'New data from old': epidemiology and record-linkage.

Record linkage in the context of epidemiology refers to the putting together of information (records) on the same person which may have been recorded on different occasions and became part of different data collections. The objective of this article is to provide an overview of the use of record linkage in epidemiology, particularly as it pertains to nutrition and food studies. A short description of epidemiological methodology is provided in order to clarify what epidemiologists expect to achieve by record linkage. A general description of the mechanics of record linkage explains how the methodology relates to the quality of the resulting data sets. The potential impact of the approach of record linkage on epidemiological research is discussed, particularly as it relates to nutritional and other food related studies. Ethical considerations related to record linkage are discussed, and in conclusion the future of record linkage is considered.