Table 3 presents the odds ratios (ORs) of weekly fast-food consumption
as a function of the perceived availability or the GISbased
presence of fast food. For models 1 and 2, univariate logistic
regression analyses of the two main exposure variables on fastfood
consumption were performed. Residents who reported higher
perceived availability of fast food were found to be at higher risk
for fast-food consumption; however, the observed associations were
not statistically significant. Interestingly, according to the GIS data,
having one or more fast-food restaurants present in the neighborhood
was significantly associated with lower odds of eating fast food
on a weekly basis compared to not being exposed to any fast-food
restaurants within the neighborhood. After adjusting for potential
confounding factors, this associationwas no longer significant. When
using a different GIS variable (distance to nearest fast-food outlet
from a respondent’s home), similar but weaker results for an inverse
association between the GIS-based presence of fast food and
fast-food consumption were observed (data not shown). The
covariates younger age, white race/ethnicity and being employed(vs. unemployed or retired) significantly increased the odds of weekly
fast-food consumption.