Learn More about Food Composition Databases for Food Frequency Questionnaires and Screeners

In order to estimate total nutrient and food pattern equivalent intake from a food frequency questionnaire (FFQ) as well as a frequency-type screener (see Food Frequency Questionnaire Profile and Screeners Profile), each line item on the questionnaire must be associated with nutrient and food group composition data. Each line item on an FFQ is typically a food group (for example, lasagna or fruit) composed of many similar individual foods (for example, various types of lasagna and fruit). As a result, a method to derive a single estimate of each nutrient and food group is required. The Willett FFQ that queries the frequency of use of particular portions of foods and beverages uses professional judgment to establish a single value for each FFQ line item. Generally, however, population data are used to generate these estimates, preferably data from the study population.

Population data consisting of dietary recalls or records have been used in a variety of ways to construct nutrient and food group databases for frequency instruments. One approach, using [glossary term:] 24-hour dietary recalls (24HR) from the [glossary term:] National Health and Nutrition Examination Survey (NHANES):

  1. Groups all foods reported into FFQ-specific food groups;
  2. Estimates the nutrient density (nutrient per 100 grams) of each food group; and
  3. Assigns density values for all FFQ food groups. This assignment may be done separately for each sex and age group, and by different portion size categories if portion size is queried in the FFQ. A [glossary term:] mean or [glossary term:] median can be used. For example, the Block FFQ uses weighted median densities multiplied by sex/age/portion size median gram weights.

Another approach is to regress nutrient intake on portion size categories, with or without consideration of other variables such as sex and age.

Using nationally representative 24HR data, Subar et al. compared different methods to compute nutrient estimates for an FFQ to determine the optimal quantitative method for the National Cancer Institute's (NCI) FFQ, the Diet History Questionnaire. They found that using means was superior to using medians, and that incorporating age is unnecessary when portion size categories are incorporated into the analysis. The nutrient composition database for NCI's Diet History Questionnaire is based on those results.

Many FFQ databases in the United States use national 24HR dietary data from NHANES. However, these nationally representative data may be suboptimal for particular study populations. For example, researchers with the Southern Community Cohort Study found that use of a database limited to the South and non-Hispanic whites and African Americans rather than the entire country led to differences in estimated dietary intake, particularly for African American women. This strategy may be useful for studies that differ substantially from the U.S. population as a whole.

For More Information

Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J, Gardner L. A data-based approach to diet questionnaire design and testing. Am J Epidemiol 1986 Sep;124(3):453-69. [View Abstract]

National Cancer Institute, Division of Cancer Control and Population Science, Applied Research Program. Diet History Questionnaire II. (Accessed 6/16/14).

Signorello LB, Munro HM, Buchowski MS, Schlundt DG, Cohen SS, Hargreaves MK, Blot WJ. Estimating nutrient intake from a food frequency questionnaire: incorporating the elements of race and geographic region. Am J Epidemiol 2009 Jul 1;170(1):104-11. [View Abstract]

Subar AF, Midthune D, Kulldorff M, Brown CC, Thompson FE, Kipnis V, Schatzkin A. Evaluation of alternative approaches to assign nutrient values to food groups in food frequency questionnaires. Am J Epidemiol 2000 Aug 1;152(3):279-86. [View Abstract]

Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985 Jul;122(1):51-65. [View Abstract]