Across 764,115 propensity-matched pairs in the TriNetX network, adults with bipolar disorder, major depression, or schizophrenia who started a GLP-1 receptor agonist had lower 4-year all-cause mortality than those starting an SGLT2 inhibitor: 4.91% versus 6.45% (hazard ratio 0.76, 95% CI 0.74-0.78), an absolute reduction of 1.54 percentage points. The design was a target trial emulation with new-user, active-comparator matching, and cardiovascular endpoints were also analyzed. The abstract is truncated, so the non-SMI results and MACE figures are not shown here.
The comparator choice matters. SGLT2 inhibitors already carry proven cardiovascular and renal benefit, so beating them is a higher bar than beating no treatment, and it blunts classic confounding by indication. Even so, this is observational electronic health record data. Residual confounding remains likely: GLP-1 prescribing tends to track insurance coverage, access to care, and engagement with clinicians, and weight loss, mental health severity, and lifestyle are poorly captured. Cohorts mostly span people with diabetes or obesity, and the 4-year mortality gap may partly reflect who gets prescribed what. A hazard ratio of 0.76 is large enough to warrant suspicion of bias, though it fits earlier GLP-1 work showing cardiovascular benefit. For patients with SMI, who die 10-20 years earlier than average, largely from cardiometabolic disease, and whose antipsychotics and mood stabilizers often drive weight gain, the finding strengthens the case for trials. Treat it as hypothesis-strengthening and confirmatory rather than practice-changing. Anyone with SMI and metabolic risk should discuss these options with their prescriber.