Disentangling overlapping influences of neighborhoods and schools on adolescent body mass index.

OBJECTIVE: To compare the simultaneous influence of schools and neighborhoods on adolescent body mass index (BMI).

METHODS: Analyzing data from a nationally representative sample of adolescents in grades 7 to 12 (n = 18,200), cross-classified multilevel modeling was used to examine the fixed and random effects of individuals, schools, and neighborhoods on adolescent BMI. Additionally, the ability of school and neighborhood demographics to explain racial/ethnic disparities in BMI was assessed.

RESULTS: There were 18,200 students nested in 128 schools and 2,259 neighborhoods, with 2,757 unique combinations of schools and neighborhoods. In girls, schools (vojk  = 0.18, CI: 0.06-0.33) contributed twice that of neighborhoods (uojk  = 0.08, CI: 0.01-0.20) to the variance in BMI, while in males, schools (uojk  = 0.15, CI: 0.05-0.30) and neighborhoods (vojk  = 0.16, CI: 0.05-0.31) had similar contributions. The interaction of the neighborhood and school random effects contributed significantly to the variance of male and female BMI. Characteristics of neighborhoods and schools explained a large portion of the racial/ethnic disparity in female BMI.

CONCLUSIONS: In an analysis of a nationally representative sample including multiple racial and ethnic groups, the BMI variance of adolescent females was associated with schools more than neighborhoods. In males, there was no difference in school or neighborhood association with BMI.