Identifying Retail Food Stores to Evaluate the Food Environment

Background: The availability of food stores is the most frequently used measure of the food environment, but identifying them poses a technical challenge. Purpose: This study evaluated eight administrative lists of retailers for identifying food stores in an urban community. Methods: Lists of inspected food stores (IFS), cigarette retailers, liquor licenses, lottery retailers, gasoline retailers, farmers' markets, and authorized WIC (Program for Women, Infants, and Children) and Supplemental Nutrition Assistance Program (SNAP) retailers for Albany NY were obtained from government agencies. Sensitivity and positive predictive value (PPV) were assessed, using ground-truthing as the validation measure. Stores were also grouped by the number of lists they were documented on, and the proportion of food stores in each group was obtained. Data were collected and analyzed in 2009. Results: A total of 166 stores, including four from ground-truthing, were identified. Forty-three stores were disqualified, as a result of having no targeted foods (n=17); being in the access-restricted area of a building (n=15); and being out of business (n=11). Sensitivity was highest in IFS (87.0%), followed by the cigarette retailers' list (76.4%). PPV was highest in WIC and farmers' markets lists (100%), followed by SNAP (97.8%). None of the lists had both sensitivity and PPV greater than 90%. All stores that were listed by four or more lists were food stores. The proportion of food stores was lowest (33.3%) for stores listed by only one list. Conclusions: Individual lists had limited utility for identifying food stores, but when they were combined, the likelihood of a retail store being a food store could be predicted by the number of lists the store was documented on. This information can be used to increase the efficiency of ground-truthing.