Large, well-designed longitudinal cohort study with up to 18 years of follow-up, prespecified statistical correction, and consistent associations across cognitive and neuroimaging endpoints, though observational design and potential confounding limit practice-changing claims.
Reported
Sample size (cognitive data)2603
Sample size (serial MRI)774
Sample size (autopsy neuropatholo…1287