Abstract
OBJECTIVES: Explore the contextual aspects of overweight in Quebec through multilevel modelling, using a purposely designed set of spatial units and a few area-based characteristics. METHODS: Data came from the Canadian Community Health Survey (CCHS Cycle 2.1). Multilevel logistic regressions were performed to test for the presence of an independent contextual effect on overweight and obesity (BMI > or = 25 kg/m2), separately for men and women. Modelling considered individual attributes, including some lifestyle aspects, and contextual characteristics. A geographic grid integrating spatial elements related to overweight and obesity in the literature was developed. Also, an area-level residuals analysis was carried out to identify spatial units presenting higher or lower odds of being overweight. RESULTS: After accounting for individual and area-level characteristics, there remain significant geographic variations in overweight in Quebec. Although this contextual effect is small for men and women, many spatial units differ significantly from the provincial average. There are differences between the geography of overweight in men and women which suggest that socio-economic mechanisms and land use patterns underlying overweight might be different between genders. Also, there is considerable variability within rural and urban areas. CONCLUSION: A complex geography of overweight is revealed. Small-scale studies, as well as methodological and data developments, are needed to deepen our understanding of this geography.