Abstract
PURPOSE: Given that the relationships between higher BMI and adverse health outcomes are nonconstant and most pronounced at either ends of the BMI distribution, we assess the association between neighborhood poverty and BMI at multiple points along the BMI distribution. METHODS: Using data from the 1999 to 2015 Panel Study of Income Dynamics of Black and White adults in the United States, we estimate quantile regression models while jointly applying a marginal structural modeling approach to account for time-varying individual-level factors that may be simultaneously mediators as well as confounders. RESULTS: Neighborhood poverty was not found to be associated with bodyweight at any point along the BMI distribution for Black or White males. However, high neighborhood poverty, compared to low neighborhood poverty, predicted increases in bodyweight for Black females at the lower end of the BMI distribution and for White females at the higher end of the BMI distribution. No association was found between neighborhood poverty and BMI at the mean. CONCLUSIONS: Results identify the most vulnerable subgroups, suggesting that White females at the higher end of the BMI distribution as well as Black females at the lower end of the BMI distribution are particularly sensitive to obesogenic environments.