Nutritional epidemiology data should be analyzed by nutritional epidemiologists

Abstract

The Energy Balance Working Group recently reported on the inaccuracy of self-report energy intake (EI) and issued a twin plea for researchers to stop using these data as outcomes and for newer, more accurate measures to be developed.1 I could not agree more with their conclusions; what concerns me is that this needs to be conveyed to the research community. The recognition that self-report EI data should not be used as outcomes or predictors in research has been known since the last century and repeatedly conveyed since.2, 3 All intake data are underreported in self-reports and the reasons go beyond just recall bias. However, as these underreports are more or less uniform across most food groups, when one adjusts for EI in a nutrient density model, the bias on the estimation of these food groups is minimized.4 The need to use these nutrient density models and the underlying rationale for their use is something that even the most basic nutritional epidemiology course teaches. Therefore, the conclusion not included in the Energy Balance Working Group’s paper is that individuals without nutritional epidemiology training should not be attempting nutritional epidemiology. That this occurs frequently has long been a source of concern to me; as a trained genetic epidemiologist I cannot recall a genetic paper or a genetic grant being put forward without the involvement of a scientist with training in genetics. However, frequently in study sections and in manuscript reviews I see studies involving nutrition data that are not treated with the same respect. It is time for the scientific community to stop this; Editors should send nutritional epidemiology manuscripts to reviewers with the necessary training, and nutritional epidemiologists should be well-represented on the relevant study sections. To summarize, I echo The Energy Balance Working Group’s urgency for new measures of EI to be developed and for self-report measures to be dropped as outcomes or predictors from the literature. However, the problems reported not only run to EI but also to other self-report nutritional intake variables and I urge the scientific community to come together and improve nutritional science using the expertise contained within.