distribution were also examined: the ninety-fifth percentile of usual intakes is just 2,413 mg (77.8 mmol)/day implying that intakes seldom exceed the UL.

When assessing the intake of populations, a number of other questions need to be considered:

• To what extent does the nutrient requirement affect intake? Such a relationship, described by WHO/FAO/IAEA (WHO, 1996), would limit the validity of the probability approach.

• What kinds of adjustments can be made, if any, for biases in the food intake data?

• What factors should be considered in interpreting the findings in different populations?

• What is the allowable level of inadequate intake in a population before concern is raised?

The EAR also may be used as a basis for planning or making recommendations for the nutrient intakes of free-living groups. If nutrient intakes are normally distributed, a target intake for a population group can be estimated based on the EAR and the variance of intake. The objective might be to set a value for the mean intake of the group that will ensure that most individuals (usually 97 to 98 percent) meet their nutrient requirement. In order that less than 2 to 3 percent of intakes fall below the EAR, a group's mean intake must be at least two SDs of intake above the EAR. Because the SD usually varies in relation to the magnitude of intake, the coefficient of variation (CV) of intake is used to calculate the target mean intake. The following formula has been derived for this calculation (see Beaton [1994] and WHO [1996] for more details):

Target mean intake for a group = EAR/(1 − [2 × CV of intake])

Where CV = SD of intake/mean intake.

For example, if a group of women in a nursing home had phosphorus intakes with a CV of 0.16, and intakes were normally distributed, achieving a group mean intake of 853 mg (27.5 mmol)/day would ensure that only 2 to 3 percent would have intakes below the EAR of 580 mg (18.7 mmol)/day (580/1−[2 × 0.16]) = 853). If intakes are not normally distributed, other mathematical approaches will be needed.