Heterogeneity was assessed using Cochran's Q and I². Observational and pharmacovigilance evidence was synthesised narratively and was not pooled with randomised trial data. Results: Across 55,385 randomised participants, DKA occurred in 74 of 30,245 SGLT2 inhibitor-treated patients (0.245%) and 24 of 25,140 placebo-treated patients (0.095%). The pooled Peto odds ratio was 2.39 (95% CI 1.60-3.57; p<0.0001), while the Mantel-Haenszel sensitivity analysis yielded an odds ratio of 2.66 (95% CI 1.67-4.24). Heterogeneity was moderate (I²=34%). In the two heart-failure trials, the pooled estimate was 1.40 (95% CI 0.45-4.35), compared with 2.58 (95% CI 1.68-3.97) in the remaining six trials; however, the subgroup interaction was not statistically significant (p=0.32). The absolute risk difference was 0.149 percentage points, corresponding to an approximate number needed to harm of 670 over the multi-year trial durations. Conclusions: In large placebo-controlled outcome trials, SGLT2 inhibitors were associated with an increased relative risk of DKA, while the absolute incidence remained low. The numerically smaller estimate in heart-failure trials should be interpreted cautiously because the subgroup interaction was not statistically significant and event counts were small. Clinicians should maintain a low threshold for checking acid-base status and beta-hydroxybutyrate in patients receiving SGLT2 inhibitors who present with unexplained metabolic acidosis, nausea, tachypnoea, or malaise, even when blood glucose is not markedly elevated.
No takes yet. Share an insight, caveat, or question.
Rana et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: