especially if interviews are conducted by trial interventionists and/or registered
dietitians. Underreporting has been shown, especially over time and in obese
individuals. Most investigators use the Nutrient Database for Research (NDS-R).
As noted, while generally accurate, for select nutrients, the database may not be
fully populated for all foods. Thus, some measurement error is introduced at the
analysis phase of measurement as well.
Food Frequency Questionnaire (FFQ)
The food frequency questionnaire (FFQ) is the more commonly administered
self-report methodology used in epidemiological research (Bingham et al. 1994 ;
Cade et al. 2002 ; Schatzkin et al. 2003 ). This is largely driven by cost savings
related to the use of scannable, bubble-format questionnaires. FFQs have evolved
over time, but the general format is the same. Respondents are asked to report
habitual/usual intake including the portion size and frequency of consumption of a
somewhat comprehensive list of food items, grouped by food group. Lists vary
from a low of about 60 to a high of about 200 food items, with most being 120– 150
items long. In addition to indicating the portion size and frequency, respondents
answer questions regarding eating behaviors such as type of milk consumed, use of
condiments,flavorings, and common cooking methods. All of this is used to further
characterize the food line items for nutrient estimations. Some food frequency
questionnaires also ask the respondent to record dietary supplement use. As with
other self-report methods, education on estimation of portion size is advised to
improve report accuracy. FFQs most commonly query usual intake over the pre-
vious 12 months, although variation in time frame for report may be implemented
Fig. 9.1 USDA multi-pass
method for 24 h dietary
recalls. Adapted from
Moshfegh et al. ( 2008 )
174 T.E. Crane and C.A. Thomson