Contributions of specific foods to absolute intake and between-person variation of nutrient consumption

Contributions of specific foods and supplements to absolute intake and between-person variance in consumption of 19 macronutrients, vitamins, and minerals were examined using dietary records of all foods consumed over 4 weeks by 194 US nurses aged 34 to 59 years. To measure their contribution to absolute nutrient intake, we ranked foods by the percentage of the population's total nutrient intake that they provided. To assess the degree to which differences in consumption of specific foods explain between-person variability in nutrient intake, we re-ranked the 20 foods contributing most to absolute intake of each nutrient as independent variables in stepwise multiple regression analyses predicting total intake of that nutrient. The increase in percentage of the variance in nutrient intake explained by the addition of a food to the progressively larger list of food items (expressed as the cumulative R²) served as the measure of contribution to variation in intake. Some nutrients had only a few major sources and were assessed relatively well by a small number of foods. For preformed vitamin A without supplements, 10 foods accounted for 82% of the absolute intake and 98% of total variance. The corresponding percentages for absolute intake and total variance, respectively, were 66% and 94% for beta-carotene; 77% and 92% for cholesterol; and 71% and 95% for vitamin C without supplements. In contrast, 20 foods accounted for only 54% of the absolute intake and 73% of the variance for total energy intake; 58% and 89%, respectively, for total carbohydrates; and 59% and 84%, respectively, for potassium. The proportions of absolute intake and variance, respectively, contributed by supplements alone were 50% and 99% for vitamin C, 43% and 88% for preformed vitamin A, and 2% and 44% for calcium. These findings indicate that the contribution of supplements and the number of foods necessary to account for the variation in nutrient intake among persons differ considerably by nutrient; such information may be helpful in the design and use of dietary questionnaires.