Assessment of food intakes: are we measuring what people eat?

Accurate assessments of dietary intake are a pre-requisite for assessing the relationships between diet and health. However, the quantification of errors in dietary data has remained largely undetected because of the absence of techniques to verify dietary survey methodology. Recent studies using doubly-labelled water estimates of total energy expenditure to validate self-reported energy intakes have demonstrated that the majority of these are systematically biased towards under-estimation of usual requirements. Bias is unlikely to be consistent in population studies. Subjects who report the highest energy intake tend to estimate their requirements more accurately than those who report low intakes. Mechanisms for cross-checking energy intake data against estimated energy requirements have been developed and provide limited guidance for checking the physiological plausibility of reported intakes, with special emphasis on under-reporting. In the absence of independent validation the evaluation of dietary survey data should be approached with caution.