Abstract
BACKGROUND: Food frequency questionnaires (FFQs) are often used to assess dietary intakes due to their ability to assess intake over extended periods, their low respondent burden, and their cost-effectiveness. A quantitative FFQ that includes locally appropriate food items for 5-year-old children in a multiethnic Asian population was developed, but its validity has not previously been evaluated. OBJECTIVE: To evaluate the relative validity of a newly developed FFQ as a dietary assessment tool for 5-year-old children in a multiethnic Asian population. DESIGN: The 112 -food item FFQ was administered by trained interviewers to caregivers of children. Frequency of food items consumed in the previous month and portion size information were collected. The FFQs were evaluated against 3-day nonweighed diet records (DRs) completed by caregivers. PARTICIPANTS/SETTING: The dietary data of 361 children aged 5 years from the Growing Up in Singapore Towards Healthy Outcomes mother-offspring cohort were collected in 2015-2016. MAIN OUTCOME MEASURES: Nutrients of interest included energy, macronutrients, fiber, cholesterol, vitamin A, beta carotene, calcium, and iron, calculated from the FFQs and DRs. STATISTICAL ANALYSES PERFORMED: Nutrient intakes according to FFQs in relation to DRs were assessed using Pearson's correlation, Lin's concordance, Bland-Altman plots, quintile joint classification, and Cohen's κ statistics. RESULTS: The highest energy-adjusted correlation (Pearson's r=0.71) and concordance (Lin's concordance=0.69) were observed for calcium. Fiber, saturated fat, polyunsaturated fatty acids (PUFA), cholesterol, and iron also showed correlation coefficients and concordance of at least 0.40. Bland-Altman plots suggested no substantial bias across ranges of intakes for the nutrients with correlations and concordance of 0.40 or above. Quintiles joint classification showed substantial agreement for calcium (κ=0.66), and moderate agreement for iron, fiber, saturated fat, polyunsaturated fat, and beta carotene (κ=0.59, 0.54, 0.49, 0.44, 0.43, respectively). CONCLUSIONS: The newly developed FFQ is in reasonable agreement with DR for estimating intakes of calcium, fiber, saturated fat, PUFA, cholesterol, and iron. In addition, the FFQ is able to classify children according to quintiles of nutrient intakes, with moderate to substantial quintile agreements between FFQ and DR for calcium, iron, fiber, saturated fat, PUFA, and beta carotene. To assess the remaining nutrients, DR method is recommended instead of the FFQ.