What you get for the measurement: A comparison of three geographic aggregations for how the built food environment influences body mass index

Introduction. A few studies have demonstrated relationships between the built food environment (which food outlets are placed where) and obesity. However, data are used at differing levels of geographic specification. Data availability varies for geographic specification. These issues make the translation of research to practice problematic. How does the level of geographic specificity influence statistical association between the food environment and BMI?

Methods. I compare the relationship between food outlets (large chain supermarkets, grocery stores, convenience stores, and fast food restaurants) and Body Mass Index (BMI) of people living in an area for counties, census tracts and zip codes. I replicate the same model three times using hierarchical linear modeling (HLM) which focus the statistical explanation on the main effects of the environmental variables on individuals, and controls for individual characteristics.

Results. The findings focus on which food outlets statistically influence BMI. The results for the three levels of geographic specificity are compared.

Discussion. HLM links people with the specific environment in which they function. A larger geographic area may have more readily available data, but the influence between the environment and the person may be lost. Conversely, data for a small geographic area may illustrate a person's immediate environment but the data are rare. Implications are presented, including which findings are consistent and other considerations for developing interventions based built environment research.

Some environmental data were provided by the Cancer Prevention and Nutrition Section, California Department of Public Health, funded by the USDA Food Stamp Nutrition Education Program.