Tirzepatide Reports in EudraVigilance

A 2026 BMJ Open analysis of EudraVigilance reports found that off-label use was the most frequently recorded preferred term in its tirzepatide dataset, followed by drug ineffective and off-label use of device. That is a description of submitted safety reports, not a measure of how often those events occur among everyone using tirzepatide.^[1]

The paper compares reporting patterns for tirzepatide with other GLP-1 receptor agonists. A GLP-1 receptor agonist is a medicine that acts on a gut-hormone pathway involved in blood-sugar control and appetite. Tirzepatide also acts at the GIP receptor, which is why it is often described as a dual GIP and GLP-1 receptor agonist.^[1]

What the researchers examined

The authors conducted a retrospective pharmacovigilance analysis of individual case safety reports in EudraVigilance, the European medicines safety-reporting database. They used descriptive counts and reporting odds ratios to compare selected preferred terms linked to suboptimal treatment outcomes and drug-use issues across tirzepatide and other GLP-1 receptor agonists.^[1]

A preferred term is a standard label used to code what was reported. It is not a clinical diagnosis and it does not supply the denominator needed to calculate a patient-level risk. The study combined reports from healthcare professionals and non-healthcare professionals for its overall dataset, then examined the healthcare-professional subset separately.^[1]

This scope matters because a report can describe a medicine being used outside a labelled indication, a perceived lack of effect, or a device-related issue without establishing why it happened. The paper asks how often particular labels appeared in the database and how their reporting odds compared. It does not test whether tirzepatide caused an individual report.^[1]

The reported terms were uneven

In the full set of preferred terms analysed, off-label use appeared 1,521 times, drug ineffective appeared 425 times, and off-label use of device appeared 99 times. Those are report counts from the study population, not estimates of the number of people affected in Europe.^[1]

For drug ineffective, tirzepatide reports had lower reporting odds than reports involving liraglutide, dulaglutide, exenatide, and semaglutide in the overall dataset. In the healthcare-professional reports, the authors found no difference in reporting odds between tirzepatide and semaglutide for that preferred term.^[1]

The pattern for off-label use differed. The overall dataset showed higher reporting odds for tirzepatide than for lixisenatide, dulaglutide, and liraglutide, but lower reporting odds than for semaglutide. A reporting odds ratio compares the relative prominence of a coded report across databases; it is not a comparison of treatment effectiveness.^[1]

Why the device signal needs caution

The largest relative signal in the paper concerned off-label use of device versus semaglutide: the reporting odds ratio was 43.47, with a 95% confidence interval from 16.00 to 118.13. The authors explicitly advised caution because the interval was wide. A wide interval signals substantial uncertainty around the estimate.^[1]

That finding should not be read as a rate of device problems, or as proof that one product is less safe than another. Spontaneous-report systems can be affected by who reports, what information is available, public attention, prescribing patterns, and changes in use over time. The study itself presents the result as a reporting pattern that complements trial evidence and clinical practice.^[1]

Readers trying to distinguish a labelled medicine from informal discussion should start with the product information and a clinician who knows their history. The US prescribing information for Mounjaro lists its approved use, warnings, contraindications, and adverse-reaction information. Those labelled materials answer a different question from a EudraVigilance disproportionality analysis.^[2]

What this paper cannot answer

The analysis cannot tell a reader the absolute chance of an off-label-use report, a device-related report, or a report coded drug ineffective. It also cannot show whether a reported outcome resulted from the medicine, the device, an underlying condition, the way the medicine was used, or another factor. Those questions need other study designs and denominator data.^[1]

It also does not replace the controlled trial evidence used to establish efficacy or the regulator-approved product information used in clinical decision-making. If you are comparing reports about tirzepatide with reports about semaglutide, the two databases of reports are not a shortcut to deciding which medicine is appropriate for you.^[1]

How to read safety-reporting headlines

The useful takeaway is narrow. That distinction is central to this paper. This paper identifies which selected drug-use and suboptimal-outcome terms were reported and compares their relative reporting patterns. It does not show that a reported event was caused by tirzepatide, and it does not convert a signal into an individual prediction. The comparison is most useful as a prompt to read the methods, check the confidence intervals, and keep report counts separate from treatment rates.^[1]

For background on the medicine itself, see our tirzepatide guide and the tirzepatide regulation section. If you have a possible adverse reaction or a concern about use, seek advice from a qualified healthcare professional and use the reporting route provided by the relevant regulator.