Click for next page ( 7

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 6
Relating Food Consumption Data and Nutritional Status Data The value of food consumption data in relation to public health is realized when the data are processed to yield estimates of nutrient intake and of exposure to food additives and contaminants. It is not possible to assess nutritional status from dietary data alone. It is possible, however, to provide an estimate of the prevalence of individuals within a population group with intakes below requirements, and, with current knowledge of human re- quirements, it is possible to assess the individual's probability of inadequate status based on intake of some (but not all) nutrients. It is also possible to consider the potential risk of excessive intake of nutrients and of natural or added food components, i.e., food ingredients and additives, if the associa- tion between level of intake and risk of toxicity is known. Health status indicators believed to be associated with food or nutrient intake can be identified if current dietary data can be linked to existing health data. Through appropriate studies of the food consumption patterns of individuals in a group, it would then be feasible to predict the prevalence of individuals who may have increased risk to their health from a particular pattern of food or nutrient intake. Environmental variables in addition to diet (such as endemic infectious disease or unusual stress) may increase the prevalence of nutritional in- adequacy from that indicated by nutrient intake. In the United States, these other factors, with the possible exception of recurrent infections in children, probably do not generally contribute to variability of requirements, and therefore a valid estimate of risk can be generated by appropriate collection and analysis of dietary data. The following discussion of an approach to the data collection system 6

OCR for page 6
Relating Food Consumption Data and Nutritional Status Data 7 focuses primarily on application of the information to a prediction of nutri- tional status. The same system could be used for assessing toxicological risk and, with some modifications that will be discussed later, for relating food consumption to various aspects of health (growth, longevity, chronic dis- eases, etc.~. The linkage between dietary intake and nutritional status is portrayed in Figure 1. Interest is in the usual situation on a particular day. Thus, the data essential to this system are usual intake of food, usual composition of foods, usual bioavailability of nutrients, and usual nutrient requirement of the individual. There is no precise definition of the time span represented by "usual," but for purpose of discussion it may be taken as the average value persisting over a period of weeks rather than a day or two. The statistical approach to analysis and to the prediction of the prevalence of individuals with usual intakes below their actual requirements has been identified and discussed by Lorstad (1971~. It has been applied in a predic- tion of the effects of iron fortification on the prevalence of inadequate intakes among menstruating women (Swiss and Beaton, 1974) and in the identification of protein:energy ratios associated with predetermined pre- valences of individuals having inadequate intakes (Beaton and Swiss, 1974~. This statistical approach is based on the bivariate distribution of intakes and of requirements among individuals in the population. It is applicable to a dietary analysis system designed to monitor conditions in the United States. Analyses based upon the bivariate distribution require knowledge of (1) the distribution of usual intakes (mean, variability), (2) the distribution of usual requirements (mean, variability), and (3) the correlation between in- take and requirement among individuals (FAD/WHO, 1973~. For many nutri- ents, requirements are at first-order approximations; and for most nutrients, but not energy, the correlation between intake and requirement appears to be very low and may be ignored provided that care is taken to express the data in a manner that avoids the effect of common variables such as body size, energy intake, etc. In the case of energy, intake and requirement of the individual are generally matched relatively closely, probably through physiologic regula- tory mechanisms. Thus, a high correlation between intake and requirement would be expected. For various micronutrients, there is little or no reason to expect a correlation between intake and requirement unless both relate to a third variable. For example, thiamin intake is likely to relate to total energy intake and thiamin requirement is believed to be related to energy utilization. If intake and requirement are both expressed as milligrams per day, a spurious correlation may appear. This error can be avoided by expressing intake in a manner (such as mg/l,OOO kcal), which controls for the confounding vari- able.

OCR for page 6
8 o ~ as ~ ~ ' o · ~ ~ ,~ ~ . 4- LU ~ ._ a) _ o ~ ~ ~ _ a ~ ~ a) ~ a, ~ CC _ 4< ~ o ~o 4 - E t',' o a' ._ C) '~ U.J Q a) Q ~ ._ ~ ~ J o ~ ·— a.) C~ ~ _ .— a) a.) tr m 0 C=~c O ~ ._ OCR for page 6
Relating Food Consumption Data and Nutritional Status Data 9 Attention must focus on the need for information about the distribution of intakes and of requirements. This can be considered in the framework of the relationship portrayed in Figure 1. Data on the observed usual intakes of food by the population may be translated to nutrient intake with the use of existing food composition data. A data bank, an appropriate coding system, and a computer program are required and could possibly be based upon one or more of the several systems now existing in the United States. However, for some nutrients, the nature of the food ingested and the total diet affect bioavailability. Reasonable information about the relationship between na- ture of the diet and relative bioavailability is available for a number of these nutrients. Information about the individual's actual requirement is needed to deter- mine the adequacy of intake relative to requirement. Clearly, such informa- tion is not available. With respect to the individual, the best estimate that can be made is the relative probability that the intake does or does not meet the individual's actual requirement. At a population level, using the bivariate distribution approach, an estimate can be made of the number of individuals (but not which individuals) with intakes below their actual re- quirements. The required data base for such estimates includes both the mean requirement for that category of individual and the variability of requirements among similar individuals. The Recommended Dietary A llow- ances (Food and Nutrition Board, 1980) is not an appropriate reference criteria* for these determinations, but, with current information, reasonable assumptions can be made about the distribution pattern for the requirements of a number of nutrients. Simplified analyses based on dietary scores or the use of indicator nutrients warrant further investigation to enhance the use- fulness of the data for various purposes. The preceding discussion has been related to the prediction of the preva- lence of intakes below actual requirements of healthy individuals. The pre- sent information base relating "requirement" and level of nutritional status is fragmentary. It is germane to recognize that neither the meaning of "re- quirement," definition of the level of nutritional health to which the re- quirement applies, nor consideration of the degree of inadequacy of intake have been addressed in this discussion. Information about the levels of intake required to maintain different levels of nutritional status is necessary if intakes are to be linked with observed severity or frequency of inadequate nutritional status. Lorstad (1974) has *By definition, the "recommended dietary allowances" (RDA) as published by the Food and Nutrition Board are intended to be sufficiently high to cover the known nutritional needs of practically all healthy people. Therefore, RDA (except for energy) are estimated to exceed the requirements of most individuals. Intakes below the RDA are not necessarily inadequate, but the risk of inadequacy increases to the extent that intake is less than the level recommended as safe.

OCR for page 6
10 ASSESSING CHANGING FOOD CONSUMPTION PATTERNS pointed out that with such information the same format of bivariate distribu- tion analysis, with different requirement figures, could offer a series of predictions of the prevalence of different levels of nutritional status. It would seem that, for many nutrients, there is now sufficient informa- tion to apply the bivariate system approach to the interpretation of dietary data. The information currently available may be imprecise and could cer- tainly be improved with additional research. Nevertheless, for many nu- trients, the quality of data available seems adequate to justify the assump- tions about bioavailability, average requirements, distribution of require- ments, and even requirements for different levels of nutritional status that are necessary for the statistical approach to analysis. Unfortunately, for some other nutrients there are major gaps in the information base that place analogous limitations on the above and other approaches to dietary data interpretation (Beaton et al., 19791. It is emphasized again that this approach does not make provision for effects of recurrent infection, of unusual environmental stress, of genetic abnormalities, of adaptabilities of host to excesses or deficiencies, or even of nutrient imbalances other than as they are built into estimates of bioavailability or requirement. This is a limitation of the model but one that should not preclude its use in the United States in a nutrition monitoring system. Dietary data may also be used in the assessment of the probable risk associated with excessive, rather than inadequate, intakes of natural or added constituents of foods. The approach and data requirements are analogous. In this case, what will be needed are data pertaining to the average level of intake at which manifestations of harm are suspected (analogous to average requirements of a nutrient) and the possible variability of this sensitivity among individuals. As before, the probable prevalence of various grades/severities of toxicity (analogous to various levels of nutri- tional status) can be estimated if information is available on the relationship between level of intake and severity of the detrimental effect. The bivariate analysis of dietary data, then, permits an examination of the population risks associated with both low and high intakes. It also per- mits an examination of the probable effects of a change in food consump- tion, such as a proposed food fortification policy or a proposed limit for a food additive, on intake and population risk. Implementation of this ap- proach to analysis poses certain study design requirements and necessitates the compilation of certain data bases. These are discussed in some detail in later sections of the report. As previously mentioned, predicting health status of individuals from dietary data alone is not possible. The role of certain foods or food compo- nents in the etiology of such chronic diseases as diabetes, coronary heart

OCR for page 6
Relating Food Consumption Data and Nutritional Status Data 11 disease, atherosclerosis, hypertension, and cancer is not well documented (Ahrens and Conner, 1979~. In addition, there appears to be a wide variation in the susceptibility of individuals within a population to these diseases. Knowledge concerning the relationship of food consumption patterns to susceptibility to chronic disease can better be determined from a careful examination of the food consumption patterns of population groups with known unusual incidences of these chronic diseases. Care must be taken to control adequately for environmental and genetic factors. A number of the health-data bases in existence could be used in conjunc- tion with food intake data to develop an appropriate monitoring system. Data from current health-data bases would need to be identified, sum- marized, and evaluated in those population subgroups used for food intake data analysis. Details are discussed in a later section of the report. .