Evaluating epidemiologic evidence of the effects of food and nutrient exposures

The objective of this paper is to discuss some of the issues to be consider ed when evaluating and interpreting epidemiologic evidence from observation al studies that collect data on dietary intake. The assessment of such evid ence should include consideration of the study design, sample selection, an d the measurements of exposure and disease. The degree and type of error in nutrient data can lead to analytic problems and potentially be a source of bias either toward or away from the null value. Because methods of statist ical correction and adjustment for error, such its energy adjustment, canno t necessarily completely compensate for sources of bias in dietary data, ad ditional research should be conducted on sources of error in dietary data. Published research using reported dietary data should include a discussion of potential sources of error and their effect on the results. The most use ful studies are likely to be those designed to address a clearly defined pr ior hypothesis about a specific diet-disease relation. Because of the poten tial for bias and confounding, observational epidemiologic studies of diet and outcome cannot generally provide decisive evidence by themselves either far or against specific hypotheses. Although randomized clinical trials of the effects of specific nutrients or dietary modifications are not always feasible, they provide more definitive results and should generally be cons idered more valid than observational studies using self-reported dietary in take. Well-designed observational epidemiologic studies using self-reported dietary intake can provide valuable data to support or challenge hypothese s derived from clinical or laboratory data and to suggest further direction s for investigation.