Lancet Psychiatry: semaglutide linked to 42 percent lower risk of worsening mental illness

Researchers in Sweden, Finland and Australia report that people with depression or anxiety had substantially less psychiatric hospital care and fewer sick leave days during periods when they were taking semaglutide compared with periods when they were not. The study was published March 19, 2026, in The Lancet Psychiatry ^[1][2].

The analysis drew on Swedish national registers covering 2009 to 2022. It included more than 95,000 people who had been diagnosed with depression or anxiety and who were prescribed some form of antidiabetic medication. Of those, 22,480 had used a GLP-1 receptor agonist at some point ^[2].

Instead of comparing users with non-users — two groups that often differ in ways that are hard to measure — the researchers compared each person against themselves. They looked at stretches of time when the same individual was using a given medication against stretches when they were not ^[2]. That approach helps control for stable personal traits such as genetics, education or long-standing illness, though it cannot account for things that change over time.

What the numbers showed

During periods of semaglutide use, the risk of sick leave or hospital care for psychiatric reasons was 42 percent lower than during non-use periods ^[2]. Broken out by diagnosis, the reduction was 44 percent for depression, 38 percent for anxiety disorders, and 47 percent for substance use ^[2].

The results were not uniform across the drug class. Liraglutide was linked to an 18 percent lower risk of psychiatric sick leave or hospital care. Exenatide and dulaglutide showed no significant risk reduction ^[2]. As a group, however, GLP-1 drugs were associated with a lower risk of self-harm ^[2].

Semaglutide is the active ingredient in Ozempic, Wegovy and Rybelsus; the Karolinska Institutet announcement refers to it as the ingredient in Ozempic ^[2]. The study did not report on tirzepatide (Mounjaro, Zepbound) or orforglipron (Foundayo), according to the available materials.

The authors are explicit about the limits of what they found. "Our findings suggest that GLP-1 drugs, particularly semaglutide, might contribute to better mental health in people with diabetes and obesity, but since this was an observational study, controlled clinical trials are needed to confirm the results," said senior author Jari Tiihonen, a specialist physician and professor at the Centre for Psychiatry Research at Karolinska Institutet ^[2].

The work was funded by the Sigrid Jusélius Foundation. Several authors have taken part in research projects funded by grants from Janssen to their institutions, and some, including Tiihonen, have ties to other pharmaceutical companies; the full disclosure is in the published paper ^[2].

Why it matters for patients

The mental health effects of GLP-1 medicines have been an open question for adults weighing these drugs. This study adds large-scale, real-world data pointing in a reassuring direction for people who already carry a depression or anxiety diagnosis, at least for semaglutide and to a lesser extent liraglutide ^[2].

Several caveats matter. First, an observational register study measures association, not cause. People may start or stay on a medication during stable stretches of life and stop during difficult ones, which can make a drug look protective. The researchers' within-person design reduces but does not eliminate that problem.

Second, the outcomes measured were hospital care and sick leave — serious, documented events ^[2]. They are not the same as day-to-day mood, sleep or anxiety symptoms, which this design would not capture.

Third, the study population was people with depression or anxiety who were prescribed antidiabetic drugs in Sweden ^[2]. Whether the same pattern would appear in US patients, or in people taking these medicines for weight management without diabetes, is not addressed in the available materials.

Finally, the differences between drugs are notable. Two GLP-1 medicines in the study showed no significant psychiatric benefit ^[2]. That suggests any effect may not be a simple class-wide property, though the reasons are not established in these sources.

What happens next

The paper appeared online March 19, 2026, with DOI 10.1016/S2215-0366(26)00014-3 ^[2]. The authors call for controlled clinical trials to test whether the association reflects a real treatment effect ^[2]. No such trials, timelines or regulatory actions are described in the sources provided, so it is not yet known when confirmatory evidence might arrive or whether any label changes would follow.

People with questions about how these medications interact with their own mental health history can raise them with the clinician managing their care.