Relative validity and reproducibility of a semi-quantitative food frequency questionnaire to assess fruit and vegetable consumption in school-aged children

BACKGROUND: Since the beneficial effects of fruit and vegetable (FV) consumption on health are well known due to the synergy of their nutrients and non-nutrients, it is crucial to have good tools to assess the FV intake. A food frequency questionnaire (FFQ) is an adequate method to estimate FV consumption, but it is necessary to relate this dietary method to the geographic and cultural environment. Therefore, this study presented the development of a semi-quantitative FFQ to estimate the FV intake in school-aged children who usually consume cooked homemade and school meals. It also aimed to evaluate the relative validity and reproducibility of the FFQ. METHODS: School-aged children (baseline age 8 years) from 14 primary schools in the city of Zagreb participated in the study during the 2019/2020 school year. Parents/caregivers, together with the children, completed the FFQs and 3-day dietary records (3DDRs). The FFQ was designed to assess the consumption of eight food categories. The FFQ was validated using the 3DDR of 141 children (51.4% of boys), whereas the reproducibility test included the FFQ of 161 children (53.4% of boys). RESULTS: Of the eight food categories, FFQ overestimates the consumption of three and underestimates the consumption of three food categories (p < 0.05; Wilcoxon signed rank test) compared to the 3DDR. De-attenuated correlation coefficients estimated a significant relationship (0.217–0.384) between the FFQ and 3DDR. Cross-classification analysis revealed that overall, 28–41% of children were classified in the same quartile, whereas less than 10% of children were extremely misclassified for all food categories obtained from 3DDR and FFQ1. κ(w) values showed fair agreement for all food categories. The Bland–Altman analysis results showed a relatively small bias for all food categories (median between -11.7 and -54.8 g), with no systematic patterns between the FFQ and 3DDR. No differences were found between food categories estimated with the FFQs on both occasions, and Spearman’s correlation coefficients ranged from 0.664 to 0.712 (p < 0.01). Cronbach’s alpha values (α > 0.700) indicate good internal consistency, and ICCs (range 0.724–0.826; p < 0.01) indicate good reproducibility of the FFQ. CONCLUSION: The results indicate reasonable relative validity and acceptable reproducibility of the FFQ for estimating FV consumption among school-aged children.