Reliability of the food retail outlet survey tool (FROST) for measuring consumer nutrition environment

INTRODUCTION: Despite the increase in the volume of food environment research, there is a general lack of reliable measurement tools. This study presents the reliability assessment of a tool for measuring consumer nutrition environment in urban food stores. METHODS: Racially diverse downtown neighborhoods in Albany, NY were the setting. A total of 123 food stores were identified through multiple lists and ground-truthing. A sample of 39 food stores was randomly selected. The Food Retail Outlet Survey Tool (FROST) was developed to quickly assess the availability (presence) of selected food and non-food items, placement, physical characteristics of the store, policy implementation, and advertisements on outside windows. For in-store items, agreement of observations between experienced and lightly trained surveyors was assessed. For window advertisement assessments, inter-method agreement (on-site sketch vs. digital photo), and inter-rater agreement (both on-site) among lightly-trained surveyors were evaluated. RESULTS: Twenty-seven of the 47 in-store items had 100% agreement. The Prevalence-Adjusted Bias-Adjusted Kappa indicated excellent agreement (≥0.90) on all items, except aisle width (0.74) and fresh vegetables (0.85). The store type (non-convenience store), the order of visits (first half), and the time to complete survey (>10 minutes) were associated with lower reliability in these two items. Both the inter-method and inter-rater agreements for window advertisements were uniformly high (Interclass Correlation Coefficients ranged 0.94 to 1.00), indicating high reliability. DISCUSSION: The FROST is a reliable tool and can be effectively used by an individual who attended a 30 minute group briefing and practiced with 3-4 stores.