requirements as defined in the DRIs does not necessarily indicate a prevalence of nutrient deficiency. For example, two different indicators for estimating an average requirement were identified for vitamin A. One was the reversal of night blindness. The other, for which an EAR was calculated, was the minimum acceptable liver vitamin A reserve. A 10 to 15 percent prevalence of usual intakes below the calculated value required to prevent night blindness would indicate a more serious public health problem than a similar prevalence of intakes below the value required to maintain liver stores in healthy individuals. Vitamin A is the only nutrient for which there are two approaches for establishing requirements to address two different endpoints. This nutrient, however, highlights the importance of considering the severity of the consequences of not meeting requirements for particular nutrients when interpreting prevalence estimates to justify the need for discretionary fortification. In addition, based on such factors as geographic location, access to food, patterns of intake, and demographics, not meeting the requirements for one nutrient (e.g., vitamin D) in a given population may pose more of a health risk than not meeting the requirements of another nutrient. Depending on the prevalence of inadequacy and the severity of the health consequences associated with inadequate intakes of a particular nutrient, regulatory agencies may wish to encourage discretionary fortification or to consider population-level interventions (similar to the approach taken with folate) rather to address identified problems.

Before an observed prevalence of nutrient inadequacy can be interpreted to scientifically justify the need for increased availability of the nutrient in the food supply, some analysis of the dietary correlates and sociodemographic characteristics associated with inadequate intakes in the population is required. Since discretionary fortification is first and foremost a strategy to increase nutrient density, it is important not to embark on this intervention without some indication that increased nutrient density might help to ameliorate the identified nutrient inadequacy. For example, if inadequate nutrient intakes are observed in the context of inadequate energy intakes, strategies to increase total food intake may be more important than strategies to increase the nutrient levels in food. An association between inadequate energy and nutrient intakes might also be indicative of an underreporting problem in the dietary intake data,