It s been an issue for as long as scientists sought to quantify what people do and don t eat. How does one get the data? According to Dr. Edward Archer of the University of Alabama, Birmingham and colleagues, the current data have been based on inaccurate methodologies that make most of these data inaccurate and useless. Published in the Mayo Clinic Proceedings, their review explicitly states that food intake data that rely on memory (memory-based methods or M-BMs) are at best inaccurate and at worst useless.
Most diet intake data are indeed based on M-BMs. Whether an interview or survey, people are asked what, when and how much they consumed of particular items, and the nutrients and calories they imbibed are calculated from that information. Then, in many cases, such intake data are used to calculate associations between consumption of particular foods, ingredients, or nutrients and a host of health-related outcomes. For example, the long-running Nurses Health Study uses survey-based information for this type of analysis.
Dr. Archer and colleagues have four main criticisms of the current reliance on M-BMs. First, they say that human memory is notoriously unreliable:
Research in several fields show that human memory is not a literal, accurate, or precise reproduction of past events.
Second, these authors state that participants in M-BMs are subject to conditions that are known to induce false recall. Third, there is no way to objectively ascertain the truth of M-BM derived data, since they are totally subjective. Finally, they attest that the lack of objective measures of physical activity makes any conclusions regarding the relationship between diet and health suspect.
ACSH s Dr. Ruth Kava agrees The validity of M-BM based data has long been an acknowledged problem in nutrition research. Various types of M-BM have been proposed to get around this perceived lack of accuracy for example 1 and 3 day dietary recalls and food frequency questionnaires. But they are still subject to the criticisms of Archer, et al. There are more objective ways to obtain dietary and activity data, such as having people live on metabolic wards where foods consumption is accurately measured, or use of the doubly-labeled water technique to ascertain energy expenditure. But these are expensive research techniques, not really suitable for wide use in the general population. Until better methods are developed, we must be more widely aware of this basic problem with nutrition research especially epidemiologic research, and demand more strongly that any results of such research be independently replicated before they can be used to set or change nutrition policy.