Pooled trials find no atrial fibrillation signal for tirzepatide, with a weaker arrhythmia finding left unresolved

An analysis pooling ten randomised trials found no clear link between tirzepatide and atrial fibrillation in adults with overweight or obesity. A broader measure covering all arrhythmias did cross the line into statistical significance, but on very few events. The analysis was published September 9, 2026 in the Journal of the American Heart Association.

Tirzepatide activates two gut hormone receptors, GIP and GLP-1, and is used for weight loss and for type 2 diabetes. Atrial fibrillation is an irregular and often fast heart rhythm that raises stroke risk, and it is more common in people carrying extra weight. Whether a medicine that drives large weight loss also changes the rate of it is a fair question, and one that had not been settled.

The researchers searched PubMed, Embase and the Cochrane Library for randomised trials comparing tirzepatide against placebo in people with overweight or obesity. Ten trials qualified, covering 6,515 people, of whom 4,491, or 68.9 percent, received tirzepatide. They pooled the results using a statistical method built for rare events, which matters here because the events being counted were uncommon.

No link to atrial fibrillation once the uncertainty range is read properly

For atrial fibrillation the pooled odds ratio was 2.20, with a 95 percent credible interval running from 0.81 to 6.75.

An odds ratio above 1 points toward higher odds. But the credible interval is the range the true value is likely to sit in, and this one runs from below 1 to well above it. Because it includes 1, which means no difference, the result does not establish a link. The point estimate of 2.20 looks alarming on its own and should not be read on its own.

For atrial arrhythmia more broadly the pooled odds ratio was 2.16, with a credible interval of 0.90 to 5.87. That range also crosses 1, so the same reading applies.

The authors' stated conclusion is that tirzepatide was not associated with atrial fibrillation or atrial arrhythmia.

The any-arrhythmia number was the only one that cleared the line

For any arrhythmia, a wider category that sweeps in rhythm disturbances beyond the atria, the pooled odds ratio was 1.84 with a credible interval of 1.04 to 3.90. That range sits just above 1 at its lower end, so by the usual convention it counts as a finding.

It is a weak one. The lower bound of 1.04 is barely past the no-difference mark, and the interval is wide, which means the data are consistent with anything from a very small increase to a near quadrupling. The authors say the event rates were low and that the finding should be interpreted cautiously. That is the right reading rather than a hedge.

Rare events are why the ranges are this wide

The whole analysis rests on a small number of actual events spread across 6,515 people. When events are rare, each additional case moves the estimate a long way, and the resulting ranges are wide enough that both reassuring and concerning interpretations remain compatible with the data.

The trials pooled here were not designed to detect heart rhythm problems. They were designed mainly around weight and blood sugar, with rhythm events captured as reported adverse events rather than hunted for with continuous monitoring. An analysis of trials like these will catch arrhythmias that were noticed and recorded, not necessarily every one that occurred.

A meta-analysis also inherits whatever the original trials did. Follow-up in obesity trials is typically measured in months to a couple of years, which is short for a condition that usually develops over a longer horizon.

None of this is a reason to start or stop a medicine. Anyone with a known rhythm problem, or symptoms such as palpitations, breathlessness or unexplained dizziness, should raise them with the clinician managing their care.