The bias resulting from the use of the cut-point method here is rather noticeable; thus, caution needs to be exercised when using the EAR cut-point method in situations in which requirements for a nutrient may be more variable than intakes of the nutrient.
The extent and direction of the bias that occurs when requirements are more variable than intakes will differ depending on whether the mean intake is above (as in Figure 4-11), equal to, or below the mean requirement. Carriquiry (1999) assessed the expected bias in several of these scenarios using a limited simulation study in which the relative sizes and standard deviations of the mean intake and the mean requirement were varied. The results suggest that in situations where the variance of requirement exceeds the variance of usual intake, the following cases arise:
When mean intake equals median requirement, use of the EAR cut-point method accurately estimates the proportion of the population with inadequate intakes.
When mean intake exceeds median requirement, use of the EAR cut-point method underestimates the proportion with inadequate nutrient intake.
When mean intake is less than median requirement, use of the EAR cut-point method overestimates the proportion with inadequate nutrient intake.
In the last two cases, the bias in the prevalence estimate can be significant even when the standard deviation of requirements is only slightly larger than the variation of usual intakes. The over- or underestimation of true prevalence is more pronounced when the true prevalence in the group is either very low or very high.
Regardless of the method chosen to assess prevalence of inadequate nutrient intakes in a group of individuals, information is required about the distribution of intakes of the nutrient in the group. Because the chronic effect of diet on an individual's well being is often of interest, the estimation of the distribution of long-term average intakes—that is, usual intakes—for the group is a concern. The usual intake distribution of a dietary component should have a spread (or variance) that reflects the individual-to-individual variation of intakes of that nutrient within the group.
Usual intake distributions can be estimated by adjusting the distribution of the mean of a few days of intake of each individual in the group. This general method was proposed by the National Research