Validity assessment of a food frequency questionnaire for school-age children in Northern Italy

We designed a food frequency questionnaire [FFQ] for Northern Italian school-children and tested its validity for the assessment of individual nutrient intake. Eighty-eight healthy children (49 females, 39 males — age range 9–12) were investigated. The FFQ we designed consists of 116 food items selected for their common use among our target population (food items found to provide up to 90 % of the following nutrients: energy, total lipids, saturate [SFA], polyunsaturate [PUFA] and monounsaturate [MUFA] fatty acids, carbohydrates, proteins, cholesterol, vitamins A, B1, B2, B6, C, D, E, calcium, iron, soluble and non-soluble fibres). To test the validity of the FFQ we collected data from a seven-day dietary record for comparison at the beginning of the study and six months later. The FFQ was distributed to parents several days before the collection of the second diary. The validity of the FFQ in the categorizing of individual children at the extreme quintiles was evaluated with contingency tables; the validity of the individual evaluations of nutrient intake was assessed by Pearson's correlation coefficient. The estimated energy intake is greater with the use of the FFQ. The intake of most nutrients is comparable, but there is a significant difference for proteins, PUFA, MUFA, cholesterol, vitamins B1, C, B6, E and for fibres. The greatest difference was found for vitamins B1 and C. The degree of correlation for most nutrients was within the 0.30 and 0.58 range. For most nutrients, the calculation of the nutrient density increased the correlation coefficient, while proteins, carbohydrates and MUFA decreased it. For some nutrients (vitamins A and B6) the coefficient of correlation was found to be low. The contingency analysis of absolute nutrient intake shows gross misclassification above all for MUFA and vitamin B1. The use of nutrient density removes the misclassification for MUFA but not for vitamin B1. Proteins are classified with less consistency when expressed as nutrient density. The comparison between a standard measurement methodology and the FFQ yields a higher consistency for individual estimates than for other paediatric populations but lower than that reported in adult populations.