Assessing the neighborhood-level determinants of obesity in New York City

Background. In addition to individual-level factors such as genetics, poverty, and gender, neighborhood-level factors such as area-wealth, healthy and unhealthy food availability and promoters and barriers to physical activity may contribute to obesity. Few studies have examined multiple neighborhood and individual-level predictors simultaneously. Methods: Individual-level obesity data from the 2005 New York City Community Health Survey (n=9,816 persons, 34 neighborhoods) was combined with neighborhood data from: (1) InfoUSA, a commercial database (availability of food stores and fitness facilities), (2) the NYC Coalition Against Hunger (emergency food programs), (3) the Department of Parks and Recreation (park space), (4) the Police Department (crime), (5) Primary Land Use Tax Lot Output data (commercial and residential land use) and (6) the 2000 U.S. Census (neighborhood demographics, racial/ethnic composition, and socio-economic status). The distribution of neighborhood features was evaluated and mapped and bivariate correlations assessed whether the availability of amenities was associated with neighborhood socio-demographic characteristics such as area-income and racial/ethnic composition. Finally, multilevel statistical models examined whether the specified neighborhood features significantly predicted an individual's odds of obesity (BMI≥30 kg/m2) above traditional individual-level factors. Results: Obesity prevalence in NYC was 20% in 2005, but neighborhood rates ranged from 7% to 32%. Neighborhood availability of food and fitness amenities varied substantially and was associated with area-income and racial composition. For example, neighborhood wealth was associated with greater availability of supermarkets, restaurants and fitness facilities, and fewer small grocers, convenience stores and crime rates. However, fast food chains and snack and beverage vendors were also more common in wealthier neighborhoods. Individual-level obesity was significantly associated with individual-level demographics, SES and lifestyle factors, but additional variance was explained by neighborhood characteristics. For example, decreased availability of large supermarkets and fitness facilities were consistently associated with obesity and may serve as potentially modifiable loci for intervention. Conclusion: This dissertation builds on the current literature on neighborhood determinants of obesity, confirming that both neighborhood and individual-level factors influence body weight status. This study has several implications for policy, intervention planning and dietetics practice and serves as the foundation for future research.