GLP-1 medicines linked to fewer heart attacks, heart failure cases, and deaths in adults with serious mental illness
A Journal of the American Heart Association study of 218,043 adults with serious mental illness, type 2 diabetes, and obesity links GLP-1 use to lower one- and three-year risks of heart attack, heart failure, and death, within the limits of health-records research.
Researchers from St. John's Episcopal Hospital and Washington University in St. Louis reported in the Journal of the American Heart Association on October 16, 2026 a retrospective, propensity score-matched cohort study of 218,043 adults drawn from the TriNetX database, which spans 73 health care organizations. Every participant had serious mental illness, type 2 diabetes, and obesity, and was taking antipsychotic or antidepressant medication, a group often left out of clinical trials. The study compared GLP-1 receptor agonist users with matched nonusers on new heart attacks (myocardial infarction), new heart failure diagnoses, and deaths from any cause at one and three years.
GLP-1 use was associated with lower risk on all three outcomes. Risk differences at one and three years were -0.007 and -0.010 for heart attack, -0.011 and -0.018 for heart failure, and -0.039 and -0.065 for death, all statistically significant (P<0.001). In absolute terms at three years, that is about 1.0, 1.8, and 6.5 fewer events per 100 people. Semaglutide and tirzepatide showed the same pattern, and findings were directionally consistent after accounting for SGLT2 inhibitor and DPP-4 inhibitor use.
The question matters because people with serious mental illness carry elevated cardiovascular risk, and the antipsychotic medications many of them take commonly drive weight gain and metabolic problems. Evidence for GLP-1 medicines has come mostly from the general population, leaving clinicians with little direct data for this group.
The authors are explicit about the limits: this was an observational study of administrative health records, so it shows association, not proof that the medicines caused the lower event rates. Serious mental illness is clinically varied, diagnoses came from billing and records data rather than structured assessment, and unmeasured differences between users and nonusers can remain after matching. Subgroup analyses by sex, race, and other drug use were exploratory.
This brief summarizes a peer-reviewed observational study. It is not medical advice, does not recommend starting or stopping any medication, and says nothing about dosing or where to obtain these drugs. Treatment decisions for people with serious mental illness belong with their clinicians.