GLP-1 drugs lowered the three-year risk of heart attack, heart failure, and death in 218,043 adults with serious mental illness, type 2 diabetes, and obesity. The trials that built the case for these drugs rarely enrolled anyone like them.
Serious mental illness means schizophrenia, bipolar disorder, or depression severe enough to need antipsychotic or antidepressant medication. People in that group die years earlier than the general population, mostly from heart disease, and the antipsychotics themselves tend to drive weight gain and diabetes. They are also among the people most often left out of the randomized trials that made GLP-1 drugs standard care, because a serious psychiatric diagnosis is a common reason to exclude a volunteer. So the evidence is thinnest exactly where the cardiometabolic risk is highest.
The study, published October 16 in the Journal of the American Heart Association ↗ and led by Samuel Governor of St. John's Episcopal Hospital with colleagues at Washington University in St. Louis, is a retrospective look at records from 73 health systems in the TriNetX database. The authors matched adults who were taking a GLP-1 drug against similar adults who were not, lining the two groups up on age, sex, and other illnesses, then counted heart attacks, heart failure, and deaths at one and three years.
At three years the GLP-1 group came out ahead on all three. Their absolute risk of death was 6.5 percentage points lower, which works out to roughly six or seven fewer deaths for every hundred patients over that stretch. Heart failure was 1.8 points lower and heart attack 1.0 point lower. The gaps were smaller at one year but already pointing the same way. Semaglutide ↗ and tirzepatide ↗, pulled out and looked at on their own, each showed the same direction.
The catch is the comparison group. These patients were measured against people who took no GLP-1 drug, not against people on a different diabetes medication. Someone who starts and stays on a weekly injectable tends to be more engaged with their care than someone taking nothing, and that difference alone can make the drug look better than it is. It is a bias with a name, healthy-user bias, and the cleaner way around it is an active comparator. Our earlier coverage of a 171,917-person Swedish study ↗ used that stronger design, measuring GLP-1 users against SGLT2-inhibitor users rather than against nonusers. This study's authors ran a check that accounted for SGLT2 and DPP-4 use, and the benefit held in the same direction, but a sensitivity analysis is not a head-to-head.
The diagnoses and the outcomes both came from billing and administrative records, not from chart review, so the mental-illness labels and the event counts carry the usual records-study noise. Serious mental illness is also not one condition, and pooling schizophrenia with treated depression hides a lot of variation. The authors flag both limits directly.
What the study does well is size and aim. It is among the largest attempts to ask whether the benefit everyone now assumes for GLP-1 drugs actually reaches the population that the trials fenced out, and the answer, with the caveats attached, is that it appears to.
Both drugs named here act on the GLP-1 receptor ↗, the gut-hormone switch that signals fullness and steadies blood sugar; tirzepatide hits a second receptor, GIP, on top of it. The cardiovascular case for the class was built in people without severe psychiatric illness. This is one of the first large signals that the same wiring pays off in the group that needed it most and was studied least.