Abstract
Background: In order to assess the nutrient intakes of a population, it is necessary to estimate the distribution of nutrient intakes in the population and to compare the distribution with the Dietary Reference Intakes (DRIs). The distribution of usual intakes must be estimated because DRIs are the reference values for only usual intakes, and not for short-term intakes. In this article, the statistical methods for estimating the distribution of usual intakes have been reviewed and some examples have been illustrated.
Existing Methods: The National Research Council (NRC) proposed a principle method based on analysis of variance, but this method was available only for normally distributed data. The Best-Power (BP) method can be applied to non-normally distributed (skewed) data. The Iowa State University method is better than the BP method as it considers the statistical property of the actual dietary data, but the improvement in performance is minor. AGE MODE and AGEVAR MODE are expected to have smaller standard errors when used for the estimation of age-specific distributions. The Iowa State University Foods method is appropriate for dietary data with many zero intakes.
Examples: In the Prefectural Health and Dietary Surveys of Nagano, Kumamoto, and Saitama, dietary surveys were conducted on multiple days. The distributions of usual intakes were estimated, and the nutrient intakes of the populations were assessed based on the DRIs.
Conclusions: These statistical methods should be used for estimating the distribution of usual intakes of nutrients and for applying the DRIs to assess the dietary intakes of a population.
