Estimation of Usual Intake Distributions

Researchers at Iowa State University developed statistical methods for estimating the distributions of usual food and nutrient intakes for populations and subpopulations. These distributions are required to determine the proportions of the population who are at risk for inadequate intake of essential nutrients or for excessive intake of undesirable dietary constituents, including pesticide, antibiotic, or chemical residues.

The approach is based on the assumption that individuals can more accurately recall the types and amounts of foods they ate yesterday than they can recall intake over any longer period of time. When at least two days of dietary information are available for individuals in the sample, it is possible to estimate long-term average intake by removing the within-person variation in intake, in order to obtain better estimates of the distribution of usual, or long-term average.

Many statistical procedures are based on the assumptions that the data under investigation are normally and identically distributed and come from a simple random sample. Dietary survey data typically do not meet these assumptions, and the method developed here does not require them. Also, nuisance effects caused by seasonality, day of week effects, and sequence of survey day can be removed from the data with this method.

Methods for estimating distributions of usual food intakes have been developed for the case when the probability of consumption is independent of the amount consumed. The key issue here is that there are many zeros in observed food intakes due to the fact that individuals do not eat the same foods each day. Work to extend the method to estimate usual intake distributions of foods when the daily consumption levels are related to frequency of consumption is underway. This extension is important to assessing exposure from pesticides on foods.

Software that implements the methods to estimate the usual intake distributions was released recently: a SAS/IML^© version, called SIDE, and a more user-friendly C Language/X Windows based version called C-SIDE. C-SIDE includes a module which implements the method of usual food intake distributions.

The development of the software was partially supported by the ARS/USDA and Iowa State University Ag. Experiment Station.

SIDE and C-SIDE are products of Iowa State University. These two programs are only available from Iowa State University.

Download a PDF version of the order form suitable for printing and return by mail or fax (e-mail orders will not be accepted).

C-SIDE and SIDE can be ordered by contacting:

Sandie Smith
Center for Survey Statistics & Methodology
219 Snedecor Hall
Ames, IA 50011

phone: 515/294-9773
FAX: 515/294-4433
e-mail: sandie@iastate.edu