Residential Segregation and the Health of African-American Infants: Does the Effect Vary by Prevalence?

Segregation effects may vary between areas (e.g., counties) of low and high low birth weight (LBW; <2,500 g) and preterm birth
(PTB; <37 weeks of gestation) rates due to interactions with area differences in risks and resources. We assess whether the
effects of residential segregation on county-level LBW and PTB rates for African-American infants vary by the prevalence of
these conditions. The study sample includes 368 counties of 100,000 or more residents and at least 50 African-American live
births in 2000. Residentially segregated counties are identified alternatively by county-level dissimilarity and isolation
indices. Quantile regression is used to assess how residential segregation affects the entire distributions of county-level
LBW and PTB rates (i.e. by prevalence). Residential segregation increases LBW and PTB rates significantly in areas of low
prevalence, but has no such effects for areas of high prevalence. As a sensitivity analysis, we use metropolitan statistical
area level data and obtain similar results. Our findings suggest that residential segregation has adverse effects mainly in
areas of low prevalence of LBW and preterm birth, which are expected overall to have fewer risk factors and more resources
for infant health, but not in high prevalence areas, which are expected to have more risk factors and fewer resources. Residential
policies aimed at area resource improvements may be more effective.