Date of Award


Document Type

Open Access Dissertation


Epidemiology and Biostatistics



First Advisor

Angela D Liese


An increasing number of studies examined the association between neighborhood characteristics and birth outcomes. However, the results can be difficult to compare because of the variety of indicators used to characterize the neighborhood. As an important neighborhood characteristic, the food environment is associated with residents' nutrition status, diet quality, and related health outcomes. In addition, the food environment has been found to influence women's diet quality during pregnancy, which is a key factor in predicting birth outcomes. However to date, studies on food environment and birth outcomes are extremely limited.

This study examined the association between food environment (evaluated by both neighborhood- and individual-level indicators) and birth outcomes using data from all South Carolina births in 2008-2009. Birth outcomes were analyzed as continuous outcomes (birth weight and gestational age) and dichotomous outcomes (low birthweight (LBW) and preterm birth (PTB)). To facilitate comparison with other studies, a Neighborhood Deprivation Index (NDI) was used to identify the association between neighborhood characteristics and birth outcomes.

First, we identified those data associated with the food desert, a community food access measure developed by US Department of Agriculture (USDA) characterizing neighborhood income and access to supermarkets, to evaluate the food environment and its relationship with the birth outcomes. We found that mothers living in food deserts did not have different birth outcomes compared to those living in areas with high neighborhood income and easy access to supermarkets. Neighborhood income is more important than food access in predicting birth outcomes.

Second, we estimated the association between mothers' accessibility (distance to the nearest store) and availability (count of stores within 1 mile around mothers' homes) to various types of food outlets and birth outcomes in an eight-county area in South Carolina. The results suggested that accessibility and availability of convenience stores were each associated with adverse birth outcomes. No significant associations were captured for healthy food outlets and limited service restaurants with birth outcomes.

In the end, we examined the relationship between NDI and adverse birth outcomes. Propensity score matching (PSM) analyses identified neighborhood deprivation as associated with increased risk of LBW among non-Hispanic whites, and with increased risk of PTB among non-Hispanic blacks. However, random effects logistic regression models identified the association between neighborhood deprivation and adverse birth outcomes only among non-Hispanic whites. PSM might be an appropriate approach to avoid off-support inferences.


© 2013, Xiaoguang Ma

Included in

Epidemiology Commons