In a propensity-score matched cohort of 57,802 adults aged 50+ with documented Alzheimer's risk factors drawn from 29 million de-identified EHR records, GLP-1-based incretin therapy was associated with a 54% reduction in first recorded Alzheimer's diagnosis (HR 0.46, 95% CI 0.29–0.73) versus non-GLP-1 antidiabetic medications. Semaglutide independently reproduced this signal (HR 0.56, N=47,350). All-cause dementia fell 34% (HR 0.66), all-cause mortality 54% (HR 0.46), heart failure 50% (HR 0.50), and chronic kidney disease 32% (HR 0.68). Negative-control outcomes showed no separation, strengthening internal validity.
These effect sizes are striking and deserve cautious interpretation. This is an observational EHR emulation, not a randomized trial — residual confounding remains plausible despite 30-variable matching, since GLP-1 users may differ in unmeasured health-seeking behaviors or metabolic phenotypes. The cohort is restricted to adults already on antidiabetic therapy, limiting generalizability to non-diabetic populations now pursuing GLP-1s off-label. Still, the breadth of endpoints, the dose-response signal from weight-loss responders, and clean negative controls make this one of the most methodologically rigorous real-world analyses of GLP-1 neuroprotection to date. It meaningfully advances the case — built on prior smaller studies linking GLP-1 receptors to neuroinflammation and amyloid clearance — that semaglutide-class drugs may constitute a genuine preventive intervention against dementia. Randomized trials in cognitively at-risk non-diabetic adults are now urgently warranted.