Total shoulder arthroplasty (TSA) is increasingly common and carries inherent perioperative risks. Diabetes is a recognized risk factor for postoperative complications after joint arthroplasty. Sodium-glucose cotransporter 2 inhibitors (SGLT2i) are widely used antihyperglycemic agents, but their safety in the perioperative period for patients undergoing TSA had not been evaluated prior to this analysis. The objective of this retrospective study was to assess the association between perioperative SGLT2i use and both short-term medical complications (90 days) and longer-term implant-related complications (2 years) in diabetic patients undergoing TSA.
The authors queried the TriNetX research network to identify diabetic patients who underwent total shoulder arthroplasty. Patients were stratified according to prescription status for SGLT2i. The study relied on available electronic health record and claims-derived data within TriNetX to assemble exposure cohorts and outcome information. Specific inclusion and exclusion criteria, and the exact temporal window defining perioperative SGLT2i exposure, were reported in the full text but are not detailed in the abstract.
To reduce confounding, the investigators performed 1:1 propensity score matching between the SGLT2i and control cohorts adjusting for relevant demographic and medical comorbidities. After matching, each cohort consisted of 2,360 patients, yielding balanced groups for comparative outcome analysis. The abstract indicates matching was used to control for key covariates, although the complete list of matching variables and post-match balance diagnostics were not provided in the abstract.
The predefined outcomes included 90-day medical and surgical complications and implant-related complications assessed up to 2 years postoperatively. The 90-day window covered acute medical events and common postoperative surgical complications. The 2-year assessment targeted implant-related complications that might necessitate revision or represent prosthesis failure or other long-term adverse outcomes.
After propensity score matching, the primary finding reported in the abstract was an increased 90-day risk of myocardial infarction in the SGLT2i cohort compared with control subjects: 2.03% versus 0.89% (P = 0.001). This difference reached statistical significance in the matched analysis.
No significant differences between the SGLT2i and control cohorts were observed in the 90-day incidence of the following outcomes as reported in the abstract: ketoacidosis, deep vein thrombosis, pulmonary embolism, stroke, pneumonia, acute kidney injury, transfusion, wound disruption, sepsis, or hospital readmission. These null findings apply to the outcomes explicitly listed in the abstract.
At 2 years postoperatively, the authors report no increased risk of implant-related complications in the SGLT2i cohort compared with controls. In other words, perioperative prescription of SGLT2i was not associated with a higher rate of implant-related adverse events within the 2-year follow-up window described in the abstract.
The study’s primary interpretation is that perioperative SGLT2i use among diabetic patients undergoing TSA was associated with a higher 90-day incidence of myocardial infarction, while not showing increased risk for the range of other early medical or surgical complications assessed, nor for implant-related complications at 2 years. The authors suggest that these findings may inform preoperative counseling and perioperative optimization for diabetic patients taking SGLT2i who are scheduled for TSA. They also state that additional research is warranted to clarify the observed associations.
The abstract reports a Level of Evidence III. The study design is a retrospective propensity-matched cohort analysis using the TriNetX database. The abstract does not provide exhaustive methodological details such as the full list of covariates included in matching, handling of medication timing relative to surgery, or potential residual confounding. Those specifics and other limitations are likely discussed in the full text but were not included in the abstract.