Relative validity of fruit and vegetable intake estimated by the food frequency questionnaire used in the Danish National Birth Cohort.

Objective: To validate the fruit and vegetable intake estimated from the Food Frequency Questionnaire (FFQ) used in the Danish National Birth Cohort (DNBC). Subjects and setting: The DNBC is a cohort of 101,042 pregnant women in Denmark, who received a FFQ by mail in gestation week 25. A validation study with 88 participants was made. A seven-day weighed food diary (FD) and three different biomarkers were employed as comparison methods. Results: Significant correlations between FFQ and FD-based estimates were found for fruit (r=0.66); vegetables (r=0.32); juice (r=0.52); fruit and vegetables (F&V) (r=0.57); and fruit, vegetables, and juice (F&V&J) (r=0.62). Sensitivities of correct classification by FFQ into the two lowest and the two highest quintiles of F&V&J intake were 58-67% and 50-74%, respectively, and specificities were 71-79% and 65-83%, respectively. F&V&J intake estimated from the FFQ correlated significantly with urinary flavonoids ( r=0.39, p<0.005), plasma beta-carotene (r=0.21, p<0.05) and erythrocyte folate (r=0.33, p<0.005); all average biomarker levels increased across increasing quintiles of F&V&J intake. Conclusion: Data on F&V&J recorded by FFQ in DNBC can be used in etiological analyses of associations between fruit and vegetable intake and health outcomes, provided that the potential misclassification of dietary exposure is kept in mind.