Evaluation of the usefulness of the cross-check dietary history method in longitudinal studies.

The data quality control of the first 4 years of an ongoing study (10 years duration) in a few hundred women and an additional control group is presented with regard to the measurements of the usual nutrient intake, covering a 1 year period, by means of the cross-check dietary history method. The results show that with regard to the disturbing time factors the cross-check dietary history method is (a) reasonably reproducible for most nutrients except vitamin A and vitamin C, (b) does not show test effects when interviews were taken only once a year, (c) does show observer effects. The method has a reasonable validity. It is concluded that, on an individual basis, the nutrient variables are not appropriate for a detailed calculation of changes per unit time but are useful as a general measure for the development of the individual (e.g. the average change per time unit over a longer period). For this reason the method can be reliably used in longitudinal studies to evaluate the role of diet in relation to chronic disease.