This article will examine issues in the use of epidemiologic and other types of data in dietary reference intake development, and offer an approach toward integration. In1994, the framework to determine revised dietary reference intakes (DRIs) was meant to incorporate the concept of risk reduction of chronic disease (Institute of Medicine, 1994). However, of the 19 estimated average requirements (EARs) that were established, none were able to be set on the basis of a chronic disease end point because of inadequacies in the available data. Adequate intake levels were partially determined by using disease end points for five nutrients: calcium (fracture rates), vitamin D (fracture rates), fluoride (dental carries), potassium (hypertension, kidney stones), and fiber (coronary heart disease). Using chronic disease end points was found to be difficult for setting individual nutrient DRIs for a number of reasons: 1. Chronic diseases have long latencies in their development, which results in almost inevitable inconsistencies in the data and makes them difficult and expensive to study. 2. Chronic diseases have numerous linkages to genetic and environmental factors. Risk of a chronic disease can never be brought to zero or reach 100% by any single nutritional intervention (Trumbo, 2008). The notion of attributable risk assumes that each factor is independent of every other factor, which is almost certainly not true. 3. Foods are complex entities. Attributing a chronic disease (e.g., cancer) linkage of a food (e.g., vegetables) to a specific nutrient (e.g., carotene) contained in that food is fraught with uncertainty. 4. Many studies on nutrients and chronic disease prevention were conducted as secondary prevention studies, rather than primary prevention studies. Extrapolation of the results of a secondary prevention study to primary prevention of a disease cannot be done and the results cannot, therefore, be translated into a public health recommendation. 5. Intervention studies on nutrients and chronic disease prevention have generally used single nutrients or a small number of combined nutrients. Such studies are expensive to conduct, and few have incorporated dose responses, which are a necessary component in DRI development. Because of the nature of the enumerated issues above, it can be concluded that prospective cohort studies can and do play a key and important role in initial hypothesis formation (as well as a confirmatory role) of chronic disease-nutrient linkages. However, such studies alone cannot yield definitive answers because of the uncertainties already mentioned that arise from long-term latency periods and from the multiple etiologic causative or modifying factors in chronic disease development. Results from prospective cohort studies (and, less importantly, observational epidemiologic studies) must be combined with evidence from experimental, animal, and small-scale, randomized, controlled clinical studies, which have used verified biomarkers of effect (i.e., biomarkers in the pathway of, or predictive of, chronic disease development) as end points (Institute of Medicine, 2008). Such biomarkers as effectors or predictors of chronic disease would have previously been demonstrated in prospective cohort and clinical intervention studies. Verified biomarkers of effect require an understanding of metabolic trafficking and can be enormously useful for decreasing the cost of studies, for providing early answers, and for studying dose-response relationships. In addition, such biomarkers should be safe to use in large-scale studies and in public health interventions. The interplay between the various types of studies will be an essential part of future DRI development. Thus, the results from several different types of studies (experimental, epidemiologic, and clinical) should be moving in the same direction with consistency, to be useful for determining DRIs for individual nutrients. In most cases, it would not be necessary to perform large-scale, randomized, and controlled nutrient-intervention trials. This gold standard type of study should be applied very selectively and only for questions of extreme public health relevance, and after a synthesis of data from all of the other various types of studies has been performed. Systematic reviews, to include meta-analysis, should always be used to complete such a synthesis (Lichtenstein et al., 2008), both in the process of designing costly intervention trials and during the DRI development process itself.
Read full abstract