Usual Dietary Intakes: Background

Usual dietary intake is the long-run average daily intake of a nutrient or food. The concept of long-term average daily intake, or "usual intake," is important because dietary recommendations are intended to be met over time and diet-health hypotheses are based on dietary intakes over the long term. Consequently, it is the usual intake that is often of most interest to policy makers -- when they want to know the proportion of the population at or below a certain level of intake -- or to researchers -- when they want to examine relationships between diet and health. However, until recently, sophisticated efforts to capture this concept have been limited at best.

24-hour recalls are considered among the most precise methods of assessing diet. They ask people to report everything eaten and drunk during the previous 24 hours and are rich in details regarding every item consumed (when, how, how much, with what). However, a single recall represents only a "snapshot in time" and one does not represent a person's long-term average daily intake. In fact, for many dietary constituents, there is greater variation in intake day-to-day within a single individual than there is person-to-person within a population.

This limitation -- excessive intra-individual variation -- can lead to serious problems in data interpretation. For example, the distribution of single-day intakes has a larger variance than the true usual intake distribution. This means using a single recall -- or even the average of two -- would lead to a biased estimate of the fraction of the population with usual intake above or below some standard.

Effect of Excessive Intra-Individual Variation on Distributions
[D]

Another problem relates to the interpretation of regression analyses in examining diet-disease relationships. The excessive intra-individual variation inherent in 24-hour recalls is a type of measurement error. All types of dietary measurement error will generally lead to attenuation of the association and a loss of statistical power in regression analyses.

Effect of Measurement Error on Regression Analyses
[D]

Fortunately, statistical modeling can be used to mitigate this limitation of 24-hour recalls. NCI has developed a method of assessing long-term average, or "usual," dietary intake that builds on the strengths of 24-hour recalls and can employ food frequency questionnaires when applicable. This research has the potential for advancing our methodological work, and has implications for our work in food guidance and policy. An improved sense of what people really are eating will help in formulating targeted recommendations and evaluating progress toward national health objectives.