People with serious mental illness (SMI) die 10 to 25 years earlier than the general population, with cardiovascular disease accounting for much of the excess mortality. Cardiometabolic risk is amplified by obesity, diabetes, smoking, psychotropic medication effects, and persistent gaps in preventive medical care. Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) can improve weight and glycemic measures, including in patients receiving antipsychotics, but whether these agents are associated with longer survival in SMI has been uncertain. This study compared mortality and cardiovascular outcomes after initiation of a GLP-1 RA versus a sodium-glucose cotransporter-2 (SGLT2) inhibitor in adults with and without SMI.

Study Design
Investigators conducted a retrospective target trial emulation using the TriNetX Analytics Network, a federated electronic health record system covering more than 120 million patients at 114 health care organizations. Adults aged 18 years or older with a first qualifying prescription for a GLP-1 RA or SGLT2 inhibitor were included in a new-user, active-comparator design after a 365-day washout. Separate cohorts were constructed for people with SMI (major depressive disorder, bipolar disorder, or schizophrenia) and people without SMI. Propensity scores incorporated demographic characteristics, cardiometabolic disease, laboratory measures, substance use indicators, diabetes severity, and psychiatric medication exposure. Participants were analyzed according to the initially prescribed class, regardless of later discontinuation, switching, or combination therapy.

The primary outcome was all-cause mortality within four years; one-year mortality was the key secondary outcome. For the primary analysis, 1,528,230 adults were propensity score matched: 195,184 GLP-1 RA initiators were matched to the same number of SGLT2 inhibitor initiators in the SMI cohort, and 568,931 pairs were included in the non-SMI cohort. Mean age was approximately 60 years. Most measured characteristics were well balanced after matching, although continuous body mass index and the prevalence of type 2 diabetes remained modestly imbalanced in the SMI cohort and required additional analyses.

Key Findings
Among participants with SMI, death occurred in 4.91% of GLP-1 RA initiators and 6.45% of SGLT2 inhibitor initiators during the four-year analysis window. GLP-1 RA initiation was associated with lower mortality (hazard ratio [HR], 0.76; 95% CI, 0.74-0.78; P < .001), an absolute risk difference of -1.54 percentage points. The association appeared early: in a separately matched one-year cohort, mortality was 1.46% with GLP-1 RAs and 2.84% with SGLT2 inhibitors (risk ratio [RR], 0.52; 95% CI, 0.49-0.54), an absolute difference of -1.38 percentage points. Median follow-up in the four-year SMI analysis was about 2.6 years in both groups.

GLP-1 RA vs SGLT2 Inhibitor: Mortality and MACE Outcomes

Comparator for all boxes: SGLT2 inhibitor initiation (an active drug, not placebo).

HR = hazard ratio (time-to-event); RR = risk ratio (cumulative risk) — not directly comparable magnitudes. ARD = absolute risk difference, in percentage points.

The absolute mortality difference was larger in the SMI cohort than in the non-SMI cohort. Among participants without SMI, four-year mortality was 2.97% with GLP-1 RAs and 3.57% with SGLT2 inhibitors (HR, 0.85; 95% CI, 0.83-0.86; absolute risk difference, -0.60 percentage points). In participants with SMI and type 2 diabetes, semaglutide initiation was also associated with lower risks of 3-point major adverse cardiovascular events (MACE; HR, 0.77), 5-point MACE (HR, 0.76), myocardial infarction (HR, 0.71), stroke (HR, 0.89), heart failure (HR, 0.73), and coronary artery bypass grafting (HR, 0.77); all comparisons were statistically significant.

Exploratory results suggested that the signal was not uniform across the GLP-1 RA class. Semaglutide and tirzepatide were associated with the clearest mortality reductions, whereas dulaglutide, liraglutide, exenatide, lixisenatide, and albiglutide did not show a consistent survival advantage across analytic methods. Semaglutide was associated with lower mortality in the major depressive disorder, bipolar disorder, and schizophrenia subgroups. Combination therapy with an SGLT2 inhibitor plus semaglutide was also associated with lower mortality than SGLT2 inhibitor monotherapy. These analyses should be considered hypothesis-generating: the windows extended up to 10 years, but observed median follow-up was much shorter, and tirzepatide initiators had a median follow-up of only 424 days.

Limitations and Caveats
The primary association persisted after stratification by diabetes, exclusion of baseline heart failure and chronic kidney disease, additional adjustment for residual metabolic imbalance, categorical body mass index balancing, and censoring at treatment crossover. Excluding baseline heart failure and chronic kidney disease was especially relevant because these conditions can channel patients toward SGLT2 inhibitors; however, diagnostic codes could not fully capture disease severity, so residual channeling bias remains possible.

The study's scale, active-comparator design, propensity matching, and convergence across mortality and cardiovascular end points strengthen confidence that the association is clinically important. Nevertheless, target trial emulation does not reproduce randomization. The electronic health record data could not fully account for medication adherence, lifestyle, socioeconomic conditions, psychiatric illness severity, care received outside participating systems, or other unmeasured differences between treatment groups.

Exposure was defined by the first prescription rather than sustained treatment, missing values were not imputed, and access barriers may limit generalizability. The study also did not establish superiority to placebo or no treatment, and it cannot determine whether starting a GLP-1 RA solely to reduce mortality is beneficial.

Reference:
McIntyre RS, Zhang-James Y, Kwan ATH. Glucagon-like peptide 1 receptor agonists, mortality, and cardiovascular outcomes in serious mental illness. JAMA Psychiatry. Published online August 26, 2026. doi:10.1001/jamapsychiatry.2026.2574. Full Article