Task 1: Key Concepts about Understanding Usual Intake and Day-to-Day Variation in Dietary Intakes

Occasionally, dietary research focuses on an acute exposure—that is, intake at a given point in time—such as when tracking the outbreak of a food borne illness.  Generally, however, for most surveillance, epidemiologic, and behavioral research purposes, dietary analyses are concerned with measuring usual intake—that is, long-term average daily intake.  This is because dietary recommendations are intended to be met over time and diet-health hypotheses are based on dietary intakes over the long term.

As noted in previous courses, the main instrument for gathering dietary intake data in the NHANES is the dietary recall.  Dietary recalls are rich in details regarding every item consumed (when, how, how much, with what), and for this reason are considered the main instrument for estimating food and nutrient intakes for the population.  However, because they only cover one or two 24-hour periods, they represent only a snapshot in time, rather than usual intake, and some of these snapshots are not typical diet days for the individual. 

Due to this day-to-day variation, one (or even a few) 24-hour recall(s) cannot be considered as an accurate assessment of an individual’s true usual intake.  Therefore, because long-run, or usual, dietary intake is most often the measurement of interest, statistical adjustments are often necessary.  One exception to this is that the mean of the population’s distribution of usual intake can be estimated from a sample of individuals’ 24-hour recalls, without sophisticated statistical adjustment.  For more advanced analyses, such as estimating the percentiles of a distribution, sophisticated adjustment techniques are needed.  See the Advanced Dietary Analyses course for more information.

The problem of estimating usual intake from 24-hour recalls can be thought of as a measurement error issue.  For more information on measurement error, see “Task 2: Key Concepts about Understanding Measurement Error.”