Selective underreporting of energy intake in women: magnitude, determinants, and effect of training.

OBJECTIVE: The aim of this study was to quantify underreporting of energy intake in Brazilian women; identify underreporting determinants; find out if underreporting was selective and; test if a motivational multimethod training, in combination with providing the subjects some results from the prior recording period, was able to reduce underreporting.
DESIGN: Energy intake (EI) was assessed by a 7-day diet record. Energy expenditure (EE) was calculated by heart rate monitoring. EI:EE ratio lower than one in subjects who did not lose weight in one month was considered underreporting. Underreporting was correlated with anthropometric, behavioral, and psychological parameters. Food and nutrient consumption was compared between underreporters and non-underreporters. A focus group investigated the main causes of underreporting. Subjects were told that the earlier food records' results were unrealistic and submitted to a motivational training. Then, they were reevaluated for underreporting.
SUBJECTS: Subjects were recruited by advertisements for a physical activity program. Thirty-eight healthy women, 13 normal-weight (34%), 13 overweight (34%), and 12 obese (32%), enrolled in the study. Three subjects (2 normal-weight and 1 obese) (8%) withdrew. STATISTICAL ANALYSIS PERFORMED: Analysis of variance, paired t tests, and simple linear regression.
RESULTS: Seventeen women (49%) underreported their EI by 21%. A significant negative correlation was found between social desirability and EI:EE. Undereating, errors in portion sizes estimation and the inconvenience of having to record everything that was eaten seemed to explain underreporting. Mean portion sizes did not differ for underreporters and non-underreporters. Fewer self-reported years of education was correlated with underreporting only among normal-weight women. Training and confrontation with earlier results reduced underreporting rate to 33%, but did not affect macronutrient densities. Applications/conclusions Subjects tended to report their intake in a socially desirable way, by eating or reporting less frequently foods considered unhealthful or fattening, like sweets and fried foods. Inclusion of social desirability score as a covariate in studies that rely on self-reports of food intake may be useful. A motivational training program, developed in such a way that subjects are comfortable reporting intake of foods considered socially undesirable, in combination with confrontation with earlier results of dietary assessment and use of portion size measurement aids, can be used to attenuate underreporting.