Chronic neighborhood stressors can modify the effect of nitrogen dioxide on term birthweight in New York City

Abstract

Urban air pollution exposures are associated with decreased newborn birthweight and developmental impairments, potentially with stronger impacts in communities of lower socioeconomic position (SEP). Here, we examined 28 community stressors - which vary with neighborhood SEP - as potential modifiers of the nitrogen dioxide (NO2)-birthweight association in New York City (NYC). We examined birth records data on 245,345 live births from 2007 to 2010 and administrative data on 28 social stressors aggregated to United Hospital Fund (UHF) areas (n = 34), categorized into quartiles. Residence-specific NO2 exposure concentrations, averaged over each pregnancy, were created using NYC Community Air Survey (NYCCAS) and daily EPA regulatory data. We applied linear mixed-effects modeling using UHF as a clustering variable and tested interactions between social factors and NO2. We observed an averaged 12.6-g (SE = 2.72) decrease in newborn birthweight per IQR increase in NO2 (IQR= 6.24 ppb). We also observed significant decrements in birthweight with higher stressor exposures (main effects), including significant decreases across quartiles for assault rates (p=.02 for trend across quartiles), child abuse (p=.02), murder (p=.007), robbery (p=.03), and perceived lack of safety (p=.02). Interaction analyses revealed weaker NO2-birthweight associations in areas with higher community stressors - including higher assault rates (p=.03), murder (p=.03), robbery (p=.04), or perceived lack of safety (p=.02). We found that birthweights in higher-pollution/higher-violence communities were substantially lower, on average, than in lower-pollution/lower-violence communities. These differences were driven, however, by strong direct associations between violence and birthweight; with very high violence exposures, impacts of pollution were dampened. Non-violent social stressors generally did not confer lower birthweights, nor did they substantially alter NO2 effects. Our results suggest complex interactions among stressors and pollutants, including potential saturation effects at the high end of violence exposures. Research in larger datasets and more flexible methods are needed to characterize non-linear interactions.