Issues in the long-term evaluation of diet in longitudinal studies

Longitudinal studies are very useful for studying diet/disease relationships. The fundamental components of a longitudinal study are that: 1) data are collected for two or more distinct time periods; 2) the subjects are the same or comparable from one time period to the next; and 3) data are compared between or among time periods in the analysis. A longitudinal study is often assumed to be synonymous with a cohort study, but there are at least four possible definitions for a longitudinal study. While focusing on cohort studies, the paper describes the nature of longitudinal studies, including a discussion of how the different definitions differ from a cohort study and a set of important assumptions necessary to cohort studies. It also highlights some of the major issues associated with such studies, including the selection of a dietary survey methodology; data collection issues in multicultural, multilingual societies; the importance of nutrient databases; measurement error and misclassification in nutrient intake and energy adjustment.