Development and validation of a shelf inventory to evaluate household food purchases among older adults with diabetes mellitus

OBJECTIVE: Valid measures of behavioral outcomes are needed to evaluate interventions. The purpose of this research was to validate a shelf inventory and evaluate food purchases using the inventory among older adults with diabetes. DESIGN: A 166-item shelf inventory of household food was assessed for face, content, and criterion validity. The sensitivity and specificity of the inventory were determined by comparing participant- and interviewer-completed inventories. A randomized pretest/post-test control group design was used to evaluate household food purchases following an intervention. Setting: The intervention was held in an outpatient setting. PARTICIPANTS: Study participants were > or = 65 years old with type 2 diabetes mellitus for > or = 1 year. INTERVENTION: The intervention included 10 weekly group sessions and emphasized applying food label information to food purchases. MAIN OUTCOME MEASURES: Each food on the inventory was defined as "encouraged" or "discouraged" according to intervention messages. ANALYSIS: Inventory sensitivity was the proportion of foods present in households that were accurately identified on the self-reported inventory as present; specificity was the proportion of foods not present that were accurately identified as not present. Two-sample t tests compared intervention scores for encouraged/discouraged foods by treatment group. RESULTS: Overall sensitivity and specificity were.90 and.97, respectively. Intervention results found more encouraged than discouraged foods present at post-test (t [91] = 2.3, P =.02). CONCLUSIONS AND IMPLICATIONS: The shelf inventory is a sensitive, specific, and valid tool for assessing household food purchases and can be used to evaluate food choice interventions among older adults with type 2 diabetes mellitus.