Intersectional Inequalities in Environmental Health Hazard across United States: Eco-Intersectional Multilevel Approach

Background and Purpose: Health outcomes from toxic environments are inequally distributed, putting vulnerable communities at greater risk. Environmental justice literature has documented these persistent inequalities, but traditional approaches often fail to capture how multiple structural dimensions interact to create patterns of vulnerability, focusing only on individual-level outcomes. The Eco-Intersectional Multilevel (EIM) approach addresses this limitation by integrating environmental justice, intersectionality theory, and ecosocial theory to analyze how structural factors combine at the community level to produce environmental health disparities. This study examines how the intersectional characteristics shape environmental health hazard distributions across US census tracts.

Methods: Using census tract-level data from 70,938 observations across the US, we constructed 215 unique intersectional strata based on combinations of race/ethnicity composition (Black and Latine residents percentages), female-headed household percentages, educational attainment and median household income level, and metropolitan status. Each stratum has a 5-digit code representing each social dimension. Based on National Air Toxics Assessment (NATA) data from 2014, the environmental health hazard index—consisting of standardized estimates of carcinogenic, respiratory, and neurological hazard from air quality—was used as a dependent variable. Dividing census tracts into the highest 20% of hazard and the remainder, we applied a logistic EIM model to quantify these disparities. The null model with random intercepts for intersectional strata, a main effects model incorporating additive effects, and a full model with covariates were tested.

Results: Results revealed substantial disparities in environmental health hazard exposure across intersectional strata. Census tracts with high percentages of Black and Latine residents showed significantly higher odds of environmental health hazard. Female-headed household percentage was positively associated with environmental risk, while higher education and income levels showed protective effects. Metropolitan status emerged as a strong predictor. The Variance Partition Coefficient (VPC) of the null model was 39%, indicating a high proportion of total variation in the sample that exists at the between-stratum level. The final model's Proportional Change in Variance (PCV) was 92.71%, which describes the between-stratum variation explained by the additive effects. This suggests the additive nature of patterns, but about 7% remained for possible interactive patterns. The tracts that had the highest likelihood to be in the hazardous group were characterized by high% Black low% Latine, high% female-headed household, low education and income levels, and metropolitan status (stratum 23111, ranked 215). Unique interaction effects from significant residuals were found in both marginalized and privileged strata. For example, stratum 33331, with low% Black high% Latine, high% female-headed household, high education and income, and metropolitan status, reported a significant random effect and ranked 210, showing very high risk despite its high education and income levels.

Conclusions and Implications: The study reveals how environmental health inequalities are shaped by the complex intersection of social dimensions rather than by single factors alone. Particularly vulnerable are urban communities with high percentages of racial/ethnic minorities and female-headed households. These findings challenge single-axis policy interventions and suggest that environmental justice initiatives should adopt intersectional frameworks to address the structural mechanisms that maintain these disparities.