the Federal Institute for Risk Assessment in Berlin, spoke on the challenges in addressing adjustment for data uncertainty.

Dr. Amy Subar, a research nutritionist at the National Cancer Institute (NCI), gave a presentation on the implications of estimating dietary intake for DRI development. Finally, Dr. Cutberto Garza, provost and dean of faculties at Boston College and former chair of the FNB, closed the session with some highlights of physiological, genomic, and environmental factors that are important to the DRI process. Discussions and comment periods were held throughout the session.

SELECTING ENDPOINTS: WHAT ARE THE ISSUES AND WHAT ARE THE OPTIONS FOR CRITERIA?

Presenter: Irwin Rosenberg


Endpoints play a pivotal role in the DRI process. They are the skeletal structure on which the Estimated Average Requirements (EARs) and tolerable upper intake levels (ULs) are draped. In essence, they are an expression of the targets or goals of the DRI development process. They should be related to quantifiable or measurable attributes that relate to the overall public health goal of the project. The key concerns from the perspective of selecting endpoints are “adequacy for what ends?” with respect to the EARs and “adverse effects as reflected by what?” for the ULs.

Experience in Selecting Endpoints

Since the 1941 National Research Council (NRC) report (1941), the selection of endpoints for nutrient reference values has evolved in response to changes in nutrition science. These advances sometimes revealed associations between an endpoint and diet and at other times identified possible endpoints through better understanding of metabolic and physiological states. Moreover, approaches for endpoint selection have been variable across the study committees responsible for the reference values. This is to be expected, given the differences in the biology and functions of essential nutrients.

Throughout the experience of developing reference values, limited data have often precluded the identification of the most appropriate endpoint for any given age/gender category. This situation in many cases results in the need to extrapolate knowledge about the endpoint used for one group that is better studied (e.g., adults) to a less well-studied group (e.g., children). This is one area of work that needs further exploration (see presentation by Dr. Atkinson in this chapter).

Importantly, limited data on dose–response relationships have always



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement