Alzheimer's disease (AD) develops over years, creating an opportunity for interventions that target modifiable risk factors before the onset of clinical dementia. Here we applied a federated electronic health record (EHR) system encompassing over 29 million de-identified patients toward a target-trial emulation framework involving 153,412 adults aged 50 years or older with at least one documented AD risk factor. New users of glucagon-like peptide-1 (GLP-1)-based incretin therapy were compared with new users of non-GLP-1 antidiabetic medications after a 12-month washout, with 1:1 propensity-score matching on 30 baseline variables and additional exact matching on index year of therapy initiation and 5-year age band. In the primary matched cohort, 28,901 patients per arm, incretin initiation was associated with lower first recorded AD diagnosis (hazard ratio [HR] 0.46, 95% confidence interval [CI] 0.29-0.73, q = 0.003), all-cause dementia (HR 0.66 [0.55-0.79], q < 0.001), and all-cause mortality (HR 0.46 [0.37-0.58], q < 0.001), with directionally-lower Mild Cognitive Impairment (HR 0.76 [0.61-0.94], q = 0.135) and Alzheimer's-related medication initiation (HR 0.82 [0.65-1.03], q = 0.217). The AD diagnosis mitigation signal was independently reproduced for semaglutide (HR 0.56; N = 23,675 per arm; q = 0.02), with the directionally consistent tirzepatide point estimate (HR 0.60) not reaching significance. GLP-1 RA initiation was also associated with substantially lower (all q < 0.001) incidence of heart failure (HR 0.50 [0.45-0.55]), cardiomyopathy (HR 0.38 [0.31-0.47]), major adverse cardiovascular events (HR 0.74 [0.67-0.82]), acute kidney injury (HR 0.67 [0.67-0.74]), and chronic kidney disease (HR 0.68 [0.62-0.75]). Negative-control outcomes showed no significant separation (all q > 0.45), including allergic rhinitis (HR 0.96 [0.87-1.07]), hemorrhoids (HR 1.07 [0.95-1.21]), and inguinal hernia (HR 1.39 [0.89-2.18]). Stricter two-code ICD definitions supported lower first recorded AD diagnosis in the incretin arm (HR 0.51 [0.29-0.89], q = 0.018) and lower first recorded all-cause dementia diagnosis (HR 0.68 [0.54-0.85], q = 0.003). In 12-month landmark analyses among semaglutide initiators, ≥5% weight-loss responders had lower 3-year AD cumulative incidence than non-responders after matching (0.07% vs. 0.40%; incidence ratio 5.76; P = .036), although this was not significant in the hazard-ratio model (HR 0.54, P = .39). Sustained-dose stratification showed no comparable gradient (high-dose vs. low-dose 0.30% vs. 0.14%; HR 2.77; P = .38), with ≤12 AD events per group. The weight-loss-specific pattern did not extend to all-cause dementia: weight-loss responders and non-responders had similar 3-year cumulative incidence (1.54% vs. 1.70%; HR 0.87, P = .53). Initiation of GLP-1 receptor agonist therapy in adults with documented AD risk factors was associated with lower recorded incidence of AD, dementia, mortality, and multiple cardiovascular and renal outcomes in this observational target-trial emulation. These findings support the hypothesis that earlier incretin therapy may contribute to AD risk modification through upstream cardiometabolic pathways, while prospective randomized prevention studies will be required to determine causality, underlying biological mechanisms, and optimal treatment timing.
