Using cross-check questions to address the problem of mis-reporting of specific food groups on Food Frequency Questionnaires

Abstract

Objective: To explore the potential mis-reporting of specific food groups from food frequency questionnaire (FFQ) data and to examine the effect of using a weighting factor on estimated nutrient intake and ranking of subjects within the cohort according to nutrient intake. Design and subjects: A weighting factor was calculated for each of the individual 6572 women aged 35–69 y for four food groups, fish, meat, vegetables and fruit, using FFQ data and cross-check responses. Results: The vegetable weighting had most effect on median intakes, particularly of fibre, vitamins A, C and E and folate. When all the weightings were applied, the median intakes of vitamins A and E were reduced by 35% and 27% respectively and the vitamin C intake was reduced by 44%. Ranking of subjects within the cohort according to nutrient intake was barely affected by the fish and meat weightings. The vegetable weighting had most effect on vitamin A with a rank correlation coefficient of 0.88. When all the weightings were applied the rank correlations for vitamins A, C and E and folate were all less than 0.90. Conclusion: Inclusion of cross-check questions in FFQs can provide an additional source of information on food group intake. This can be compared with FFQ data to help identify possible over-reporting and then to adjust frequency of intake accordingly. Sponsorship: World Cancer Research Fund.