related to the uncertainty surrounding the value and ensuring appropriate discussions about the variation in requirements. A challenge lies in obtaining adequate data to allow a reasonable approximation of the variability in requirements and hence the distribution of the requirement among individuals (Taylor, 2008).

The RDA is calculated from the EAR. It is dependent upon estimating the variance around the EAR and reflects a point estimate defined generally as two standard deviations above the EAR (Taylor, 2008). Although some refer to this reference value as “the requirement plus a safety factor,” this is potentially misleading in that it underplays the importance of the variability around the median. The RDA is intended to reflect the EAR plus two standard deviations.

This RDA calculation starts with the assumption that the distribution of a nutrient requirement is generally normal. However, this is not the case for a number of nutrients. There is also the need to describe the variance around the EAR. Such data are usually limited; when the variance is not known, the coefficient of variation is assumed, commonly as 10 percent. There is concern expressed by some that RDAs cannot be considered to be scientifically derived because too often the variance around the EAR cannot be determined precisely from the available data, and is therefore unknown, and the assumptions made about the variance may be inappropriate (Taylor, 2008).

The estimation of the RDA results in a value that is above the intake required for about 97.5 percent of the population. The RDA thus exceeds the requirements of nearly all members of the life stage group. Current guidance (IOM, 2000a, 2003) stipulates that the RDA is useful for some applications with individuals, but it is not appropriate when working with groups of persons for the purposes of assessing and planning for nutrient intake (Taylor, 2008).

The possibility of the AI—except for reference values for infants—was not considered when the DRI framework was first developed in 1994 (IOM, 2008). The AIs emerged as a result of the deliberations of the early study committees during the implementation of the initial DRI process. When the available data were judged lacking for the purposes of estimating an EAR, an AI was set. The value was seen as filling the gap that would have existed had no value been issued (Taylor, 2008).