Where 'More Serious Adverse Events' With Tirzepatide Comes From: Three Trials, Mostly One

What happened / Why it matters

What happened. A 2026 meta-analysis in Clinical Obesity that pooled head-to-head studies of tirzepatide and semaglutide reported that serious adverse events were more frequent with tirzepatide (relative risk 1.83, 95% CI 1.18 to 2.85). A separate network meta-analysis, which compares drugs indirectly, found no significant difference in serious adverse events.

Why it matters. The first number can be read as “tirzepatide has nearly twice the serious adverse events.” The paper’s own data show a narrower picture: the figure rests on only three studies, and most of the events come from one of them. This article explains where the number comes from. It does not say that either drug is more or less dangerous, and it is not medical advice.

What did the meta-analysis include?

Paccola and colleagues searched for studies comparing the two drugs directly, with at least 24 weeks of follow-up. They included 10 studies: three randomized trials and seven retrospective cohort studies. By our count from its table of studies, more than 90% of participants were in the observational studies.

The serious adverse event result is different from the weight results: it was pooled from only three of the ten studies. The authors report 5.7% of people on tirzepatide versus 2.9% on semaglutide had a serious adverse event. The heterogeneity statistic for this outcome was 0%, meaning the three trials did not visibly disagree, though with three trials that says little.

The paper does not grade the certainty of its evidence, and its authors note high variability in the efficacy results and the limits of retrospective data. We do not repeat the weight-loss results here; our guide on semaglutide versus tirzepatide covers them.

Which three trials produced the serious-event figure?

Figure 4B of the full text lists the three: SURPASS-2, SURMOUNT-5 and a smaller trial by Heise and colleagues. Together they give the totals the paper reports (104 of 1,828 people on tirzepatide and 26 of 889 on semaglutide), and the event counts match the trial registries.

The figure gives each trial’s events and relative risk:

What does that breakdown show?

• One trial supplies most of the events. In the paper’s figure, 85 of the 104 serious events on tirzepatide (about 82%) come from SURPASS-2, which carries 58.6% of the statistical weight. SURMOUNT-5 carries 39.5%.
• SURPASS-2 was in people with diabetes and used a lower semaglutide dose. Its comparison arm was semaglutide 1 mg, a dose below the one used for obesity.
• The one trial in obesity without diabetes did not show a significant difference. In SURMOUNT-5, the difference was not statistically significant, but the interval runs up to 2.80 and the trial was not designed to answer this question, so it does not rule out a difference either.
• Within SURPASS-2, there was no clear pattern by tirzepatide dose (7.0%, 5.3% and 5.7% across the three doses, by our calculation from the registry).
• These are serious events of any cause. They are not events attributed to a drug. We found no obvious pattern in the most frequent event types listed in the registry.
• We cannot say the difference applies only to diabetes. That reading comes from three trials without a formal test for whether the population changes the result.

The SURPASS-2 registry entry also lists 12 deaths among 1,409 people on tirzepatide and 1 among 469 on semaglutide; the other two trials list none. That count is small and imprecise, we did not review the causes, and it should not be read as a signal.

Why does the network meta-analysis disagree?

It answers a different question. The network meta-analysis by Pałka and colleagues compared liraglutide, semaglutide and tirzepatide indirectly, mostly through placebo-controlled trials of at least 52 weeks, and found no significant differences in serious adverse events between treatments. Its focus was a higher semaglutide dose, not tirzepatide versus semaglutide, and according to its full text it did not include SURPASS-2 or SURMOUNT-5 in the network. Its results come from indirect comparisons, which are weaker evidence than a direct trial.

The two analyses use different studies, different semaglutide doses and different methods. Neither settles the comparison of safety between the two drugs.

Who paid for these studies?

According to the trial registries, all three randomized trials (SURPASS-2, SURMOUNT-5 and the smaller trial) were sponsored by Eli Lilly, which makes tirzepatide. Among the retrospective studies, one was funded by Lilly, another had authors who were employees and shareholders of Novo Nordisk, which makes semaglutide, and a third had authors employed by the weight-management company Second Nature. The meta-analysis authors report no conflicts of interest. The meta-analysis does not analyze the funders’ influence.

What is the regulatory status?

Both drugs are FDA-approved for weight management in adults, in specific products with their own labels: Wegovy (semaglutide) and Zepbound (tirzepatide). The labels list their own safety information and warnings.

What does this mean for a reader?

• The meta-analysis reports more serious adverse events with tirzepatide across three trials. That does not establish that one drug is more dangerous than the other, and the evidence does not support calling them equally safe either.
• People taking or considering either medicine should talk with their prescriber about their own health history and about the label’s safety information. Do not stop or change a medicine because of a news article.
• This article gives no dosing guidance and does not link to or recommend sellers.

U.S. regulatory status checked Nov. 1, 2026. Based on the published meta-analyses, Figure 4 of the Paccola paper and trial registry entries; we did not read the supplements of either paper, the figures of the Pałka paper or the full text of the individual trials.