Modeling the impact of exposure reductions using multi-stressor epidemiology, exposure models and synthetic microdata: an application to birthweight in two environmental justice communities

Abstract

BACKGROUND: Many vulnerable populations experience elevated exposures to environmental and social stressors, with deleterious effects on health. Multi-stressor epidemiological models can be used to assess benefits of exposure reductions. However, requisite individual-level risk factor data are often unavailable at adequate spatial resolution. OBJECTIVE: To leverage public data and novel simulation methods to estimate birthweight changes following simulated environmental interventions in two environmental justice communities in Massachusetts, US. METHODS: We gathered risk factor data from public sources (US Census, Behavioral Risk Factor Surveillance System, and Massachusetts Department of Health). We then created synthetic individual-level datasets using combinatorial optimization, and probabilistic and logistic modeling. Finally, we used coefficients from a multi-stressor epidemiological model to estimate birthweight and birthweight improvement associated with simulated environmental interventions. RESULTS: We created geographically-resolved synthetic microdata. Mothers with the lowest predicted birthweight were those identifying as Black or Hispanic, with parity > 1, utilization of government prenatal support, and lower educational attainment. Birthweight improvements following greenness and temperature improvements were similar for all high-risk groups and were larger than benefits from smoking cessation. SIGNIFICANCE: Absent private health data, this methodology allows for assessment of cumulative risk and health inequities, and comparison of individual-level impacts of localized health interventions.