Adults who started semaglutide or liraglutide to lose weight were later diagnosed with depression far more often than matched people who never started the drugs, a study across 13 South Korean hospitals found. Semaglutide users carried about three and a half times the rate of new depression diagnoses.
That figure (a hazard ratio of 3.42, meaning roughly 3.4 times the rate, with a 95 percent confidence interval running from 1.51 to 7.74) is the loudest signal in a cohort study published July 6 in Diabetes, Obesity and Metabolism ↗. A hazard ratio of 1.0 would mean no difference between the groups. Semaglutide ↗, sold for weight loss as Wegovy, and liraglutide ↗, sold as Saxenda, both target the GLP-1 receptor ↗, the gut-hormone switch these drugs flip to blunt appetite.
The researchers pulled electronic records from 2018 to 2025 and mapped them to the OMOP common data model, a shared format that lets separate hospitals pool records. They used a new-user design. Everyone in the drug group was starting the medication for the first time, matched by propensity score against people with similar profiles who did not. That gave 2,357 semaglutide starters against 22,602 non-starters, and 6,953 liraglutide starters against 68,001. Ten safety outcomes were tracked at each hospital, then combined by meta-analysis.
The mental-health numbers dominated. Beyond depression, semaglutide users were more than twice as likely to receive any psychiatric diagnosis (2.02) and about 2.4 times as likely to be diagnosed with an anxiety disorder. They were also nearly four times as likely to be coded for a gut-motility problem or obstruction (3.91), the kind of stalled-stomach complaint the drugs are already known to cause, and modestly more likely to have a vision problem recorded (1.58).
Liraglutide's panel was broader but milder. Its users saw elevated rates of psychiatric diagnoses overall (1.66), anxiety (1.68), and depression (1.51), plus liver trouble (1.46), pancreatobiliary disorders (1.33), the pancreatitis-and-gallstone cluster (1.40), and vision problems (1.34). Almost every arrow pointed the same way: more diagnoses in the people on the drug.
Here is the catch, and it is a big one. The comparison group was people who never took the drug, not people on a different weight-loss treatment. That is the weakest comparator available, because the reasons someone seeks out a GLP-1 (higher weight, more clinic visits, more baseline distress) are also reasons they get diagnosed with more of everything. Someone who starts an injectable and sees a doctor every few weeks has more chances to be coded for depression or a stomach complaint than someone who never came in. Detection can masquerade as risk.
This section has watched the comparator decide the headline before. A Korean cohort earlier this year found that GLP-1 drugs raised depression risk against one diabetes drug and not another ↗; switch the reference group and the verdict flipped. And in people with bipolar disorder, semaglutide was tied to fewer psychiatric hospitalizations ↗, not more. Randomized trials, where the comparison is fixed by design, have not shown these drugs cause depression, which is why regulators reviewing suicidality reports found no causal link. The authors themselves flag that the signal wobbled: semaglutide's vision result and liraglutide's liver result did not survive every sensitivity analysis.
What the study does add is a population. Most GLP-1 safety data comes from diabetes trials, where patients are sicker and watched closely. This is a look at the weight-loss indication specifically, in ordinary hospital records, and it says the psychiatric column deserves a proper prospective study rather than reassurance borrowed from diabetes cohorts. It is a flag, not a verdict. Read it as a reason to measure, not a reason to stop.