Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 99
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary 4 General Guidance for Users of DRIs: Session 31 Session 3 participants were asked to develop their presentations keeping in mind several questions: Is the general guidance for users of Dietary Reference Intakes (DRIs) appropriate as well as consistent with the purpose and goals of the DRIs? What more needs to be done? In addition, specific topics of interest included: How should Adequate Intakes (AIs) be used for planning and assessing, especially within the context of the total diet? Can clarification be provided on the differences between groups and individuals and between applications for small groups and those for large groups? What “practical guidance” and tools can be provided to assist practitioners? The moderator of this session was Mary Bush of Health Canada. The session opened with a presentation by Dr. Christine Taylor, the Study Director, on the wide-ranging issues surrounding general guidance for users. This was followed by a joint presentation by Dr. Suzanne Murphy of the University of Hawaii and Dr. Susan Barr of the University of British Columbia on the issues and options for enhanced guidance in terms of planning and assessing the total diet. The final presentation was given by Dr. Valerie Tarasuk of the University of Toronto, who discussed issues related to a framework for individual- and group-level applications. Discussions open to all audience members were held after each presen- 1 This chapter is an edited version of remarks presented by Drs. Taylor, Murphy, Barr, and Tarasuk at the workshop. Discussions are composites of input from various discussants, presenters, moderators, panelists, and audience members.
OCR for page 100
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary tation. Panel discussions exploring how guidance for users of DRIs could be enhanced closed the session. OVERVIEW: ISSUES RAISED ABOUT GENERAL GUIDANCE FOR USERS Presenter: Christine Taylor The issues surrounding general guidance for users are, overall, both wide ranging and overlapping. The purpose of this presentation is to offer some context for this session’s discussions. It is important at the outset to highlight at least two different interpretations of the term “uses” as they relate to DRIs because these interpretations have caused confusion. Some refer to the “uses of the DRIs” to mean their general purpose and intent, consistent with the conceptual underpinnings of DRIs. Others refer to “uses of the DRIs” when they are referring to the specific guidance for assessing and planning for individuals and groups. While these are not necessarily disparate, they each have a different focus. The purpose of DRIs is an important topic that properly belongs with those discussions relevant to conceptual underpinnings that occurred in Session 1. The discussions for this session on general guidance for users relate to the approaches outlined for applying the various DRI values to accomplish certain tasks. From the 1940s through the late 1980s, the Recommended Dietary Allowances (RDAs) and Recommended Nutrient Intakes (RNIs) were issued with little or no guidance for users. With the 1994 expansion to a more complex set of reference values, the need to provide general guidance was evident. This guidance was offered in two publications: Applications in Dietary Assessment and Applications in Dietary Planning (IOM, 2000a, 2003a). These documents were the work of the Subcommittee on Interpretation and Uses of Dietary Reference Intakes. The starting point for the guidance was the general categories of tasks commonly carried out using a nutrient reference value: assessing and planning dietary intakes for groups and individuals. These activities are often illustrated using a two-by-two table, shown as Figure 4-1. In essence, as outlined in the two guidance publications, the DRI value to be used will differ depending on which activity is being carried out. For instance, in the case of guidance for assessing groups, the focus is on using the Estimated Average Requirement (EAR) (not the RDA), cutpoint methods, and probability approaches. For guidance for planning for groups, the goal is identified as a low prevalence of inadequate intakes and consideration is given to definitions of acceptable prevalence of inadequate intakes. For assessing individuals, the guidance contains both qualitative and quan-
OCR for page 101
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary FIGURE 4-1 “Two-by-two table”: General applications of Dietary Reference Intakes. titative discussions. Guidance for planning for individuals emphasizes that the use of the RDA is preferable to the use of the EAR; also, AIs can be used in place of the RDA, but there is a greater level of uncertainty. This guidance derives from a statistical foundation and is based on understandings of distributions, normality, and probability. Whether the concept reflected in the two-by-two table works both conceptually and during the implementation process has been questioned. Some have asked whether there are special issues if a program fits within more than one box of the two-by-two table. Others have noted that guidance for planning seems to have been less well implemented than guidance for assessment. Broader questions for this session’s discussions include the following: Given the expansion of the DRI values and the subsequent development of guidance for their use, what have we learned about the needed guidance? What are its pros and cons? What, if anything, needs rethinking, further elaboration, or more work? How would changes in the conceptual framework for DRIs that have been suggested at this workshop impact DRI guidance in the future? A helpful set of background papers was developed for this workshop, targeted specifically to guidance for users. The U.S. government, Health Canada, the American Dietetics Association, and Dietitians of Canada all offered input on the topic. Numerous issues were highlighted in these documents, but they generally fell into two categories: Further methodological work, gaps, and conceptual evolutionary changes that need to be addressed or need some type of revamping A set of questions about ease of use, practicality, and the need for simpler guidance Specific questions raised included what further work needs to be pursued to address the emerging world of statistical methods and their application to DRI guidance for users (particularly individuals), and to what extent guidance is limited by lack of research on relevant methodologies.
OCR for page 102
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary In gathering input on issues related to guidance for users, two special challenges emerged. One is that the DRIs have to operate in the context of the total diet. They are created for individual nutrients, but when the user applies them, it is done in the context of the total diet. One snag is that for some nutrients they are using EARs, whereas for others they are using AIs. A second special challenge surfaced relative to the framework for individual- and group-level applications. It was asked whether it was worthwhile to distinguish between these and how their distinctions could be made clear. Regarding practitioners such as dietitians who work in a wide variety of settings in which DRIs are used, the main theme voiced was the difficulty in understanding DRIs (separate from applying the DRIs). Practitioners are interested in more clarity, more “practical” guidance, more guidance on use with individuals, and more tools, such as software. An interest in “simplicity” was expressed. As mentioned earlier, the issues for this topic area are quite diverse, and they provide a rich background for the session’s discussions. The overall questions are the following: Is the guidance heading in the right direction? Is it consistent with the purpose and goals of the DRIs? What more needs to be done? DISCUSSION Discussant: Johanna Dwyer The session moderator, Ms. Bush, introduced Dr. Dwyer and invited her to offer an opening remark. Discussant Opening Remarks Dr. Dwyer mentioned several practical issues concerning the use of the DRIs. The first focuses on the level of precision needed for assessment and planning for individuals. The precision needed may be more than for food-based guidance, but less than the prescriptive recommendations needed in medical nutrition therapy. A second concern is harmonizing the DRIs, food labels, and MyPyramid advice with respect to chronic degenerative disease risk, especially when the focus is on counseling individuals. A third issue is the application of DRIs to those with treated diseases (e.g., those taking high blood pressure medication) and who are apparently healthy. For most nutrients, the DRIs may be appropriate for these persons, with the exception of specific nutrients that may be affected by the disease in question. Dr. Dwyer also suggested that assessment and planning require estimates of total nutrient intakes from all sources, including fortificants and
OCR for page 103
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary dietary supplements. The ability to link intake with health consequences depends on usual total dietary intakes from all sources of nutrients, not just food. However, estimating the distribution of total intake for a population is challenging, and current estimates of intake suggesting low or high intakes in North America may be questionable. Are these findings due to uncertainties in estimating the requirements, difficulties in obtaining accurate dietary intake information, the unintended consequences of applying the values incorrectly, or some other factor? Finally, some of the problems experienced by users may be addressed if the DRI process incorporated systematic evidence-based reviews (SEBRs) on specific questions and a risk assessment approach. However, the lack of data may be challenging. Regarding the development of future DRIs, it will be vital to identify the most critical gaps in knowledge and methodologies and to obtain a broad consensus on how best to resolve these. However, this should not justify endless procrastination. General Discussion Dr. Taylor joined Dr. Dwyer on the dais. When asked for any initial thoughts, Dr. Taylor responded that there seemed to be an emerging theme that practitioners need assistance in understanding and applying DRIs, and that a starting point may be educational efforts. Dr. Dwyer added that Dietary Reference Intakes: The Essential Guide to Nutrient Requirements (IOM, 2006) was accessible to all. She also noted an online course produced by Dietitians of Canada as well as an array of helpful articles. She suggested that the Institute of Medicine (IOM), Canadian and U.S. governments, and professional societies should partner with various user groups, such as dietitians and family practitioners, to develop additional tools. In short, there is a need to train practitioners to interpret values rather than blindly using reference values. Meaning of “Uses” An audience member remarked that the term “uses” seemed to have various meanings. “Uses” in the context of the DRIs is a generic term that focuses on planning and assessment for the general population. However, the workshop participants seem to be referring to other “uses”; for example, are these DRIs to be used for fortification or for food programs? She queried whether, in the current context, Dr. Taylor was referring to uses generically or seeking information that is much more specific in terms of its application. Dr. Taylor responded that the focus of this session was general guidance for users in the context of planning and assessment. She added that the overall perspective is that specific applications that require
OCR for page 104
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary more in-depth or focused scientific considerations would be best addressed on a case-by-case basis, perhaps by special study committees if the IOM were to be involved. She commented, however, that the developers of this workshop had also noted the various meanings associated with the term “uses” in the context of DRIs. Another audience member commented that during her work on the United Nations University Harmonization Report, there had been a vigorous discussion of the uses of dietary standards and recommendations. It became clear that there were two ways of looking at “uses”: one way is the applications of the DRIs to specific programs, and the second is the methods for planning and assessing total diets for groups and individuals. A comment was made that application of the risk analysis paradigm to DRI development elevates the question of “uses.” She suggested it will frame the set of issues important to sponsors and other stakeholders. Chronic Disease Considerations A participant asked whether planning for groups should return to focusing solely on non-chronic disease endpoints or whether it should continue to embrace the chronic disease question. Examples of the challenges included the inclusion of macronutrient recommendations and tolerable upper intake level (UL) considerations in planning school lunches and establishing the basis for nutrition labeling. Dr. Dwyer responded that in planning for groups, it may be better to split adequacy and chronic disease into two reports. She added, however, that eventually they must come together in a single recommendation because a population is subject to both risks. Simplification A participant issued a word of caution about simplification, suggesting the issues are not simple. While the most common user may be private practitioners, their use of the DRIs may not result in proper application of values. It is more important to ensure that DRIs are used correctly. He suggested it may be better to rank the importance of the usage rather than the frequency. SPECIAL CHALLENGES: PLANNING AND ASSESSING THE TOTAL DIET—WHAT ARE THE ISSUES AND WHAT ARE THE OPTIONS FOR ENHANCED GUIDANCE? This presentation is divided along the lines of the paradigm, into assessing and planning intakes. Several crosscutting issues apply to both.
OCR for page 105
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary Assessing the Total Diet Presenter: Suzanne Murphy To use the DRIs to assess usual intake for an individual, one needs to know the probabilities of inadequacy and excess. For a group, one needs to know the prevalence of inadequacy within the group and the prevalence of intakes at risk of being excessive. Although simple in concept, this application is difficult in practice because of three challenges. Challenges and Possible Solutions The first challenge is associated with combining AIs and EARs. For most nutrients, the EAR and RDA are only about 20 percent apart. Generally, as shown in Figure 4-2, in which the left-hand curve is drawn more steeply than usual to reflect the assumption, the AI is assumed to lie beyond the RDA, but is potentially slightly below the RDA. It is a vaguely defined number and, in many ways, not very useful. For nutrients with an EAR, the probability of inadequacy for an individual as well as the prevalence of inadequacy for groups can be calculated. For nutrients with an AI, neither probability nor prevalence can be FIGURE 4-2 Relationship of the AI to the EAR and RDA. NOTE: EAR = Estimated Average Requirement; RDA = Recommended Dietary Allowance; AI = Adequate Intake; UL = tolerable upper intake level.
OCR for page 106
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary estimated; it can be stated only that intake at or above the AI should have a low probability or a low prevalence of inadequacy. The reference value for calcium is an AI. In practice, one option for estimating the probability of inadequacy is to assign approximate probabilities for a range of intake levels, based on an evaluation of the data used to set the AI. For example, the following probabilities shown in Table 4-1 might be used to assess persons over 50 years of age for calcium. Someone with an intake of calcium of less than 300 mg/day probably has nearly 100 percent probability of inadequacy. Someone whose intake is at the AI of 1,200 mg/day should have a very low probability of inadequacy. A second challenge relates to the assessment of individuals. Theoretically, the probability of adequacy or inadequacy for an individual can be calculated using the distribution curve and considering the standard deviation (SD) of the requirement. However, the methodology recommended in the IOM assessment report (IOM, 2000a) was to calculate a confidence of adequacy. That is, instead of comparing the intake of the individual with the EAR and its SD to get the probability, the SD should be increased to consider the number of days of intake and the within-person variance of intake. The resulting confidence of adequacy is typically less than the probability of adequacy. Using thiamin as an example, suppose that the average intake for an individual over 3 days is 1.3 mg. The thiamin RDA for an adult is 1.2 mg, which is below this person’s actual intake. If this was the person’s usual intake, the probability of adequacy would be about 98 percent. However, if the probability of adequacy is adjusted for the number of days and the within-person variability in thiamin intake, the confidence of adequacy is only 85 percent. This value is difficult to interpret because it does not necessarily mean a person should increase his or her intake. Rather, it probably means the number of days of intake data should be increased. By combining these two concepts, the confidence of adequacy often cannot be interpreted. A better method would be to calculate both the probability of adequacy and the confidence of adequacy. TABLE 4-1 Examples of Approximate Probabilities That May Be Used to Estimate Probability of Inadequacy Calcium Intake (mg/day) Probability of Inadequacy (percent) 0–300 Probability = 100 301–600 Probability = 75 601–900 Probability = 50 901–1,200 Probability = 25 Above 1,200 Probability = 0
OCR for page 107
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary Alternatively, a qualitative evaluation of the individual’s intake can be performed. If data on several days of intake show the usual intake is below the EAR, an increase in intake is probably needed. If the intake is between the EAR and the RDA, it probably needs to be improved. If it is at or above the RDA, it is probably adequate. A third challenge focuses on the need for better communication of the concepts associated with “uses” of the DRIs. For example, separating dietary collection methods from dietary evaluation methods would be helpful. The term “uses” should be called “evaluating intakes,” and would include guidance on calculating the probability of inadequacy for individuals and the prevalence of inadequacy for groups using the EAR as well as information for correctly using other DRI values (e.g., the Estimated Energy Requirement, AI, UL, Acceptable Macronutrient Distribution Range [AMDR]). Collecting intakes would cover the choice of methods used to collect the data, the number of days of data needed, and adjusting the data to reflect usual intake distributions. Moreover, the concept of uses should be separated from applications. Uses can be considered as the theoretical applications of the DRIs, or how the DRIs are used to evaluate and plan intakes. The term “applications” should be saved for the ways in which people apply the DRIs for specific problems (e.g., dietary counseling, food labels, Special Supplemental Nutrition Program for Women, Infants and Children [WIC] vouchers, dietary guidelines). Each type of problem might be better addressed by its own study committee and report. Dietitians also need a separate report on how to use the DRIs for dietary counseling; the American Dietetics Association and the Dietitians of Canada should consider collaborating on such a report. Implications General guidance on uses of the DRIs should focus only on evaluating and planning intakes. A separate group should consider dietary collection methods, and separate reports might be written for various applications of the DRIs (as was done for the WIC package revisions and the changes to the food labels). Collaborative groups, as well as IOM study committees, could address some of these topics. Planning the Total Diet Presenter: Susan Barr Given that the goals of dietary planning for individuals using DRIs are to achieve intakes with a low risk of prevalence of inadequacy or excess, the suggested approach is fairly straightforward. Unless there are special
OCR for page 108
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary considerations (e.g., vitamin C for a smoker or iron for a vegetarian marathon runner), the planning goals can be met by intakes that meet the RDA or AI, are below the UL, meet energy needs, and are within the AMDR. These goals have been operationalized using food guides. In contrast, proposed approaches to planning for groups are considerably more complex, depending on whether the groups are homogeneous or heterogeneous in terms of energy and nutrient requirements, whether there are vulnerable subgroups that can be identified and targeted for intervention, and whether the requirement distribution is skewed or symmetrical. The simplest case is planning for a homogeneous group with a symmetrical requirement distribution (e.g., vitamin C for nonsmoking young men). Planning in this case is accomplished by the EAR cutpoint method. The goal of a low prevalence of inadequacy is met if the proportion with usual intake below the EAR is minimal. If the baseline assessment has revealed a high prevalence of inadequacy, as estimated by the proportion below the EAR, planning would be done to shift the distribution upward to minimize the prevalence of inadequacy. Because variability in intakes is generally greater than variability in requirements, the mean or median intake of that distribution with the low prevalence of inadequacy would invariably be above the RDA. At this point, however, it is not known whether implementing this approach by changing the foods offered would actually shift the distribution or simply change its shape, so the approach remains largely theoretical. Key Considerations and Challenges The methods described in the IOM planning report (IOM, 2003a) focused on planning for single nutrients, whereas planners typically need to plan total diets. This process presents a number of challenges: One of the first challenges is “where to start” when planning for a total diet versus a single nutrient. This will vary depending on whether there is a preexisting set of menus and what the results of a baseline assessment show. If there are preexisting menus and the assessment showed a high prevalence of inadequacy for only one or two nutrients, it would be logical to start the process by planning to modify intakes of those specific nutrients. If, however, planning were to occur de novo (e.g., for a new retirement facility for 300 seniors of Chinese ancestry), more guidance would be helpful. Establishing the energy “budget” within which nutrients need to be incorporated would be a logical starting point, but other approaches could be considered. A second challenge relates to the need to establish priorities in situations where not all the planning goals can be met. This situation is
OCR for page 109
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary most likely to arise when planning for those with low energy needs, such as seniors in a retirement facility. However, it also arose in developing Canada’s Food Guide, where it was not feasible to reach targets for linoleic acid, potassium, vitamin D, and fiber, among others. Considerations in setting priorities might include whether to focus on nutrients that have an EAR/RDA rather than an AI, whether to focus on nutrients that have more direct consequences of inadequacy, and whether supplementation or fortification is feasible. Third, planning must sometimes occur for nutrients for which no specific guidance has been provided. This is the case for saturated fatty acids, trans fatty acids, and cholesterol, which lack ULs. Although the rationale for not having a UL is clearly explained in the macronutrient report, users are left with the need to pick their own benchmarks or use guidance from other groups. Fourth, “conflicting” DRIs must be addressed. Several macronutrients have an EAR/RDA (or AI) and an AMDR. For fatty acids, the AIs for the N3 and N6 fatty acids were based on median intakes of the U.S. population, while the AMDR was set with the AI as the lower boundary. For nutrients with an AI, one normally plans for a low prevalence of inadequacy by planning for median intakes to equal the AI. However, if this were to be done for N3 or N6 fatty acids, 50 percent of the group would be below the AMDR. It is difficult for users to know how to proceed in such a case. Fifth, it should be recognized that food guides traditionally have been used for individuals (by being designed to meet the RDAs), but could potentially be used for groups as well if they were developed with that objective. Canada’s Food Guide was designed to lead to a low prevalence of diets with nutrient levels below the EAR, to meet most AIs, etc. This means it could be used as a starting point to plan diets for groups as well as for individuals (although, as indicated above, it was not possible to meet all planning goals). Although the approach used to develop Canada’s Food Guide was based in part on the planning paradigm outlined in the planning report, it did not use a baseline usual intake distribution as the starting point. Instead, it was based on the assumption that intakes would comply with the food guide. If menus were planned using the food guide and a high prevalence of inadequacy was found after intakes were assessed, planners would still need to try to shift the distribution upward. Implications Users have faced many challenges in adopting and understanding the new paradigm for planning for total diets. One is knowledge gaps, which
OCR for page 112
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary General Discussion DRIs in Clinical Practice In the opening discussion among the presenters and discussants, one presenter emphasized the desirability of convening a working group of dietitians and others to discuss DRI application in clinical practice. She pointed out that the background papers by the Dietitians of Canada and the American Dietetic Association indicated that dietitians want to use the DRIs, but are uncertain how to apply them. She suggested that dietitians do not just use food guides; some situations require nutrient reference values, such as when working with a population that has specific nutrient needs, so they must be able to adjust the nutrient intake recommendations. A second presenter agreed, but a discussant disagreed, saying that when she was a clinical dietitian counseling patients with a need for dietary modification (e.g., someone with renal disease), she did not tell her patients how much of the various nutrients to consume, but instead suggested the amounts and types of foods they should consume. Interface Between Individuals and Groups When the discussion was opened to the audience, a participant commented that the developers of the DRI guidance for users failed to recognize the interface between individuals and groups. Existing guidance indicates that one plans for the individual using the RDA and assesses plans for a group using the EAR. This means, on one hand, that each individual has to be seen at low risk and, on the other hand, that the population at low risk is a goal. These are very different concepts. That clash between individuals and groups needs to be resolved. In response, one of the speakers indicated that if an individual is at the RDA, by definition he or she has a 2–3 percent risk of not meeting the individual requirement. For example, in a population of 1,000 individuals, all of whom had an intake equal to the RDA, there would be a 2–3 percent prevalence of inadequacy. No one in the population would be below the EAR, so the cut point method would not be applicable in that situation because there would be no variability in intake. The original questioner disagreed with these calculations,2 but it was noted that the issue would be further covered in the talk by Dr. Tarasuk on individual- and group-level applications. 2 The questioner later explained his disagreement as follows: Given the impossible situation of a group of 1,000 individuals each consuming an intake intended to cover all but 2.5 percent (risk 0.025), the expected prevalence of inadequacy would approximate 0, and thus the risk to a randomly selected individual would be approximately 0, not 0.025.
OCR for page 113
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary Food Supply Changes and Intake Distribution An audience member noted that one of the speakers said there were no data concerning the distribution of intake when the food supply changes. She suggested it would have been possible to use intake data published before and after iron and folic acid fortification—showing both a shift and a skew in distribution—as well as data from studies that evaluated intakes before and after the introduction of food stamps. In response, the speaker indicated that most of the studies referred to by the commenter were published after the work of the presenters. In addition, the published papers generally did not show percentiles, which are needed to determine how the shape of the distribution changed. Skewed Distributions An audience member remarked on the conceptual advances and new tools offered by the Subcommittee on Interpretation and Uses of Dietary Reference Intakes. However, he highlighted a problem in the approach for predicting the shape of the intake distribution given an intervention that shifts the distribution—the assumption that there will be a shift and no skew when the intent of the intervention is to skew the distribution results in a target value well above the RDA for many nutrients. He suggested this will cause problems in planning diets for group feeding situations, and work to address this is needed. Other Comments A participant emphasized the challenges in developing a food guide based on the DRIs while maintaining recommended caloric intakes. Another commented that the U.S. MyPyramid and Canada’s Food Guide are both useful, but there is also a place for the use of food exchange lists and other tools. Another asked who is responsible for developing software that would allow practitioners to make better use of DRIs. A discussant responded that more consensus on the methodology is needed before more software is developed. A final commenter focused on an earlier suggestion to develop an AI-based approach to examine the probability of inadequacy. She noted that a careful read of the DRI reports could reveal the point at which one sees an indicator of adequacy in a group (albeit not necessarily an indicator used to create an EAR) and through that process find a way to establish cutoff values.
OCR for page 114
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary SPECIAL CHALLENGES: WHAT ARE THE ISSUES RELATED TO A FRAMEWORK FOR INDIVIDUAL-LEVEL AND GROUP-LEVEL APPLICATIONS? Presenter: Valerie Tarasuk The two-by-two table that divides DRI applications into assessment and planning for individuals and groups (see Figure 4-1) is the organizing framework for the discussions that appear in the DRI applications documents (IOM, 2000a, 2003a). Four core concepts underlie the framework and are articulated in these documents: Requirements are recognized as distributions, whether or not they can be mapped. If the distribution of requirements is known, there is a known probability of inadequacy or adequacy associated with each intake level in relation to that distribution. The relevant point of comparison for nutrient requirements is usual dietary intakes. Planning—either for the usual intake with a low risk of inadequacy for an individual or for a distribution of usual intakes with an acceptably low prevalence of inadequacy for groups—is fundamentally an extension of assessment. Challenges Associated with Implementing the Core Concepts Implementation of these concepts has presented challenges. One challenge is that the framework functions only where probability theory can be applied. Therefore, we cannot apply AIs, AMDRs, or ULs using the framework (as we can for EARs and RDAs). These values do not allow us to differentiate between individuals and groups or between assessment and planning. Their application is arbitrary. Moreover, there is no application guidance for report recommendations such as “saturated and trans fat intakes should be as low as possible.” A second challenge is that the strength of the assessment paradigm is at the population level. Assessment of the adequacy of an individual’s usual intake is severely limited by our inability to assess that intake with any degree of precision. A related issue is the questionable applicability of the framework to small groups. Assessment and planning for small groups hinge on our ability to estimate distributions of usual intake; factors to be considered include sample size, number of replicate observations, and the representativeness of the subsample from which they are drawn. The reliability of the estimated prevalence of inadequacy is a function of the reliability of the estimated distribution of usual intakes. Part of the challenge is the difficulty in deriving
OCR for page 115
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary stable estimates of usual intake when there are only a few observations and a small sample size. For example, our research found a prevalence estimate of 26 percent for folate in a small group of women (n ≈ 300, with replicate observations for 80), but the confidence intervals around that prevalence estimate ranged from 0 to 85 percent. The other issue with small-group applications is the nature of those groups. They may include nursing home residents, the elderly, or other subgroups of the population that fall outside the core assumptions of DRI development. The appropriateness of applying the DRIs to those groups can be questioned. Assessment and planning activities for small groups would benefit from clearer guidance on the conditions under which group-level assessment and planning methods can be applied with reasonable confidence; alternative approaches to assess and plan for nutrient adequacy among small groups; and different reference standards for use in clinical settings. The third challenge relates to the interface between populations and individuals. The two-by-two table (Figure 4-1) draws a firm line between individuals and populations. However, attempts to apply the DRIs, particularly for regulatory affairs and public policy purposes, show that this distinction is a gross oversimplification. The goal of planning for individuals is “to ensure that the diet as eaten has an acceptably low probability of nutrient inadequacy while simultaneously minimizing the risk of nutrient excess” (IOM, 2003a). We assume the RDA is a point on a distribution of requirements that lies at the upper tail, which will encompass the needs of 97 or 98 percent of the population. If someone is consuming a dietary intake at the level of the RDA, he or she will have a low probability of inadequacy. Using the RDA as the goal of planning for an individual assumes that one will achieve a usual intake approximately equal to the RDA (i.e., there is no between-person variation in intakes), but this is not realistic for population-level applications. The midpoint of the distribution of usual intakes that achieves that low prevalence of inadequacy will be higher than the RDA. Thus, the goal in planning for groups is to achieve a distribution of usual intakes that has a low prevalence of nutrient inadequacy and a low proportion of intakes above the UL (IOM, 2003a). This takes into account between-person variation in usual intakes. The implication of these differences is that, as currently defined (IOM, 2003a), population- and individual-level approaches to planning are not interchangeable, but yield different nutrient targets with potentially different outcomes. An Illustration of the Challenge Examination of the U.S. and Canadian processes to update their food guidance by incorporating the DRIs provides an illustration of the current controversy surrounding the interface of individual- and population-
OCR for page 116
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary level applications. The U.S. group, when faced with the task of reframing MyPyramid, decided that its purpose was to plan for individuals using the RDA. To develop food intake patterns, it calculated the nutrient profile for each food group or subgroup from the weighted average of the nutrient content of foods in the group, based on National Health and Nutrition Examination Survey (NHANES) food consumption data, then used an interactive process of modeling to determine the amounts from the food group composites required to meet nutrient and energy goals. The purpose of Canada’s Food Guide is “to assist the people of Canada in making food choices that promote health and reduce the risk of nutrition-related chronic diseases” (Katamay et al., 2007). The Canadian goal in the updating process was a low prevalence of diets with nutrient content below the EARs and a median nutrient content approximately equal to the AIs. They first modeled food composites to obtain a food intake pattern for each age and gender group with satisfactory average nutrient levels, similar to the U.S. group. However, once they obtained some food composites with reasonable nutrient values, they assessed the nutrient distributions from 500 simulated diets designed to comply with the test pattern for each age and gender group (i.e., recognizing variation in food selection). They then used an iterative process to identify the dietary pattern with a low prevalence of inadequacy and a median approximately equal to the AI. Although these were two totally different approaches to the implementation of the DRIs for food guidance purposes, the results appear to have been similar (Table 4-2). Perhaps this can be explained by the facts that nutrients with AIs were treated similarly in both modeling exercises and that translation of multiple nutrient targets into numbers of servings from different food groups blurred the distinctions between the methods. Following either guide results in intakes well in excess of the RDAs for most nutrients. Although these results might suggest that the frameworks for individuals and populations are interchangeable, they will result in different nutrient targets in many applications. TABLE 4-2 Food Guide Recommendations for a Sedentary Adult Woman ≤ 50 Years of Age in Canada and the United States Number of Servings Food Group United States Canada Grain products 6 6 Fruit and vegetables 7 6 Milk products 3 2 Meat and alternatives 2 2 SOURCE: Murphy and Barr (2007).
OCR for page 117
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary Conclusion In conclusion, the core concepts that underpin the framework are sound. However, the two-by-two table (Figure 4-1) is an oversimplification of what is needed to apply those core concepts appropriately, and its interface between individuals and populations must be evolved further. We need to separate the needs of the dietetic profession and others who engage in nutritional counseling from applications of the DRIs for public policy purposes (e.g., food guides, fortification, labeling). Furthermore, better supporting documents for nutrition professionals must be developed. In the end, all applications of the DRIs have the individual as their end user—because people eat food. The question of whether the application relates to individuals or populations is not useful. We must, as we move forward, think about the applications of the concepts, not the numbers. That is the next wave of activity and of thinking in terms of applications of the DRIs. DISCUSSION Discussant: Gerard Dallal The session moderator, Ms. Bush, introduced the discussant and invited him to offer an opening remark. Discussant Opening Remarks Dr. Dallal noted that debates about populations and individuals are not useful. From a statistician’s point of view, the focus should be on the question to be answered, how the question is approached, the techniques to use, the assumptions to be made, and whether the assumptions and methods are reasonable for the problem at hand. Dr. Dallal also suggested that distinguishing between individuals and populations was primarily an argument about terminology that in the end would not be useful and only distracted from the underlying problem. General Discussion Two-by-Two Table A participant commented that the two-by-two table (Figure 4-1), while useful conceptually, creates problems for users. Policy issues often do not fit neatly into just one box, and some are appropriately placed in all four boxes. The relevant task is to make useful decisions with clear documenta-
OCR for page 118
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary tion concerning the use of EARs, RDAs, and ULs. In this case, the useful “tools” are efforts to understand the science and the public health ramifications while also incorporating the conceptual and statistical underpinnings relevant to the needed application. Furthermore, she remarked that the decisions are frequently case specific. Thus, by making the primary tool (i.e., the two-by-two table) too simplistic, one loses the ability to be flexible in a rational way. Another participant noted the use of the two-by-two table as a basic organizing structure and suggested it should not be abandoned. Moreover, she suggested that the use of different DRI values for different applications aligns with the concept of intent-to-treat versus treated-as-intended. When one is working with an individual, as many dietitians do, the intent-to-treat and the treated-as-intended are presumably the same. When one is planning for a large group, such as a food stamp program or school lunch program, the intent-to-treat is not the same as treated-as-intended. This difference explains why one would use a different DRI under different circumstances. Feedback Loop Dr. Tarasuk remarked that there are special challenges when a group of experts is asked to address questions raised by government agencies, particularly in terms of clarity about the needs for and ramifications of the outcomes. She expressed interest in ensuring that such efforts include a “feedback loop” to avoid addressing the wrong question or finding that the “real” question had not been specified. She suggested the value of dialogue when these efforts are undertaken to ensure that the potential of the evolving science impacts public policy application. Development of Dietary Guidance An audience member asked Dr. Tarasuk about her comparison of the dietary guidance in Canada and the United States. Because the countries have common dietary patterns, how similar or dissimilar were the foods and the end result? Dr. Tarasuk responded that there are several explanations for why the outcomes were similar, despite the use of different approaches. One is that modeling food patterns to achieve the AIs for selected nutrients had a strong impact on the final guidance in both countries. In addition, multiple nutrient goals were translated into food servings for five population groups, so there was a blunting effect for some of the distinctions. Moreover, for both the Canadian and U.S. guides, the nutrient intake achieved by following those guidelines would be far above the RDA for most nutrients. She remarked that this is an interesting example of how a question could be addressed quite differently by two government
OCR for page 119
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary bodies—depending on whether the perspective was at a public health level or an individual level—and yet the outcome is not very different despite different starting points and assumptions. Application of the RDAs An audience member asked about the importance of sensitivity analysis during planning activities, given the uncertainty involved, especially with the coefficient of variation. He also asked how the analysis could be prioritized given the cost and complexity of implementation. Dr. Tarasuk suggested that the use of RDAs for public policy purposes should be reexamined because, in her view, many current applications of the RDAs may not be appropriate; they seem to be carryovers from an older era. Moreover, their use has been driven largely by the question of who the end user is—and the answer is always the individual. However, Dr. Tarasuk noted, this does not mean there is not work that could be carried out for the purpose of counseling individuals, including exploration of the use of ranges or percentages. PANEL DISCUSSION: IN WHAT WAYS COULD THE GUIDANCE FOR USERS OF DRIs BE ENHANCED? Panel Members: Danielle Brulé, Mary Frances Picciano, William Rand, and Linda Van Horn The session moderator, Ms. Bush, introduced the panel members and asked each panelist to offer an opening remark. Panelist Opening Remarks Dr. Brulé emphasized the importance of identifying the target audience or users of the DRIs. User needs differ, and the challenge faced by risk managers is to communicate policies and programs to health professionals and consumers who hopefully understand and can appropriately apply DRIs. She further commented that the achievements of the existing framework are impressive and have set the stage for efforts to improve the process, notably documentation and transparency. Dr. Brulé further pointed to the ability of the risk assessment approach to articulate the roles and responsibilities of risk assessors, risk managers, and others. She endorsed a continuing dialogue between risk managers and risk assessors. Finally, she underscored her perspective that ensuring reality checks—for instance, using examples of policies or programs to validate newly revised DRI values—is needed and would be valuable in accomplishing risk communication.
OCR for page 120
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary Dr. Picciano expressed a desire to keep the RDAs in the report as goals because social, cultural, or political forces could result in the EARs being misused as goals. There should be more documentation of the assumptions made and their influence on the values derived, particularly for pregnant and lactating women. She highlighted an interest in including discussions in DRI reports that focus on how the EARs and RDAs relate to intake data within subgroups, notably those extrapolated with and without consideration of supplement intake. Finally, she noted that pediatric, geriatric, and reproductive nutritionists should be included in future DRI study committees. Dr. Rand commented that the interest in “keeping it simple” should be balanced by the value in preserving the inherent complexity. He said it is important to explore the complexities and to avoid plans that ignore these realities or fail to recognize key interactions. Clearly, nutrients have multiple effects, and chronic diseases have multiple risks; such complexities will grow over time. He suggested that when simplification is desired, it would be best to split the tasks more formally so that each has its own focus, including the distinction between science and policy. Simplification could also involve deriving the best estimates of the risk curves together with descriptions of how they have been derived and the assumptions used. In turn, users should provide information on what risk curves are required and, based on their own needs, determine how best to use this information. Finally, Dr. Rand expressed a hope that the future process will include explicit guidelines for risk assessment and SEBR as well as more attention to the “mathematics,” such as statistics, modeling, probability, and simulation. In addition, an explicit validation step is desirable. Dr. Van Horn remarked that ideally the DRIs would be well understood by those who intend to use them. Regarding the role of the registered dietitian and other health-care professionals, she highlighted training and the development of practical tools for implementing and using the DRIs. She also suggested that there were special challenges for practitioners who work within a constant overlay of obesity concerns and must in turn interpret and use the DRIs in the absence of specific interdisciplinary involvement. In terms of reaching out to practitioners and consumers, she stated that, in her own experience with the National Cholesterol Education Program, the “know your number” public education campaign was extremely successful. Perhaps a similar approach could be used for teaching about DRIs. Finally, Dr. Van Horn expressed concern that an exclusive focus on DRIs could detract from efforts to help children select and consume healthful foods and food patterns. She cautioned against the potential for marketing and encouraging the consumption of less nutritionally desirable foods to which nutrients had been added.
OCR for page 121
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary General Discussion Cross-Panel Discussion Two-by-two table The usefulness of maintaining the two-by-two framework was raised as an issue during the cross-panel discussion. It was pointed out that so many factors are embedded within discussions related to groups versus individuals and planning versus assessment that it is difficult to reflect or “constrain” them within a table. Yet as a broad means of differentiating across the general applications, it is a useful place to begin. Another panel member found it impossible to place her government agency’s policies and programs into the four boxes. It was noted that there is a need to have the choice to use the two-by-two table or not use it, as appropriate. Another participant commented that assessment and planning are different ends of the same continuum: If you plan a diet, one of the first things you would do is assess whether it was any good. Another pointed out that the definition of what constitutes a group is also an issue: What are two individuals? Target population Panel members discussed whether the guidance to users should be enhanced by considering not just healthy populations, but also other populations, such as the institutionalized, the elderly, and “nonhealthy” populations. It was noted that there were sections in most DRI reports that focused on special populations and at times also addressed specific disease conditions. However, a further comment was that “healthy” is difficult to define. Moreover, as more than 50 percent of the North American population is overweight or obese, an important question is how obesity fits into these considerations. Open Discussion Guidance for use with individuals With respect to the two-by-two table (Figure 4-1), one participant advocated that individuals remain a component of the approach, and that a decision to omit them should occur only after considerable thought and discussion. She emphasized that it was inappropriate to guide individuals based on a population mean. She further suggested that the needed value does not have to be labeled as an RDA, but it should be a requirement plus a safety factor. She added that a reexamination of safety factors to be used may be in order. A panel member posited that the safety factor need not be two SDs as currently used to develop the RDA, but could be one SD or a value anywhere between the EAR and the UL. Another panel member suggested that once a risk curve is established the determination can be made, and remarked that the key point is that the DRI process should specify risk curves.
OCR for page 122
The Development of DRIs 1994–2004: Lessons Learned and New Challenges - Workshop Summary One person noted there had not been much discussion of “family,” as opposed to individuals, small groups, and populations, and that special advice might be needed in this context. In response, it was pointed out that families could be considered small groups of individuals. An audience member indicated her concurrence with the interest in developing more specific guidance for users, but also agreed such guidance should be application specific, not profession specific. She also asked how much time she should spend explaining the current framework to individuals given that changes are likely in the future. A panelist responded that “you have to go with what you have” because we do not know what will happen in the future. Target population As a follow-up to the discussion about the definition of a healthy individual within the context of DRI development, the comment was offered that the focus on a healthy person may no longer be needed or useful because a large percentage of the population is overweight or obese, hypertensive, or hypercholesterolemic. One participant noted that the focus was on an “apparently healthy” person to reduce confounders, such as those who need blood pressure medication. Another audience member said using healthy persons as study subjects simplifies research protocols. Consumer messages Dr. Van Horn’s analogy to the National Cholesterol Education Program prompted a question on whether a “know your number” target could focus on the concept of a distribution shift. It would be a simple message, but based on distributions and populations; marker nutrients could be used to point persons in the right direction. Another attendee cautioned that the “know your number” concept may work when you have one number, but not when you have multiple numbers. DRIs do not represent all nutrient needs, and pulling out just a few of them has the potential for harm if it becomes a public education effort. People will not turn to food sources of nutrients, but to supplements. Concern was expressed that we need to be careful to target foods rather than nutrients. Dr. Van Horn agreed completely and indicated that her point had been that a well-designed public education campaign worked for raising cholesterol awareness. The lessons learned may be valuable for DRI education and outreach, including awareness of how DRIs can be met by eating the recommended foods.