Measuring intake in free-living human subjects: a question of bias

Experimental as well as epidemiological nutrition studies often aim at measuring food intake and its correlates in 'free-living' human subjects, and the identification of optimal ways to do this is a challenge in both types of research.The purpose of the present paper is, first, to give a brief overview of general problems encountered in measuring intake in free-living human subjects.Focus will then be placed on issues involving assessment of nutritional exposures in observational epidemiology, a field that is vulnerable to many sources of measurement error.Finally, obesity will be presented as an example of a condition that can introduce bias in population-based surveys, as well as in clinical studies describing the obese condition.