A Short Dietary Method for Epidemiologic Studies

Abstract

Statistical analyses were performed on the measured 7-day food records of 93 Japanese-American men to investigate the variability of their daily nutrient intakes before developing a short dietary method for epidemiologic studies. Individual daily intakes of calories, protein, fat, carbohydrate, and sodium were analyzed for the following characteristics: 1) Equality of variance of the daily intakes and the transformed logarithmic values. 2) Normality of the distribution of the daily errors. 3) Comparison of daily means, standard deviations, and pair-wise correlation. 4) Differences between persons and days. The variances of the five dietary factors were heterogeneous, indicating large individual differences in daily eating patterns. In addition, analyses of the sodium data revealed that the variance was dependent on the mean intake, the daily residual errors deviated from a normal distribution, and the daily means of the sample differed significantly. Possible sources of error in the sodium data were the measurements of seasonings in food preparation, amounts of salt added to prepared foods, ingredients of restaurant dishes and new commercial products, and average values in food composition tables. A study that includes measured 7-day records, laboratory analysis of all food items of questionable sodium content, weight of salt additions at the table, and analysis of 24-hr urinary outputs is needed to determine if accurate estimates of sodium intakes can be obtained by current dietary methods.