In a huge U.S. health‑record study, people with type‑2 diabetes who started the weight‑loss drug tirzepatide (or semaglutide) were about half as likely to develop a new alcohol‑use disorder over the next 18 months compared to those on older diabetes pills. The effect was strongest for tirzepatide, which also beat liraglutide in a head‑to‑head look.

There is growing evidence that the GLP-1 system is implicated in alcohol and other substance use disorders, and that GLP-1-based therapies may have therapeutic relevance in alcohol use disorder (AUD). We aimed to determine the impact of GLP-1 based therapies on incident AUDs in a real-world setting in patients with T2D. We conducted emulation target trials based on a real-world network of electronic health records (EHRs) from over 120 million patients in the United States of America. Four target trials were emulated among eligible patients with type 2 diabetes (T2D) who had no prior AUD diagnosis by comparing tirzepatide, semaglutide, liraglutide, and dulaglutide with DPP4 inhibitors (DPP4i). First-ever diagnosis of AUD occurred within an 18-month follow-up period and was examined using Kaplan-Meier survival analyses. Four target trial cohorts were generated and compared with a reference arm of patients treated with DPP4i: cohort (1) treatment with tirzepatide; cohort (2) treatment with semaglutide; cohort (3) treatment with liraglutide; and cohort (4) treatment with dulaglutide. Cohorts underwent propensity score matching 1:1 for confounders. We examined rates of incident AUD (ICD-10 code F10) and performed head-to-head analyses of the incretin-based therapies. We also performed sensitivity analyses relating to whether treatment was adjunctive therapy with metformin and by treatment adherence. After propensity-score matching, we identified four target trials of patients treated with tirzepatide (n = 7165), semaglutide (n = 20 198), liraglutide (n = 6565), and dulaglutide (n = 19 061); 1:1 with the reference (DPP4i) patients. Tirzepatide and semaglutide (but not liraglutide or dulaglutide) were associated with significant risk reduction of incident AUD compared to DPP4i (hazard ratio 0.47 [95% confidence interval 0.29, 0.75] and 0.68 [0.52, 0.89], respectively). Head-to-head comparison revealed tirzepatide had a significant risk reduction compared to liraglutide in incident AUD (0.47 [0.24, 0.92]). In patients with T2D, tirzepatide and semaglutide treatment is associated with a lower incidence of AUD; robust randomised, controlled evidence for the use of these drugs for this novel indication is appropriate.