In the context of a deprived neighborhood, does self‐identified race cause dementia?

Abstract Background Although accumulating evidence has shown that individual socioeconomic status affects dementia incidence, it remains unknown whether a higher degree of neighborhood disadvantage which is independent of individual socioeconomic status contributes to the excess risk of dementia among African Americans (AA). Method In the Monongahela‐Youghiogheny Healthy Aging Team and 15104 Senior Project, two population‐based prospective cohort studies in an economically disadvantaged Rust Belt community, 2733 participants aged 65+ years were assessed for their health history, lifestyle, and cognitive functioning. Mild cognitive impairment (MCI) or dementia was defined as Clinical Dementia Rating score ≥0.5. Each participant was assigned an Area Deprivation Index (ADI) value according to the census block group where one resided, indicating the level of neighborhood disadvantage. In cross‐sectional data, racial differences in the odds of MCI or dementia were assessed using weighted logistic regressions. Inverse probability weighting was used to balance AA and European Americans (EA) with respect to 1) ADI, 2) other confounders (age, sex, APOE*4 genotype, cardiometabolic diseases, individual socioeconomic status, health‐related behaviors, use of regular health services, functional health literacy, and air pollution [PM 2.5 ]), and 3) ADI and other confounders. Result The frequency of MCI and dementia were 648 (24%) and 25 (0.9%), respectively. AA had significantly higher proportion of MCI or dementia (35% vs. 24%, p = 0.001) and higher mean ADI (96 vs. 79, p<0.0001) than EA. Being AA was associated with higher odds of MCI or dementia compared to being EA (odds ratio [OR] = 1.71, 95% confidence interval [CI] = 1.26 ‐ 2.33, unadjusted model). After balancing racial groups with respect to ADI only, the difference between AA and EA no longer reached the level of significance. When balancing all other confounders only, the OR was 1.66 (95% CI = 1.04 ‐ 2.67) and the confounders explained 5.2% of the observed disparity. After balancing both ADI and other confounders, the difference again became non‐significant (OR = 1.61, 95% CI = 0.93 ‐ 2.80) and resulted in 11.1% of the MCI or dementia disparity being explained. Conclusion A higher degree of neighborhood disadvantage may contribute to the excess risk of MCI or dementia found among AA.