Abstract

Using data obtained from 1,473 adults aged 18-68 yrs, residing in large cities and by use of a semi-quantitative food frequency questionnaire, we assessed the relative importance of various foods as indicators of both the amount and the variability of selected nutrient intake to develop a simple food frequency questionnaire. Since Cronbach's alpha value of the questionnaire including 78 food items was 0.76, the reliability of this questionnaire was acceptable. A large fraction of the variability of nutrient intake in this population could be explained by the small number of food items. The estimation of dietary nutrient intake such as total calories or protein content, which are derived from almost all foods, will require more food items with nutrients such as calcium or vitamin A, which are concentrated in a few food items. A dietary history ascertaining the intake of as few as 5-19 food items mighted be all needed in order to determent the association between disease outcome and the intake of a single nutrient. There was certainly a high level of agreement with nutrient intake by the sbujects who were cross-classified by quartiles of nutrient indices based on all the food items(78) and by quartiles of nutrient indices based on food items selected by stepwise multiple regression for selected nutrients. The data provided further evidence that useful information on dietary intake over an extended period can be obtained by a simple and relatively inexpensive food frequency questionnaire.

• Keywords: semi-quantitative food frequency questionnaire; variability; reliability; variance; food items; agreement