The state of nutritional epidemiology: why we are still unsure of what we should eat?

Most people want to know how they can improve their health by implementation of a proper diet. Therefore nutritional epidemiology studies, which correlate the intake of specific nutrients, food items, or dietary patterns with health outcomes, receive substantial interest in the media. However, the results of many nutritional epidemiology studies have not been replicated in subsequent studies. This paper reviews the primary methodological problems in nutritional epidemiology that result in lack of replicability and inconsistency across studies. The problematic methodological issues include substantial measurement error, confounding, the variable effects of food items, variable reference groups, interactions, and multiple testing.