Modifiable Contributors to Socioeconomic Inequality in Brain Aging

Abstract

Importance Socioeconomic disadvantage is associated with accelerated brain aging. However, the modifiable factors accounting for this association, and whether they differ across socioeconomic indicators, remains unclear. Objective To determine whether modifiable risk factors account for socioeconomic differences in the brain age gap, and whether the contributions of these risk factors differ between individual-level and area-level socioeconomic indicators. Design, Setting, and Participants This cohort study used data from participants in the UK Biobank, a population-based cohort recruited at ages 40 to 70 years from 2006 to 2010. Participants with T1-weighted and T2-FLAIR brain magnetic resonance imaging at first imaging visit were eligible; 7700 participants used for model development in previous work were excluded, yielding 36 878. Data were analyzed from April to July 2026. Exposures Household income, highest educational attainment, and area-level deprivation (Townsend Deprivation Index). Main Outcomes and Measures Brain Age Gap (predicted minus chronological age, years) from T1-weighted (primary) and T2-FLAIR (secondary) magnetic resonance imaging, derived with a deep learning model. Eleven risk factors and the Life's Essential 8 composite cardiovascular health score were modeled as mediators; indirect effects were estimated in single-mediator and parallel mediation models with 95% confidence intervals from 5000 bootstrap resamples. Results Among 36 878 participants (mean [SD] age, 65.1 [7.7] years; 20 360 [55.2%] female), lower income and greater area deprivation were associated with a larger brain age gap: lowest vs highest income group, 0.35 years (95% CI, 0.22-0.47); most vs least deprived quartile, 0.33 years (95% CI, 0.23-0.43). Education showed weaker associations that differed in direction between imaging contrasts. Life's Essential 8 score mediated 36% of the income (indirect effect, 0.031 [95% CI, 0.025-0.036]) and 17.9% of the area-deprivation (0.024 [95% CI, 0.019-0.029]) associations. In parallel models, where all risk factors were entered simultaneously, smoking was the largest mediator for both income (0.022 [95% CI, 0.016-0.028]) and area-deprivation (0.030 [95% CI, 0.023-0.037]). Higher income was associated with higher alcohol intake, offsetting part of the income association. Conclusions and Relevance Socioeconomic differences in brain age gap were partly accounted for by modifiable cardiovascular and lifestyle risk factors, with smoking being the single largest contributor. These findings identify modifiable cardiovascular risk factors as a substantial component of socioeconomic inequalities in brain aging.