Introduction:
Community food environments (FE) are an important correlate of diet- and weight-related CV health. Conventional approaches to measuring the FE focus on residential neighborhoods, and do not assess the full extent of food sources regularly encountered and used. Further, little attention has been given to how individual diet-related experiences, like food insecurity, may interact with features of the FE to affect health. To address these limitations, we use an activity space approach, defined by the locations women routinely visit, to measure FE exposure and use, and assess differences by food security status.
Hypothesis:
Food-related spatial behavior and features of the FE differ between a) conventional and activity space definitions, and b) food secure and insecure women.
Methods:
We present initial results (n=51) from an ongoing clinic-based study of low-income African American women in Atlanta, GA. Data are collected in-person using a Google Map-powered activity space questionnaire. USDA’s 10-item adult scale is used to measure food insecurity. Retail FE data are from Dun & Bradstreet. ArcGIS 10.5 was used to define three environments: residential census tract (CT), and convex hull polygons of
overall
and
food-specific
activity spaces. We tested differences, by food security status, in mean behaviors and FE features with one-way ANOVAs.
Results:
Eighty-eight percent of women were food insecure. Food insecure women were lower income, less often employed, and less often had access to a car. CTs contained fewer supermarkets (μ=1.2
SD
=1.4) and fast food restaurants (μ=3.9
SD
=3.2) than activity spaces (μ=7.9
SD
=7.0; μ=55.5
SD
=44.1, respectively). On average, 6.7% (
SD
=13.5) of utilized food sources fell within CT bounds, while 53.4% (
SD
=35.5) fell within activity spaces. Compared to food secure women, food insecure women had smaller overall (μ=329.8km
2
SD
=340.4 vs. μ=548.3km
2
SD
=422.4;
p
=0.16) and food-specific (μ=48.1km
2
SD
=74.3 vs. μ=85.6km
2
SD
=106.4;
p
=0.28) activity spaces, and a smaller proportion of their utilized supermarkets fell within their activity spaces (μ=60.9%
SD
=42.4 vs. μ=81.9%
SD
=21.4;
p
=0.24). FE features did not differ by food security status.
Conclusions:
Conventional FE definitions likely underestimate the number of food sources women encounter, and do not capture the majority of sources used. Smaller activity spaces among food insecure women suggest that routine spatial mobility may be constrained by factors like transportation access. Still, food insecure women more often traveled
outside
of their activity spaces to utilize supermarkets, suggesting a dual burden of constrained spatial mobility and access. Interestingly, FE features did
not
differ by food security status. In planned future analyses, any observed differences in diet and weight may indicate variation in how women
interact
with the FE, rather than differences in exposure.
Food sources and access strategies in Ugandan secondary cities: an intersectional analysis
