1-year outcomes of semaglutide, tirzepatide, and sleeve gastrectomy in obesity in type 2 diabetes: a retrospective cohort study

BACKGROUND Type 2 diabetes and obesity are prevalent, inter-related conditions for which treatment options have expanded substantially, yet direct real-world comparisons of modern pharmacological and surgical interventions remain scarce. We aimed to compare 1-year weight loss, glycaemic outcomes, and selected safety outcomes associated with semaglutide, tirzepatide, and sleeve gastrectomy among adults with obesity and type 2 diabetes in real-world clinical practice. METHODS We conducted a retrospective cohort study using Epic Cosmos, a de-identified electronic health record dataset from 1633 hospitals and 37 900 clinics across 280 health systems in the USA. Adults with a BMI of at least 35 kg/m2 and HbA1c above 6·4% who initiated at least 12 months of therapy with semaglutide or tirzepatide and attaining maintenance doses, or who underwent sleeve gastrectomy, were eligible for inclusion. The primary outcome was attainment of at least 20% bodyweight loss and HbA1c below 5·7% at 1 year after the index date. Multivariable logistic regression estimated adjusted probabilities and relative risks, adjusting for age, sex, race, baseline HbA1c, BMI, Charlson Comorbidity Index, Social Vulnerability Index (SVI 2020), and baseline medication burden for diabetes, lipids, and hypertension. Prespecified safety outcomes included emergency department visits and new prescriptions for gastro-oesophageal reflux disease or nausea within 1 year of the index date. The primary composite outcome was analysed among patients with both follow-up weight and HbA1c available in the 9-month to 15-month window. Emergency department visits were analysed in the final analysis cohort (ie, patients meeting all inclusion and exclusion criteria and with available baseline and 9-month to 15-month weight and HbA1c measurements) and medication-based safety outcomes were analysed descriptively using outcome-specific available data denominators without imputation. FINDINGS A total of 468 712 patients with at least 365 consecutive days of semaglutide or tirzepatide prescriptions and 238 028 patients undergoing sleeve gastrectomy were identified in Epic Cosmos. After applying prespecified inclusion and exclusion criteria, 45 649 patients met inclusion criteria; after exclusion of 556 patients with missing Charlson Comorbidity Index data, 45 093 were included in the primary outcome analysis: 33 482 (74·3%) treated with semaglutide, 4178 (9·3%) treated with tirzepatide, and 7433 (16·5%) who underwent sleeve gastrectomy. The adjusted probability of attaining the primary composite outcome of at least 20% bodyweight loss and HbA1c below 5·7% at 1 year was 3·0% (95% CI 2·8-3·2) for semaglutide, 13·2% (12·2-14·3) for tirzepatide, and 24·0% (22·9-25·1) for sleeve gastrectomy. Emergency department visits within 1 year were more frequent after sleeve gastrectomy than after semaglutide or tirzepatide. New prescriptions for gastro-oesophageal reflux disease and nausea occurred in all treatment groups but varied in frequency, with higher absolute rates observed after sleeve gastrectomy. INTERPRETATION At 1 year after the index date, sleeve gastrectomy was associated with the highest probability of attaining combined weight and glycaemic targets, followed by tirzepatide and semaglutide. Interpretation should consider baseline differences-including higher BMI, younger age, and lower baseline HbA1c in the sleeve gastrectomy group-as well as residual confounding and the limited scope of safety assessment. FUNDING None.