Abstract/Project Summary Studies of physical and cognitive decline in persons older than 85 years are limited but suggest that late measures of traditional risk factors have little predictive value and some inverse associations with risk. Rigorous evidence regarding modifiable risk factors in midlife and older age could improve dementia prevention efforts and health among the oldest old. In ARIC-NCS, with its large population and >35 years of follow up, including the longest followed cohort of African-Americans for cognition, midlife vascular risk factors (VRF) strongly predict dementia, frailty, physical function, and both cognitive and physical decline occurring before age 85. Only cognitive status and physical function reliably predict dementia after 85. This raises a compelling need to better understand risk among the oldest-old (85+ years) including deeper phenotyping to discover markers which may explain the attenuation and distortion of risk associations at older age. The use of blood-based biomarkers in dementia diagnosis and prediction requires assessments across the lifespan and extension to diverse cohorts. ARIC allows exploration of differences in predictors of dementia (measured from midlife to older age) among distinct risk groups (<85 vs 85+ years, sex, multimorbidity, and race). The proposed fourth cycle of ARIC-NCS builds on high productivity (245 papers since 2009) to continue rigorous data collection (proposed visits 11-13 and dementia surveillance on ~4,000 living participants ages 81-101, ~70% female and ~25% African-American; accumulating 4732 dementia cases, half over age 85 by 2028) and address pressing questions. ARIC-NCS offers major advantages over most published studies, with over 30+ years of data in a diverse cohort starting in midlife on: (1)VRF, multimorbidity and medications; (2)behaviors and social determinants of health (SDOH); (3)genetic, genomic, and metabolic markers; (4) longitudinal plasma biomarker assessment; (5)quantitative brain imaging (5438 MRIs and 1146 amyloid PET; an additional ~400 of each expected in 2023); and (6)clinical outcomes, to which we will collect additional measurement of multimorbidity, repeated biomarkers, cognitive and physical function, sleep and activity. ARIC-NCS can control for an extensive range of covariates and modifiers, and we will consider race- and sex-specific analyses. The specific aims are: 1) Longitudinally study blood biomarkers (~5000 proteins and targeted blood AD biomarkers (amyloid, p-tau, NfL and GFAP) across 4 decades) in relation to: (i) dementia; (ii) mild cognitive impairment; (iii) multimorbidity and frailty; (iv) cognitive and physical function (e.g. short physical function battery, gait speed) decline. 2) To evaluate the association and interaction of midlife VRF, multimorbidity (including disturbed sleep), and SDOH with blood (proteomic and targeted AD and neurodegeneration markers) and brain imaging (MRI & amyloid PET) dementia biomarkers and their prog...