Preoperative SGLT2 Inhibitor Use and Postoperative Diabetic Ketoacidosis

Abstract

Importance: Case reports of postoperative diabetic ketoacidosis in patients using sodium-glucose cotransporter 2 inhibitor (SGLT2i) medications underlie guidance by the US Food and Drug Administration to withhold SGLT2i medication for at least 3 days prior to surgery. Given the potential negative consequences associated with preoperative medication withholding, a large-scale evaluation of the risk of diabetic ketoacidosis in this population is needed.

Objective: To estimate the association between preoperative SGLT2i medication use and postoperative diabetic ketoacidosis in a population of patients who underwent a variety of emergency surgeries. Emergency surgery was chosen given the assumption that a patient would be unable to withhold their SGLT2i medication per the current guidance.

Design, setting, and participants: This retrospective cohort study was conducted among a nationwide sample of patients aged 18 years or older with type 2 diabetes who were enrolled in commercial or Medicare fee-for-service insurance plans and who underwent 1 of 13 emergency surgeries between January 1, 2016, and December 15, 2022. Emergency surgeries were defined as those occurring on the same day or the 1 to 2 days after an emergency department claim. Data were analyzed from November 2023 through December 2024.

Exposure: SGLT2i medication use.

Main outcomes and measures: Diabetic ketoacidosis, defined by diagnosis codes, in the 0 to 14 days after surgery.

Results: Among 34 671 patients with type 2 diabetes who underwent emergency surgery (mean [SD] age, 63.9 [14.0] years; 19 175 female [55.3%] and 15 496 male [44.7%]), the most common surgeries were laparoscopic cholecystectomy (9385 patients) and transurethral procedures (12 246 patients). There were 2607 patients (7.5%) who used SGLT2i medications and 32 064 patients (92.5%) who did not. Unadjusted incidence of diabetic ketoacidosis was 127 patients (4.9%) for those exposed to SGLT2i medications and 1115 patients (3.5%) for those unexposed. After accounting for covariates, including demographic characteristics, indicators of diabetic severity, comorbidities, and surgery type, the incidence of the outcome was 3.8% for those exposed to SGLT2i medications and 3.5% for those unexposed. The average treatment effect [ATE] was 0.2% (95% CI, -1.7% to 2.2%). Results were robust to alternate specifications (eg, intensive care unit-level care as the outcome: ATE, -1.0%; 95% CI, -2.9% to 1.1%).

Conclusions and relevance: This study found that preoperative use of SGLT2i medications in patients undergoing emergency surgery was not associated with an increased risk for postoperative diabetic ketoacidosis compared with no use of SGLT2i medications. These findings may justify liberalizing current guidance on preoperative SGLT2i medication withholding periods.