Abstract
A validation study of an interviewer-administered, seven-day semi-quantitative food frequency questionnaire (7-d SQFFQ) was conducted in Bangladeshi rural preschool age children. Using a cross-sectional study design, 105 children from 103 households were randomly selected. For the SQFFQ, a list of commonly consumed foods was adapted from the Bangladesh national micronutrient survey 2011-12. The data on the actual number of times and the amount of the children's consumption of the foods in the preceding 1 week were collected by interviewing the mothers. The intake was compared with two non-consecutive days 24-h dietary recalls conducted within 2 weeks after the SQFFQ. Validity was assessed by the standard statistical tests. After adjusting for the energy intake and de-attenuation for within-subject variation, the food groups (cereals, animal source foods, milk and the processed foods) had 'good' correlations between the methods (rho 0⋅65-0⋅93; P < 0⋅001). Similarly, the macronutrients (carbohydrate, protein and fats) had 'good' correlations (rho 0⋅50-0⋅75; P < 0⋅001) and the key micronutrients (iron, zinc, calcium, vitamin A, etc.) demonstrated 'good' correlations (rho 0⋅46-0⋅85; P < 0⋅001). The variation in classifying the two extreme quintiles by the SQFFQ and the 24-h recalls was <10 %. The results from Lin's concordance coefficients showed a 'moderate' to 'excellent' absolute agreement between the two methods for food groups, and nutrients (0⋅21-0⋅90; P < 0⋅001). This interviewer-administered, 7-d SQFFQ with an open-ended intake frequency demonstrated adequate validity to assess the dietary intake for most nutrients and suitable for dietary assessments of young children in Bangladesh.