Preoperative GLP-1 receptor agonist continuation and perioperative safety after thyroidectomy in type 2 diabetes: a target trial emulation using a retrospective multicenter cohort

Glucagon-like peptide-1 receptor agonists (GLP-1 RA) are increasingly prescribed in adults with type 2 diabetes (T2D) presenting for elective surgery, yet procedure-specific perioperative safety data for thyroidectomy remain limited. We evaluated whether preoperative GLP-1 RA continuation is associated with aspiration, thyroidectomy-specific complications, or metabolic events. Using the TriNetX US Collaborative Network (2017 to 2025), we emulated three active-comparator target trials in adults with T2D undergoing thyroidectomy, comparing preoperative GLP-1 RA with DPP-4 inhibitors (DPP-4i), SGLT2 inhibitors (SGLT2i), or usual care. Cohorts were 1:1 propensity score matched; Cox models estimated hazard ratios (HRs). The primary outcome was 24-month all-cause mortality; secondary outcomes included diabetic ketoacidosis (DKA), readmission, hypocalcemia, recurrent laryngeal nerve injury, and class-relevant safety events. Aspiration-related events and a broader postoperative respiratory composite were prespecified and analysed over three postoperative windows. Negative-control outcomes probed residual confounding. Matched cohorts comprised 722, 736, and 611 patients per arm. Aspiration-specific events were below the minimum reportable cell size in all comparisons and windows; a broader postoperative respiratory composite showed no excess. No excess recurrent laryngeal nerve injury was observed. Hypocalcemia was more frequent with GLP-1 RA than with SGLT2i in the primary analysis without reaching significance (HR 1.43, 95% CI 0.96–2.12) and reached nominal significance in the 2-year exposure-window and new-user analyses. GLP-1 RA was associated with lower 24-month DKA versus DPP-4i (HR 0.56, 95% CI 0.34–0.91) and lower all-cause mortality versus DPP-4i (0.49, 0.26–0.92) and SGLT2i (0.39, 0.19–0.78). The lower DKA risk was reproduced in the exposure-window and new-user designs; the mortality findings were not. All were significant at the nominal level only and none survived correction for multiple comparisons. Mortality over postoperative days 1 to 30 and 1 to 90 was below the minimum reportable cell size in every comparison. Negative-control outcomes were null. In adults with T2D undergoing thyroidectomy, preoperative GLP-1 RA continuation was not associated with excess perioperative harm. The lower DKA and all-cause mortality risks were significant at the nominal level only, did not survive correction for multiple comparisons, and are hypothesis-generating. Not applicable. This study emulated target trials using retrospectively collected, de-identified electronic health records and was not a prospective clinical trial, so no trial registration number applies.