This dissertation examined the association between neighborhood-level deprivation and perinatal outcomes. We studied the association between neighborhood poverty rate and pre-term birth (PTB; birth < 37 weeks) using longitudinal maternally-linked natality files of women and their infants in Michigan during the period 1990-2012. This study examined the embodiment of place and role of maternal characteristics during pregnancy in an effort to understand how selection into neighborhood may bias our understanding of neighborhood level associations. We looked at pregnancy outcomes across multiple pregnancies for the same woman (the mother) as she changed neighborhoods, and levels of poverty between pregnancies. In the first study examining residential mobility between pregnancies, we reported that approximately half of our sample changed residences between pregnancies. We further exploited our data structure to examine the association with prior PTB on subsequent mobility in two sub-samples restricted by parity: births 1 and 2, and births 2 and 3. We found the strongest risk factors for mobility were related to marital change (Divorce: births 1 to 2 OR: 2.5 95% CI: 2.4-2.6, births 2 to 3 OR: 3.3, 95% CI: 3.1-3.6); Married: births 1 to 2 OR: 2.8, 95% CI: 2.7-2.8, births 2 to 3 OR: 1.9, 95% CI:1.9-2.0) but not prior PTB (prior PTB: births 1 to 2 OR: 1.0, 95% CI:1.0-1.0, births 2 to 3 OR: 1.1 95%CI: 1.0-1.1). In the second study, we report that most women did not experience a change in the level of neighborhood poverty, based on quartile of neighborhood poverty. Women who remained in the poorest neighborhoods experienced the highest percentage of PTB across two births samples, Births 1 to 2 (11.4% PTB) and Births 2 to 3 (12.3% PTB). We found increased odds of PTB for births 1 to 2 with strong downward neighborhood trajectory (OR 1.2, 95% CI 1.0-1.3) but also increased odds of PTB among strong upward neighborhood poverty trajectory (OR 1.1, 95%CI: 1.1- 1.2) compared to the static trajectory group of lowest neighborhood poverty quartile. In Study 3, we then employed a novel approach, maternal fixed effects, utilizing data linked over time to compare birth outcomes for the same mother under different exposures which allows the mother to act as her own control, analogous to a case-crossover design, while comparing the contextual effects of neighborhood deprivation on PTB. We conducted logistic regression, random effects and fixed effects analysis to evaluate n=2,191,063 eligible births during our study period. Because a fixed effects model relies on variation over time within a mother to identify the estimated association of neighborhood deprivation and PTB, the primary analytic sample was restricted (n=280,277 births to 103,328 women).We found a null association between neighborhood poverty and PTB when using a maternal fixed effects analysis (OR: 1.0, 95% CI: 1.0-1.0). This was one of the first studies to profile the maternal neighborhood mobility patterns over a long period of time, between successive pregnancies and evaluated by neighborhood poverty rate.
Read
