Validity of a food-frequency questionnaire for a large prospective cohort study in Bangladesh

We have developed a thirty-nine-item semi-quantitative food-frequency questionnaire (FFQ) to assess the dietary consumption of 11 746 men and women in a prospective cohort study that evaluates the health effects of As from drinking water in Bangladesh. In order to validate the FFQ, two 7 d food diaries (FD) were completed for 189 randomly selected cohort participants in two different seasons of the year. Nutrient values were converted based on both the United States Department of Agriculture's National Nutrient Database and a food composition table for the Indian subcontinent. Pearson product-moment and Spearman non-parametric rank correlation coefficients comparing food and nutrient consumptions estimated from FFQ and 7 d FD were calculated based on log-transformed consumption values with or without adjustment for total energy and correction for within-individual variation. Correlations of macronutrients and common micronutrients including total fat, monounsaturated fat, polyunsaturated fat, saturated fat, protein, carbohydrate, dietary fibre, Na, K, vitamin B₆, vitamin B₁₂, riboflavin, Mn, thiamin and Fe were moderately good, ranging from 0.30 to 0.76. However, correlations of other micronutrients were weak (<0.30). Large seasonal variations in intakes of retinol equivalents and vitamin C were observed. This analysis documents the degree of validity of the FFQ in measuring specific nutrient intakes in the study population. To our knowledge, the present study is the first to document the validity of a FFQ with the use of 7 d FD in a Bangladeshi population.