Reproducibility and relative validity of a semi-quantitative food and beverage frequency questionnaire for Spanish children aged 3 to 11 years: the COME-Kids F&B-FQ

A reliable food and beverage frequency questionnaire (F&B-FQ) to measure dietary intakes for children across Spain is currently unavailable. Thus, we designed and assessed the reproducibility and relative validity of a new F&B-FQ in 210 Spanish children aged 3-11 years. COME-Kids F&B-FQ contained 125 items to assess the usual diet intake in the past year among children. To explore the reproducibility, caregivers answered COME-Kids F&B-FQ twice over a 15-day period (± 1 week). To evaluate the relative validity, estimates from a third COME-Kids F&B-FQ administered at 1 year of follow-up were compared with the mean estimates from 3-day dietary records (3d-DR) collected at baseline, 6 months, and after 1 year of follow-up. Reproducibility and relative validity of the COME-Kids F&B-FQ in estimating food groups and nutrients were assessed using Pearson (r) and intra-class (ICC) correlation coefficients. We used the kappa index to evaluate the agreement in repeat administrations or with the 3d-DR. We used Bland-Altman plots to identify bias across levels of intake. A total of 195 children (105 boys, 90 girls) completed the study. The reproducibility of data estimated from COME-Kids F&B-FQ was substantial with mean r and ICC being 0.65 and 0.64 for food groups and 0.63 and 0.62 for nutrients, respectively. Validation assessments comparing the FFQ and 3d-DRs showed r = 0.36 and ICC = 0.30 for food groups and r = 0.29 and ICC = 0.24 for nutrients. The mean agreement for food group reproducibility and relative validity was 86% and 65%, respectively. These estimates were 85% for reproducibility and 64% for relative validity in the case of nutrients. For reproducibility and relative validity, the overall mean kappa index was 63% and 37% for all food groups and 52% and 27% for nutrients, respectively. Bland-Altman plots showed no specific bias relating to the level of intake of nutrients and several food groups.

Conclusion: COME-Kids F&B-FQ showed substantial reproducibility and acceptable relative validity to assess food and beverage intake in Spanish children aged 3 to 11 years. Most children were correctly classified in relation to the intake of food groups and nutrients, and misclassification was unlikely with reference to 3d-DR.

What is known: • The estimation of dietary intake in children is complex, especially in large cohorts. • The food frequency questionnaire is a well-recognized and the most frequently used method for assessing food consumption.

What is new: • A new food and beverage frequency questionnaire including a beverage section and novel plant-based food items has been validated in Spanish children aged 3-11 years.