A cohort study from 13 South Korean hospitals, published in July in Diabetes, Obesity and Metabolism, found semaglutide associated with a hazard ratio of 3.42 for depression — meaning users were 3.42 times more likely to receive a depression diagnosis than matched non-users.

That number will generate headlines. It deserves careful interpretation before those headlines run.

Here's what the study actually showed, what it didn't show, why it conflicts with other evidence, and what the conflict means for anyone on a GLP-1 protocol.

What the Korean Study Found

Study design: Multicentre retrospective new-user cohort, electronic health records from 13 South Korean hospitals (2018–2025), propensity-score matching, ten prespecified safety outcomes Population: 2,357 semaglutide and 6,953 liraglutide initiators for weight management, matched to 22,602 and 68,001 non-initiators respectively Published: Park J, et al. Diabetes Obes Metab. Published online July 6, 2026. PMID 42410329. DOI 10.1111/dom.71065.

Semaglutide findings (selected):

Depressive disorder: HR 3.42 (95% CI 1.51–7.74)
Anxiety disorder: HR 2.39 (1.22–4.71)
Psychiatric disorders overall: HR 2.02 (1.30–3.14)
Gastrointestinal dysmotility or obstruction: HR 3.91 (1.42–10.82)
Vision impairment: HR 1.58 (1.04–2.41), not significant in some sensitivity analyses

Liraglutide findings: Elevated associations with psychiatric disorders overall, anxiety, depression, hepatic impairment, pancreatobiliary disorders, and vision impairment.

Source: Park J, et al. "GLP-1 Receptor Agonists for Weight Loss and Risk of Major Safety Outcomes: A Multicentre Cohort Study." Diabetes Obes Metab. 2026. DOI 10.1111/dom.71065.

What This Study Does and Doesn't Show

What it shows: An association — people on semaglutide in this dataset received depression diagnoses at higher rates than propensity-matched non-users — with a wide confidence interval (1.51 to 7.74) that reflects a small number of events.

What it cannot show: Whether semaglutide caused those depression diagnoses. This is observational research. Causation requires a randomized design that holds all other variables constant. Observational studies cannot do that, regardless of how well they match populations.

The follow-up problem: New-user cohorts built from hospital records observe most patients for months, not years. Depression is a condition that develops over time, responds to treatment over time, and has complex relationships with weight, metabolic status, and life circumstances. A short window may not capture the full picture.

The confounding problem: People who receive GLP-1 prescriptions differ from those who don't in ways that are difficult to fully account for. They may have more clinical contact (increasing the probability of receiving a psychiatric diagnosis). They may have more severe metabolic disease. They may be at different points in their health journey that create vulnerability to mood changes independent of the drug.

Propensity matching controls for measured variables. It cannot control for unmeasured ones.

Why This Conflicts With Other Evidence

The Korean study's findings are in direct tension with several other data sources.

The FDA's January 2026 review: On January 13, 2026, FDA asked the manufacturers of Wegovy, Saxenda and Zepbound to remove the suicidal ideation and behavior warning from their labels, citing a meta-analysis of 91 placebo-controlled trials (107,910 patients) and a Sentinel System cohort of 2,243,138 GLP-1 RA users compared with SGLT2-inhibitor users. Behind that sits SELECT — 17,604 adults with cardiovascular disease and obesity followed for a mean of 39.8 months on semaglutide or placebo — one of the largest controlled semaglutide datasets available.

The TriNetX network analysis (192 million records): Chen and colleagues analysed the TriNetX Global Federated Network (more than 192 million patients): 85,546 matched pairs starting tirzepatide versus semaglutide showed comparable composite psychiatric risk in Year 1 (HR 0.98) and Year 2 (HR 1.00); against earlier-generation GLP-1 RAs, semaglutide was associated with lower depression (HR 0.81), anxiety (HR 0.92) and suicidal ideation (HR 0.49) in Year 1.

The nference cohort (63,215 patients with pre-existing neuropsychiatric conditions): Published September 1, 2026 in npj Metabolic Health and Disease, this US electronic-record study found semaglutide associated with broadly lower neuropsychiatric event risk over two years compared with metformin, SGLT2 inhibitors, and DPP-4 inhibitors — and, within semaglutide users, lower subsequent incidence of mood, anxiety and substance-related disorders at higher attained doses.

That's three large data sources pointing in the opposite direction from the Korean hospital study published in July.

How to Think About Conflicting Observational Evidence

When well-designed observational studies conflict, the honest answer is: the evidence isn't settled.

Several factors can produce conflicting observational findings:

Population differences: The Korean hospital study enrolled a specific patient population in a specific healthcare system with specific prescribing patterns. That population may differ from U.S. or European populations in ways that affect results.

Surveillance differences: Patients on GLP-1 therapy receive more clinical contact than those not on any obesity medication. More contact means more diagnostic events — including psychiatric diagnoses — regardless of drug effect.

Duration differences: A months-long hospital-record cohort captures a very different slice of the patient trajectory than a 2-year analysis. Early GLP-1 initiation may be associated with adjustment-related mood changes that differ from longer-term effects.

Comparator differences: The Korean study compared GLP-1 users to non-GLP-1 users. The nference study compared semaglutide users to users of other diabetes medications. These are different questions.

None of this means the Korean findings are wrong. It means they're one piece of a complex picture that research is still assembling.

What This Means for Anyone on a GLP-1 Protocol

The honest clinical message isn't "semaglutide causes depression" — the evidence doesn't support that claim.

It isn't "semaglutide is definitively safe for mood" either — the evidence isn't settled enough to make that claim.

What the evidence does support:

Monitor mood as part of any GLP-1 protocol. This is already a clinical recommendation. Track it systematically — not because the drug definitely affects mood, but because the interaction between significant weight loss, changes in appetite, changes in reward signaling, metabolic shifts, and psychological response to body change is complex. Mood data matters regardless of mechanism.

Track daily, not just at appointments. A patient who develops mood changes between appointments may not connect those changes to their protocol at a 3-month follow-up. Daily check-ins create a timeline that makes those connections visible.

Bring the data to your provider. The question "has my mood been affected by this protocol?" is answerable with data. It's much harder to answer from memory at an appointment four months after the change began.

The Broader GLP-1 2.0 Context

GLP-1 receptors are expressed in the brain's reward and motivation circuitry. We've covered this extensively — the anhedonia signal, the "wanting vs. liking" distinction, the reward flattening that some users experience.

The Korean study doesn't resolve whether GLP-1 drugs affect mood through receptor signaling, through indirect effects of caloric restriction, through changes in metabolic state, or through some other mechanism.

What it does contribute is a signal that deserves ongoing attention — and a reminder that "this drug is approved and generally well-tolerated" is not the same as "this drug has no effect on brain chemistry."

The full picture of GLP-1 effects on mood, motivation, cognition, and neurological function is still being assembled. Studies like this one — with all its limitations — are part of how that picture gets built.

What Would Actually Settle This

A randomized controlled trial with validated psychiatric outcomes measures administered prospectively — not retrospectively from diagnostic codes — in a sample large enough to detect effects across dose groups, with follow-up long enough to capture both early adjustment and longer-term trajectory.

That trial would require years to complete. The Korean study took months to publish. The tension between research speed and research quality is real.

In the meantime: track your mood. Bring the data to your provider. Don't make medication decisions based on a single observational study in either direction.

The Bottom Line

A Korean hospital study found semaglutide associated with a 3.42x higher depression hazard ratio in an observational cohort. That finding is real data from a real study. It is also in direct conflict with three other large data sources pointing in the opposite direction.

The answer to "does semaglutide cause depression?" is not yes. It's not no. It's: the evidence is actively conflicted and the question isn't settled.

What's settled: mood should be tracked systematically on any GLP-1 protocol. Daily, over time, version by version alongside your protocol changes.

That's exactly what PeptidesGPT is built for.

→ Track your protocol at PeptidesGPT.com

Sources:

Park J, et al. "GLP-1 Receptor Agonists for Weight Loss and Risk of Major Safety Outcomes: A Multicentre Cohort Study." Diabetes Obes Metab. Published online July 6, 2026. PMID 42410329. DOI 10.1111/dom.71065.
Lincoff AM, et al. "Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes" (SELECT). N Engl J Med. 2023;389:2221–2232. PMID 37952131. DOI 10.1056/NEJMoa2307563.
Chen SC, et al. "Neuropsychiatric Outcomes With Tirzepatide, Semaglutide, and Other GLP-1 Receptor Agonists." Diabetes Obes Metab. 2026;28(10):8895–8904. PMID 42420795. DOI 10.1111/dom.71088.
Murugadoss K, et al. "Higher semaglutide dose is associated with lower neuropsychiatric event incidence independent of weight loss." npj Metab Health Dis. 2026;4(1). Published September 1, 2026. PMID 42680805. DOI 10.1038/s44324-026-00131-3.
U.S. Food and Drug Administration. "FDA Requests Removal of Suicidal Behavior and Ideation Warning from Glucagon-Like Peptide-1 Receptor Agonist (GLP-1 RA) Medications." Drug Safety Communication, January 13, 2026.

PeptidesGPT is an educational platform. This content is for informational purposes only and does not constitute medical advice. If you are experiencing significant mood changes, depression, or thoughts of self-harm, contact a qualified healthcare provider immediately. Do not change medications without consulting your provider.

Does Semaglutide Increase Depression Risk? A Korean Study Says Yes — But the Answer Isn't Settled.