초록
Intake frequency of one food is often associated with other food items, but few studies examined the relationship of food intake frequency among food items. Finding the relationship among intake frequencies of different food might be useful to understand the food intake patterns of population and correlated foods would be used as an indicators of another food intake. Relationship of food intake can be also applied to make a more simple and useful form of food frequency questionnaire to assess the association between diet and various diseases. The objective of this study was to examine the correlation of intake frequency among food items in food frequency questionnaire (FFQ). A FFQ with 84 food items was administered to 14533 Korean males who 40-65 years of age participating Korean cancer research survey. Data fromm 7647 subjects who completed FFQ were used to examine correlation among food items with three different methods-log linear regression models, Spearman correlation coefficients and cell frequency distribution. To examine the rank correlation, coefficients were calculated by Spearman correlation after scoring the frequency categories. Three most correlated foods were selected in every food intems by three methods each. In most food items, there was positive correlatin, except cooked rice and cooked brown rice, in intake frequency between foods that belonged to similar food groups. But serveral food items-Sausage (processed fish, cheese), Milk (whilte bread, orange juice), Soymilk(other juices), Cheese (pizza, butter), and Coffee(thick beef soup)-showed correlation among totally different food groups. Two sets of food items which were selected by log linear regression model and Sperman correlation coefficients were compared. There were exactly three common foods in 18 food items, 2 common foods in 47 items, 1 common food in 16 items and no common food in 3 items among 3 ranked foods. Three sets of selected food were compared. There were exactly three common foods in 5 food items, 2 common foods in 21 items, 1 common food in 34items and no common food in 24 items among 3 ranked foods. These results indicate that certain patterns exist among intake frequencies of specific food items in the FFQ. More researches are suggested to understand the relationshiops among the intakes of foods so that this information can be used in developing better FFQ or analyzing missing items from self-administered FFQ.
