Validity and Reproducibility of FFQ in Measuring Food and Food Group Intakes

Abstract

Abstract Objectives We evaluated the validity and reproducibility of a 149-item semi-quantitative Food Frequency Questionnaire (FFQ) in measuring food and food group intakes among 736 participants of the Women's Lifestyle Validation Study (WLVS) and 649 participants of the Men's Lifestyle Validation Study (MLVS). Methods The WLVS was conducted among a subset of the Nurses' Health Study (NHS) and NHSII participants aged 45–80 years in 2010. Participants of the MLVS aged 46–82 years between 2011 and 2013 were recruited from the Health Professionals Follow-up Study and members of Harvard Pilgrim Health Care. We matched 10,128 foods with unique food codes collected from the 7-day dietary records (7DDRs) to the 149 foods listed on the FFQ based on the similarity of food descriptions. After the matching, we converted gram intakes of the 7DDR-measured foods into servings using the amounts specified on the FFQ. Food group intakes were calculated as the sum of the individual component foods in servings. Validity of the FFQ compared to repeated 7DDRs measured six months apart and the reproducibility between two FFQs administered one year apart were assessed using Spearman correlations. Results Foods generally considered to be healthy were often overreported through FFQs, while those perceived unhealthy were underreported. The average 1-year reproducibility of all foods listed on the FFQ was 0.64 in both the WLVS and MLVS. Reproducibility of the food groups (mean: 0.70 among women and 0.71 among men) was generally higher than that for individual foods. Among women, the average validity correlation for individual foods was 0.59 when comparing the FFQ completed at the end of the 1-year study period with the 7DDRs and correcting for within-person variations in 7DDRs. Among men, the corresponding average validity correlation was 0.61. Food groups assessed by the FFQ had slightly higher validity correlations than did individual foods in both women (range: 0.44–0.92, mean: 0.61) and men (0.46–0.88, mean: 0.64). Conclusions This study reaffirms that the FFQ performs well in measuring foods and food groups and provides the data enabling investigators to adjust for measurement error in epidemiologic studies of foods and food groups in relation to health outcomes. Funding Sources The Women's Lifestyle Validation Study and Men's Lifestyle Validation Study were funded by National Cancer Institute.