Semaglutide: 6 Years of EU Reports, Two Unsettled Signals

A disproportionality analysis published in Naunyn-Schmiedeberg’s Archives of Pharmacology on 22 August 2026 went through six years of European adverse reaction reports, 2020 to 2025, and compared three GLP-1 receptor agonists against each other. Semaglutide came out with more eye disorder reports and more suicidal ideation reports than dulaglutide.

Before that sentence travels any further, it needs the thing a headline will not carry: a disproportionality analysis measures what gets reported, not what happens.

A framing note. Semaglutide is an approved medicine, this analysis concerns patients, and semaglutide is not in the Peptra Labs catalogue. Nothing here is guidance about human use, and nothing here should be read as a claim that any product causes harm.

What the semaglutide analysis actually did

The authors, working from the University of Messina and several Italian hospitals, pulled suspected adverse reaction reports for liraglutide, semaglutide and dulaglutide out of EudraVigilance, the European Medicines Agency database of spontaneously reported suspected side effects.

Spontaneous reporting is worth understanding before reading any number that comes out of it. A report enters EudraVigilance when a clinician, a company or a patient decides to file one. Nobody counts the people who took the drug and reported nothing, and nobody verifies that the drug caused the event. The database holds suspicions.

What you can do with it is compare drugs to each other. If two medicines are used by broadly similar populations and one accumulates a particular kind of report far more often, that disproportion is a signal worth chasing. That is the entire logic of the reporting odds ratio, or ROR, and it is what this paper computes.

The two signals

The two rows are not equally solid, and the confidence intervals say so plainly.

The eye disorder signals are reasonably tight. An interval of 1.30 to 2.75 describes a difference that is probably real in the reporting data, whatever its cause.

The suicidal ideation intervals are enormous. A point estimate of 6.49 with an interval running from 1.57 to 26.81 is compatible with a modest difference and with a seventeenfold one. Intervals that wide come from small report counts, and a small change in the numerator moves them a long way. The authors report the finding and do not overstate it.

There is also a detail inside the psychiatric row that cuts against the obvious reading. Liraglutide scored higher than semaglutide, 8.61 against 6.49. If this were a story about one drug, the older and less discussed molecule would not be at the top of it.

The cohort study that points the other way

This is where the paper needs company, and it has some.

In January 2024, Nature Medicine published a real-world cohort analysis of electronic health records covering 240,618 patients with overweight or obesity, replicated in a further 1,589,855 patients with type 2 diabetes. Compared with non-GLP-1 anti-obesity medications, semaglutide was associated with a lower risk of both incident suicidal ideation, hazard ratio 0.27, and recurrent suicidal ideation, hazard ratio 0.44. The authors state their findings do not support higher risk.

Two studies, opposite directions, and both defensible, because they are asking different questions. The cohort study asks how often the event occurred among people who took the drug. The disproportionality analysis asks how often the event was reported relative to other drugs in the same database. A drug under intense public discussion attracts reports for reasons that have nothing to do with pharmacology. Reporting bias is not a flaw in EudraVigilance, it is a property of it.

The authors themselves note that drug agencies established no causal relationship while stating the risk should not be underestimated. That is the European regulatory position, reported accurately, and it is more careful than most coverage of this topic.

The eye finding has independent support

The ophthalmic side is on firmer ground, and not only because the intervals are narrower.

A systematic review and meta-analysis of ocular adverse events with semaglutide appeared in JAMA Ophthalmology in September 2025. A separate cohort of US veterans with type 2 diabetes, published in the same journal in March 2026, examined new-onset non-arteritic anterior ischaemic optic neuropathy among people starting semaglutide.

When a spontaneous-reporting signal, a meta-analysis and a large cohort all circle the same organ system, the signal has moved past the noise floor of any one method. That is not the same as established causation, and the mechanism remains unsettled, but it is a different quality of evidence from the psychiatric row in the same table.

What this means for the class, and for the catalogue

Neither of the compounds in the Peptra Labs catalogue appears in this analysis, and that gap is itself informative.

Tirzepatide is a dual GIP and GLP-1 agonist and was not among the three drugs studied. Retatrutide, the triple agonist, is investigational and has no comparable body of post-marketing reports at all, because post-marketing data requires a market. Our triple-agonist research guide sets out what the published record does and does not contain.

The absence matters analytically. Disproportionality analysis compares drugs within a database, so a molecule with fewer years of exposure and fewer prescriptions cannot generate a comparable signal, and its silence should never be read as safety. It is a sample size problem wearing the costume of a reassuring result.

Something similar is true in the other direction for identification. We reported earlier on a validated method that separates nine of them, and the existence of that method is a reminder that this class is now large enough that telling members apart is a laboratory task in its own right, well before anyone attributes an adverse event to one of them.

It is also worth setting this paper against the one we covered yesterday. In a cohort of 13,007 psychiatric patients, semaglutide initiation was associated with fewer recorded cognitive signs and symptoms. Same drug, same broad domain, opposite valence, different method. Two papers published a day apart, and any honest account of semaglutide and the brain has to hold both.

For laboratories, everything in our catalogue is supplied on a research use only basis, and our ranking of the most-studied research peptides tracks how much published evidence each compound actually carries.

References

• Ammendolia I, Mondello C, Esposito E, et al. Psychiatric and eye disorders associated with use of glucagon-like peptide 1 receptor agonists (GLP-1 RAs). Naunyn-Schmiedeberg’s Archives of Pharmacology, 22 August 2026. doi 10.1007/s00210-026-05813-9
• Wang W, Volkow ND, Berger NA, Davis PB. Association of semaglutide with risk of suicidal ideation in a real-world cohort. Nature Medicine, 5 January 2024
• Natividade GR, Spiazzi BF, Baumgarten MW, et al. Ocular Adverse Events With Semaglutide: A Systematic Review and Meta-Analysis. JAMA Ophthalmology, 1 September 2025
• Heberer K, Bress AP, Cogill S, Maldonado AI. New-Onset Nonarteritic Anterior Ischemic Optic Neuropathy and Initiators of Semaglutide in US Veterans With Type 2 Diabetes. JAMA Ophthalmology, 1 March 2026

The products referenced on this site are supplied for laboratory research use only. They are not medicines and are not intended for human or veterinary use. This article summarises published research for informational purposes and is not medical advice. Statements about third-party studies belong to their authors.