Measuring dietary patterns in surveys.

This paper reviews the strengths, weaknesses, and limits to interpretation of dietary intake methods used in national nutrition surveys, including per capita consumption of food and nutrients and the 24-hour recall. These dietary assessment methods make up but one of five categories of nutritional assessment techniques. The other four are body measurements, hematological and biochemical tests, medical examination for the presence of clinical signs of deficiency or toxicity, and medical history. No one method alone is sufficient for assessing the nutritional status of individuals or groups. In the hypothesis-generating phases of research on diet and cardiovascular diseases and diet and cancer, correlations of per capita availability of food and nutrients with mortality from these diseases in several countries proved useful. However, the approach is limited to the extent that similar data are available from several other countries and that the association observed is true and not spurious. The 24-hour recall method has proved to provide accurate and reproducible estimates of the mean intakes of population groups, but multiple-day information is necessary for characterizing an individual's usual nutrient intake. Food frequency questionnaires are useful when one is interested in food rather than nutrient consumption of individuals or groups. The 24-hour recall, food diary, and food frequency methods can be used to develop a descriptive epidemiology of food intake within U.S. population groups with specified fitness and activity characteristics or to monitor the prevalence of dietary risk factors within the population. They are limited to the extent that food composition data are available. When selecting a dietary assessment method to be used in a nutrition survey, three points must be kept in mind: Practicality in terms of respondent burden and analysis resources, reliability, and validity.