Two large real-world studies find little separating tirzepatide and semaglutide

Tirzepatide and semaglutide have never been compared against each other in a cardiovascular outcomes trial, and probably never will be. Two papers published within a week of each other in November try to fill that gap using routine care data — one on cardiovascular events, one on severe gastrointestinal harms. Both land in roughly the same place: the drugs look more alike than different.

The cardiovascular comparison

The Nature Medicine study, published 9 November, ran five cohort studies in patients with elevated cardiovascular risk, including obesity and type 2 diabetes, enrolled in US insurance programmes between 2018 and 2025 [s1].

The design is the part worth understanding. Before comparing the drugs directly, the authors first emulated two completed cardiovascular outcome trials — SUSTAIN-6 (semaglutide versus sitagliptin as a placebo proxy) and SURPASS-CVOT (tirzepatide versus dulaglutide) — inside the claims data, to see whether their method could reproduce results already known [s1]. That benchmarking step found high agreement between the reference trials and their emulations for all individual endpoints except all-cause mortality in SUSTAIN-6, and that finding was used to shape the subsequent analyses [s1]. Baseline confounders were balanced using propensity score matching [s1].

In expanded populations meant to reflect patients routinely seen in practice rather than trial enrollees, semaglutide versus sitagliptin gave a hazard ratio of 0.82 (95% CI 0.74 to 0.91) for the composite of myocardial infarction or stroke [s1]. Tirzepatide versus dulaglutide gave a hazard ratio of 0.87 (95% CI 0.75 to 1.01) for the composite that also included mortality [s1].

The head-to-head comparison of tirzepatide against semaglutide returned a hazard ratio of 1.06 (95% CI 0.95 to 1.18) [s1]. The confidence interval spans 1, and the authors read the result as supporting comparable cardiovascular benefit for the two drugs in clinical practice [s1].

The gastrointestinal comparison

Five days earlier, Annals of Internal Medicine published a new-user, active-comparator cohort study of severe gastrointestinal adverse events across dulaglutide, subcutaneous semaglutide, and tirzepatide in adults with type 2 diabetes [s2].

Patients initiated one of the three drugs between 1 January 2019 and 30 August 2024, in three cohorts corresponding to three pairwise comparisons, with 1:1 propensity score matching within each [s2]. The primary outcome was a composite of acute pancreatitis, biliary disease, bowel obstruction, gastroparesis, and severe constipation [s2].

The matched cohorts were large: 65,238 pairs for semaglutide versus dulaglutide, 20,893 pairs for tirzepatide versus dulaglutide, and 46,620 pairs for tirzepatide versus semaglutide [s2].

The hazard ratios for the composite gastrointestinal outcome were 0.96 (95% CI 0.87 to 1.06) for semaglutide versus dulaglutide, 0.96 (0.77 to 1.20) for tirzepatide versus dulaglutide, and 1.07 (0.90 to 1.26) for tirzepatide versus semaglutide [s2]. All three intervals cross 1. The authors conclude the three drugs have similar gastrointestinal safety profiles in this population [s2].

What these designs can and cannot do

Neither study randomised anyone. Both used propensity score matching to make treated groups comparable on measured characteristics, which does nothing about characteristics that were not measured. The Annals authors name the specific gap: possible residual confounding by glycaemic control and body mass index [s2] — two variables that plausibly influence both which drug a clinician chooses and what happens afterwards, and that are poorly captured in claims data.

The Nature Medicine team's response to that general problem is the benchmarking step: if a method can reproduce trials whose answers are already known, it earns some credibility when applied to a question no trial has answered [s1]. That is a genuine methodological advance over simply reporting an observational hazard ratio, and it is also where the study's one discordant result sits — all-cause mortality in the SUSTAIN-6 emulation did not agree with the trial [s1].

The Nature Medicine cohorts are drawn from US insurance programmes [s1], which is not the same as the populations these drugs are prescribed to elsewhere; the Annals study is described as population-based [s2].

What it adds up to

For cardiovascular events, the evidence now includes a direct comparison, and it is null: a hazard ratio of 1.06 with an interval from 0.95 to 1.18 does not support one drug over the other on that outcome [s1]. For severe gastrointestinal events, the same pattern holds across all three pairwise comparisons [s2].

That is a useful negative. Much of the public conversation about these two drugs is framed as a contest, and on weight loss the trial evidence does separate them. On the two harder outcomes examined here — cardiovascular events and severe gut complications — the November evidence does not.

What neither study addresses is longer-term outcomes, populations outside US insurance coverage, or the many patients now taking these drugs without type 2 diabetes. Those remain open.

This article is informational and does not constitute medical advice.

Sources

• [s1] Cardiovascular outcomes of semaglutide and tirzepatide for patients with type 2 diabetes in clinical practice. Nature Medicine, 9 November 2025. https://doi.org/10.1038/s41591-025-04102-x
• [s2] Comparative Gastrointestinal Safety of Dulaglutide, Semaglutide, and Tirzepatide in Adults With Type 2 Diabetes. Annals of Internal Medicine, 4 November 2025. https://doi.org/10.7326/ANNALS-25-01724