Neighborhood Factors Related to Disparities in Low Birth Weight Infants in Virginia

BACKGROUND: Racial disparities in low birth weight (LBW) deliveries have long persisted. Efforts to understand and improve LBW outcomes have largely focused on individual-level maternal risk factors but have failed to account for all of the variation in birthweight. With many metropolitan areas in the United States exhibiting geographic separation of blacks and whites and a disproportionately large percentage of black women residing in disadvantaged urban neighborhoods, it is increasingly important to examine explanatory risk factors at the neighborhood level to target interventions and reduce disparities in LBW births. The purpose of this study is to spatially map and examine the spatially varying relationship between the distribution of low birth weight deliveries and neighborhood factors in Virginia at the census tract level.

METHODS: The prevalence of low birth weight infants was calculated for all census tracts in Virginia using Birth Certificate data from 2005-2009. Census tract level neighborhood data was obtained from the 2009 American Community Survey. Hotspot analysis identified statistically significant spatial clusters of high rates of LBW deliveries at the census tract level. LBW deliveries were then modeled using global Ordinary Least Squares (OLS) linear regression looking at neighborhood indices such as education, affordability, household income diversity, racial residential diversity, Townsend deprivation, environmental hazard, population density, population churning, job participation, and job imbalance. Additional population characteristics, including percent black, unemployed, living below poverty, receiving public assistance, female headed household, and under age 18, were also included. Significant predictors of LBW were further explored for spatial nonstationarity using Geographically Weighted Regression (GWR) to provide a model explaining the possible influence of local spatial effects on the pattern of LBW distribution at the census tract.

RESULTS: The prevalence of LBW deliveries at the census tract level in Virginia ranged from 1.2% to 24.2%. The OLS regression revealed that racial residential diversity, black race, education, female headed households, and receipt of public assistance were all independent predictors of LBW in Virginia (R2=0.42). The GWR analysis revealed spatial variation in the model fit, with higher Local R2 values and better model fit in areas with a greater prevalence of LBW infants.

CONCLUSIONS: Neighborhoods with a large proportion of black residents, poor racial residential diversity, and low SES were more likely to have higher rates of LBW infants. These analyses demonstrate how researchers, policymakers and the community may utilize Geographic Information Systems to develop geographically informed interventions.