Search

Objectives. We compared the association between advancing maternal age and risk of preterm delivery across 4 groups (Black smokers, Black nonsmokers, White smokers, White nonsmokers) and within the context of neighborhood deprivation levels. Methods. We obtained data from linked census and birth records for singletons (n=182938) delivered by women aged 20 to 39 years in Philadelphia, Pennsylvania; Baltimore, Maryland; 16 Michigan cities; 3 Maryland counties; and 2 North Carolina counties. Results from area-specific multilevel logistic regression models were combined to obtain pooled estimates of relations between maternal age and risk of preterm delivery. We repeated the models after categorizing women by neighborhood deprivation level (low, medium, and high). Results. Among multiparous women, there was a significant age-related increase in preterm delivery in 3 of the 4 groups. The adjusted odds ratio per 5-year age increase was 1.31 in Black smokers, 1.11 in Black nonsmokers, and 1.16 in White smokers. In each group, the odds ratio increased as neighborhood deprivation increased. Conclusions. These results support the 'weathering' hypothesis, suggesting that Black women, women with high-risk behaviors, and women living in highdeprivation neighborhoods may develop 'accelerated aging' that increases preterm delivery risk. (Am J Public Health. 2009;99:1864-1871. doi:10.2105/AJPH. 2008.151589)

Neighborhood socioeconomic effects on health have been estimated using multiple variables and indices. This inconsistent estimation approach makes comparison across geographic areas challenging. In this paper, we developed indices representing specific socioeconomic domains that can be reproduced in other areas to estimate elements of the neighborhood socioeconomic environment on health outcomes, specifically preterm birth. Using year 2000 U.S. census data and principal components analysis, socioeconomic indices were developed representing a priori--defined domains of education, employment, housing, occupation, poverty and residential stability. These socioeconomic indices were subsequently used in race-stratified multilevel logistic regression models of preterm birth in eight socioeconomically distinct study areas in the U.S. Maternal residence was obtained from birth records and was geocoded to census tracts. In maternal age and education adjusted models, living in tracts with high unemployment, low education, poor housing, low proportion of managerial or professional occupation and high poverty was associated with increased odds of preterm birth for non-Hispanic white women at most sites. Among non-Hispanic black women, similar associations were noted for tract-level low education, high unemployment, low occupation, and high poverty, but the effect estimates were generally smaller than those seen for white women. Increasing amounts of residential stability were not associated with preterm birth in these analyses. We combined the domain estimates across the eight study sites to produce pooled effect estimates for the socioeconomic domains on preterm birth. The research reported here Keywords: USA Neighborhood effects Racial inequalities Preterm birth Socioeconomic domains Multilevel modeling