Council (NRC, 1986) and was further developed by Nusser et al. (1996). As described below, to apply these methods of adjusting intake distributions it is necessary to have at least two independent 24-hour recalls or diet records for at least some individuals in the group (or at least three days when data are collected over consecutive days). Independent observations are obtained by collecting intake data over nonconsecutive days.
Several characteristics of dietary intake data make estimating the distribution of usual intakes for a group a challenging problem. This section focuses on the need for adjustment of distributions, illustrates the use of two of the most widely used approaches, and discusses the consequences of poorly estimating usual intake distributions.
Dietary intake data have characteristics that need to be taken into account when estimating the usual intake distribution of a nutrient for a group of individuals. If intake distributions are not properly adjusted, the prevalence of nutrient inadequacy will either be overestimated or underestimated, regardless of whether the probability approach or the cut-point method is chosen.
Should the distribution of observed intakes be used as an estimate of the usual intake distribution?
No. Although the mean of the distribution of observed intakes in the group is an unbiased estimate of the mean usual intake in that group (assuming that intakes have been accurately measured), the variance of the distribution of observed intakes is almost always too large (NRC, 1986; Nusser et al., 1996). This is because it includes both the within-person (day-to-day) variation and the individual-to-individual variation, thus leading to estimates of prevalence of inadequacy or excess that are likely to be higher than the true prevalence. In order to get accurate prevalence estimates, the distribution of observed intakes must be adjusted to more closely reflect only the individual-to-individual variability in intakes.
Individuals usually vary the types and amounts of the foods they consume from day to day. This translates into a large variability in