Housing instability, air pollution, and health: Three studies from the United States
Access StatusFull text of the requested work is not available in DASH at this time ("dark deposit"). For more information on dark deposits, see our FAQ.
MetadataShow full item record
CitationKhadka, Aayush. 2021. Housing instability, air pollution, and health: Three studies from the United States. Doctoral dissertation, Harvard University Graduate School of Arts and Sciences.
AbstractThe United States has consistently ranked poorly in terms of the maternal and child health outcomes relative to other economically developed countries. In addition, there are vast disparities in maternal and child health outcomes within the country by race, ethnicity, and class. A large and active body of literature suggests that differential exposure to social and environmental determinants across high income countries and within the United States may partly explain the existence of these disparities. This dissertation contributes to this field of social and environmental determinants of poor maternal and child health outcomes in the United States. Specifically, it investigates the role of housing instability – a social determinant – and air pollution – an environmental determinant – in impacting the risk of preterm birth, infant death, and pregnancy loss across the country.
Chapter 2 brings together lessons from the maternal health literature – which shows that prenatal psychosocial stress is a risk factor for preterm birth – and the housing literature – which demonstrates that threatened evictions are a major source of stress – to investigate if prenatal exposure to threatened evictions increases the risk of preterm birth. To answer this question, my co-authors and I combined over seven million live birth records from 1,633 counties between 2009 and 2016 with the largest, county-level dataset on threatened evictions from The Eviction Lab at Princeton University. Using a retrospective cohort study design, we fit regression models with several control variables including county fixed effects and find that increased prenatal exposure to threatened evictions was positively associated with the risk of prematurity over the study period.
Chapter 3 analyzes the relationship of prenatal and post-birth air pollution exposure with infant death. Although this is a well-studied topic, the evidence base is mixed for a variety of reasons. My co-author and I contribute to the existing literature by using a Structural Equation Modeling framework to estimate direct paths from average prenatal and post-birth PM2.5 exposure to infant mortality as well as indirect paths from prenatal PM2.5 exposure to infant death via preterm birth and low birth weight. We fit the Structural Equation Model on over ten million linked birth-infant death records from 2011 to 2013 merged with daily, county-level average concentration of particulate matter less than 2.5 μm in diameter (PM2.5). Our results suggest that increased exposure to PM2.5 prenatally was positively associated with the risk of infant mortality with the majority of this association being driven by the direct path from prenatal air pollution to infant death. Our results for the association between post-birth PM2.5 exposure and infant death were less precisely estimated in our primary analysis; however, robustness checks indicate a strong, positive association between post-birth air pollution exposure and infant death as well.
Chapter 4 investigates if higher levels of prenatal exposure to air pollution is associated with pregnancy loss. We use a novel analytic framework which allows us to infer the relationship between prenatal air pollution and pregnancy loss by instead analyzing the relationship between the same exposure and conceptions leading to live births, a metric which we can calculate using live birth records. To operationalize this framework, we used birth certificate data between 2001 and 2014 combined with daily, county-level concentration of PM2.5, and daily, county-level data on temperature, precipitation, and relative humidity. For our primary analysis, we fit quasi-Poisson models of the total number of conceptions leading to live births on average, month-by-month PM2.5 exposure over a nine-month gestation period adjusting for county-month of year fixed effects and various meteorological and temporal confounders. We conducted several sensitivity analyses as well. Overall, we find inconclusive evidence of an association between prenatal PM2.5 exposure and pregnancy loss at any point during gestation.
Citable link to this pagehttps://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37368395
- FAS Theses and Dissertations