Characteristics and Main Findings of Included Studies and Regulatory Reviews.
Characteristics and Main Findings of Included Studies and Regulatory Reviews.
| Evidence Type | Study/ Source | Population/Sample or Scope | Comparator | Main Outcome(s) | Key Estimate(s) |
|---|---|---|---|---|---|
| RCT pooled analysis | O’Neil et al. (2017)16 | 5,325 adults with overweight or obesity, pooled across the phase 2/3a SCALE trials | Placebo | Depression, anxiety, suicidality (CSSRS, PHQ-9) | Depression 2.1 vs 2.1/100 PY; SI 1.0% vs 1.0% |
| RCT meta-analysis | Silverii et al. (2024)17 | 31 RCTs; 84,713 participants (22 trials in T2DM, 9 in obesity) | Placebo / active comparators | Any psychiatric disorder; SB | OR 0.97; OR 0.86 |
| RCT meta-analysis | Pierret et al. (2025)18 | 80 RCTs; 107,860 participants with overweight or obesity and/or T2DM | Placebo | Psychiatric AEs; depressive symptoms; QoL | No increase; QoL favored GLP-1 RAs |
| RCT meta-analysis | Ebrahimi et al. (2025)19 | 27 placebo-controlled RCTs; 59,396 participants (32,354 GLP-1 RA, 27,042 placebo) | Placebo | Suicide / self-harm | RR 0.76 (0.48–1.21); 0.047 vs 0.042/100 PY |
| RCT post hoc pooled | Wadden et al. (2024)20 | 3,377 participants pooled from STEP 1, 2, and 3, and 304 participants from STEP 5 (STEP 2 participants also had T2DM) | Placebo | PHQ-9; C-SSRS; AEs | PHQ-9 −0.56 at week 68; no SI/SB excess |
| RCT post hoc pooled | Wadden et al. (2026)21 | 4,056 participants with obesity or overweight (SURMOUNT-1, -2 and -3) | Placebo | PHQ-9; C-SSRS; AEs | SI 0.6% vs 0.6%; PHQ-9 −0.6 |
| Target-trial emulation | Tang et al. (2024)22 | GLP-1 RA initiators: 21,807 matched with SGLT2i initiators; 21,402 matched with DPP-4i initiators | Active | SI/SB | HR 1.07 vs SGLT2i; HR 0.94 vs DPP-4i |
| Active-comparator cohort | Ueda et al. (2024)23 | 124,517 GLP-1 RA initiators, compared with SGLT2i initiators | SGLT2i | Suicide death; composite | HR 1.25 (0.83–1.88); composite HR 0.83 |
| Active-comparator cohort | Shapiro et al. (2025)24 | 36,082 GLP-1 RA versus 234,028 DPP-4i initiators; 32,336 GLP-1 RA versus 96,212 SGLT2i initiators | DPP-4i / SGLT2i | Suicidality composite | HR 1.02 vs DPP-4i; 0.91 vs SGLT2i |
| Target-trial emulation | Tang et al. (2025)25 | GLP-1 RA initiators: 14,665 matched with SGLT2i initiators; 13,711 matched with DPP-4i initiators | Active | Incident depression | HR 1.07 vs SGLT2i; HR 0.90 vs DPP-4i |
| Propensity-matched cohort | De Giorgi et al. (2024)26 | Propensity-matched pairs of semaglutide versus comparator: 23,386 versus sitagliptin; 22,584 versus empagliflozin; 19,206 versus glipizide | Active | 22 neuropsychiatric outcomes | No psychiatric harm; favorable cognitive/nicotine signals |
| Broad EHR cohort | Wang et al. (2024)27 | 240,618 patients with obesity; replication cohort of 1,589,855 patients with T2DM | Non-GLP-1 anti-obesity / antidiabetic | Incident/recurrent SI | Incident HR 0.27; recurrent HR 0.44 (obesity) |
| Earlier cohort | Gamble et al. (2018)28 | 501 GLP-1 RA users in the UK CPRD, compared with users of other glucose-lowering drugs | Other glucose-lowering | Depression/self-harm | No clear excess risk |
| Case-time-control | Bezin et al. (2025)29 | 1,102 cases of suicide or suicide attempt (French SNDS) | Self-controlled with time controls | Suicide/suicide attempt | Adjusted case-time-control ratio 0.62 (0.51–0.75) |
| Tirzepatide real-world cohort | Yu et al. (2025)30 | 16,321 propensity-matched pairs, tirzepatide versus non-GLP-1 anti-obesity drugs (US TriNetX; 29% with T2DM) | Non-GLP-1 anti-obesity drugs | SI/attempts | aHR 0.52 (0.28–0.91); 17 vs 33 events |
| Outlier observation al cohort | Kornelius et al. (2024)31 | 162,253 patients with obesity treated with liraglutide or semaglutide, compared with non-users | Non-users | Depression, anxiety, SB | Markedly elevated risks reported |
| EudraVigilance analysis | Tobaiqy and Elkout (2024)32 | 31,444 individual case safety reports, of which 372 were psychiatric | None | Psychiatric AEs | Depression 50%; anxiety 39%; SI 19.6% of psychiatric reports |
| FAERS analysis | Chen et al. (2024)33 | 181,238 GLP-1 RA reports, of which 8,240 were psychiatric | None | Psychiatric AEs | Multiple signals; median onset 31 days |
| WHO disproportionality | Schoretsanitis et al. (2024)34 | Reports for semaglutide and liraglutide in the WHO global pharmacovigilance database | All drugs | SI; self-injurious ADRs | Semaglutide SI ROR 1.45; +antidepressant ROR 4.45; +benzodiazepine ROR 4.07 |
| FAERS neuropsychiatric analysis | Lu et al. (2025)35 | 25,110 GLP-1 RA-related neuropsychiatric adverse-event reports | None | Neuropsychiatric AEs | Semaglutide weight-loss suicide-related ROR 2.55 |
| Regulatory review | European Medicines Agency (2024)4 | Class-wide review of GLP-1 RAs | Regulatory synthesis | Suicidal/self-injurious thoughts/actions | No causal association supported |
| Regulatory review | US Food and Drug Administration (2026)5 | 91 placebo-controlled trials; n=107,910 plus Sentinel cohort, n=2,243,138 | Placebo and active comparator | SI/behavior; psychiatric AEs | No increased risk; warning removal requested |
ADR(s), adverse drug reaction(s); AE(s), adverse event(s); aHR, adjusted hazard ratio; C-SSRS, Columbia Suicide Severity Rating Scale; DPP-4i, dipeptidyl peptidase-4 inhibitor; EHR, electronic health record; EudraVigilance, European Union adverse event reporting database; FAERS, FDA Adverse Event Reporting System; GLP-1 RAs, glucagon-like peptide-1 receptor agonists; HR, hazard ratio; OR, odds ratio; PHQ-9, Patient Health Questionnaire-9; PY, person-years; QoL, quality of life; RCT(s), randomized controlled trial(s); ROR, reporting odds ratio; RR, risk ratio; SB, suicidal behavior; SCALE, liraglutide weight-management trial program; SGLT2i, sodium-glucose co-transporter 2 inhibitor; SI, suicidal ideation; SNDS, Système National des Données de Santé (French national health data system); T2DM, type 2 diabetes mellitus; UK CPRD, United Kingdom Clinical Practice Research Datalink; WHO, World Health Organization.