Adolescents with obesity who started semaglutide, liraglutide or tirzepatide were less likely to have a recorded suicide-related event over up to three years than matched adolescents managed with lifestyle intervention, according to an electronic health records study published on 14 September in Diabetes, Obesity and Metabolism. It points away from psychiatric harm in the age group where that question has been hardest to settle. The authors themselves caution against reading it as protection: against a different comparison group, metformin, the difference was not significant.
What the study compared
The team, based at Chi Mei Medical Center in Taiwan and collaborating universities, used de-identified records from the TriNetX Global Collaborative Network. They identified adolescents with overweight or obesity who were new users of one of the three drugs and compared them with adolescents receiving lifestyle interventions, balancing the two groups by propensity score matching. After matching there were 9,222 adolescents in each group, 18,444 in total. The primary outcome was a suicide-related event, meaning recorded suicidal ideation or a suicide attempt, counted from 30 days to 1,095 days after the start date.
The hazard ratio for suicide-related events was 0.74 (95% CI 0.57 to 0.95). Put simply, at any point during follow-up the recorded rate in the drug group was about a quarter lower than in the matched lifestyle group. The upper end of that interval, 0.95, sits close to 1, so the data are also compatible with a much smaller difference. Suicidal ideation on its own gave a similar figure, 0.73 (0.56 to 0.96), and anxiety a slight one, 0.92 (0.84 to 0.99). There was no significant difference for suicide attempts, insomnia, depression or eating disorders.
The lower risk was concentrated in two subgroups: female adolescents (0.61, 0.46 to 0.82) and those without type 2 diabetes (0.63, 0.47 to 0.85). It was not significant in males or in adolescents with type 2 diabetes. The published abstract does not give event counts or absolute rates, so the result cannot be turned into a number of events per thousand adolescents.
Where this sits, and what it cannot show
US regulators have already moved on this question for the drug class as a whole. In January 2024 the FDA said its preliminary evaluation had not found evidence that GLP-1 drugs cause suicidal thoughts or actions, while noting it could not rule out a small risk. On 13 January 2026, after what it called a comprehensive review, it asked the makers of Saxenda, Wegovy and Zepbound to remove the suicidal behaviour and ideation warning from their labels, having found no increased risk. This study adds adolescent-specific records to that picture rather than overturning anything, and it is consistent with our summary of semaglutide’s known side effects.
A second, much smaller paper published the same day points the same way. A chart review at the Medical College of Wisconsin followed 44 adolescents treated with liraglutide or semaglutide. Among the 15 with paired depression questionnaires, the median PHQ-9 score fell from 6 to 1; among the 23 with paired suicide screening, no increase in suicidality and no new suicidal ideation was recorded. With numbers that small it shows an absence of observed harm, not evidence of benefit.
The main study is observational, and its comparison group is the weak point. Adolescents who are prescribed a GLP-1 drug may differ from those offered lifestyle intervention alone in ways matching cannot capture: closer follow-up, better insurance, or clinicians who screen for mood problems before prescribing. The authors name selection bias directly, and the null result against metformin, a comparison closer to like-for-like, is the reason to hold the headline figure loosely. Health records also capture only what a clinician codes, so suicidal ideation that is never disclosed goes uncounted in both groups. The three drugs were pooled, so the study cannot say whether any one of them differs. We read the published abstract; the full paper is behind publisher access, and the abstract does not state funding.
Sources: Liu TH, Shen YL, Kuo TH, Wu JY, Lin CH, Lai CC. Psychiatric Safety of GLP-1 and Dual GIP/GLP-1 Receptor Agonists in Adolescents With Obesity. Diabetes, Obesity and Metabolism, published online 14 September 2026. doi:10.1111/dom.71353 · Said J, Liegl M, Pan AY, Dabrowski E. GLP-1 receptor agonist use and mental health outcomes in adolescents with obesity: a retrospective review. Journal of Pediatric Endocrinology and Metabolism, published online 14 September 2026. doi:10.1515/jpem-2026-0227 · FDA Drug Safety Communication, FDA Requests Removal of Suicidal Behavior and Ideation Warning from GLP-1 RA Medications, 13 January 2026 · FDA Drug Safety Communication, Update on FDA’s ongoing evaluation of reports of suicidal thoughts or actions, January 2024
