Saved in this browser. Sign in to sync across devices.
The critical distinction: The study involved adults with both schizophrenia and type 2 diabetes. GLP-1 medicines are used for metabolic indications, not FDA-approved to treat schizophrenia. These results do not justify prescribing one solely to prevent psychosis, or changing an antipsychotic or diabetes plan without the treating team.
What did researchers compare?
The investigators studied electronic health records from 70,257 adults in the United States with schizophrenia and type 2 diabetes, looking at whether starting GLP-1 receptor agonist–based therapy was associated with psychiatric acute-care use during follow-up. Of those eligible, 9,093 initiated a GLP-1–based medicine, 4,968 initiated a DPP-4 inhibitor, 5,328 initiated an SGLT2 inhibitor, and 50,868 did not initiate one of these medication classes.
Researchers used statistical weighting to account for recorded health and medication characteristics, then compared psychiatric hospitalizations and emergency department visits. It was not a randomized experiment.
PSYCHIATRIC HOSPITALIZATION
Hazard ratio 0.76
Starting GLP-1–based therapy was associated with a 24% lower adjusted relative hazard than not starting one of the compared medicines (95% CI 0.71–0.82).
PSYCHIATRIC EMERGENCY VISITS
Hazard ratio 0.71
The adjusted hazard was 29% lower relative to non-initiation (95% CI 0.65–0.77). These percentages are relative associations, not an individual's absolute chance of avoiding care.
ACTIVE COMPARISONS
Also compared with other diabetes medications
GLP-1 initiation was associated with lower risks than initiation of DPP-4 inhibitors or SGLT2 inhibitors in weighted analyses. That helps contextualize the association but cannot eliminate all treatment-selection differences.
LIMITATION
No causal conclusion
Residual confounding, differences in care, selection bias and limitations of electronic records could explain part of the result. The study did not randomly assign treatments.
Does this mean GLP-1 medicines improve hallucinations or delusions?
No direct conclusion can be drawn about that. The study measured hospitalizations and emergency visits, not whether an individual person's hallucinations, delusions, negative symptoms or daily functioning improved because of a GLP-1 medicine. A reduction in acute-care use might reflect multiple factors, including physical health management and differences in clinical care.
The study also cannot be generalized to people with schizophrenia who do not have type 2 diabetes. Randomized trials are needed before GLP-1 medicines can be evaluated as primary psychiatric treatments.
Why should physical and psychiatric health be discussed together?
Metabolic health is a major concern for many people with schizophrenia, including people who experience weight, blood glucose or lipid changes during antipsychotic treatment. Coordinated psychiatry and primary-care or endocrinology follow-up can address both mental and physical health without assuming a diabetes medicine will treat psychosis.
How this differs from an approved schizophrenia medication
WELTRUZA, a monthly form of olanzapine, is FDA-approved to treat schizophrenia in adults. GLP-1 medicines have metabolic indications, but are not approved to treat schizophrenia. This observational cohort did not specifically study WELTRUZA or show that taking a GLP-1 medicine prevents side effects from olanzapine. Discuss physical and psychiatric health together rather than treating the two findings as interchangeable.
Read the WELTRUZA approval and safety explainer →Questions to bring to a clinician
- What metabolic measurements and risk factors should we review with the psychiatric treatment plan?
- If diabetes treatment is indicated, how should the diabetes and psychiatric teams coordinate?
- What benefits, side effects, contraindications, drug interactions, cost and access issues matter for a particular medication?
- How should changes in psychiatric symptoms, emergency visits, weight and overall functioning be tracked separately?
- Which findings are established for the patient's medical condition, and which are still hypotheses?
What to watch next
Randomized trials measuring psychiatric symptoms, metabolic outcomes, safety and day-to-day functioning—particularly in people without diabetes—would help determine whether the association represents a direct treatment benefit. For now, this is useful research on integrated care, not new prescribing guidance for schizophrenia.
Original research and clinical context
- The Lancet Psychiatry — original observational cohort (Davoodian and colleagues) ↗
Peer-reviewed; October 2026 online reporting, November 2026 issue
- PubMed — study abstract, full methods summary and effect estimates ↗
PMID 42874603
- NIMH — Understanding schizophrenia ↗
Established condition information and treatment context
Evidence and links rechecked October 20, 2026. Educational editorial draft; independent clinical review pending. Information here is not a diagnosis or individualized medication recommendation.