OBJECTIVE:

Underreporting of energy intake is prevalent in food surveys, but there is controversy about which dietary assessment method provides greater underreporting rates. Our objective is to compare validity of self-reported energy intake obtained by three dietary assessment methods with total energy expenditure (TEE) obtained by doubly labeled water (DLW) among Brazilian women.

DESIGN:

We used a cross-sectional study.

SUBJECTS/SETTING:

Sixty-five females aged 18 to 57 years (28 normal-weight, 10 overweight, and 27 obese) were recruited from two universities to participate.

MAIN OUTCOME MEASURES:

TEE determined by DLW, energy intake estimated by three 24-hour recalls, 3-day food record, and a food frequency questionnaire (FFQ).

STATISTICAL ANALYSES PERFORMED:

Regression and analysis of variance with repeated measures compared TEE and energy intake values, and energy intake-to-TEE ratios and energy intake-TEE values between dietary assessment methods. Bland and Altman plots were provided for each method. chi(2) test compared proportion of underreporters between the methods.

RESULTS:

Mean TEE was 2,622 kcal (standard deviation [SD]=490 kcal), while mean energy intake was 2,078 kcal (SD=430 kcal) for the diet recalls; 2,044 kcal (SD=479 kcal) for the food record and 1,984 kcal (SD=832 kcal) for the FFQ (all energy intake values significantly differed from TEE; P<0.0001). Bland and Altman plots indicated great dispersion, negative mean differences between measurements, and wide limits of agreement. Obese subjects underreported more than normal-weight subjects in the diet recalls and in the food records, but not in the FFQ. Years of education, income and ethnicity were associated with reporting accuracy.

CONCLUSIONS:

The FFQ produced greater under- and overestimation of energy intake. Underreporting of energy intake is a serious and prevalent error in dietary self-reports provided by Brazilian women, as has been described in studies conducted in developed countries.