Relative Validity and Reliability of a 1-Week, Semiquantitative Food Frequency Questionnaire for Women Participating in the Supplemental Nutrition Assistance Program

Background The Supplemental Nutrition Assistance Program (SNAP) plays a critical role in reducing food insecurity by distribution of benefits at a monthly interval to participants. Households that receive assistance from SNAP spend at least three-quarters of benefits within the first 2 weeks of receipt. Because this expenditure pattern may be associated with lower food intake toward the end of the month, it is important to develop a tool that can assess the weekly diets of SNAP participants. Objective The goal of this study was to develop and assess the relative validity and reliability of a semiquantitative 1-week food frequency questionnaire (FFQ) tailored to a population of women participating in SNAP. Design The FFQ was derived from an existing 195-item FFQ that was based on a reference period of 1 month. This 195-item FFQ has been validated in a population of low-income postpartum women who were recruited from central Texas during 2004. Mean daily servings of each food item in the 195-item FFQ completed by women who took part in the 2004 validation study were calculated to determine the most frequently consumed food items. Emphasis on these items led to the creation of a shorter, 1-week FFQ of only 95 items. This new 1-week instrument was compared with 3-day diet records to evaluate relative validity in a sample of women participating in SNAP. For reliability, the FFQ was administered a second time, separated by a 1-month time interval. Participants/setting The validity study included 70 female SNAP participants who were recruited from the partner agencies of the Central Texas Food Bank from March to June 2015. A subsample of 40 women participated in the reliability study. Main outcome measures Outcome measures were mean nutrient intake values obtained from the two tests of the 95-item FFQ and 3-day diet records. Statistical analysis performed Deattenuated Pearson correlation coefficients examined relationships in nutrient intake between the 95-item FFQ and 3-day diet records, and a paired samples t test determined differences in mean nutrient intake. Weighted Cohen's κ indicated agreement in quartile classification of study participants by the 95-item FFQ and 3-day diet records, according to nutrient intake. Test–retest reliability was assessed by intraclass correlations and weighted Cohen's κ. Results Mean deattenuated Pearson correlation between the FFQ and 3-day diet records was 0.61, and the weighted Cohen's κ=0.39. Finally, the average test–retest correlation and weighted Cohen's κ of the FFQ was 0.66 and 0.50, respectively. Conclusions These results suggest that the 1-week, 95-item FFQ demonstrated acceptable relative validity and reliability in low-income women participating in SNAP in southwestern United States.