The Lancet Psychiatry's April issue published the largest within-person cohort study yet on GLP-1 medications and serious mental-health outcomes. Heidi Taipale and colleagues at Karolinska Institutet, the University of Eastern Finland, and Griffith University analyzed Swedish national health records on 95,490 individuals diagnosed with depression or anxiety between 2009 and 2022. Of those, 22,480 took GLP-1 receptor agonists during the follow-up. The within-person design — comparing each patient against periods when they were not on the drug — controls for the confounding that has bedeviled the GLP-1-and-mood literature. [1] The paper reported the headline 44% depression reduction on May 5. The follow-on this week is what makes the result load-bearing for clinical practice.
Semaglutide — Ozempic for diabetes, Wegovy for obesity — was associated with a 42% lower risk of psychiatric hospitalization or extended sick leave during treatment periods (adjusted hazard ratio 0.58). Depression-related outcomes fell 44%. Anxiety fell 38%. Substance-use-disorder outcomes fell 47%. Liraglutide showed a smaller effect — 18% lower for overall mental-illness risk, 26% lower for depression specifically. [1] Exenatide and dulaglutide showed no effect at all (adjusted hazard ratios of 1.01 and 1.01, statistically indistinguishable from no treatment).
The class-effect interpretation is dead. The benefit is specific to the GLP-1 receptor agonists semaglutide and liraglutide, with semaglutide carrying nearly all the mental-health signal. Among 3,705 individuals who used both semaglutide and another GLP-1 during the follow-up, semaglutide produced a 33% lower risk versus the alternative — a within-patient comparison that isolates the molecule from confounding by indication.
The clinical context that makes this Wednesday's news is the scaling of the Wegovy oral pill. The paper has been tracking the Foundayos and Wegovy oral dosing rollout this quarter; the pill print this week registers Wegovy oral at roughly 1.3 million U.S. weekly prescriptions in Q1, capturing 65% of new obesity prescriptions. The Lancet Psychiatry result reframes that print: the same molecule reducing depression-related hospitalization 44% in a Swedish national cohort is now being prescribed to roughly 100 million Americans of obesity-eligible BMI without any psychiatric monitoring labeling, screening protocol, or coordinated psychiatric care infrastructure.
This is structurally different from the SSRI scale-up. SSRIs entered the U.S. market with FDA labeling, black-box warnings (added later), and a primary-care prescribing workflow that included depression screening tools. The Wegovy oral pill is being prescribed for obesity. The mental-health benefit is incidental — beneficial, but unscreened. The 47% substance-use-disorder reduction is the most clinically interesting result in the paper because the dopamine-reward modulation of GLP-1 medications has been the leading neurobiological hypothesis for the addiction-treatment effects observed in smaller studies. None of this is on the Wegovy label.
The authors call for a randomized controlled trial in the diabetes-plus-depression population. That trial — Taipale told the Karolinska press office — is being scoped at Karolinska. The result it would test is straightforward: does semaglutide work as a depression treatment in non-diabetic, non-obese patients with major depressive disorder. The within-person cohort cannot answer that question. The class-specificity of the finding — semaglutide working, exenatide not — points to a molecular mechanism that may or may not generalize.
The earlier September 2024 eClinicalMedicine propensity-matched study from Riccardo De Giorgi's group at Oxford had already shown semaglutide associated with a lower risk of depression diagnosis and dementia compared with sitagliptin and glipizide in a U.S. EHR cohort of 100,000 type-2 diabetics. [2] The Lancet Psychiatry paper extends that finding to people who already had depression or anxiety, with within-person controls and a 14-year follow-up.
What the paper does not yet know is the long-term picture. The 171 deaths by suicide in the cohort — only one during semaglutide use — contributed 0.2% to the primary outcome; the suicide signal is too sparse to draw strong inferences. The 2024 Scientific Reports TriNetX study found a 1.98 hazard ratio for any psychiatric disorder over five years among GLP-1 users versus controls in obesity, but with a different design and patient population. [3] The within-person Swedish design is a stronger inference for safety; it is not a final word.
The clinical takeaway for primary care is narrower than the headline suggests. Semaglutide does not appear to worsen depression or anxiety in patients who already have those diagnoses. It may reduce psychiatric hospitalizations and sick leave during treatment periods. Whether it should be used as primary depression treatment is what the RCT will test. Whether the Wegovy oral pill scale-up should carry a psychiatric monitoring framework is a question the FDA has not yet engaged. May 6 finds these two questions on the same desk.
-- NORA WHITFIELD, Chicago