Abstract
Neighborhoods or residential environments have physical and social attributes which may contribute to inequalities in the overweight and obesity pandemic. We examined the longitudinal associations of baseline neighborhood-level income and racial residential segregation (using the Gi* statistic: low, medium, high) with changes in body mass index (BMI in kg/m(2)), using geocoded data from 1821 civil servants in the municipality of Rio de Janeiro, Brazil, followed-up for approximately 13 years (baseline wave 1: 1999, wave 2: 2001-2002, wave 3: 2006-2007, wave 4: 2012-2013). Linear mixed effects models using BMI measured in all four study waves were performed, accounting for gender, race, length of residence, education and time-dependent age, and per capita family income. After adjustments, both income and racial segregation were positively associated with BMI differences (but not BMI changes) over time, in a dose-response pattern. For income segregation, mean differences in BMI for participants living in high and medium vs. low segregated neighborhoods were 1.04 kg/m(2) (β = 1.04; 95% CI 0.47, 1.62) and 0.86 kg/m(2) (0.86; 0.33, 1.39), respectively. For racial segregation, mean differences in BMI for participants living in high and medium vs low segregated neighborhoods were 0.71 kg/m(2) (0.71; 0.14, 1.29) and 0.30 kg/m(2) (0.30; - 0.24, 0.83), respectively. We also showed a moderate to strong correlation between racial and income segregation at baseline. Strategies to reduce BMI and obesity-related health inequalities should include special efforts aimed at segregated neighborhoods and its obesogenic environments.