Understanding Tolerable Upper Intake Levels Evaluation of Dietary Intake Data Using the Tolerable Upper Intake Levels 1–3 Alicia L. Carriquiry 4 and Gabriel Caman ˜ o-Garcia Department of Statistics, Iowa State University, Ames, IA 50011 ABSTRACT We discuss the problem of assessing nutrient intake relative to the tolerable upper intake levels (UL) for the nutrient proposed by the Institute of Medicine and focus on 2 important topics: the estimation of usual nutrient intake distributions and the extent to which intakes above the UL can be considered risky. With the information that is currently available for most nutrients, it is not possible to estimate the proportion of individuals in a group with intakes that place them at risk. This is because the shape of the dose-response curve needed to carry out a risk assessment is unknown for most nutrients. Thus, intakes above UL cannot be declared to be unsafe. Intakes below the UL, however, are likely to pose no risk to individuals in the group. Because determining the proportion of individuals with intakes below the UL requires estimation of an upper-tail percentile of the intake distribution, the use of 1-d intake data or otherwise unadjusted intake data are likely to lead to severely biased estimates. It is important to remove within-individual variance in intakes from daily intakes so that the tails of the usual intake distribution are accurately estimated. Underreporting of the amount of nutrients consumed will tend to shift the estimated usual nutrient intake distribution downwards. In this case, the true proportion of individuals with intakes below the UL is likely to be overestimated. J. Nutr. 136: 507S–513S, 2006. KEY WORDS:  upper level intakes  usual intake distributions  total nutrient intake  day-to-day variability in intake  under-reporting of calories The United States government has regularly collected dietary intake data for .60 y. Over time, the design of nation- wide food consumption surveys has changed significantly, as have the uses of the survey information and the methods for analyses of the data. Today, many of the consumers of the information provided by nationwide food consumption surveys are interested in public health issues and, in particular, in the links between diet and health. Many health outcomes are the result of exposure to foodborne pathogens (e.g., Salmonella enterica serovar Enteritidis), and such exposures are manifested as acute and deleterious effects on health. Although we are not minimizing the effects of those acute exposures on health, we focus here on the problem of assessing chronic exposure to nu- trients and other components in food. That is, we consider the problem of estimating the distribution of the usual intakes of nutrients and other food components and of evaluating that dis- tribution relative to the new dietary reference intakes (DRIs) 5 (1). In particular, we discuss approaches for evaluating usual intakes relative to the tolerable upper intake levels (ULs) and discuss the appropriate interpretation of the results of such analyses. We distinguish between daily intake and usual intake in the next section. At the national level, collection of dietary intake data has been the purview of two departments. The United States Department of Agriculture (USDA) has been responsible for implementing the Continuing Survey of Food Intakes by In- dividuals (CSFII), whereas the United States Department of Health and Human Services (DHHS) has been in charge of the National Health and Nutrition Examination Survey (NHANES). The CSFII surveys have focused exclusively on food consump- tion, whereas the NHANES surveys have also focused on health and have included a comprehensive questionnaire on health and well-being, as well as a complete medical examination of all survey respondents at a mobile examination center. Because of the significant differences in the designs of the surveys, the use of combinations of data from both surveys for statistical ana- lyses has been impractical. Late in the 1990s, CSFII was dis- continued, and today only NHANES is conducted. To address the effects of the discontinuation of CSFII on the quality of information about food intake that is available to policy makers, researchers, and other practitioners, the design of NHANES was revised to include several of the design features of CSFII. In particular, the design of the new NHANES surveys calls for the collection of replicated observations for a randomly chosen subset of all respondents, a feature that was not present in the earlier NHANES surveys. In addition, DHHS also adopted the USDA five-pass method to collect dietary intake data during NHANES. 1 Published in a supplement to The Journal of Nutrition. Presented as part of the Workshop on Understanding Tolerable Upper Intake Levels held in Washington, DC, April 23 and 24, 2003. This workshop and publication were supported by the Project Committee of the North American Branch of the Inter- national Life Sciences Institute. Guest editors for the supplement publication were Johanna Dwyer, Louise A. Berner, and Ian C. Munro. Guest Editor Disclosure: J. Dwyer, no relationships to disclose; L. A. Berner, no relationships to disclose; I. C. Munro, no relationships to disclose. 2 Author Disclosure: No relationships to disclose. 3 This work was partially funded by the US Department of Agriculture ERS/ FRED/FAB. 4 To whom correspondence should be addressed. E-mail: alicia@iastate.edu. 5 Abbreviations used: CSFII, Continuing Survey of Food Intakes by Individuals; DRI, dietary reference intake; NHANES; National Health and Nutrition Examination Survey; UL, tolerable upper intake level. 0022-3166/06 $8.00 Ó 2006 American Society for Nutrition. 507S by guest on January 21, 2016 jn.nutrition.org Downloaded from