Household, Housing, and Neighborhood Drivers of Environmental Health Disparities

Abstract

Attention to both environmental hazards and social factors is required to understand environmental health disparities. In the past, researchers raised two challenges with regard to this assertion. One is to understand the intercorrelation between environmental and social exposures. The other is to understand the pathway that leads to differential susceptibility. However, it remains unclear how environmental and social exposures are clustered at a fine spatial resolution among low to medium-income urban households. Differential susceptibility could be missed due to differential measurement error resulting from exposure heterogeneity that is not correctly captured. This dissertation is a comprehensive discussion of these research gaps with the input of household, housing, and neighborhood data, and additionally asked how can filed data assist the exposure measurement to understand exposure better. In our first project (Chapter 2), we explored patterns of environmental and social exposure at the household level in two areas in greater Boston with Latent Class Analysis and examined potential predictors of these joint exposures. Our analyses confirm that environmental and social stressors cluster in socially disadvantaged households characterized by their race/ethnicity, income, and education. Housing type (single-family versus multifamily), ownership, and location of the residence were also strong predictors of cluster membership, with renter and multifamily residents at risk of high exposure to environmental and social stressors. In the second project (Chapter 3), we addressed the exposure heterogeneity through a building ventilation factor, Air Exchange Rate (AER), that determines the infiltration of ambient air pollution indoors. We incorporated parcel-level structural AER estimates as a modifier of the association between short-term ambient PM2.5 exposure and cause-specific mortality in Commonwealth of Massachusetts (MA) with a case-crossover study design. We found that parcel-level structural AER was an effect modifier during warm seasons. Including parcel-level AER may help us to better quantify associations with pollution by reducing measurement error especially when it is differential. In our last project (Chapter 4), we explored the relative contribution of meteorology and occupant behaviors on the residential AER, in addition to the structural estimates we used in Chapter 3, with high temporal resolution data and linear quantile mixed models. We confirmed the current knowledge that meteorology and behaviors were critical for AER estimates. We additionally highlighted that these influences were more significant at higher quantiles of residential AER distributions. Heterogeneous associations between single-family and multifamily homes were observed after stratifying by housing types. The significance of this dissertation is that 1) it informs future exposure assessment to consider a cumulative exposure structure, 2) it identifies critical household and housing characteristics to target and reduce joint exposures, 3) it adds empirical evidence to the understanding of the effect heterogeneity of ambient exposures through building factors that have sociodemographic distributions, and 4) it highlighted a high temporal variability of residential AER in addition to structural estimates when activities happened at home and suggested the factors that need to be additional controlled for future epidemiology studies.