As of October 17, 2026, the day this article was written, a new GLP-1 schizophrenia study appeared in The Lancet Psychiatry. It tracked 70,257 U.S. adults with both schizophrenia and type 2 diabetes. Patients who started a GLP-1 drug had a 24% lower adjusted hazard of psychiatric hospitalization than non-initiators, a comparison that accounts for event timing.
They also had a 29% lower adjusted hazard of a psychiatric emergency visit than non-initiators. That is an encouraging signal, but it does not prove diabetes shots treat psychosis. Researchers reviewed routine health records rather than assigning treatment by chance.
GLP-1 medicines are not FDA-approved to treat schizophrenia. No one should change an antipsychotic medication on their own.
What the GLP-1 schizophrenia study found on hospitalization
Najmeh Davoodian and colleagues analyzed U.S. NeuroBlu health records for 70,257 adults aged 18 and older. Slated for the November 2026 issue, the paper went online at 23:30 UK time on Wednesday, October 14, 2026, per the UK Science Media Centre. All patients had schizophrenia and type 2 diabetes, with a mean age of 53.1 years (SD 12.32) and 33,882 women (48.2%).
Researchers compared four groups after an index visit from January 1, 2018. First, 9,093 adults (12.9%) started a GLP-1 drug between January 1, 2018 and October 1, 2025. Second, 4,968 (7.1%) started a DPP-4 inhibitor and 5,328 (7.6%) started an SGLT2 inhibitor, two other diabetes drug classes.
The remaining 50,868 adults (72.4%) were non-initiators, which does not mean receiving no diabetes care. Follow-up lasted up to two years, ending at the last observation or March 24, 2026.
An October 15, 2026 Lancet-supplied summary on Medical Xpress named semaglutide, liraglutide, tirzepatide, and dulaglutide. The primary abstract gives class-level figures only and ends first prescriptions on October 1, 2025, not the summary's October 2026 date.
Inverse-probability weighting balanced measured clinical and medication differences between groups. An intention-to-treat framework grouped patients by initial drug, without random assignment. Weighted hazard ratios compare how often an event occurs over time between groups.
For GLP-1 psychiatric hospitalization, the hazard ratio was 0.76 (95% CI 0.71–0.82) versus non-initiation and 0.82 (0.75–0.90) versus DPP-4 inhibitors. Against SGLT2 inhibitors, the hospitalization hazard ratio was 0.83 (0.78–0.88). For psychiatric emergency visits, the hazard ratio was 0.71 (95% CI 0.65–0.77) versus non-initiation and 0.87 (0.80–0.95) versus DPP-4 inhibitors.
Against SGLT2 inhibitors, the emergency visit hazard ratio was 0.80 (0.73–0.88). These are comparisons of event timing between groups, not 24 or 29 fewer affected people per 100. Absolute event rates were not reported in the abstract.
GLP-1 antipsychotic weight gain and randomized trials
The study's authors describe the association as a possible secondary benefit when a GLP-1 drug is already indicated for metabolic care. Antipsychotics are central to care, per the 2024 National Institute of Mental Health guide. Yet drugs like clozapine and olanzapine often contribute to weight gain and diabetes.
The UK National Institute for Health and Care Excellence guideline CG178 highlights metabolic risks including weight gain and diabetes. Published in February 2014 with a September 4, 2026 header update, it recommends baseline and ongoing physical checks. Clinicians track weight, waist size, blood pressure, lipids, and glucose or HbA1c (two-to-three-month average blood sugar).
Two Danish randomized trials—where treatment is assigned by chance—tested weekly semaglutide for antipsychotic weight gain. The 30-week double-blind HISTORI trial in JAMA Psychiatry, online September 3, 2025, enrolled 154 adults aged 18 to 60.
Participants (77 per group) had schizophrenia, prediabetes and overweight or obesity, with a body mass index of at least 27. They had received stable second-generation antipsychotic treatment for at least six months. They took semaglutide up to 1.0 mg weekly or placebo alongside regular psychiatric care, with funding from the Novo Nordisk Foundation and study drug from Novo Nordisk A/S.
| HISTORI, change from baseline at week 30 | Semaglutide | Placebo | Between-group difference (95% CI) |
|---|---|---|---|
| Body weight, kg | −9.04 | 0.17 | −9.21 (−11.68 to −6.75) |
| HbA1c, percentage points | −0.38 | 0.08 | −0.46 (−0.56 to −0.36) |
Semaglutide lowered weight by 9.04 kg within its arm, or 9.21 kg more than placebo. Scores on the PANSS-6, a six-item psychotic-symptom scale, showed no significant difference between groups.
A 26-week 2026 trial in JAMA Psychiatry by Marie R. Sass randomized 73 adults aged 18 to 65 at three Danish sites. All took clozapine or olanzapine started within five years and had early glycemic abnormalities; 57 (78%) completed.
At 26 weeks, the semaglutide group's HbA1c change was 0.25 percentage points lower than placebo (95% CI, 0.16 to 0.33 lower). Its weight change was 9.2 kg lower (95% CI, 5.1 to 13.3 lower). Neither trial detected a significant difference in psychiatric symptoms, and neither established relapse prevention.
What the data do not mean for Ozempic and schizophrenia
Headlines about fewer hospital stays can make these drugs sound like psychiatric treatments. This study did not establish that. Davoodian and colleagues note that "Potential residual confounding and selection bias restrict generalisability of the results" in routine records.
In the October 14, 2026 UK Science Media Centre reaction, Dr. Jon van Niekerk of the Royal College of Psychiatrists agreed. He also declared his chief medical officer role at Elysium Healthcare. He cautioned that "it would be premature to regard GLP-1 medications as treatments for schizophrenia" today.
Van Niekerk pointed to unmeasured differences in illness severity, antipsychotic prescribing, and healthcare engagement. Better metabolic health, steadier antipsychotic adherence, or closer clinical contact remain hypotheses rather than proven brain mechanisms. The study also cannot show a weight-independent effect or whether results hold in schizophrenia without type 2 diabetes.
GLP-1 drugs are not FDA-approved to treat schizophrenia. Labels for Ozempic (revised 5/2026) and Wegovy (revised 6/2026) list diabetes, weight, cardiovascular, kidney, or liver (MASH) indications only. The Zepbound label (tirzepatide, a dual GIP/GLP-1 agonist, revised 8/2026) covers weight management and obstructive sleep apnea.
Similarly, Saxenda (revised 6/2026) and Trulicity (revised 3/2026) cover weight, diabetes, or cardiovascular indications only. Our guides on whether semaglutide causes depression and whether tirzepatide causes depression cover the FDA's January 13, 2026 communication (current April 3, 2026). That review "found no increased risk of SI/B associated with the use of GLP-1 RA medications" for suicidal ideation or behavior.
It is a safety conclusion, not proof of psychiatric efficacy.
Medication safety, compounded options, and what to watch next
Anyone taking antipsychotics should keep their prescriber involved. The NIMH guide advises against stopping medication without first talking with a healthcare provider. GLP-1 drugs also slow gastric emptying, meaning pills leave the stomach more slowly.
Wegovy's label states to "Monitor the effects of oral medications concomitantly administered with WEGOVY" during care. It adds to "Consider increased clinical or laboratory monitoring" for narrow-therapeutic-index drugs, where small blood-level changes matter. Zepbound's label states to "Monitor patients on oral medications dependent on threshold concentrations for efficacy and those with a narrow therapeutic index" similarly.
In HISTORI, "Gastrointestinal symptoms were more frequent in semaglutide-treated patients" than on placebo, while serious adverse events were 11.7% versus 18.2%. Admissions numbered 68 across 15 semaglutide patients versus 47 across 24 placebo patients, driven by a few frequently admitted participants. In the Sass trial, "Gastrointestinal adverse events were common but mild and transient; psychiatric adverse events were similar across groups" over 26 weeks.
On Ozempic's label, "OZEMPIC is contraindicated in patients with:" "A personal or family history of MTC or in patients with MEN 2" syndromes. Those terms mean medullary thyroid carcinoma and Multiple Endocrine Neoplasia syndrome type 2.
The FDA's October 1, 2026 statement notes, "However, compounded drugs are not FDA approved" by the agency. Compounded drugs are not reviewed for safety, effectiveness, or quality before marketing. The FDA adds that "Patients should obtain a prescription from their doctor and fill the prescription at a state-licensed pharmacy" instead.
The randomized trials did not test Promise's compounded preparations, and the record study does not establish outcomes for them. Promise does not offer liraglutide or dulaglutide.
Semaglutide and tirzepatide prescribed through Promise are dispensed by a licensed U.S. compounding pharmacy. Each is dispensed as a compounded medication, which is different from an FDA-approved product: the formulation offered here is not FDA-approved.
Even so, a licensed provider may still prescribe — that decision is between you and your doctor. At Promise, a licensed provider reviews every request and not everyone qualifies. Next, researchers need randomized trials testing psychiatric hospitalization directly and checking whether results hold in schizophrenia without type 2 diabetes.
