Abstract
This study examines the area-based variations in obesity from a community-based epidemiologic survey of Boston, MA, USA, using a geographic information system and multilevel modeling techniques. A combination of quantitative and qualitative methods was used to assess whether a function of the food and the physical activity (PA) environment can explain the body weight of residents. First, a series of multilevel analyses was conducted after accounting for the well-established individual determinants and capturing a wide range of environmental attributes to represent a more realistic portrayal of urban typology. Second, the results of multilevel analysis were framed into the theoretical model of area-based variations in obesity to qualitatively summarize the association of contextual factors with the body weight of residents. Based on the overall correlation, the area-based variations defined by a function of the food and PA environment seem to be insufficient in explaining the body weight of residents. By testing the cross-level interactions of gender and race/ethnicity with contextual factors, the results suggest that the concept of area-based variations in obesity will have to consider how residents behave differently within a given environment. More research is needed to better understand the contextual determinants of obesity so as to put forth population-wide interventions.