Portal de Periódicos da CAPES

Interpreting and using the Dietary References Intakes in dietary assessment of individuals and groups

2002; Elsevier BV; Volume: 102; Issue: 6 Linguagem: Inglês

10.1016/s0002-8223(02)90177-x

1878-3570

Susan I. Barr, Suzanne P. Murphy, Mary I. Poos,

Nutrition, Genetics, and Disease

The term Dietary Reference Intakes (DRIs) refers to a set of 4 nutrient-based reference values intended primarily for use in assessing and planning diets. The DRIs, the outcome of the expanded approach adopted by the Food and Nutrition Board of the National Academies' Institute of Medicine, replace the periodic updates and revisions of the Recommended Dietary Allowances (RDAs) for the United States ((1)National Research Council.Recommended Dietary Allowances. 10th Ed. National Academy Press, Washington, DC1989Google Scholar) and the Recommended Nutrient Intakes (RNIs) for Canada ((2)Health and Welfare Canada.Nutrition Recommendations. The Report of the Scientific Review Committee. Canadian Government Publishing Centre, Ottawa1990Google Scholar). To date, DRIs have been developed for vitamins and minerals ((3)Institute of Medicine.Dietary Reference Intakes for Calcium>, Phosphorus, Magnesium, Vitamin D, and Fluoride. National Academy Presslocation>Washington, DC, 1997Google Scholar, (4)Institute of Medicine.Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pantothenic Acid, Biotin, and Choline. National Academy Press, Washington, DC1998Google Scholar, (5)Institute of Medicine.Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids. National Academy Press, Washington, DC2000Google Scholar, (6)Institute of Medicine.Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium and Zinc. National Academy Press, Washington, DC2001Google Scholar). It is anticipated that DRIs for energy and macronutrients will be released in 2002. The authors would like to thank the members of the Subcommittee on Interpretation and Use of Dietary Reference Intakes for their outstanding work in preparing the report on which this commentary is based: Lenore Arab, PhD, professor of Nutrition and Epidemiology, University of North Carolina, Chapel Hill; Susan T. Borra, BS, senior vice-president/director of Nutrition, International Food Information Council, Washington, D.C.; Alicia Carriquiry, PhD, professor of Statistics and Associate Provost, Iowa State University; Barbara Devaney, PhD, senior fellow, Mathematica Policy Research Institute, Princeton, N.J.; Johanna Dwyer DSc, RD, professor of Medicine and Community Health, Tufts Medical School and School of Nutrition Science and Policy, Boston; Jean-Pierre Habicht, MD, PhD, James Jamison, Professor of Nutritional Sciences, Cornell University; Janet King, PhD, RD, director, USDA Western Human Nutrition Research Center, Davis, CA; and Harriet Kuhnlein, PhD, RD, professor of Human Nutrition and Founding Director, Centre for Indigenous Peoples' Nutrition and Environment, McGill University, Quebec. We also appreciate the leadership provided by Allison Yates, director of the Institute of Medicine's Food and Nutrition Board (FNB); Vernon Young, chair of the FNB Standing Committee on the Scientific Evaluation of Dietary Reference Intakes (DRI Committee); and Ian Munro, chair of the Subcommittee on Upper Levels of Nutrients (UL Subcommittee), in addition to the many scientists from both the United States and Canada who developed the reports on DRIs for nutrients. The members of both the DRI Committee and the UL Subcommittee also contributed substantially to the process, as did the external reviewers who commented the report. George Beaton served as a consultant throughout most of the deliberations and provided many important contributions. We wish to acknowledge those agencies that have funded the DRI process. In particular, the authors thank Health Canada and the Economic Research Service of the U.S. Department of Agriculture which have supported much of the work of the Subcommittee on Interpretation and Use of the DRIs; and the Dannon Institute, a private foundation which also provided partial support. The DRIs represent a new conceptual approach to nutrient-based reference standards, necessitating new approaches by practitioners and researchers. Guidance on interpretation and use of the DRIs has been provided by the Food and Nutrition Board through publication of the report titled Dietary Reference Intakes: Applications in Dietary Assessment((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar). The purpose of this article is to highlight and summarize some of the information presented in that report, with an emphasis on the implications for dietetics professionals. Specifically, each of the DRIs is briefly defined, and then how they can be interpreted and used in the dietary assessment of individuals and groups is being discussed. Most nutrients have a set of DRIs (Table 1). Usually, a nutrient has an Estimated Average Requirement (EAR) and a corresponding RDA. When an EAR for the nutrient cannot be determined (and therefore, neither can the RDA), then an Adequate Intake (AI) is set. Many nutrients also have a Tolerable Upper Intake Level (UL). Like RDAs and RNIs in the past, all of the DRIs refer to long-term average daily nutrient intakes, although intakes on a given day may vary substantially from this average without apparent ill effect. Each DRI is defined briefly as follows:Table 1Dietary Reference Intakes that apply to vitamins and minerals for people aged 1 year and olderaChildren age 1 year or younger have AI for all vitamins and minerals except iron and zinc, which have an EAR and RDA for infants aged 7 to 12 months.NutrientEARbEAR=Estimated Average Requirement.and RDAcRDA=Recommended Dietary Allowance.AIdAI=Adequate Intake.AI is meaneFor some nutrients, the AI was set as the mean or median intake of a healthy group. In other cases, a combination of observed and experimentally determined intake data was used.intakeULfUL=Tolerable Upper Intake Level.Specific intake sources to which UL appliesThiamin✓——Riboflavin✓——Niacin✓—✓Supplements, fortificants, and medicationsVitamin B-6✓—✓Supplements, fortificants, and medicationsFolate✓—✓Supplements and fortificantsVitamin B-12✓——Pantothenic acid—✓Yes—Biotin—✓No—Choline—✓No✓All sourcesVitamin C✓—✓All sourcesVitamin A✓—✓Preformed retinol from food, supplements, and fortificantsVitamin D—✓No✓Does not include sunlight exposureVitamin E✓—✓All forms of supplemental α-tocopherolVitamin K—✓Yes—Boron——✓All sourcesCalcium—✓No✓All sourcesChromium—✓YesgAI for chromium is based on the mean chromium content per 1,000 kcal of 22 well-balanced daily diets designed by nutritionists.—Copper✓—✓All sourcesFluoride—✓No✓All sourcesIodine✓—✓All sourcesIron✓—✓All sourcesMagnesium✓—✓Supplements and medications onlyManganese—✓Yes✓All sourcesMolybdenum✓—✓All sourcesNickel——✓Soluble nickel salts (supplements)Phosphorus✓—✓All sourcesSelenium✓—✓All sourcesVanadium——✓Elemental vanadium from supplements and fortificantshData to establish a UL were adequate only for adults; for all younger age groups, vanadium intake should be from food only.Zinc✓—✓All sourcesa Children age 1 year or younger have AI for all vitamins and minerals except iron and zinc, which have an EAR and RDA for infants aged 7 to 12 months.b EAR=Estimated Average Requirement.c RDA=Recommended Dietary Allowance.d AI=Adequate Intake.e For some nutrients, the AI was set as the mean or median intake of a healthy group. In other cases, a combination of observed and experimentally determined intake data was used.f UL=Tolerable Upper Intake Level.g AI for chromium is based on the mean chromium content per 1,000 kcal of 22 well-balanced daily diets designed by nutritionists.h Data to establish a UL were adequate only for adults; for all younger age groups, vanadium intake should be from food only. Open table in a new tab Estimated Average Requirement (EAR): A nutrient intake estimated to meet the requirement for a specified indicator of adequacy of half the healthy individuals in a particular life stage and gender group. Although the term "average" is used, the EAR actually represents an estimated median requirement. By definition, the EAR exceeds the requirements of half the group, and falls below the requirements of the other half. Recommended Dietary Allowance (RDA): The average daily dietary intake level sufficient to meet the nutrient requirement (for the specified indicator of adequacy) of nearly all (97% to 98%) healthy individuals in a particular life stage and gender group. The RDA thus exceeds the requirements of nearly all members of the group. For nutrients that have a statistically normal requirement distribution, the RDA is set by adding 2 standard deviations to the EAR. For nutrients with skewed requirement distributions (most notably, iron), the RDA is set between the 97th and 98th percentile of the requirement distribution. The RDA is a "recommended intake" for individuals. Adequate Intake (AI): A recommended intake level based on observed or experimentally determined approximations or estimates of nutrient intake by a group (or groups) of healthy people that are assumed to be adequate. Although an AI is used when data are not sufficient to estimate an EAR (and therefore an RDA), the AI is set at a level thought to meet or exceed the needs of virtually all members of a life stage/gender group. Tolerable Upper Intake Level (UL): The highest average daily nutrient intake level likely to pose no risk of adverse health effects for almost all individuals in the general population. As intake increases above the UL, the potential risk of adverse effects increases. For each nutrient, the chosen criterion of nutritional adequacy used to set the EAR (and RDA) or AI is specified in the nutrient reports ((3)Institute of Medicine.Dietary Reference Intakes for Calcium>, Phosphorus, Magnesium, Vitamin D, and Fluoride. National Academy Presslocation>Washington, DC, 1997Google Scholar, (4)Institute of Medicine.Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pantothenic Acid, Biotin, and Choline. National Academy Press, Washington, DC1998Google Scholar, (5)Institute of Medicine.Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids. National Academy Press, Washington, DC2000Google Scholar, (6)Institute of Medicine.Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium and Zinc. National Academy Press, Washington, DC2001Google Scholar), as is the adverse health effect used to set the UL. For example, the primary indicator of adequacy used to set the vitamin C requirement was near-maximal saturation of leukocytes with ascorbate, reflecting antioxidant protection ((6)Institute of Medicine.Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium and Zinc. National Academy Press, Washington, DC2001Google Scholar). Someone who did not meet their vitamin C requirement would have suboptimal leukocyte ascorbate concentrations, but would likely not show clinical signs of vitamin C deficiency. When dietetic professionals assess the nutritional status of individuals, they frequently have access to information on food or nutrient intakes, as well as clinical, biochemical, and anthropometric data. Food or nutrient intake may be assessed qualitatively (e.g., comparing typical food intake to the Food Guide Pyramid ((8)United States Department of Agriculture.The Food Guide Pyramid. Home and Garden Bulletin no. 252. U.S. Government Printing Office, Washington DC1992Google Scholar)) or quantitatively (comparing nutrient intake determined from analysis of a food recall or record to reference standards). Ideally, information from all areas is combined to provide a valid assessment of an individual's nutritional status. It should be emphasized that quantitative assessment of nutrient intakes is not an essential component of a nutrition assessment. In many cases, qualitative assessment of food intake may be at least as informative. However, when quantitative assessments of nutrient intake are conducted, it is critical that accurate dietary intake data are collected, the correct DRI is selected for the assessment, and the results are interpreted appropriately. It is not easy to accurately compare an individual's intake to his or her requirement for a nutrient. One reason is that a given individual's actual requirement is almost never known. In addition, measuring an individual's long-term usual intake of the nutrient is very difficult, due to day-to-day variation in intake. Because access to food intake data is typically limited to intake on a single or small number of days, this observed or reported intake rarely represents the individual's usual intake ((9)Basiotis P.P. Welsh S.O. Cronin F.J. Kelsay J.L. Mertz W. Number of days of food intake records needed to estimate individual and group nutrient intakes with defined confidence.J Nutr. 1987; 117: 1638-1641Google Scholar). For these reasons, it is not possible to state with complete certainty whether or not an individual's intake meets his or her requirement. However, a statistical approach has been developed that provides an estimate of the level of confidence that an individual's usual intake meets their requirement ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar). This approach compares the difference between the individual's reported intake (the best estimate of their usual intake) and the EAR (the best estimate of their requirement). The equation also accounts for the variability of the requirement (because the individual's requirement could differ from the EAR) and the day-to-day variability of nutrient intake within an individual (because the observed intake may differ from the person's usual intake). The "result" of the equation is a z score, from which one can determine a probability value reflecting the degree of confidence that the individual's usual intake meets their requirement. For example, when z=0.50, P=70%; when z=1.0, P=85%, and when z=2.0, P=98%. Z scores represent a difference from a mean value divided by its standard deviation, and correspond to probability values associated with the normal distribution. (For details of the equations, see pages 185–197, Reference ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar)). Practitioners should be aware that this method of assessing individuals' nutrient intakes has the same sources of error in intake data (e.g. under-and overreporting, inadequate food composition data) as does intake data from group surveys. By definition, the EAR is unknown for nutrients that have an AI, and the approach described above to estimate the degree of confidence that the individual's intake exceeds their requirement cannot be used for nutrients with an AI. However, a statistically based hypothesis testing procedure can be used to compare intake to the AI (Reference ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar), pages 198–200). This equation accounts for the uncertainty in the reported nutrient intake, as it includes the day-to-day (within-person) variation of nutrient intake. If there is a high degree of confidence that an individual's usual intake equals or exceeds the AI after applying this statistical test, it can be concluded that the diet is almost certainly adequate, as the AI is thought to exceed the requirements of almost everyone. If, however, intake falls below the AI, no quantitative (or qualitative) estimate can be made of the probability of nutrient inadequacy, because the requirement is not known. Thus, nutrition professionals should use their judgment, after considering additional types of information about the individual, when interpreting intakes below the AI. For example, without biochemical testing, one would not be able to determine whether an elderly adult with a Vitamin D intake below the AI of 15μg was meeting his requirement or not. However, to remove the possibility of inadequacy, he would be recommended to increase his intake so that it met the AI. Similarly, it is not possible to determine whether a teenager with a calcium intake of 900mg/d (below the AI of 1,300mg/d) is obtaining enough calcium to support optimal rates of bone mineral deposition, and it would be inappropriate to state that she was "deficient." Nevertheless, she would be recommended to increase her intake to meet the AI so that adequacy could be ensured. The UL may be used to assess whether an individual's usual nutrient intake is so high that it poses a potential risk of adverse health effects. A statistical test similar to the one proposed for the AI can be used to determine the level of confidence that usual intake is below the UL (Reference ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar), pages 201–202). For some nutrients, the UL applies only to intake from supplements, fortificants, or medications, while for other nutrients, total intake from all sources is considered (Table 1). Nutrient analysis software programs may eventually be modified to perform the above calculations, with the results included in the output of the analysis. While not as statistically valid, a more pragmatic approach may be used. That is, reported intakes below the EAR very likely need to be improved (because the probability of adequacy is 50% or less); those between the EAR and the RDA probably need to be improved (because the probability of adequacy is less than 97% to 98%); and intakes at or above the RDA can be considered adequate only if they have been observed for a large number of days. Along similar lines, intakes above the AI can be considered adequate. Although the probable adequacy of intakes below the AI cannot be determined, they should likely be improved to meet the AI. Finally, intakes below the UL can be considered at no risk of adverse effects. A hypothetical example of a dietary assessment for a woman aged 73 years is shown in Table 2. This individual reported 3 days of dietary data, and nutrient intake has been calculated for 5 nutrients (riboflavin, folate, calcium, vitamin D, and zinc). The appropriate statistical equations were used to determine the level of confidence that her usual intake meets her requirement (for nutrients with an EAR), or exceeds the AI, and is below the UL. Several points are apparent from this example: 1) Although riboflavin intake was well above the EAR, and even above the RDA of 1.1mg/d, because of day-to-day variation in intake, there is still a 15% probability of inadequacy (probability of inadequacy is equal to 100% minus the probability of adequacy). A similar situation exists for zinc, where her intake equals the RDA of 8mg; 2) Folate intake is well below the EAR, and the confidence of adequacy is only 5%; 3) Calcium does not have an EAR or RDA, but only an AI of 1,200mg/d. The woman's intake was 1,250mg/d, above the AI, and therefore adequate if true long-term intake was accurately captured. However, because of the day-to-day variation in calcium intake, the confidence that intake was above the AI is about 60%; and 4) the opposite situation is illustrated for vitamin D, where intake is below the AI. In this case, no assessment can be made because the distribution of requirements for vitamin D is not known. Evaluation of vitamin D intakes is further complicated by the difficulty of assessing sunlight exposure.Table 2Evaluation of a 73-year-old woman's diet, based on 3 days of intakeNutrientMean intakeEARaEAR=Estimated average requirement.or AIbAI=Adequate intakes.ULcUL=Tolerable upper intake level.Confidence that intake isConfidence that intake is Al)dEquations to estimate the degree of confidence that intake is adequate or meets the AI and is below the UL are available in Appendix B (pages 185–202) of Dietary Reference Intakes. Applications in Dietary Assessment (source: reference (7)). For example, the confidence that this woman's riboflavin intake is adequate is estimated as follows: z=D/SDD, where D=y−EAR=the difference between the individual's reported intake and the EAR, and SDD=[Vr+(Vwithin/n]0.5=the square root of the sum of the variance of the requirement distribution (where variance is SD2 [standard deviation]) plus the within-person day-to-day variance of nutrient intake divided by the number of days for which intake was reported. In this example, D=1.3mg–0.9mg=0.4mg.The coefficient of variation for riboflavin requirements is 10%; thus, the standard deviation is equal to 0.09mg and the variance of the requirement distribution (Vr) is equal to (0.09)2, or 0.0081. The within-person day-to-day variance of intake (Vwithjn) is estimated using data from the Continuing Survey of Food Intake of Individuals (available in Appendix B); for riboflavin intake in older women, the standard deviation is 0.6mg/d, so the variance is 0.62 or 0.36. This is divided by the number of days of records (n=3), to obtain a value of 0.12. SDD is thus the square root of (0.0081+0.12), which equals 0.36. The z score thus is 0.4/0.36=1.1, and the P-value associated with this z score is about 0.85. Thus, we are 85% confident that her intake meets her requirement.Riboflavin (mg)1.30.9 (EAR)Not established∼85%No ULFolate (μg DFEeDFE=Dietary folate equivalent.)200320 (EAR)1,000 (synthetic folate only)∼ 5%She does not use supplements, and has little intake from fortified foods. Confidence would be very high (>98%)Calcium (mg)1,2501,200(AI)2,50060% (reflects confidence that intake >AI, not confidence of adequacy)Very high (>98%)Vitamin D (μg)515 (AI)50Cannot determine; intake is below the AIVery high (>98%)Zinc (mg)86.8 (EAR)4065%Very high (>98%)a EAR=Estimated average requirement.b AI=Adequate intakes.c UL=Tolerable upper intake level.d Equations to estimate the degree of confidence that intake is adequate or meets the AI and is below the UL are available in Appendix B (pages 185–202) of Dietary Reference Intakes. Applications in Dietary Assessment (source: reference (7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar). For example, the confidence that this woman's riboflavin intake is adequate is estimated as follows: z=D/SDD, where D=y−EAR=the difference between the individual's reported intake and the EAR, and SDD=[Vr+(Vwithin/n]0.5=the square root of the sum of the variance of the requirement distribution (where variance is SD2 [standard deviation]) plus the within-person day-to-day variance of nutrient intake divided by the number of days for which intake was reported. In this example, D=1.3mg–0.9mg=0.4mg.The coefficient of variation for riboflavin requirements is 10%; thus, the standard deviation is equal to 0.09mg and the variance of the requirement distribution (Vr) is equal to (0.09)2, or 0.0081. The within-person day-to-day variance of intake (Vwithjn) is estimated using data from the Continuing Survey of Food Intake of Individuals (available in Appendix B); for riboflavin intake in older women, the standard deviation is 0.6mg/d, so the variance is 0.62 or 0.36. This is divided by the number of days of records (n=3), to obtain a value of 0.12. SDD is thus the square root of (0.0081+0.12), which equals 0.36. The z score thus is 0.4/0.36=1.1, and the P-value associated with this z score is about 0.85. Thus, we are 85% confident that her intake meets her requirement.e DFE=Dietary folate equivalent. Open table in a new tab Thus, based on these analyses, the dietary assessment is as follows: Folate intake clearly should be improved, and calcium, riboflavin and zinc intakes could be increased, depending on what level of confidence one wants that intake meets requirement or exceeds the AI. Although the confidence of adequacy cannot be calculated for vitamin D, vitamin D intake is not at a desirable level and should be improved to meet the AI. A similar approach would be used to determine if this woman's intakes are below the UL. For the nutrients in the illustration, ULs have been set for 4: Folate (from supplements and fortificants only), calcium, vitamin D, and zinc. Because none of her intakes is close to the corresponding UL, excessive intake is not a concern. Nutritionists often want to know what proportion of a group has usual intake of a nutrient below their requirement. The level of concern would be quite different if 50% vs 5% of individuals have inadequate intakes. An estimate of the proportion of the group with excessive nutrient intake is also important, as it reflects those at potential risk of adverse effects. There are 2 general methods of assessing the prevalence of inadequate intakes for groups: The probability approach ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar, (10)National Research Council.Nutrient Adequacy. Assessment Using Food Consumption Surveys. National Academy Press, Washington, DC1986Google Scholar) and the EAR cut-point method ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar, (11)Carriquiry A.L. Assessing the prevalence of nutrient inadequacy.Public Health Nutr. 1999; 2: 23-33Google Scholar). The probability approach The probability approach is one way to produce an estimate of the expected proportion of individuals at risk for inadequacy ((7)Institute of Medicine.Dietary Reference Intakes: Applications in Dietary Assessment. National Academy Press, Washington DC2000Google Scholar, (10)National Research Council.Nutrient Adequacy. Assessment Using Food Consumption Surveys. National Academy Press, Washington, DC1986Google Scholar). It is a statistical method that combines the distributions of requirements and intakes for the group. It is necessary that the entire distribution of requirements is known, and that little or no correlation exists between usual intakes and requirements in the group. A probability of inadequacy can be calculated for any level of usual intake. When intake is very low the risk of inadequacy is very high, whereas when intake is very high, the risk of inadequacy is very small. The risk of inadequacy associated with the usual intake of each member of the group is determined, and the prevalence of inadequacy in the group is estimated as the average of the risks to each individual. For example, if the average risk of inadequacy for iron in a large group of women was 12%, one would conclude that 12% did not meet their requirements. The EAR Cut-Point Method The EAR cut-point method is a simpler version of the probability approach that can be applied to most nutrients. Like the probability approach, it can only be used if no correlation exists between usual intakes and requirements. However, it is not necessary to know the exact requirement distribution but only that it is symmetrical around the EAR (this is thought to be true for most nutrients, but is known not to be true for iron); and, that the variance of intakes in the population group is greater than the variance of their requirements (thought to be true in most situations in Canada and the United States). The EAR cut-point method is easier to use because one simply counts how many individuals in the group have usual intakes below the EAR (see Figure 1). That proportion is an estimate of the proportion of individuals in the group with inadequate intakes. It does not, however, identify the individuals whose intakes are below their requirements. Some individuals with usual intakes below the EAR will meet their individual (lower-than-average) requirements. However, provided the assumptions