Valid data on energy and nutrient consumption are necessary in human nutrition research, mainly in epidemiologic studies concerning diet and health. Until the advent of doubly labeled water, it was not possible to validate dietary assessment methods using an independent procedure and this marker showed an important underreporting of energy intake. This paper aims to discuss the major aspects related to this phenomenon: prevalence, determinants, nutrient specificity and attenuation. Underreporting seems to be associated with obesity and with psychological and social factors, such as dietary restraint and social desirability. Error in the dietary assessments may lead to under- or over-estimation (less common). Underreporting seems to be specific to lipid- and carbohydrate-rich foods, introducing then a serious bias on conclusions based on food surveys. This paper also discusses how to deal with this error and recommends that energy intake estimates be validated against energy expenditure values, in order to detect underreporting.
food surveys; energy intake; validity; doubly labeled water
