Racial residential segregation is associated with ambient air pollution exposure after adjustment for multilevel sociodemographic factors: Evidence from eight US-based cohorts

Abstract

Objective: We examined if racial residential segregation (RRS) – a fundamental cause of disease – is independently associated with air pollution after accounting for other neighborhood and individual-level sociodemographic factors, to better understand its potential role as a confounder of air pollution-health studies. Methods: We compiled data from eight large cohorts, restricting to non-Hispanic Black and White urban-residing participants observed at least once between 1999 and 2005. We used 2000 decennial census data to derive a spatial RRS measure (divergence index) and neighborhood socioeconomic status (NSES) index for participants’ residing Census tracts, in addition to participant baseline data, to examine associations between RRS and sociodemographic factors (NSES, education, race) and residential exposure to spatiotemporal model-predicted PM2.5 and NO2 levels. We fit random-effects meta-analysis models to pool estimates across adjusted cohort-specific multilevel models. Results: Analytic sample included eligible participants in CHS (N = 3,605), MESA (4,785), REGARDS (22,649), NHS (90,415), NHSII (91,654), HPFS (32,625), WHI-OS (77,680), and WHI-CT (56,639). In adjusted univariate models, a quartile higher RRS was associated with 3.73% higher PM2.5 exposure (95% CI: 2.14%, 5.32%), and an 11.53% higher (95% CI: 10.83%, 12.22%) NO2 exposure on average. In fully adjusted models, higher RRS was associated with 3.25% higher PM2.5 exposure (95% CI: 1.45%, 5.05%; P < 0.05) and 10.22% higher NO2 exposure (95% CI: 6.69%, 13.74%; P < 0.001) on average. Conclusions: Our findings indicate that RRS is associated with the differential distribution of poor air quality independent of NSES or individual race, suggesting it may be a relevant confounder to be considered in future air pollution epidemiology studies.