Semaglutide use tied to 26% lower predicted 5-year dementia risk in 1.2 million patient cohort

The study examined 1,183,000 adults with type 2 diabetes from the TriNetX network between 2017 and 2024. Propensity-matched cohorts compared semaglutide initiators against users of other antidiabetic agents. Predicted dementia risk was derived from validated machine-learning models trained on ICD codes and cognitive assessments.

Data showed hazard ratios of 0.74 for any dementia and 0.71 for Alzheimer's-type codes at five years. Effect size remained stable after adjustment for BMI, HbA1c, and cardiovascular events. No similar signal appeared for other GLP-1 agonists at equivalent exposure duration.

Operational impact centers on prescription patterns rather than immediate clinical guidelines. Current evidence remains observational; confounding by indication and unmeasured lifestyle factors cannot be excluded. Neurology and endocrinology services will monitor pending results from the ongoing EVOKE and SELECT-MRI trials.

Next milestones include readout of cognitive endpoints in the semaglutide cardiovascular outcomes extension by late 2026 and potential FDA label review if RCT data confirm incidence reduction above 15%.