Metropolitan-level ethnic residential segregation, racial identity, and body mass index among U.S. Hispanic adults: a multilevel cross-sectional study

BACKGROUND The few studies that have examined whether metropolitan-level ethnic residential segregation is associated with obesity among Hispanics are mixed. The segmented assimilation theory, which suggests patterns of integration for immigrant groups varies by social factors, may provide an explanation for these mixed findings. In this study we examined whether one social factor, racial identity, modified the association between ethnic residential segregation and body mass index (BMI) among Hispanics. METHODS We used data on 22,901 male and 37,335 non-pregnant female Hispanic adult participants of the 2003-2008 U.S. Behavioral Risk Factor Surveillance System living in 227 metropolitan or micropolitan areas (MMSAs). Participants self-identified as White, Black, and 'some other race'. BMI was calculated using self-reported height and weight; the Hispanic isolation index was used to measure Hispanic residential segregation. Using multi-level linear regression models, we examined the association of Hispanic residential segregation with BMI, and we investigated whether this relationship varied by race. RESULTS Among men, Hispanic segregation was unassociated with BMI after adjusting for age, race, MMSA-level poverty, and MMSA-level population size; there was no variation in this relationship by race. Among women, significant associations between Hispanic segregation and BMI in models adjusted for demographics and MMSA-level confounders became attenuated with further adjustment for education and language of exam. However, there was statistically significant variation by race (Pinteraction = 0.03 and 0.09 for Hispanic Blacks and Hispanics who identified as some other race, respectively, vs. Hispanic Whites). Specifically, higher segregation was associated with higher mean BMI among Hispanic Whites, but it was associated with lower mean BMI among Hispanic Blacks. Segregation was unassociated with BMI among Hispanic women identifying as some other race. CONCLUSIONS This heterogeneity highlights the persistent influence of race on structural processes that can have downstream consequences on health. As Hispanics grow as a proportion of the U.S. population, especially across urban centers, understanding the health consequences of residence in segregated areas, and whether or not these impacts vary across different groups, will be important for the design of more comprehensive solutions to prevent adverse health outcomes.