Reproducibility and validity of a food frequency questionnaire among pregnant women in a Mediterranean area

Background Studies exploring the role of diet during pregnancy are still scarce, in part due to the complexity of measuring diet and to the lack of valid instruments. The aim of this study was to examine the reproducibility and validity (against biochemical biomarkers) of a semi-quantitative food frequency questionnaire (FFQ) in pregnant women. Methods Participants were 740 pregnant women from a population-based birth cohort study in Valencia (INMA Study). We compared nutrient and food intakes from FFQs estimated for two periods of pregnancy (reproducibility), and compared energy-adjusted intake of several carotenoids, folate, vitamin B12, vitamin C and α-tocopherol of the FFQ in the first trimester with their concentration in blood specimens (validity). Results Significant correlations for reproducibility were found for major food groups and nutrients but not for lycopene (r=0.06); the average correlation coefficients for daily intake were 0.51 for food groups and 0.61 for nutrients. For validity, statistically significant correlations were observed for vitamin C (0.18), α-carotene (0.32), β-carotene (0.22), lutein-zeaxantin (0.29) and β-cryptoxantin(0.26); non-significant correlations were observed for retinol, lycopene, α-tocopherol, vitamin B12 and folate (r≤0.12). When dietary supplement use was considered, correlations were substantially improved for folate (0.53) and to a lesser extent for vitamin B12 (0.12) and vitamin C (0.20). Conclusion This study supports that the FFQ has a good reproducibility for nutrient and food intake, and can provide a valid estimate of several important nutrients during pregnancy.