Validation of Dietary Assessment Methods in Adolescent Populations
Table 6.1 presents validation studies on adolescent populations, including
the studies included in the comprehensive reviews of the literature through
2000 by McPherson and colleagues (1), and by Livingston and colleagues
(161) discussed in Exhibit
Food Record (FR) and Diet History (DH) Methods
records, either weighed or estimated, and diet
histories have received very limited validation in adolescent populations.
Energy intake reported on weighed food records underestimated total
energy expenditure (TEE)
by 18 to 24 percent in two European studies of adolescents (138;196).
Likewise, estimated food records underestimated TEE by more than 20%, with obese teens underreporting energy intake by 42% (195).
An intensive diet history interview (2 hours) was found to underestimate
energy intake by 10 to 14% compared with TEE measurements by the DLW (Doubly
Labeled Water) method in both vegans and omnivore adolescents (253;254).
Another study found no significant differences at the group level between
energy intake reported in a diet history interview and TEE measurements (138).
A reanalysis of a 1990 dataset of concurrent measurements of TEE by the
DLW method, 7-day weighed food records (FR), and diet history (DH) by Livingston
et al. found only 25% of children who underreported energy intake on weighed
food records were identified by blanket cut offs of 1.55 x BMR, based on
WHO physical activity levels (199).
However, only 6 adolescents were included in the dataset. A 3-day weighed
FR correlated significantly with biomarkers of folate intake in a population
of older (more than 15 years old) Canadian female adolescents (255).
24-Hour Recall (24HR) Interviews
Only two studies have examined the validity of 24-hour
recall (24HR) interviews in populations containing adolescents. A comparison
between energy intake from three 24HR interviews over a 2-week period and
TEE measurements by the DLW method in a large, diverse group of Alabama
youth age 9 to 14 years achieved close agreement at the group level, but
wide individual variability (200).
In this study, data for adolescents were not analyzed separately and study
participants were admitted to a metabolic unit for DLW dosing and the first
recall. In Minnesota middle school students, a parent-assisted telephone-administered
24HR interview underestimated energy intake reported on a parent-assisted
3-day FR by 12% (256).
Food Frequency Questionnaires (FFQs)
The Youth Adolescent Questionnaire (YAQ), which is an adaptation for adolescents
of the Harvard
Frequency Questionnaire (HFFQ), found close agreement at the group level
between energy intake reported on the YAQ and TEE measurements by the DLW
method, but not at the individual level (208).
Fifty percent of the adolescents misreported intake; misreporting was related
to body weight and percent body fat. The YAQ has also been validated with
multiple 24HR interviews in mainly white middle-class adolescent populations
The 27 fruit and vegetable questions on the questionnaire have been studied
in a diverse population (24% black and 15% Hispanic) of high school students;
a shorter (6 items) questionnaire had higher correlations with fruit and
vegetable intake reported on three 24HR interviews (257).
Small validation studies have assessed a variety of other FFQs in adolescent
populations. The differences in the instrument as well as the study designs
and adolescent populations limit generalizations of results. In 2002 Van
reviewed 26 validation studies of food frequency questionnaires in adults
and 8 studies in children/adolescents (167;180;183-185;257;259;260)
and concluded available FFQs have only limited capability to make valid
assessments of fruit and vegetable intake from children of various socioeconomic
status and ethnic groups, although older children and adolescents were generally
more accurate than younger children. Larger studies on more diverse groups
Validation studies on a variety of short questionnaires targeting specific
nutrients or food groups are presented in Table 6.1.