Variability in planning and assessment: statistical considerations and models

Using Criteria to Establish Nutrient Intake Values (NIVs)

One of the most important of the nutrient intake values (NIVs) is the average nutrient requirement (ANR). The ANR is defined as an intake value that will be adequate for half of the individuals in a group of people with similar characteristics. It is used to estimate the prevalence of adequacy, and it serves as the basis for the individual nutrient level (INLx). The determination of adequacy is a complex process, with the resulting value of the ANR dependent on the criterion or functional outcome chosen to define nutrient adequacy. Because nutrients have multiple sites of action in human metabolism, it is possible to demonstrate abnormal function in one parameter measured or observed as a result of inadequate intake of a nutrient, while other parameters requiring the same nutrient appear normal or within normal ranges. Thus, depending on the criterion of adequacy selected, the requirement for a given nutrient may be at a lower or a higher intake amount. In harmonizing development of NIVs, it is important to clearly identify the criterion of adequacy selected and the rationale for its selection. Rarely are available data sufficient to provide dose—response information from which to select a level of intake at which half of the individuals demonstrate adequacy and half appear to demonstrate inadequacy. Three levels of intake, of which at least one level of intake is below the requirement for most of the individuals in the sample, and one level of intake is above their requirement, are useful for establishing a level at which half of the group might be considered to demonstrate adequacy. Types of human nutrient studies that may be used to obtain data are discussed, as well as characteristics of the sample size needed to demonstrate adequacy. The variation in requirements is also an important aspect in predicting levels of intake that will have defined probabilities of adequacy for groups (to develop the INLx, where x is the defined probability chosen). An analysis of the origins of different types of variability is presented. When estimating energy requirements, a special case of NIVs, important issues must be considered. Additionally, an example of evaluating data used to establish an ANR for vitamin A, and the effect of variability in requirements for vitamin A, is provided.