Assessment of a short dietary method for prospective study on cancer.

This study was a trial to develop a short dietary questionnaire for estimating individual calorie and nutrient intakes among Japanese-American women in Hawaii participating in a prospective cancer study. One hundred five women recorded 4-day measured food records, and their average individual daily intakes of calories, protein, fat, carbohydrate, saturated fat, and cholesterol were computed. Individual food items of the 105 records were grouped according to common food categories and types of usage. The amount and frequency of use, and the average serving weights of the food groups were then calculated. Fifty food groups were selected for stepwise regression analyses of the individual nutrient intakes on the daily frequencies of the recorded items. Regression equations for each nutrient were computed from the 4-day records of 53 women and validated against the known intakes of the remaining 52 participants. The equations yielded high multiple R's and, except for protein and carbohydrate, there was resonable agreement between the mean observed and predicted intakes of the 52 women. However, the coefficients of determination of the observed and predicted intakes of the 52 women revealed considerable unexplained variation. Before designing a questionnaire, further studies are needed to develop and validate a precise short method for estimating individual nutrient intakes in large prospective studies.