Between January 2021 and December 2025, Europe’s pharmacovigilance database logged 42,941 adverse-event reports naming liraglutide, semaglutide or tirzepatide as a suspect drug in diabetes or weight management. Of those, 225 involved suicide or self-injury: 141 for semaglutide, 47 for tirzepatide, 37 for liraglutide. A new analysis sets the three against each other and finds such events make up a larger share of semaglutide and liraglutide reports than of tirzepatide reports. The authors are explicit that this does not establish that the drugs cause them.
What a reporting odds ratio actually compares
The team, based at the University of Campania and the Campania Regional Centre for Pharmacovigilance, pulled individual case safety reports from EudraVigilance, kept those whose indication was consistent with type 2 diabetes or weight management, and identified suicide and self-injury events using the standardised MedDRA query built for that purpose. Gastrointestinal disorders were the most frequently reported category overall, followed by injury, poisoning and procedural complications. Suicide and self-injury made up roughly half a percent of the total. Compared with tirzepatide, the reporting odds ratio was 2.69 for semaglutide (95% CI 1.91 to 3.83) and 2.54 for liraglutide (95% CI 1.60 to 4.01). A reporting odds ratio asks a narrow question: among the reports filed for drug A, does this event make up a bigger slice than among the reports filed for drug B? It is a ratio of proportions inside a voluntary reporting system. There is no count of how many people took each drug, so it cannot be turned into a risk, an incidence, or a number needed to harm.
How this sits with what regulators have said
Both European and US regulators have reviewed suicidality on GLP-1 drugs and concluded the available evidence does not support a causal link. This paper does not challenge that, and says so in its own conclusions: it presents a comparative disproportionality analysis and argues that monitoring should continue as weight-management use expands. The value is in the comparison rather than the verdict. Until now the public argument has mostly been GLP-1 drugs versus everything else; this puts three drugs in the same class side by side using one database and one case definition, which is a cleaner question than it sounds, because the three are prescribed for overlapping reasons to overlapping populations.
Several things could produce this pattern without any difference in the drugs. The strongest is notoriety bias: semaglutide’s possible link to suicidal thoughts was heavily covered from 2023 onward, and publicity reliably increases reporting for the drug that was named, which is exactly the shape of a semaglutide-versus-tirzepatide gap in a spontaneous system. Liraglutide has been on the market far longer and, as a weight-management product, is licensed down into adolescence, so its reporting population is not tirzepatide’s. Spontaneous reports are unverified, can be duplicated, and carry no medical confirmation of causality. The confidence intervals are wide, particularly liraglutide’s, which spans 1.60 to 4.01 on 37 events. And none of this speaks to whether the underlying rate in any of the three groups is higher than would be expected among people with obesity or type 2 diabetes, who have elevated rates of depression before any prescription is written.
Sources: Scavone C, Cerasuolo M, Gaio M, Lettera N, Bernardi FF, Capuano A, Rinaldi B. Reporting patterns of suicide- and self-injury-related events involving liraglutide, semaglutide, and tirzepatide: data from the European pharmacovigilance database. Frontiers in Pharmacology, 27 August 2026;17:1882225. doi:10.3389/fphar.2026.1882225.
