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.
Nutrient Requirements as ~ Basis for Dietary Evaluation When using nutrient requirements as a basis for dietary evaluation, a number of factors must be considered. For example, there is great variability among similar people as well as different interpretations of adequacy and deficiency. All these factors must be considered in order to use nutrient requirements most effectively. VARLa8ILITY OF NUTRIENT REQUIREMENTS Like all other biological features, nutrient requ~re- ments vary among se ~ ngly similar persons. Although variability generally applies to all nutrients, its nature is known to be specific to certain nutrients. Dietary standards are usually described as recommended dietary allowances, safe levels of intake, or other simi- lar terms. A distinction must be made between these and the term requirements, which is used in this report. Over the past few decades, groups charged with the development of standards for dietary intake recognized the variability of nutrient requirements; nevertheless, they designated a single point in the distribution as the recommended dietary allowance (RDA) (FAO/WHO/UNU, in press; Health and Welfare, Canada, 1983; NRC, 1980). For protein, for example, a sin- gle point was chosen to estimate the dietary intake level adequate to meet the needs of almost all healthy persons in a specified age or sex group. The point was established after examining the data base for each nutrient and then making scientific judgments about the position and nature of the requirement distribution. By definition, the point chosen to meet the needs of almost all persons lies near the upper tail of the requirement distribution (NRC, 1974). 10
11 When there is information on which to base inferences about the actual distribution, as for protein, the mean plus 2 SD has been identif fed as the recommended intake ( FAO/WHO/UNU, -in press; Health and Welfare, Canada, 1983; NRC, 1980). For most nutrients, the distribution has not been explic- itly described, and the relationship between recommended intake and requirement distribution has not been explored in detail. me principle still holds, however, that the recommended intake level generally exceeds the estimated requirements and, hence, the needs of almost all persons. Dietary standards for energy intake are different from those for specific nutrients because the level published for energy is usually the estimated mean requirement, i.e., one-half of the persons are expected to have higher needs and one-half, lower needs (FAO/WHO/UNU, in press; Health and Welfare, Canada, 1983). Some reports provide a range around the median energy requirement (NRC, 1980); others present an estimate of the variance of energy requirements (FAO/WHO/UNU, in press; Health and Welfare, Canada, 1983). Nutrient requirements of specific persons can only be expressed by referring to the probability (FAO/WHO/UNU, in press) or likelihood that each level of observed intake is inadequate. In a probability approach, therefore, the underlying distribution of requirements among similar persons must be recognized. This is in contrast to using a fixed cutoff point delineating inadequate from adequate nutrient intake, which fails to recognize those persons whose intake and requirement may both be below the cutoff point. If the cutoff point is set below the RDA, both intake and requirement of some persons may be higher than the cutoff point. The period during which a specified requirement must be met is seldom defined (FAO/WHO/UNU, in press; Health and Welfare, Canada, 1983). Requirement estimates are usually related more generally to levels of usual or habitual intake (FAO/WHO/UNU, in press; Health and Welfare, Canada, 1983; ARC, 1980). They do not refer to intake on a particular day, unless that is a reliable measure of the usual intake. Nutrient requirements as used for the analysis in this report consist of a aistri- bution of usual dietary intakes required to maintain an adequate or acceptable physiological or nutritional state. For a few nutrients, requirement information simply is not available, and meaningful analysis of the adequacy
12 of dietary intake must await the development of further knowledge of requirement distributions. For others, there is some information, but its precision may be low. Nevertheless, coupling available information about requirement with other information about factors expected to affect requirement can permit the development of informed judgments about requirements for various age and sex groups. As a result, it should be feasible to use a probability approach to improve present inferences about the adequa Q of dietary intake. It is important that priority be given to the nutrients that are most likely to present public pow cy problems in the United States. For a few nutrients and age groups, better information about requirements appears to be available, and one can be more confident in their application. On the whole, however, there is a clear need for research on nutrient requirements. It is important that priority be given to shone nutrients which are most likely to present public policy problems in the United States. Such refinement should permit the development of improved statistical approaches to survey interpretation through the use of information about nutrient requirements. Until exact information about requirements is available, the resultant inferences about prevalence must be considered imprecise, although the probability approach is superior to other possible methods. Information about mean requirements and characteristics of their distributions for some but not all nutrients may be found in the reports of committees charged with devel- oping recommended intakes. However, the development and presentation of information needed for all nutrients have not been included in the mandate to such committees. This does not mean that the info` - ation is unavailable but, rather, that it may be necessary to undertake a special effort to examine the literature and develop the required data bare. m e present subcommittee die not attempt such a search and notes that skills and experience not repre- sented in its membership would be needed to perform the requisite task. LEVELS OF REQUIREMENT Multiple definitions of adequacy are possible ( ee ge r the prevention of clinical deficiency symptoms, the main-
13 tenance of specified levels of the nutrients or their metabolites in tissues, the maintenance of enzyme activity at specified levels), and each of these could be associ- ated with a different dietary requirement. Thus, it is possible to establish a family of requirement curves marking different definitions of adequacy. Estimates of average requirement based on these dif- ferent criteria can be derived from the early nutrition literature. The earliest marker for nutritional adequacy was the prevention of clinically detectable signs of mal- function. Estimates based on deficiency avoidance were provided in some early dietary standards, along with recommended intakes for improved nutrition, e.g., in the 1963 Dietary Standard for Canada (Committee on Revision of the Canadian Dietary Standard, 1964). Recent RDA reports (NRC, 1980) provide some of the information needed for the proposed approach. Such information can also be found in reports issued by FAO and WHO (e.g., FAD/WHO, 1961, 1970). Throughout the world, dietary standards and recommended intakes are based on basic philosophies that differ in detail ( IONS, 1983a,b) but have ~ similar goal of estab- lishing levels of intake that will maintain a state of nutriture beyond the mere prevention of clinical deficiency disease. For example, adequate iron intakes are regarded as those that maintain reasonable iron stores rather than those that merely stabilize mild anemia or maintain hemo- globin at physiologically normal levels (FAD/WHO, 1970; Health and Welfare Canada, 1983; NRC, 1980). Similarly, vitamin C requirements have been set at a level that is sufficient to establish and maintain metabolic pools (Health and Welfare Canada, 1983; NRC, 1980) or to main- tain tissue levels (FAD/WHO, 1970) rather than just to prevent scurvy. Requirements for other nutrients are determined in analogous ways. The criteria selected for the same nutrient often vary between reports, even within the some country (e.g., ascorbic acid in Canada). Levels of intake and requirement estimates lower than those given in many recently published dietary standards appear to be consistent with the absence of clinical signs of ill health. Thus, some definitions of requirement may be more desirable that others, depending on the purposes of the dietary assessment. Interpretations of dietary intake data in relation to estimated requirements require consideration of the particular biochemical, physical,
14 clinical, or functional criteria that were used to establish the requirement. A multilevel assessment procedure can be used by devel- oping a series of requirement distribution estimates, each referring to a defined criterion of adequacy. By using the several requirements, one can calculate a series of estimates of the prevalence of inadequate intakes. A1- though the RDA reports may not provide the appropriate information for determining such a family of requirement estimates, this absence does not mean that the information is not available. Such information may not have been pre- sented because those reports are intended primarily for use in developing a single requirement to meet the needs of all healthy individuals. FIXED CUTOFF POrNTS It is a common practice to use fixed cutoff points to estimate the adequacy of nutrient intake. In this method, estimates of the prevalence of inadequate intake have been based on a fixed proportion of the RDA. The rationale has been based on the recognition that the RDAs, designed to include virtually all healthy individuals, have included margins of safety--often generous margins. Hence, if applied as a criterion, an RDA would clearly lead to an overestimate of the prevalence of inadequacy. Therefore, the proportions selected as cutoffs have varied--somet~mes two-thirds, sometimes three-quarters, and sometimes 70~. There has not been a clear rationale for the selection of the level. The use of fixed cutoff points is conceptually similar to population-based screening for unrecognized disease (e.g., Rogan and Gladen, 1978). The well-known terms used to describe problems of medical screening are similar to those encountered with the use of cutoff points (Habicht, 1980). Thus, both the teems and the screening approach have been used to examine the fixed cutoff point method. Regardless of the cutoff point selected, some persons who meet their nutrient requirement will be identified as having inadequate intake. Conversely, some who do not meet their requirement will be identified as adequately nourished. The term sensitivity is applied to the ability of a test to detect truly affected individuals; specifi-
15 city refers to the ability of a test to identify truly unaffected individuals. Misclassification occurs when people are designated not at risk when they are truly affected by the condition (false negatives) or at risk when they are actually unaffected (false positives). statistical decision~making, a similar concept is used, with sensitivity and specificity corresponding to Type I (a ) and Type II (D ) errors in hypothesis testing. In Figure 3-l, the distribution of those who truly fail to meet their requirement and the distribution of those who truly meet their requirement are plotted. For the purpose of illustration, it is assumed that persons can be cla~si- fied with an absolutely accurate test. AREA OF OVER LAP I NAD EQUATE ADEQUATE INTAKE OF NUTR I ENT X FIGURE 3-1. The distributions of people who truly fail to meet their requirement (inadequate) and those who truly meet it (adequate) for a hypothetical nutrient X.
16 This figure is useful in gaining an understanding of the implications of the fixed cutoff point approach. Presumably, a cutoff point would be selected somewhere in the area of.overlap between the distribution of truly. adequate and truly inadequate intake. For this tube done, the sensitivity and specificity of the cutoff point would have to be detenmlned (Habicht et al., 1982). Brownie and Habicht (1984) have developed a strategy for selecting the optimal cutoff point under certain conditions. To most important conclusion from these considerations {Brownie and Habicht, 1984) is that the choice of the optimal cutoff point to estimate prevalence or changer in prevalence depends upon a rather exact estimate of the prevalence being sought--an impossibility. Estimates of prevalence using less than optimal cutoff points can be corrected (Rogan and Gladen, 1978) by taking the sensitiv- ity and specificity of the cutoff point into account . Although this approach is theoretically possible, no such data are presently available and, more importantly, it is probably impracticable to acquire such data with the pre- cision required. It i. not rational to select the cutoff as a proportion of the highest requirement in the popula- tion. When fixed cutoff points are used without these correc- tions, estimation of the prevalence of inadequate intake is in error and the magnitude, the extent, and even the direction of the error cannot be estimated. Recogniz- ing the drawbacks in the use of fixed cutoff points, the subcommittee concluded that a different approach was required to analyze the adequacy of dietary intake. The probability approach proposed in this report avoids the limitations of the fixed cutoff points.
