Invited commentary: Dietary misreporting as a potential source of bias in diet-disease associations: future directions in nutritional epidemiology research

Error and bias in self-reported intakes make estimating relationships among dietary factors, obesity, and related health outcomes a complex challenge in observational studies. In the absence of measures that can be applied in calibration adjustments of dietary data, simple methods to identify persons who misreport their intakes have been used to assess the impact of screening out reports characterized by energy intakes that are implausible when compared with estimated energy needs. Sensitivity analyses in cross-sectional studies have shown these methods to yield more plausible associations between diet and obesity, but few longitudinal studies have evaluated this approach. In this issue of the Journal, findings reported by Rhee et al. (Am J Epidemiol. 2015;181(4):237) underscore the need for caution in drawing conclusions on how self-reported diet may influence such outcomes based on cross-sectional associations but suggest that this approach might have little impact on the more credible associations derived from prospective analyses. However, other prospective studies have found that diet-disease relationships emerge or are substantially strengthened with the use of calibration adjustments using recovery biomarkers. To better understand the influence of diet on obesity-related health outcomes, efforts to reduce dietary measurement error through improved collection, evaluation, and analysis of consumption data are still urgently needed.