GLP-1 drugs appear psychiatrically safe, but one finding still raises questions

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Sep 6 2026

With millions of patients now taking GLP-1 drugs, researchers examined whether concerns about depression and suicidality are supported by the evidence and whether these treatments could also influence mood.

Study: Impact of Glucagon-Like Peptide-1 Receptor Agonists on Mental Health: A Systematic Review of Psychiatric Safety and Antidepressant Potential. Image Credit: zimmytws / Shutterstock

In a recent study published in the journal Cureus, researchers evaluated the psychiatric safety of glucagon-like peptide-1 receptor agonists (GLP-1RAs) and their potential effects on depressive symptoms, psychiatric worsening, suicidality, and psychiatric hospitalization.

Background

Over 4.9 million patients were represented in 11 primary studies examining GLP-1RAs and mental health. This topic is significant since these drugs are prescribed for type 2 diabetes mellitus (T2DM) as well as obesity, both of which can be associated with major depressive disorder (MDD).

Reports of suicidal ideation and worsening depressive episodes emerged early in the post-marketing stages of the drugs. However, the U.S. Food and Drug Administration (FDA) subsequently investigated the signal and concluded that available evidence did not support a causal link between GLP-1RAs and suicidal behavior.

The presence of GLP-1 receptors in brain regions involved in metabolic and mood regulation has prompted interest in possible neuroprotective or antidepressant effects. Evidence differed in how mental health was measured. Further research is needed to directly measure mood changes with validated scales.

About the study

The systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Searches covered studies published from 2021 through May 2026 in PubMed/MEDLINE, Google Scholar, and ScienceDirect. PubMed Central (PMC) records were grouped with PubMed/MEDLINE.

Search terms covered GLP-1RAs, including semaglutide and liraglutide, depression and other mental disorders, suicidal thoughts, psychiatric assessment scales, adverse effects, and hospitalization, and the initial search identified 617 records. The review was not prospectively registered.

Peer-reviewed English-language studies involving Food and Drug Administration (FDA)-approved GLP-1RAs were eligible. Tirzepatide, a dual glucose-dependent insulinotropic polypeptide (GIP) and GLP-1 receptor agonist, met the inclusion criteria.

Eligibility required studies to report at least one relevant outcome involving changes in depressive symptoms or rating scales, psychiatric safety, or psychiatric or mental health-related hospitalization. Randomized controlled trials (RCTs), observational cohort studies, and cross-sectional studies in diabetes or obesity populations were eligible.

Animal, in vitro, pregnancy, lactation, non-peer-reviewed, gray-literature, and studies without relevant psychiatric outcomes were excluded. Researchers identified studies and gathered information on their designs, psychiatric measures, interventions, comparator groups, and results.

Cohort studies were assessed using the Newcastle-Ottawa Scale (NOS), RCTs using the Cochrane Risk of Bias 2 (RoB 2) tool, and the cross-sectional study using the Joanna Briggs Institute (JBI) tool. Populations, comparators, and outcome definitions varied; hence, findings were synthesized narratively rather than statistically pooled.

Study results

The search identified 617 records across the three databases: 352 from PubMed/MEDLINE (including PMC), 125 from Google Scholar, and 140 from ScienceDirect.

After removing 75 duplicates, 542 records underwent title and abstract screening. Of these, 361 were excluded, leaving 181 reports for full-text assessment. Another 163 were excluded because of ineligible designs, psychiatric outcomes, or insufficient data.

A total of 18 papers remained in the broader review catalog, and 7 were excluded after quality appraisal against predefined design and scope criteria, leaving 11 primary studies for the final synthesis. These studies represented more than 4.9 million patients receiving treatment for T2DM or obesity.

All 11 studies in the final synthesis were observational, comprising 10 cohort studies and 1 cross-sectional study; no RCTs with psychiatric safety or antidepressant efficacy as a primary endpoint were included.

The final studies included large populations from the United States, the United Kingdom, Scandinavia, and Poland. Mental health assessment differed substantially: most studies used International Classification of Diseases (ICD) diagnostic codes, whereas only one included study directly used validated psychometric scales.

Overall, the cohort evidence showed no consistent increase in psychiatric risk with GLP-1RAs. Nine of 10 cohort studies found no elevated neuropsychiatric risk or suggested a protective association when compared with active antidiabetic treatments.

One multinational registry study found GLP-1RA use was associated with about a 17% lower risk of the combined outcome of suicide death and nonfatal self-harm, while another large study reported about a 10% lower risk of suicidal ideation, suicide attempt, or intentional self-harm. 

Stanislaus et al. found no increased risk compared with sodium-glucose cotransporter-2 (SGLT-2) inhibitors and a lower risk compared with dipeptidyl peptidase-4 (DPP-4) inhibitors.

One obesity-focused cohort produced a divergent finding. Kornelius et al. reported nearly three times the risk of MDD and about twice the risk of anxiety and suicidal ideation or attempts among GLP-1RA users.

The review noted that greater clinical scrutiny in specialized weight-management settings might partly contribute to increased diagnostic capture, although surveillance bias was not directly tested. The authors also noted that surveillance bias may not fully account for the elevated risk observed in obesity-focused cohorts.

Evidence suggesting possible antidepressant effects was limited, with direct psychometric evidence coming from a single cross-sectional study.

Witaszek et al. used the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7) among 1,105 adult women. In this cross-sectional analysis, semaglutide users had modestly lower depression and anxiety scores than the rest of the study sample, with both differences reaching statistical significance.

Taipale et al. also found that semaglutide and liraglutide were associated with reduced psychiatric worsening in patients with pre-existing depression or anxiety. Semaglutide was associated with a 28% lower risk of hospitalization for a psychiatric cause or self-harm. 

Only one study directly assessed inpatient events, finding no increased risk, while leaving psychiatric hospitalization evidence limited because its combined endpoint did not separate psychiatric admissions from self-harm hospitalizations.

Conclusions

The review found no consistent evidence of an increased class-wide neuropsychiatric risk with GLP-1RAs. Across the included evidence, psychiatric outcomes were generally neutral or protective, although one obesity-focused cohort reported increased diagnostic risks.

Semaglutide was associated with lower depression and anxiety scores on validated psychometric measures and with reduced psychiatric worsening in patients with pre-existing mood disorders. However, antidepressant potential remains incompletely characterized because scale-based evidence was limited to a single included study, and all 11 studies in the final synthesis were observational, so causation cannot be established. 

The review also did not apply a formal GRADE certainty-of-evidence framework. Psychiatric hospitalization was also insufficiently assessed. Prospective research using validated continuous mood measures and clearly defined hospitalization endpoints is needed.