GLP-1 agonists linked to less psychiatric acute care in schizophrenia and diabetes

Background

Many people with schizophrenia continue to have psychotic symptoms despite antipsychotic therapy, and the condition carries a disproportionately heavy burden of metabolic dysregulation¹. It is not known whether GLP-1 receptor agonists provide psychiatric benefit in addition to their metabolic effects¹.

Study design

The investigators used electronic health record records held in the NeuroBlu database (USA) to build an observational cohort of people aged 18 years or older with recorded schizophrenia plus type 2 diabetes¹. They compared starting a GLP-1 receptor agonist with three alternatives: starting none of the study drugs, starting a DPP-4 inhibitor or starting an SGLT2 inhibitor¹.

Time zero was an initial prescription for one of these drugs from Jan 1, 2018, and Oct 1, 2025. For non-initiators, it was the index date¹. Follow-up lasted up to 2 years. It ended sooner at the final clinical observation in the repository or at the close of the observation window (March 24, 2026)¹.

Inverse probability weighting balanced the groups on demographic, pharmacological and clinical characteristics, and the investigators fitted weighted regression models¹. They used an intention-to-treat approach, classifying exposure by the drug started at time zero¹.

Results

Of 70,257 eligible individuals, the mean age was 53.1 years (SD 12.32) and 33 882 (48.2%) were women¹. Treatment starts were distributed as follows¹:

• 9,093 (12.9%) started a GLP-1 receptor agonist.
• 4,968 (7.1%) started a DPP-4 inhibitor.
• 5,328 (7.6%) started an SGLT2 inhibitor.
• 50,868 (72.4%) started none of these drugs.

Psychiatric hospitalisation: starting a GLP-1 receptor agonist was associated with lower risk. The weighted HR was 0.76 (95% CI 0.71-0.82) versus non-initiation, 0.82 (0.75-0.90) versus DPP-4 inhibitors and 0.83 (0.78-0.88) versus SGLT2 inhibitors¹.

Emergency department presentation: the corresponding weighted HRs were 0.71 (0.65-0.77), 0.87 (0.80-0.95) and 0.80 (0.73-0.88)¹.

Interpretation and limitations

The authors note that possible selection bias as well as residual confounding limit how far the results can be generalised¹. They suggest that reduced psychiatric acute-care use might represent a welcome additional benefit where GLP-1-based treatment is clinically indicated for metabolic control¹. Any role as a primary psychiatric treatment would need to be tested through randomised trials and mechanistic research¹.

The work was funded by the SYNDICAT project (GNT2035912)¹.