Neighborhood Types and Body Weight in a National Sample of Adults
thesisposted on 01.02.2019, 00:00 authored by Kelly Kathleen Jones
Overweight and obesity continue to be challenges in the United States. Interventions targeting the environments where people live, work, and play may prove valuable because they have wide reach and the potential to affect the entire population. In order to design effective population-based approaches to reduce overweight and obesity, causal effects of neighborhood environments on health must be identified. Mixed results from previous research may be partly due to analyses that do not adequately account for bias (e.g., neighborhood self-selection) or the wider neighborhood context. This dissertation addresses these shortfalls by considering neighborhood environments holistically and making use of a geographically varied, longitudinal dataset of health and neighborhood environment data. Using data from the Weight and Veterans’ Environments Study, I first classified 838,215 veterans into nine neighborhood types based on their residential food, physical activity, and socioeconomic environments using latent class analysis. Then, in the first study, I estimated cross-sectional and longitudinal regressions to identify differences in body mass index (BMI) between neighborhood types. In the second study, I performed a difference-in-difference analysis to identify whether the effectiveness of a weight management program differed by neighborhood type. In the first study, I found that in cross-sectional analysis, neighborhood types characterized by greater access to food and physical activity resources were consistently associated with lower BMI, and that the association between the neighborhood socioeconomic environment and BMI differed between men and women. However, many of the significant associations dissipated in longitudinal analysis. In the second study, I found that participation in a weight management program was associated with smaller decreases in BMI among those living in neighborhoods with lower access to food and physical activity resources that were additionally characterized by low socioeconomic status for men, while women living in neighborhood types characterized by low access to food and physical activity resources and middle socioeconomic status lost the most weight. The dissertation findings suggest that neighborhood policies ought to take both the built and socioeconomic environments into account, and that consideration of particular challenges faced by residents of neighborhoods characterized by low socioeconomic status might improve weight management program outcomes.