[A dietary method for epidemiological study of cancer]

A semiquantitative frequency questionnaire was developed to estimate the average intake of foods and nutrients of individuals with the aim of utilization in cancer epidemiology. Validity of the questionnaire was evaluated in comparison with four 7-day dietary records carried on 25 spouses over one year. Absolute mean intakes estimated by the questionnaire and by the records were roughly comparable at the group level. However, the individual's estimated intakes generally showed poor agreements between the two methods in terms of rank correlation, except for a few foods such as fish and milk and for alcoholic beverages. Further efforts should be exercised to develop a more accurate dietary method.