Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
14 Uses of Dietary Reference Intakes OVERVIEW The Dietary Reference Intakes (DRIs) may be used for many pur- poses, most of which fall into two broad categories: assessing existing nutrient intakes and planning for future nutrient intakes. Each cat- egory may be further subdivided into uses for individual diets and uses for diets of groups (Figure 14-1). For example, the Recommended Dietary Allowance (RDA), Esti- mated Average Requirement (EAR), and Tolerable Upper Intake Level (UL) may be used as one aspect in the assessment of the diet of an individual. The RDA and Adequate Intake (AI) may be used as a basis for planning an improved diet for the same individual. Likewise, the EARs and ULs may be used to assess the nutrient intakes of a group of individuals, such as those participating in a dietary survey regularly conducted as part of the National Nutrition Monitoring System. The EAR and UL can also be used to plan nutritionally adequate diets for groups of people receiving meals in nursing homes, schools, and other group-feeding settings. In the past, RDAs in the United States and Recommended Nutrient Intakes (RNIs) in Canada were the primary values available to health professionals for assessing and planning the diets of individuals and groups, and for making judgments about inadequate and excessive intake. However, the RDAs and RNIs alone were not ideally suited for many of these purposes (IOM, 1994). The DRIs provide a more complete set of reference values. The transition from using RDAs 554
USES OF DIETARY REFERENCE INTAKES 555 Nutrient Nutrient Requirements Intakesa Planning Assessing Diets Diets Group Individual Group Individual FIGURE 14-1 Conceptual frameworkâuses of Dietary Reference Intakes. a Food plus supplements. and RNIs alone to appropriately using all of the DRIs will require time and effort by health professionals and others. Appropriate uses of each of the new DRIs are described briefly in this chapter and in more detail in two reports on the applications of the DRIs in assessment (IOM, 2000) and planning. Also included in this chapter are specific applications to the nutrients discussed in this report. Details on how the DRIs are set with reference to specific life stage and gender groups and the primary criterion that defines adequacy for each of these nutrients are given in Chapters 4 through 13. ASSESSING NUTRIENT INTAKES OF INDIVIDUALS Using the Recommended Dietary Allowance and the Estimated Average Requirement for Individuals The Dietary Reference Intakes (DRIs) were not designed to be used alone in assessing the adequacy of the diet of a specific individ- ual because there is variability in the requirement estimate. The
556 DIETARY REFERENCE INTAKES Estimated Average Requirement (EAR) estimates the median of a distribution of requirements for a life stage and gender group, but it is not possible to know where an individual falls on this distribu- tion without further physiological or biochemical measures. Thus, from dietary data alone, it is only possible to estimate the likelihood of nutrient adequacy or inadequacy. Furthermore, the rarity of hav- ing precise and representative data on the usual intake of an indi- vidual adds further uncertainty to the evaluation of an individualâs dietary adequacy. Dietary assessment methods have several inherent inaccuracies. One is that individuals underreport their intakes (Mertz et al., 1991; Schoeller, 1999), and it appears that obese individuals often do so to a greater extent than do normal-weight individuals (Heitmann and Lissner, 1995). Currently, a method for adjusting intakes based on underreporting by individuals is not available and much work is needed to develop an acceptable method. Furthermore, large day-to-day variations in intake, which occur for almost all individuals, mean that it often takes a prohibitively large number of days of intake measurement to approximate usual intake (Basiotis et al., 1987). As a result, substantial caution must be used when interpreting nutrient assessments based on self-reported dietary data covering relatively few days of intake. Data on nutrient intakes should almost always be interpreted in combination with typical food usage patterns. An approach for using data from dietary records or recalls to estimate the likelihood that an individualâs nutrient intake is ade- quate is presented in the report Dietary Reference Intakes: Applications in Dietary Assessment (IOM, 2000). This approach, which is appropri- ate for nutrients with symmetrical requirement distributions, re- quires the following data: â¢ the individualâs mean nutrient intakes during a given number of days; â¢ the day-to-day standard deviation of intakes for each nutrient of interest, as estimated from larger data sets for the appropriate life stage and gender group; â¢ the EAR; and â¢ the standard deviation of the nutrient requirement in the indi- vidualâs life stage and gender group. From this information, a ratio is computed that compares the magnitude of difference between the individualâs intake and the EAR to an estimate of variability of intake and requirements. The
USES OF DIETARY REFERENCE INTAKES 557 bigger the difference between intake and EAR and the lower the variability of intakes and requirements, the greater the degree of certainty one has in assessing whether the individualâs nutrient in- take is adequate or inadequate. This approach is preferred because of its relative accuracy and should be used when the data indicated above are available. However, when the estimate of usual intake is not based on specific recalls or records, a more qualitative interpretation of intakes could be used. For example, many practitioners use the diet history method to construct a usual dayâs intake, but the error structure associated with this method is unknown. Thus, a practitioner should be cau- tious when using this method to approximate usual intakes. For practical purposes, many users of the DRIs may find it useful to consider that observed intakes below the EAR very likely need to be improved (because the probability of adequacy is 50 percent or less), and those between the EAR and the Recommended Dietary Allowance (RDA) probably need to be improved (because the prob- ability of adequacy is less than 97.5 percent). Only if intakes have been observed for a large number of days and are at or above the RDA, or observed intakes for fewer days are well above the RDA, should one have a high level of confidence that the intake is ade- quate. For example, a 40-year-old man who usually consumes 8 mg/day of zinc from his food and who takes a multiple vitamin and mineral supplement containing 15 mg of zinc 3 days a week would average 14.4 mg/day (8 mg + [15 mg Ã 3 Ã· 7]). Thus, his diet alone (8 mg/ day) would put him at a high risk of inadequacy since it is below the EAR of 9.4 mg/day. The addition of the supplement, however, pro- vides an amount above the RDA of 11 mg/day, thus suggesting little likelihood that intake is inadequate if dietary assessment represents his true usual food and supplement intake. If this same man took his multiple vitamin and mineral supplement every day, his usual intake from supplements alone would exceed the RDA, and one could conclude that he has little likelihood of inadequate zinc in- take even without knowledge of his intake of zinc from food. Using the Adequate Intake for Individuals Adequate Intakes (AIs) have been set for all nutrients for infants through 6 months of age. By definition and observation, infants born at term who are exclusively fed human milk by healthy mothers are consuming an adequate nutrient intake. Infants who are con- suming formulas with a nutrient profile similar to human milk (after
558 DIETARY REFERENCE INTAKES adjustment for differences in bioavailability) are also consuming adequate levels. When an infant formula contains lower nutrient levels than human milk, the likelihood of nutrient adequacy for infants consuming this formula cannot be determined because data on infants at lower concentrations of intake are not available for review. AIs have also been established for older individuals for several nutrients. Of the nutrients considered in this report, AIs have been developed for vitamin K, chromium, and manganese. Usual individ- ual intakes equal to or above the AI can be assumed adequate. However, the likelihood of inadequacy of usual intakes below the AI cannot be determined. Using the Tolerable Upper Intake Level for Individuals The Tolerable Upper Intake Level (UL) is used to examine the possibility of overconsumption of a nutrient. If an individualâs usual nutrient intake remains below the UL, there is little or no risk of adverse effects from excessive intake. At intakes above the UL, the risk of adverse effects may increase. However, the intake at which a given individual will develop adverse effects as a result of taking large amounts of a nutrient is not known with certainty. For exam- ple, an adult with usual zinc intakes that exceed the UL (40 mg/ day) may be at increased risk of the adverse effect of reduced copper status. There is no established benefit for healthy individuals in consuming amounts of nutrients that exceed the RDA or AI. ASSESSING NUTRIENT INTAKES OF GROUPS Using the Estimated Average Requirement for Groups The prevalence of nutrient inadequacy for a group of individuals may be estimated by comparing the distribution of usual intakes with the distribution of requirements. The Estimated Average Re- quirement (EAR) is the appropriate Dietary Reference Intake (DRI) to use for this purpose. In most situations, a cut-point approach may be used to estimate the prevalence of inadequate intakes. This approach is a simplification of the full probability method of calculat- ing the prevalence of inadequacy described by the National Research Council (NRC, 1986). The cut-point approach allows the prevalence of inadequate intakes in a population to be approximated by deter- mining the percentage of individuals in the group whose usual intakes are less than the EAR for the nutrient of interest. This
USES OF DIETARY REFERENCE INTAKES 559 approach assumes that the intake and requirement distributions are independent, an assumption made for the nutrients addressed in this report. It further assumes that the variability of intakes among individuals within the group under study is at least as large as the variability of their requirements. This assumption is warranted in free-living populations. Finally, it assumes that the requirement dis- tribution is symmetrical. This is thought to be true for all nutrients discussed in this report except iron, for which requirement distri- butions are skewed. Additional information on assessing the ade- quacy of group intakes of iron is provided in the section on âNutrient- Specific Considerationsâ. Before determining the percentage of the group whose intake is below the EAR, the intake distribution should be adjusted to re- move the effect of day-to-day variation in intake. This can be accom- plished either by collecting dietary data for each individual over a large number of days or by statistical adjustments to the intake dis- tribution that are based on assumptions about the day-to-day varia- tion (derived from repeat measurements of a representative subset of the group under study) (Nusser et al., 1996). When this adjust- ment is performed and observed intakes are thus more representa- tive of the usual diet, the intake distribution narrows, giving a more precise estimate of the proportion of the group with usual intakes below the EAR (Figure 14-2). An explanation of this adjustment procedure has been presented in two previous reports (IOM, 2000; NRC, 1986). An example of using the EAR cut-point approach to assess the dietary zinc adequacy of women aged 51 to 70 years follows. Dietary intake data are available from the Third National Health and Nutri- tion Examination Survey (NHANES III), which includes intakes from both food and supplements. Although the NHANES food- intake data were based on a single 24-hour recall for all individuals, replicate 24-hour recalls were conducted on a subset of the partici- pants, and these estimates of day-to-day variation have been used to adjust the intake distributions (see Appendix Tables C-25 and C- 26). The EAR for zinc for women is 6.8 mg/day. In the U.S. popula- tion, about 25 percent of adult women aged 51 to 70 years did not consume adequate amounts of zinc from food sources alone (Ap- pendix Table C-25), as this proportion had estimated intakes below the EAR. When dietary supplements were included, there was little difference in the proportion below the EAR, suggesting that few individuals with low zinc intakes use zinc supplements. The assessment of nutrient adequacy for groups of people re- quires unbiased, quantitative information on the intake of the nu-
560 DIETARY REFERENCE INTAKES Percentage of Individuals âUsual intakesâ observed over several days EAR 1-day observations Usual Intake (amount/day) FIGURE 14-2 Comparison of 1-day and usual intakes for estimating the propor- tion of a group consuming intakes below the Estimated Average Requirement (EAR). trient of interest by individuals in the group. Care must be taken to ensure the quality of the information upon which assessments are made so that they are not underestimates or overestimates of total nutrient intake. Estimates of total nutrient intake, including amounts from supplements, should be obtained. For some of the nutrients discussed in this report, drinking water may also be a sig- nificant nutrient source. It is also important to use appropriate food composition tables with accurate nutrient values for the foods as consumed. In the example for zinc, both a database of representa- tive zinc values for all foods that contribute substantially to the intakes of this nutrient and a supplement database with the zinc composition of the specific supplements consumed by the popula- tion under study are required. Overestimates of the prevalence of inadequate intakes could result if the data used are based on intakes that are systematically under- reported or if foods rich in zinc are underreported. Such under- reporting is common in national surveys (Briefel et al., 1997). Cur- rently, a method for adjusting intakes based on underreporting by individuals is not available and much work is needed to develop an acceptable method. Conversely, underestimates of the prevalence of inadequacy would result if zinc-rich foods were overreported. A more extensive discussion of potential sources of error in self-reported
USES OF DIETARY REFERENCE INTAKES 561 dietary data can be found in the report Dietary Reference Intakes: Applications in Dietary Assessment (IOM, 2000). Using the Recommended Dietary Allowance for Groups The Recommended Dietary Allowances (RDAs) are not useful in estimating the prevalence of inadequate intakes for groups. As described above, the EAR should be used for this purpose. Using the Adequate Intake for Groups In this report, Adequate Intakes (AIs) are assigned for infants, and they reflect the average intake for infants receiving human milk through 6 or 12 months of age. Human milk and formulas with the same nutrient composition as human milk (after adjustment for bioavailability) provide the appropriate levels of nutrients for full- term infants of healthy, well-nourished mothers. Groups of infants consuming formulas with lower levels of nutrients than human milk may be at some risk of inadequacy, but the prevalence of inadequacy cannot be quantified. AIs are assigned to all age groups in this report for vitamin K, chromium, and manganese. For vitamin K and manganese, AIs were based on median intakes of apparently healthy populations as assessed in large national surveys. By definition, this means that groups with median intakes equal to or above the AI can be assumed to have a low prevalence of inadequacy (provided that variability in intake does not exceed that of the healthy group used to establish the AI). However, it should be noted that group median intakes below the AI cannot be assumed to be inadequate. For chromium, the AI was established by using an estimated amount of chromium/ 1,000 kcal of nutritionally balanced meals and median energy in- takes from NHANES III (Briefel et al., 1997). Thus, there is less certainty about concluding that the prevalence of nutrient inade- quacy is low in groups with mean usual intakes equal to or above the AI for chromium. Using the Tolerable Upper Intake Level for Groups The proportion of the population with usual intakes below the Tolerable Upper Intake Level (UL) is likely to be at no risk of adverse effects due to overconsumption, but the proportion above the UL may be at some risk. In the case of zinc, for example, the UL for adults is 40 mg/day. The NHANES III data in Appendix Table
562 DIETARY REFERENCE INTAKES C-26, which include reported intakes from supplements, illustrate that the proportion of U.S. women aged 51 to 70 years who exceed this UL is just over 1 percent. In contrast, when data from food alone are examined, the proportion of the population with intakes above the UL is less than 1 percent (Appendix Table C-25). In typical North American food-based diets, ULs for vitamin A, iodine, iron, manganese, molybdenum, and zinc can rarely be ex- ceeded. The UL for copper pertains to food sources and copper supplements. Use of dietary supplements containing these nutri- ents would be the primary reason for exceeding the ULs. The mean intake of a population cannot be used to evaluate the prevalence of intakes above the UL. A distribution of usual intakes, including intakes from supplements and drinking water, is required to assess the proportion of the population that might be at risk of overconsumption. PLANNING NUTRIENT INTAKES OF INDIVIDUALS Using the Recommended Dietary Allowance for Individuals Individuals should use the Recommended Dietary Allowance (RDA) as the target for their daily nutrient intakes if an RDA has been established. For example, to increase their vitamin A consump- tion to meet the RDA (900 and 700 Âµg/day for men and women, respectively), adults can increase their intake of foods that provide preformed vitamin A (including dairy products, eggs, margarine, liver) and carotenoids like Î²-carotene (deep green and yellow fruits and vegetables). An 8-ounce glass of milk contains about 65 Âµg of preformed vitamin A, and a half-cup serving of carrots contains the equivalent of approximately 950 Âµg of vitamin A as Î²-carotene. Using the Adequate Intake for Individuals Adequate Intakes (AIs) are set for infants through 6 months of age for all nutrients, and for all nutrients except iron and zinc, for infants 7 through 12 months of age. Human milk will supply the AI for a nutrient for term infants through 6 months of age, and so it is not necessary to plan additional sources of intakes for infants exclu- sively fed human milk. Likewise, an infant formula with a nutrient profile similar to human milk (after adjustment for differences in bioavailability) should supply adequate nutrients for an infant. In this report, AIs are also set for children, adolescents, and adults
USES OF DIETARY REFERENCE INTAKES 563 for vitamin K, chromium, and manganese. Accordingly, individuals should use the AI as their goal for intake of these nutrients. PLANNING NUTRIENT INTAKES OF GROUPS Using the Estimated Average Requirement and the Tolerable Upper Intake Level for Groups For those nutrients with Estimated Average Requirements (EARs), the EAR may also be used as a basis for planning or making recom- mendations for the nutrient intakes of groups. The mean intake of a group should be high enough so that only a small percent of the group would have intakes below the EAR, thus indicating a low prevalence of dietary inadequacy. Traditionally, a prevalence of in- adequacy below 2 to 3 percent has been used as a target. For nutri- ents with a statistically normal requirement distribution, this goal would be attained by planning for a group mean intake equal to the EAR plus 2 standard deviations (SD) of the intake distribution. Because the variability of intakes generally exceeds the variability of requirements, this target group mean intake will usually exceed the Recommended Dietary Allowance (RDA) (which equals the EAR plus 2 SDs of the requirement distribution). Prevalences of inadequacy more or less than 2 to 3 percent could also be considered, and mean intakes needed to attain these prevalences would be estimated by determining the number of SDs of intake that would result in the desired prevalence below the EAR. This is done by consulting tables that list areas of the standard normal distribution in relation to standard deviation scores (z scores). When it is known that requirements for a nutrient are not nor- mally distributed (for example, iron requirements) and one wants to ensure a low group prevalence of inadequacy, examination of the distributions of both intakes and requirements would be needed to determine a median intake at which the proportion of individuals below the EAR is low. Using the EAR and Tolerable Upper Intake Level (UL) in plan- ning intakes of groups involves a number of key decisions and the analysis of issues such as the following: â¢ determination of the current nutrient intake distribution of the group of interest; â¢ an evaluation of possible interventions to shift the current dis- tribution, if necessary, so there is an acceptably low prevalence of intakes below the EAR, as well as an acceptably low prevalence of
564 DIETARY REFERENCE INTAKES intakes above the UL; some interventions may increase the intake of only those most at risk (usually by individual intervention), while others may increase the intake of the entire group (such as fortifica- tion of the food supply); and â¢ the selection of the degree of risk that can be tolerated when planning for the group (for example, a 2 to 3 percent prevalence versus a higher or lower prevalence). Using the Adequate Intake for Groups Adequate Intakes (AIs) have been established as mean or median intakes of healthy groups for some nutrients discussed in this report. This includes all nutrients for infants fed human milk through 6 months of age and the nutrients vitamin K and manganese for adults. Planning a group intake that meets the AI should, by defini- tion, be associated with a low prevalence of inadequacy. This, of course, assumes that the group being planned for has similar char- acteristics to the group used to establish the AI. For chromium, the only nutrient in this report with an AI that is not based on the mean or median intake of healthy groups, there is less certainty that group mean intakes equal to or above the AI will be associated with a low prevalence of inadequacy. NUTRIENT-SPECIFIC CONSIDERATIONS Vitamin A A major change in the extent to which provitamin A carotenoids can be used to form vitamin A is the replacement of retinol equiva- lents (Âµg RE) with retinol activity equivalents (Âµg RAE) for the pro- vitamin A carotenoids. The RAEs for dietary Î²-carotene, Î±-carotene, and Î²-cryptoxanthin are 12, 24, and 24 Âµg, respectively, compared to corresponding REs of 6, 12, and 12 Âµg reported by the National Research Council (NRC, 1989). It is recommended that future food composition and intake tables use actual weights (Âµg) of provitamin A carotenoids rather than use converted data based on the equivalency to vitamin A. This prevents confusion as to whether the RE or RAE has been used for determining the total vitamin A content of a food or for estimating total vitamin A intakes. This change raises two issues: (1) how vitamin A intakes can be determined using the cur- rently available U.S. Department of Agricultureâs (USDA) Nutrient Database for Standard Reference, and (2) how to interpret pub- lished data on vitamin A intakes of various population groups.
USES OF DIETARY REFERENCE INTAKES 565 Determining the Vitamin A Content of Foods with Current Nutrient Databases Nutrient databases will need to be revised to give total vitamin A activity in Âµg RAE. In addition, developers of nutrient databases may choose to provide the amount (Âµg) of preformed vitamin A and of individual carotenoids. Thus, if the vitamin A activity for provitamin A carotenoids changes in the future, it will be possible to recalculate total vitamin A. In the meantime, it is possible to estimate total vitamin A activity in Âµg RAE from existing tables in Âµg RE. For foods, such as liver, containing only vitamin A activity from preformed vitamin A (retinol), no adjustment is necessary. Vitamin A values for foods (e.g., carrots) containing only plant sources (pro- vitamin A carotenoids) of vitamin A can be adjusted by dividing the Âµg RE by two. For foods that are mixtures containing both plant and animal sources of vitamin A (e.g., a casserole containing meat and vegetables), the adjustment process is more complex. If the recipe for a mixture is known, the new vitamin A value may be calculated after adjusting the vitamin A content of each ingredient, as necessary. Alternatively, if the nutrient database contains values as Âµg RE for both total vitamin A and carotenoids, then it is possible to calculate a new value for both carotenoids and for total vitamin A. For example, USDAâs Nutrient Database for Individual Surveys contains both these variables. To determine a revised total vitamin A value, the retinol value is calculated as the difference between the original total vitamin A value and the original carotenoid value. The revised total vitamin A content is then calculated as the sum of the retinol value and the adjusted carotenoid value, which is the original carotenoid value in Âµg RE divided by two. As discussed in the following section, this same procedure may be used to adjust intake data that have been analyzed using other databases. As shown in Figure 4-2, supplemental Î²-carotene has a higher bioconversion to vitamin A than does dietary Î²-carotene. With low doses, the conversion is as high as 2:1, and developers of composi- tion information for dietary supplements should use this higher conversion factor. Little is known about the bioconversion of the forms of Î²-carotene that are added to foods, so fortification forms of Î²-carotene should be assumed to have the same bioconversion as food forms, 12:1. Food and supplement labels usually state vitamin A levels in International Units (IU). One IU of retinol is equivalent to 0.3 Âµg of retinol, or 0.3 Âµg RAE. One IU of Î²-carotene in supple- ments is equivalent to 0.5 IU of retinol or 0.15 Âµg RAE (0.3 Ã 0.5). One IU of dietary Î²-carotene is equivalent to 0.165 IU retinol or
566 DIETARY REFERENCE INTAKES 0.05 Âµg RAE (0.3 Ã 0.165). One IU of other dietary provitamin A carotenoids is equivalent to 0.025 Âµg RAE. Interpreting Published Data on Vitamin A Intakes of Various Population Groups Existing data on vitamin A intakes of individuals and groups will need to be reinterpreted because of the changes in the retinol molar equivalency ratios for carotenoids to Âµg RAE. Two scenarios are possible: (1) the existing data provide values for both total vita- min A and carotenoid intake, and (2) the existing data provide values only for total vitamin A intake. Existing Data Provide Values for Both Total Vitamin A and Carotenoids. Data from some dietary surveys, such as the 1994â1996 Continuing Survey of Food Intakes of Individuals (CSFII) and the Third National Health and Nutrition Examination Survey (NHANES III), include REs for both total vitamin A and for carotenoids. The data manipulations required depend on the type of information that is sought (for example, mean intakes versus the proportion of a group with inadequate intakes). A way to approximate the mean intake of a group follows: 1a. Find the group mean intake for total vitamin A intake (e.g., for women aged 30 to 39 years in the CSFII, mean intake was 895 Âµg RE). Subtract the group mean intake of carotenoids (e.g., for women aged 30 to 39 years in the CSFII, mean carotene intake was 500 Âµg RE). Thus, preformed vitamin A intake would be estimated as 395 Âµg (895 â 500). 1b. Divide the group mean intake of carotenoids by 2 (in this example, 500 Ã· 2 = 250 Âµg RAE). This represents the corrected value for provitamin A intake. 1c. Add the corrected provitamin A intake determined in Step 1b to the preformed vitamin A intake determined in Step 1a. In this example, the mean vitamin A intake of women aged 30 to 39 years in the CSFII would be 645 Âµg RAE (250 + 395). To determine the group prevalence of inadequate vitamin A in- takes, one would need to have access to the individual intake data from which the group means were determined. For each person in the group, Steps 1a through 1c would be followed. Then the pro- portion of individuals with intakes below the Estimated Average
USES OF DIETARY REFERENCE INTAKES 567 Requirement (EAR) can be determined. For vitamin A, the EAR for women is 500 Âµg, and thus the proportion of the group with intakes below this level would reflect the group prevalence of inadequate intakes. Existing Data Provide Values for Only Total Vitamin A Intake. In this situation, there will be more uncertainty associated with estimates of both group mean intakes and the proportion of a group with inadequate intakes. This is because of the lack of information on the proportion of the total vitamin A intake that was derived from carotenoids. In this situation, a possible approach to approximating group mean intakes follows: 2a. Use other published data from a similar subject life stage and gender group that provide intakes of both total vitamin A and caro- tenoids to perform the calculations in Steps 1a through 1c above. For example, if the group of interest was 30- to 39-year-old women, data for this group from the CSFII could be used. 2b. Calculate the adjusted vitamin A intake for this group as a percentage of the unadjusted mean intake. For the example of 30- to 39-year-old women, the adjusted mean intake was 645 Âµg, and the unadjusted mean was 895 Âµg. Thus the adjusted vitamin A intake would be 0.72 (645 Ã· 895), or 72 percent. 2c. Apply the adjustment factor to the mean intake of the group of interest. For example, if the groupâs mean intake had been re- ported as 1,100 Âµg, the adjusted intake would be 792 Âµg (1,100 Ã 0.72). This method could also be used to estimate the group prevalence of inadequacy, again with access to the individual intake data from which the group means were determined. In this case, the adjust- ment factor (in the example above, 72 percent) would be applied to data for each individual, and used in Steps 2a through 2c. The resulting distribution of intakes would then be examined to approx- imate the proportion of intakes below the EAR. In this situation, the approximate nature of this approach should be emphasized to an even greater extent. Given that the proportion of vitamin A de- rived from preformed vitamin A and from carotenoids will differ among individuals, use of an âaverageâ adjustment factor has the potential to introduce errors that may not be random.
568 DIETARY REFERENCE INTAKES Implications Arising from the Development of Retinol Activity Equivalents (RAE) The vitamin A activity of provitamin A carotenoids found in darkly colored fruits and green leafy vegetables is half that previously assumed. Consequently, individuals who rely on plant foods for the majority of their vitamin A needs should ensure that they consume foods that are rich in carotenoids (specifically, deep yellow and green vegetables and fruits) on a regular basis. Another implication of the reduced contribution from the pro- vitamin A carotenoids is that vitamin A intakes of most population groups are lower than was previously believed. For example, in the CSFII survey, the reported mean proportion of vitamin A derived from carotenoids was 47 percent. Using the new conversion factors would thus reduce the population mean vitamin A intake by about 23 to 24 percent, or from 982 Âµg RE to 751 Âµg RAE. Multiple EARs for Vitamin A A second (lower) EAR has been set for vitamin A for each age group and gender, primarily for those populations with limited access to vitamin A-rich foods. The functional endpoint for this EAR is the correction of abnormal dark adaptation, rather than assuring adequate stores. Users may wish to utilize this lower EAR to assess the population prevalence of intakes that are inadequate to support normal dark adaptation, but this EAR is not intended to be used for planning intakes of groups in the United States and Canada. Vitamin K Because habitual vitamin K intake may modulate warfarin dosage in patients using this anticoagulant, these individuals should main- tain their normal dietary and supplementation patterns once an effective dose has been established. Short-term, day-to-day variability in the intake of vitamin K from food sources does not appear to interfere with anticoagulant status and therefore does not need to be carefully monitored. However, changes in supplemental vitamin K intake should be avoided, as bioavailability of synthetic (supple- mental) phylloquinone is considerably greater than bioavailability of phylloquinone from food sources.
USES OF DIETARY REFERENCE INTAKES 569 Chromium Because the chromium content of foods is not included in exist- ing food composition databases, the intakes of individuals and groups cannot be assessed unless duplicate portions of ingested meals are directly analyzed. The lack of a readily available, accurate biochemical method of assessing chromium status further compli- cates the use of the Adequate Intake (AI) for assessment and plan- ning purposes. Iron As described in Chapters 1 and 9, iron requirements were esti- mated through the use of factorial models involving the summation of estimates of component losses and deposition of iron. Since it is expected that the distribution of requirements would not fit the normal distribution, a process involving Monte Carlo simulation of a very large (100,000 individuals) data set was undertaken, with each âpersonâ assigned a random value of the distribution of each com- ponent of iron need. These needs were summed at the level of the individual to yield an estimated distribution of total requirements among individuals (see Appendix Tables I-3 and I-4). The EAR and Recommended Dietary Allowance (RDA) estimates were then derived from those distributions, as the fiftieth and ninety-seventh and one-half percen- tiles, respectively. Very few data were available with which to estimate the distribution of basal iron losses. For this reason, variability in body size (weight in adults, surface area in children) was used as a proxy for direct measurement of variability in basal losses. The rationale for this approach stems from the fact that dermal, intestinal, and urinary losses (components of basal losses) are related to body size. How- ever, even though body weight was used to obtain information on the distribution of losses, it is not appropriate to adjust the iron EAR or RDA for body size because variability of body weight was taken into account when these numbers were derived. Although populations with a body size distribution that differs substantially from those in the United States and Canada may have different iron requirements, a method for making such adjustments is not avail- able.
570 DIETARY REFERENCE INTAKES Assessing the Adequacy of Intakes of Groups Information on the distribution of iron requirements (rather than knowledge of the EAR) is needed to estimate the prevalence of inadequate intakes in a population. Because iron is one nutrient for which it is known with certainty that the requirement distributions are not symmetrical for all life stage and gender groups, the pro- portion of individuals with intakes below the EAR will not reflect the population prevalence of nutrient inadequacy. Instead, the full probability approach must be used. The Probability Approach. Using the probability approach requires knowledge of both the distribution of requirements and the distri- bution of usual intakes for the population of interest. As described previously (IOM, 2000; NRC, 1986), the probability approach in- volves (1) determining the risk of inadequacy for each individual in the population, and then (2) averaging the individual probabilities across the group. For iron, Appendix Tables I-5, I-6, and I-7 give the probability of inadequacy at various intakes. These tables may be used to calculate the risk of inadequacy for each individual, and then the estimated prevalence of inadequacy for a population. In addition, Appendix C of Dietary Reference Intakes: Applications in Dietary Assessment (IOM, 2000) demonstrates how to carry out the necessary calculations to obtain a prevalence estimate for a group, and statistical programs (SAS or similar software) can be used to carry out these procedures. A simplified estimate that could also be determined manually is illustrated in Table 14-1 for a hypothetical group of 1,000 menstru- ating women not taking oral contraceptives and consuming a typi- cal omnivorous diet. The first and second columns of this table are based on information in Appendix Tables I-4 and I-7. Intakes below 4.42 mg/day are assumed to have a 100 percent probability of inad- equacy (risk = 1.0). Those with intakes above 18.23 mg/day are assumed to have a zero risk of inadequacy. For intakes between these two extremes, the risk of inadequacy is calculated as 100 minus the midpoint of the percentile of requirement. For example, in- takes between 4.42 and 4.88 fall between the 2.5 and 5th percentile of requirement. The midpoint is 3.75, and the probability of inade- quacy is 100 â 3.75 â 96.3 percent, or a risk of 0.96. The appropriate risk of inadequacy is then multiplied by the number of women with intakes in that range. In this example, only one woman had an intake between 4.42 and 4.88 mg/day, so the number of women with inadequate intake is 0.96 (1 Ã 0.96). In the next range (4.89
USES OF DIETARY REFERENCE INTAKES 571 TABLE 14-1 Illustration of the Full Probability Approach to Estimate the Prevalence of Dietary Iron Inadequacy in a Group of 1,000 Menstruating Women (Not Using Oral Contraceptives and Following an Omnivorous Diet) Range of Usual Intake Associated Number of Percentiles with Number of Women of Requirement Risk of Women with Requirement Percentiles Inadequate with Intake Inadequate Distribution (mg/d) Intake in Range Intake < 2.5 < 4.42 1.0 1 1 2.5â5.0 4.42â4.88 0.96 1 0.96 5â10 4.89â5.45 0.93 3 2.79 10â20 5.46â6.22 0.85 10 8.5 20â30 6.23â6.87 0.75 15 11.25 30â40 6.88â7.46 0.65 20 13 40â50 7.47â8.07 0.55 23 12.65 50â60 8.08â8.76 0.45 27 12.15 60â70 8.77â 9.63 0.35 50 17.5 70â80 9.64â10.82 0.25 150 37.5 80â90 10.83â13.05 0.15 200 30.0 90â95 13.06â15.49 0.08 175 14 95â97.5 15.50â18.23 0.04 125 5 > 97.5 > 18.23 0.0 200 0 Total 1,000 165 mg/day to 5.45 mg/day, or between the fifth and tenth percentiles) there were three women, with an associated number of women with inadequate intake of 2.79 (3 Ã 0.93). If this is done for each intake range, the total number of women with inadequate intakes can be determined. In this example, 165 of the 1,000 women have inade- quate intakes, for an estimated prevalence of inadequacy of 16.5 percent. It is important to remember that this approach does not identify the specific women with inadequate intakes, but is rather a statistical calculation of the prevalence of inadequate intakes. Thus, it cannot be used to screen individuals at risk of inadequacy. Note that the prevalence of nutrient inadequacy that is estimated by the full probability approach differs considerably from that esti- mated by the cut-point method (the proportion with intakes below the EAR). In this example, the EAR (median requirement) is 8.07 mg/day, and only 73 women have intakes below this amount. Thus,
572 DIETARY REFERENCE INTAKES the cut-point method would lead to an estimated prevalence of in- adequacy of 7.3 percent, which differs considerably from the estimate of 16.5 percent obtained by using the full probability approach. The reason for the discrepancy is that one of the conditions needed for the cut-point approach (a symmetrical requirement distribution) is not true for iron requirements of menstruating women. Comparison of Assessments Using the Probability Approach to Biochemical Assessment. If requirement estimates are correct and both the dietary data and biochemical measures are reliable estimates of true usual intake and true blood concentrations in the same population, then the prevalence of apparently inadequate dietary intakes and bio- chemical deficiency should be similar, as discussed in Chapter 9. In the example above, one would expect to observe a prevalence of low serum ferritin concentrations (< 15 Âµg/L) that approximates the prevalence of inadequate intakes, or about 16.5 percent. The individuals with low serum ferritin concentrations are not necessarily the same as the individuals with low intake values, so the probability approach is not appropriate for identifying specific individuals with low serum ferritin values. Special Situations in Which the EAR and RDA May Vary Special situations in which iron requirements may vary are sum- marized in Table 14-2 along with suggestions on how to adjust esti- mates of requirements. Zinc Bioavailability of zinc is known to vary greatly, depending on the intakes of other dietary components, most notably phytate, that inhibit absorption. The World Health Organization (WHO, 1996) suggested that bioavailability of zinc might range from 15 percent in a diet with low bioavailability to a high of 50 percent in diets with high bioavailability. Characteristics associated with diets varying in bioavailability are summarized in Table 14-3. Gibson and Ferguson (1998) have reviewed the use of the phytate:zinc ratio for assessing dietary zinc intake. Table 14-3 indicates that diets of most North Americans would have âmediumâ bioavailability, approximating the fractional absorption rate of 38 percent that was used in estimating the EAR for adults. It also indicates that diets of some strict vegetar- ians may have low bioavailability, with the result that their dietary requirements for zinc would be increased. A quantitative estimate
USES OF DIETARY REFERENCE INTAKES 573 TABLE 14-2 Situations in which the Iron Requirement May Vary Special Consideration Recommended Iron Intake Infants who do not The Adequate Intake (AI) of 0.27 mg/day does not receive human milk, apply. For infants who do not receive human milk, 0 through 6 months it is recommended that iron-containing formula (4â12 mg/L) be used from birth through 12 months. Preterm infants Even if they receive human milk, the AI is not adequate for preterm infants as their iron stores are low. Supplementation is recommended. Menarche before (or The Estimated Average Requirement (EAR) and after) age 14 in girls Recommended Dietary Allowance (RDA) for girls ages 9 to 13 years make no allowance for menstrual losses. Girls who reach menarche before age 14 years should consume an additional 2.5 mg/day. Conversely, the RDA for girls ages 14 to 18 years assumes that menstruation is occurring. It thus follows that girls 14 years and older who have not reached menarche would have a lower recommended intake of iron. Teens/preteens in Because the rate of growth during the adolescent the growth spurt growth spurt can be more than double the average rate for boys, and up to 50 percent higher for girls, it is recommended that boysâ intakes during the growth spurt increase by 2.9 mg/day and girlsâ intakes by 1.1 mg/day. Oral contraceptive Because blood losses are reduced by approximately 60 users percent in women who habitually use oral contraceptive agents, the iron requirement and thus recommended intake for adolescent girls and women taking oral contraceptives would be lower. Postmenopausal women Postmenopausal women who use HRT may be treated using cyclic hormone with use of either cyclic (a given number of days on replacement therapy active hormones followed by a week or so without (HRT) hormones) or continuous protocols. Women using cyclic protocols frequently experience withdrawal bleeding in the week without hormones and thus would have higher iron requirements than women not using HRT or using continuous HRT. Few data are available on the magnitude and variability of HRT-associated blood loss, but it is probably between the losses experienced by premenopausal women who use oral contraceptives and those of postmenopausal women who do not bleed. continued
574 DIETARY REFERENCE INTAKES TABLE 14-2 Continued Special Consideration Recommended Iron Intake Vegetarians Iron bioavailability is reduced in vegetarian diets, both because of the absence of easily absorbed heme iron and because of the presence of inhibitors of iron absorption. The percent bioavailability was estimated at 10 percent (versus 18 percent in omnivorous diets). Thus, the iron requirement for vegetarians would be approximately 1.8 times higher than the values established for omnivores, and recommended intakes could be adjusted using a similar factor. Athletes Basal losses of iron by athletes performing intense exercise on a daily basis are elevated, with estimates ranging from a 30 to 70 percent increase. Therefore, the iron requirement is increased for those who exercise intensely on a daily basis. It should be noted, however, that much of the research conducted with respect to iron needs of athletes has been done with runners. The postulated mechanisms of increased basal losses (hematuria and fecal blood loss) may not occur to as great an extent in athletes who participate in other sports. Blood donors The donation of 1 unit of blood/year is estimated to increase the need for absorbed iron by 0.6 to 0.7 mg/day, which, assuming 18 percent absorption, suggests that intake would need to be 3 to 4 mg/ day higher. Thus, individuals who donate blood on a regular basis will have an increased iron requirement. Presumably, iron needs of frequent donors would increase in proportion to the amount of blood donated. of the average requirement of individuals consuming diets with low zinc bioavailability cannot be made at this time. However, it seems reasonable to suggest that such individuals should be counseled to consume intakes that are at least equal to the RDA, and perhaps up to as much as twice the RDA. The Tolerable Upper Intake Level (UL) for zinc for adults is 40 mg, which exceeds the RDA for men by somewhat less than four- fold and for women by five-fold. Although intakes of zinc above 40 mg/day from food alone are uncommon (the ninety-ninth percen-
USES OF DIETARY REFERENCE INTAKES 575 TABLE 14-3 Qualitative Bioavailability of Zinc According to Diet Characteristicsa Bioavailability Dietary Characteristics High Refined diets low in cereal fiber and phytic acid, with adequate protein primarily from meats and fish Phytate/zinc molar ratio < 5 Medium Mixed diets containing animal or fish protein Vegetarian diets not based primarily on unrefined, unfermented cereal grains Phytate/zinc molar ratio 5â15 Low Diets high in unrefined, unfermented, and ungerminated cereal grains, especially when animal protein intake is negligible High-phytate soy protein products are the primary protein source Diets in which â¥ 50 percent of energy is provided by high phytate foods (high extraction rate [90 percent] flours and grains, legumes) Phytate/zinc molar ratio > 15 High intake of inorganic calcium (> 1 g/day) potentiates the inhibitory effects of these diets, especially when animal protein intake is low a The phytate content of foods is provided by Hallberg and Hulthen (2000). The zinc content of foods is available from the U.S. Department of Agriculture at http:// www.nal.usda.gov/fnic/foodcomp. SOURCE: Modified from WHO (1996). tiles for intake were less than 40 mg/day for all adults in both the NHANES III and CSFII surveys), when intake from supplements is added, higher proportions are above the UL. This is not unexpected, as many multiple vitamin-mineral supplements contain 15 mg of zinc. On the other hand, zinc intakes below the EAR are also fairly common. The dilemma, then, is how to ensure adequate zinc nutri- ture in the population while avoiding intakes in excess of the UL. Even in populations with low mean zinc intakes, care must be taken not to intervene in ways that would move a substantial proportion of the population above the UL. For example, widespread fortifica- tion of the food supply with zinc may not be appropriate, even if the prevalence of inadequacy in a population is high. More targeted approaches, such as increased consumption of zinc-rich foods by those at a high risk of inadequacy, should be considered.
576 DIETARY REFERENCE INTAKES Trace Elements Previous editions of the RDAs in the United States (NRC, 1980, 1989) established a category of estimated safe and adequate daily dietary intakes (ESADDI) for essential nutrients with databases that were insufficient for developing an RDA, but where evidence of potentially toxic intakes was known. The values for ESADDI typically were ranges of intakes. The DRI process has taken a different approach. If the data required to establish intake recommendations (either an EAR and RDA, or an AI) were not available, as is the case for arsenic, boron, nickel, silicon, and vanadium in this report, no requirement is set. However, the database was adequate to establish ULs for boron, nickel, and vanadium. Therefore, for these nutri- ents an upper limit has been set, but not a lower limit. Accordingly, when intake data are available, users can estimate the proportion of the population that may be at risk from excessive intakes of these elements. SUMMARY The Dietary Reference Intakes (DRIs) may be used to assess nutri- ent intakes as well as for planning nutrient intakes. Box 14-1 sum- marizes the appropriate uses of the DRIs for individuals and groups. For the nutrients presented in this report, only iron requires the use of the full probability approach to estimate the prevalence of inadequacy due to skewedness of the requirement distributions. Guidance is provided for adjustments to the iron requirement for several special situations, including onset of menarche before age 14 years in girls, onset of the growth spurt in adolescent males and females, use of oral contraceptives, athletes, vegetarians, and fre- quent blood donors. Adjustments to zinc requirements are also recommended on the basis of the impact of the zinc:phytate ratio on bioavailability of zinc. Examples are provided of ways to determine the appropriate ad- justments to estimates of the usual intake of vitamin A considering the change in the vitamin A activity of provitamin A carotenoids.
USES OF DIETARY REFERENCE INTAKES 577 BOX 14-1 Uses of Dietary Reference Intakes for Healthy Individuals and Groups Type of Use For an Individual a For a Group b Assessment EAR: use to examine the EAR: use to estimate the probability that usual prevalence of inadequate intakes intake is inadequate. within a group. RDA: usual intake at or RDA: do not use to assess intakes above this level has a low of groups. probability of inadequacy. AI c : usual intake at or AI c : mean usual intake at or above this level has a low above this level implies a low probability of inadequacy. prevalence of inadequate intakes. UL: usual intake above UL: use to estimate the this level may place an percentage of the population at individual at risk of potential risk of adverse effects adverse effects from from excess nutrient intake. excessive nutrient intake. Planning RDA: aim for this intake. EAR: use to plan an intake distribution with a low prevalence of inadequate intakes. AI c : aim for this intake. AI c : use to plan mean intakes. UL: use as a guide to limit UL: use to plan intake intake; chronic intake of distributions with a low higher amounts may prevalence of intakes potentially increase the potential risk at risk of adverse effects. of adverse effects. RDA = Recommended Dietary Allowance EAR = Estimated Average Requirement AI = Adequate Intake UL = Tolerable Upper Level a Evaluation of true status requires clinical, biochemical, and anthropometric data. continued
578 DIETARY REFERENCE INTAKES BOX 14-1 Continued b Requires statistically valid approximation of distribution of usual intakes. c Forthe nutrients in this report, AIs are set for infants for all nutrients, and for other age groups for vitamin K, chromium, and manganese. The AI may be used as a guide for infants as it reflects the average intake from human milk. Infants consuming formulas with the same nutrient composi- tion as human milk are consuming an adequate amount after adjustments are made for differences in bioavailability. When the AI for a nutrient is not based on mean intakes of healthy populations, this assessment of adequacy is made with less confidence. REFERENCES Basiotis PP, Welsh SO, Cronin FJ, Kelsay JL, Mertz W. 1987. Number of days of food intake records required to estimate individual and group nutrient in- takes with defined confidence. J Nutr 117:1638â1641. Briefel RR, Sempos CT, McDowell MA, Chien S, Alaimo K. 1997. Dietary methods research in the Third National Health and Examination Survey: Under- reporting of energy intake. Am J Clin Nutr 65:1203Sâ1209S. Gibson RS, Ferguson EL. 1998. Assessment of dietary zinc in a population. Am J Clin Nutr 68:430Sâ434S. Hallberg L, Hulthen L. 2000. Prediction of dietary iron absorption: An algorithm for calculating absorption and bioavailability of dietary iron. Am J Clin Nutr 71:1147â1160. Heitmann BL, Lissner L. 1995. Dietary underreporting by obese individualsâIs it specific or non-specific? Br Med J 311:986â989. IOM (Institute of Medicine). 1994. How Should the Recommended Dietary Allowances be Revised? Washington, DC: National Academy Press. IOM. 2000. Dietary Reference Intakes: Applications in Dietary Assessment. Washington, DC: National Academy Press. Mertz W, Tsui JC, Judd JT, Reiser S, Hallfrisch J, Morris ER, Steele PD, Lashley E. 1991. What are people really eating? The relation between energy intake de- rived from estimated diet records and intake determined to maintain body weight. Am J Clin Nutr 54:291â295. NRC (National Research Council). 1980. Recommended Dietary Allowances, 9th ed. Washington, DC: National Academy Press. NRC. 1986. Nutrient Adequacy. Assessment Using Food Consumption Surveys. Washing- ton, DC: National Academy Press. NRC. 1989. Recommended Dietary Allowances, 10th ed. Washington, DC: National Academy Press.
USES OF DIETARY REFERENCE INTAKES 579 Nusser SM, Carriquiry AL, Dodd KW, Fuller WA. 1996. A semiparametric transfor- mation approach to estimating usual daily intake distributions. J Am Stat Assoc 91:1440â1449. Schoeller DA. 1999. Recent advances from application of doubly labeled water to measurement of human energy expenditure. J Nutr 129:1765â1768. USDA (U.S. Department of Agriculture). 1999. USDA Nutrient Database for Standard Reference, Release 13. [Online.] Available: http://www.nal.usda.gov/fnic/ foodcomp [accessed February 2000]. WHO (World Health Organization). 1996. Zinc. In: Trace Elements in Human Nutri- tion and Health. Geneva: WHO. Pp. 72â104.