Food consumption methods : development, reproducibility and validation of a food frequency questionnaire for vitamin B6

Objective: The correct classification of adults with a habitual low vitamin B6 intake by means of a food frequency questionnaire (FFQ). Assessment of reproducibility of the food frequency questionnaire and of the relative validity as compared to 3 d diet records (DR). Design: A self-administered food frequency questionnaire (FFQ) to measure habitual vitamin B6 take was developed using the data base of the first Dutch National Food Consumption Survey 1987/88 (DNFCS). The frequency and quantity of consumption of 92 (combinations of) food products, contributing to about 95% of vitamin B6 intake, were asked for. Recent food consumption was measured with 3 d diet records. Reproducibility of the FFQ was assessed after 10 months (n = 330). Subjects: Randomly selected men and women aged 20-79 y, stratified for 10 y age classes and sex. Results: A total number of 2867 FFQs was adequately completed (response rate 46%). A low vitamin B6 group was selected from the lowest tertile of age-gender specific vitamin B6 intake (36 men and 36 women, aged 20-49 and 36 men and 36 women, aged 50-79) and a so-called reference group (n = 300) was selected at random. The ratio of vitamin B6 intake according to FFQ and DR (FFQ/DR) varied between 0.77 and 0.99 for the various age-gender-study groups; for vitamin B6/g protein the FFQ/DR ratio varied between 1.00 and 1.10. After correction for intraindividual variance Pearson correlation coefficient between vitamin B6 intake according to FFQ and DR varied between 0.27 and 0.95 for the various age-gender-study groups, for vitamin B6/g protein the corrected Pearson correlation coefficients varied between 0.27 and 0.65. Of the subjects selected in the low B6 groups, 61-81% fell in the lowest tertile of vitamin B6 intake according to diet records and 3-6% fell in the extreme opposite tertile. The percentage subjects selected in the same/opposite quartile of vitamin B6 intake according to FFQ and DR was 50/11 for men, and 45/18 for women. Pearson correlation coefficients for vitamin B6 intake according to the first and the second FFQ were 0.64 and 0.72 for men and women, respectively; for vitamin B6/g protein these values were 0.70 and 0.51 for men and women, respectively. The proportion of subjects classified in the same/extreme opposite tertile of vitamin B6 intake was 56/6 and 62/5 for men and women, respectively. Conclusions: It is concluded that the FFQ was adequate for the selection of subjects with a habitually low vitamin B6 intake. Sponsorship: Ministry of Health, Welfare and Sports, the Netherlands.