No trial has randomized semaglutide against tirzepatide on cardiovascular outcomes, so the comparison runs on claims data. A target trial emulation matched 217,920 adults starting one or the other and followed them for up to three years, reporting atrial fibrillation, heart failure and myocardial infarction as primary outcomes [1]. The same two drugs on a different endpoint are compared in the liver comparison.
Tirzepatide came out ahead on every outcome at both time points, and the associations strengthened over the three years. That is the finding most summaries will report, and taken at face value it is a straightforward win.
That asymmetry is worth internalizing because it recurs throughout this literature. Baseline risk determines how much a relative reduction is worth, which is why the same drug can look transformative in a high-risk population and marginal in a healthy one while behaving identically in both. The interaction tests here were significant for atrial fibrillation (P = 0.003), heart failure (P < 0.001) and myocardial infarction (P = 0.019), so the difference between the groups is not noise.
The design carries the limitation that matters most. These are two drugs that entered the market at different times and get prescribed to different people for reasons a claims database records only partially, so the comparison is vulnerable to who was chosen for what rather than to what the drugs did. Propensity matching on recorded characteristics narrows that without closing it, the same caveat that applies to the glaucoma comparison and the carpal tunnel cohort, both of which also found tirzepatide ahead of its class on an unexpected endpoint.
Read alongside the randomized evidence on weight in the network meta-analysis of nineteen drugs, the picture is consistent in direction and much weaker in quality here. A pattern of tirzepatide outperforming across several unrelated outcomes could reflect a genuinely more effective drug, or it could reflect who gets prescribed the newer and more expensive option.