Analysis of the WHO pharmacovigilance database found increased reporting of suicidal ideation with semaglutide, liraglutide, and tirzepatide, but paradoxically decreased reporting of actual suicide attempts and completed suicides — and causation cannot be established from this data.
Suicidal ideation ROR 5.82 but completed suicide ROR 0.01 for semaglutideThe striking paradox: semaglutide had nearly 6 times more reports of suicidal ideation than expected, but 100 times fewer completed suicides than expected — a contradictory pattern that cannot be explained by a simple causal drug effect.
What the researchers found
Using the WHO VigiBase database, reporting odds ratios (ROR) showed a mixed pattern:
**Increased ROR (more reports than expected):**
- Suicidal ideation: semaglutide (5.82), liraglutide (4.03), tirzepatide (2.25)
- Depression/suicidal: semaglutide (14.74), liraglutide (5.86)
- Suicidal behavior: semaglutide (6.52), liraglutide (3.90)
**Decreased ROR (fewer reports than expected):**
- Suicide attempts: semaglutide (0.11), dulaglutide (0.075), exenatide (0.047), liraglutide (0.15)
- Completed suicide: semaglutide (0.01), dulaglutide (0.003), exenatide (0.002), liraglutide (0.008)
This paradoxical pattern — more reports of suicidal thoughts but dramatically fewer actual attempts and deaths — differs from what was found in the FDA's FAERS database.
Why it matters
With tens of millions of people now taking GLP-1 drugs like semaglutide, any potential link to suicidality would be a major public health concern. Both the EMA and FDA have investigated this signal. This WHO database analysis — the largest global pharmacovigilance database — provides important context by revealing a paradoxical pattern that neither clearly supports nor refutes a causal link, highlighting the limitations of spontaneous reporting data for answering this critical safety question.
How the study worked
Researchers searched the WHO global pharmacovigilance database (VigiBase) from inception through January 2024 for reports of suicidality associated with GLP-1 receptor agonists. Reporting odds ratios (ROR) were calculated for different suicidality outcomes (ideation, depression/suicidal, suicidal behavior, suicide attempts, completed suicide) for each GLP-1 RA. An ROR was considered significant when the lower limit of the 95% confidence interval exceeded 1.0.
What this study cannot tell us
Pharmacovigilance databases like VigiBase capture spontaneous reports, which are subject to severe reporting biases — including stimulated reporting (increased reports due to media attention), underreporting, and inability to establish causation. The study cannot account for underlying depression or suicidality risk in the patient populations taking GLP-1 drugs (obesity and diabetes both independently increase suicide risk). The paradoxical findings (more ideation, fewer attempts/deaths) suggest that reporting artifacts, rather than true drug effects, may be driving the signal.
How to read the evidence
This is a pharmacovigilance analysis of a spontaneous adverse event reporting database, which is the lowest level of safety evidence. While useful for signal detection, reporting databases cannot establish causation due to inherent biases including stimulated reporting, underreporting, and lack of denominator data (total number of patients exposed). The contradictory findings further limit interpretability.
When this study was published
Published in 2025 using data through January 2024, this is a very current and timely analysis directly relevant to the ongoing regulatory and clinical debate about GLP-1 drug safety and neuropsychiatric effects.
The bigger picture
The GLP-1 drug-suicidality question is one of the most consequential pharmacovigilance investigations ongoing in medicine today. This study's paradoxical findings — elevated suicidal ideation but reduced actual suicides — highlight a fundamental challenge of adverse event reporting databases: they capture what gets reported, not what actually happens. The high media attention on GLP-1 drugs may increase reporting of suicidal thoughts (reporting bias), while the weight loss and metabolic improvements may actually reduce overall suicide risk. Dedicated prospective studies are needed to resolve this question.
Questions still open
- Is the elevated suicidal ideation reporting driven by stimulated reporting (patients and doctors reporting more after media coverage) rather than a true drug effect?
- Could GLP-1 drugs actually be protective against suicide through weight loss-related improvements in mood, self-esteem, and quality of life?
- Will prospective clinical trials specifically designed to assess neuropsychiatric outcomes resolve the conflicting signals from different adverse event databases?
Common questions
Should I worry about suicidal thoughts if I'm taking semaglutide or liraglutide?
Why are the findings so contradictory — more suicidal thoughts but fewer actual suicides?
Read the original research
Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and suicidality: A replication study using reports to the World Health Organization pharmacovigilance database (VigiBase®).
Journal of affective disorders, 369, 922-927
Citation
McIntyre, Roger S; Mansur, Rodrigo B; Rosenblat, Joshua D; Rhee, Taeho Greg; Cao, Bing; Teopiz, Kayla M; Wong, Sabrina; Le, Gia Han; Ho, Roger; Kwan, Angela T H. (2025). Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and suicidality: A replication study using reports to the World Health Organization pharmacovigilance database (VigiBase®).. Journal of affective disorders, 369, 922-927. https://doi.org/10.1016/j.jad.2024.10.062