cient to maintain a desirable body store or reserve. In developing an RDA (and AI, see below), emphasis is placed instead on the reasons underlying the choice of the criterion of nutritional adequacy used to establish the requirement. It is not designated as basal or normative.

When the distribution of a requirement for a nutrient among individuals in a group can be assumed to be approximately normal (or symmetrical) and a standard deviation (SD) of requirement (SD_requirement) can be determined, the EAR can be used to set the RDA as follows:

If data about variability in requirements are insufficient to calculate an SD_requirement for that specific nutrient in that population group, but normality or symmetry can be assumed, then a coefficient of variation (CV) of 10 percent is assumed and the calculation becomes:

The assumption of a 10 percent CV is based on extensive data on the variation in basal metabolic rate (FAO/WHO/UNA, 1985; Garby and Lammert, 1984) and the CV of 12.5 percent estimated for the protein requirements in adults (FAO/WHO/UNA, 1985). If there is evidence of greater variation, a larger CV is used. In all cases, the method used to derive the RDA from the EAR is stated in the DRI reports.

Since it is derived from the EAR, the RDA’s usefulness as a goal depends on the choice of nutritional status indicator or criterion and the type and amount of data available. Its applicability also depends on the accuracy of the form of the requirement distribution and the estimate of the variance of requirements for the nutrient in the population subgroup for which it is developed. For many of the nutrients there are few direct data on the requirements of children and the elderly. In the case of children, EARs and RDAs are based on extrapolations from adult values.