Click for next page ( 30


The National Academies | 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 29
Data Base Requirements The proposed monitoring and evaluation system requires the use of sources of data on food composition, nutrient requirements, aggregate food disap- pearance, and health. There is considerable variation in the adequacy of these data sources. The information may be incomplete, inaccurate, and out of date. It is necessary, however, to deal with the data that are available as the cost of collecting new data specifically for use in the proposed system would be prohibitive. There must be a continuing effort to upgrade these basic data sources in terms of quantity and quality of information and ease of access for use. This chapter discusses the sources required and many of the limitations that exist in the information they currently provide. FOOD COMPOSITION DATA The U.S. Department of Agriculture maintains the most extensive bank of data on food composition. The quantity and quality of the data are variable and in the case of many of the less studied nutrients are not adequate for an accurate calculation of nutrient intake. Data for the nutrients for which need has long been established are rela- tively complete for food commodities. Data are much less complete for commercially prepared items, although much more information is becoming available as a result of nutrition labeling of many food products. For a number of the nutrients for which dietary requirements have more recently been identified, analytical data are fragmentary. For some nutrients, in- cluding several of the trace elements, content in food varies extensively with geographic region and growing conditions. Therefore, meaningful analyti- 29

OCR for page 29
30 ASSESSING CHANGING FOOD CONSUMPTION PATTERNS cat descriptions may be difficult. However, when interest is in usual intake and hence usual composition, sample-to-sample variation in composition or small differences attributable to various processing or preparation practices may not be of concern. For some studies, data on the composition of particular brands; changes caused by processing, packaging, and storage; and variations in composition due to changes in selection of ingredients (such as type of fat) may be required. In other studies, information on the composition of the water added to or used in preparation of the food items may be desirable. This type of information is very limited. Conventional tables of food composition do not provide information about amounts of nonnutritive food additives or food contaminants present. If concern exists as to potential toxicity of any nonnutritive components or if dietary data are to be applied for prediction of risk of excessive intake, an increased data base describing the levels or potential levels in food is needed. Information on additive content in foods is currently available from GRAS (additives Generally Regarded As Safe) survey data prepared by the Committee on GRAS List Survey, Phase III, Food and Nutrition Board, the Food and Drug Administration, and other sources. As the needs of each study are different, decisions on the desirable detail of description of foods must be made at the time of data collection. It is, however, necessary to keep in mind potential future uses for the data. The incomplete nature of available data banks currently limits the attain- ment of accurate and useful nutrient intake data. Current data bases must be expanded to include some information on formulated, processed, and ethnic foods as well as mass-produced fast food items and products prepared for institutional use. More data on a wider number of nutrients for many foods should be provided. The ideal data base would (1) be current, reliable, and valid; (2) be responsive to changes in the food supply; (3) contain information on all the nutrients of interest; (4) have complete data (unavailable data should be extrapolated until analytical values are obtained); (5) be expandable as new data become available; (6) reflect differences associated with brands; and (7) be in a physical form that facilitates coding and analysis. Coding System Concurrent with the further development of data banks for nutrient analysis must be the development of a coding system that will allow maximum flexibility in analyzing data and identifying and tracking trends in food consumption patterns. Coding by the various food consumption data banks must be standardized so information from several sources can be efficiently

OCR for page 29
Data Base Requirements and easily combined. The coding system must be designed to mesh with health data systems. Simplified analyses based on dietary scores or the use of indicator nutrients warrant further investigation to enhance the usefulness of the data for various purposes. 31 Nutrient Requirement Estimates It is customary in evaluating data on dietary intake to compare calculated nutrient intakes to the Food and Nutrition Board's Recommended Dietary Allowances (RDA). This practice leads to misinterpretation of nutrient intake data. Intake data may be interpreted more accurately by a bivariate distribu- tion approach that requires a description of the distribution of nutrient re- quirements (median plus variance). For many nutrients, these data are avail- able. For other nutrients, appropriate informed judgment can be made about the distribution of requirements and, thus, the extent of variance. For a few nutrients, there are insufficient data to permit even an informed judgment, and interpretation of observed intake is not possible. It will be necessary to compile existing data and prepare descriptions of the distributions of nutrient requirements. Therefore, data on average re- quirements and their variances should be collected and published. At the same time it will be necessary to aggregate data on nutrient requirements for different levels of nutritional status if an assessment of both prevalence and severity of nutritional risk is to be made. Such data are available for many nutrients. For example, the average requirement of thiamin to prevent neurological manifestations of deficiency is estimated to be about 0.2~0.22 mg/1,000 kcal (FAD/WHO, 1967~. The average requirement for physiologic saturation of tissue needs, judged by the pattern of urinary excretion, is about 0.33 mg/1, 000 kcal with a coefficient of variation of requirement of 10 percent of the mean (FAD/WHO, 19671. Data are now available relating thiamin intake to erythrocyte transketolase activity. These types of informa- tion provide estimates of requirements for different levels of nutritional status and can be used in the bivariate distribution analysis. Thus, in at- tempting to relate dietary status to health status, it is necessary to express requirement in terms of purpose: e.g., prevention of clinical deficiency, maintenance of a level of enzyme activity, etc. To what extent all of these criteria relate to health status remains to be determined. AGGREGATE DATA A number of ''food use" government and commercial data systems cur- rently collect data on a continuing or periodic basis. These data series can be classified by the degree of aggregation of the data that they present and,

OCR for page 29
32 ASSESSING CHANGING FOOD CONSUMPTION PATTERNS therefore, by the applications to which they can be put. Avenues for their effective use in conjunction with individual data should be identified. Production and shipment data are usually collected from the producers or the distributors of the raw commodities of the processed foods. The primary source of agricultural output data is the U.S. Department of Agriculture. Other sources include the Department of Commerce, the Bureau of Census, various growers or manufacturer associations, the state agricultural depart- ments, and the alcoholic beverage production and tax control commissions of the states and the federal government. Most of these data are published in standardized time series. Warehouse withdrawal data are collected from the data processing records of large warehouses of supermarket chains, or food and grocery product brokers. The data are aggregated and reported for a large number of markets or cities. Information is presented on individual brands, by type of product, by size of package, and for canned, packaged, refrigerated, and frozen foods. Most fresh fruits and vegetables, fresh meat, and similar items that do not move through the chain warehouses or the computerized inventory management systems of the stores are not included. The data are usually reported in raw unprojected form, covering sales in each market for the cooperating chains. The system in many areas accounts for 50 to 75 percent of all foods sold. The information is usually sold to food manufacturers and advertisers on a product-class basis for confidential use and not for resale or publication. Standard reports are usually issued, but customized processing is possible. Audits of in-store movement of food items are made by a number of commercial services. These services collect their information from a selected sample of supermarkets throughout the United States. A physical audit of the quantities of various products on the shelf and delivered into each store is made bimonthly. This information is then projected from the sample to total market levels and aggregated into regional subtotals and a U.S. total. Per-capita availability can be determined from these audits and warehouse withdrawal data. Aggregated consumer food purchase and usage data are collected by consumer purchase panels. These syndicated data services collect informa- tion via the mail from continuing nationwide samples of households and project these data to regional and U.S. totals. Diaries are collected from consumers weekly or monthly. Occasionally, custom samples are set up in selected cities for test marketing purposes, operated for a period of from t/2 to 2 years, and then disbanded. Consumer records include mail-order, door-to-door, and specialty store purchases in addition to grocery store purchases. Reports are issued to subscribers on a monthly or quarterly basis. Reports include data on the total quantity of the product bought, the average

OCR for page 29
Data Base Requirements 33 price paid, the percent of households by which the product was bought, and the average quantity bought per buying household. Data on consumption by individuals within the household are not provided. A "Menu Census Service" provides detailed information on food prep- aration and on all food consumed at and away from home from regional or nationwide samples. Data are collected by mail following maintenance of consecutive days of diaries of individual food intake. Data for each food product are summarized and reported quarterly and annually. Usage of each food product is cross-tabulated by the demographic characteristics of the household and the individual eaters. Persons living away from home or in institutions (i.e., schools, hospitals, armed forces) are not included in con- sumer purchase panels or menu census services. Food consumption surveys of selected products or sources are conducted by various companies. The data are collected by mail from consumers on a continuing syndicated basis. Information on, for example, the consumption of all foods at restaurants, or the consumption of all beverages at home or away from home, is summarized into standard reports designed to satisfy selected information needs of a given industry, such as soft drink manufac- turers or restaurant operators. HEALTH STATUS DATA BASES Indicators of health status are objective measures of the state of health of a population. Health status indicators with a nutritional status component fall into several categories comprising the continuum from "perfect'' health to death from a variety of illnesses and disease. The first category, morbidity, includes specific diseases with an apparent nutritional component and ones where evaluation of nutritional status would provide an indication of poten- tial risk of the disease. Among these diseases, but not limited to them, are noninsulin-dependent diabetes, atherosclerosis, hypertension, cirrhosis of the liver, gout, osteoporosis, obesity, coronary heart disease, dental dis- ease, anemia, cerbrovascular disease, and, possibly, cancer. Of less impor- tance, but still significant, would be others, including gastrointestinal disor- ders, respiratory disease, a variety of conditions subsumed within aging, nutrient inadequacy and toxicity, chance contaminations, and food-borne illnesses. The next group of health status indicators may be drawn from mortality statistics. Of particular nutritional importance are infant mortality, perinatal mortality, and age-specific mortality rates indicative of longevity. The third group are those indicators that are characteristics of the popula- tion and are nutrition related. They include the incidence of low birth weight, the age at menarche, the outcome of early adolescent pregnancies, \

OCR for page 29
34 ASSESSING CHANGING FOOD CONSUMPTION PATTERNS birth defects, obesity, stunted growth, and behavioral or functional condi- tions (including those related to learning and to performance). Nutritional status indicators are considered a category of health status indicators. They include the various types of measures by which nutritional status is determined: (1) biochemical (including blood and urine samples), (2) anthropometric, and (3) clinical (including behavioral measures). The data derived from such measurements are used in two ways: to identify individuals at risk who may then be recalled for follow-up examinations, and to identify groups that are at risk and design appropriate programs to alleviate the risk. Nutritional status determinations are complex, time- consuming, and expensive. The methodology of relating food consumption to information on health, including nutrition status indicators, encompasses a range of data-gathering, coding, and analytical procedures. The general types of food consumption information desired for useful analyses can be divided into three groups: 1. Feeding behavior, which includes the pattern of meals within a family, the type of infant feeding, the frequency of eating, the proportion and location of meals eaten outside the home, snacking patterns, food prepara- tion, and many others; 2. Food groupings, which can be as finely divided as the investigator thinks necessary; 3. Food components: saturated and unsaturated fats, cholesterol, sugars, other carbohydrates, fiber, animal and vegetable proteins, specific nutri- ents, additives and contaminants. Some aspects of a system for relating food consumption to health status do exist, but the linkages can seldom be made due to inconsistencies in methods and in populations studied. Existing information is often not fully used, and studies are not often compatible with existing information sources or with other new studies. Monitoring for the conditions in populations can be at several levels and may utilize already existing health or individual data or may require field survey methods. Table 3 indicates sources of information currently available on defined populations of suitable size to detect small area or regional differences. The sources of information are varied, at present are not inte- grated, may be redundant, and have varying degrees of validity. Develop- ment of these information sources into a complementary, if not single, information system should be undertaken. Medical record linkage as a method to utilize the diverse sources of information should be developed. Such development will require that cur- rent concern about privacy and confidentiality of personal records be satis- factorily resolved in favor of the ability to carry out epidemiologic studies.

OCR for page 29
Data Base Requirements TABLE 3 Types and Sources of Health Status Data Vital Statistics Birth records National Center for Health Statistics Death records National Center for Health Statistics Linked birth and death records Public health agencies Special studies Medical Records Hospital discharge data Commission on Professional and Hospital Activities Professional Standards Review Organization Insurance data banks Tumor registry National data set Local registries Physician records Group practices National Disease and Therapeutic Index Insurance Claims Institutions 35 Health Screening Program Voluntary agencies Public health agencies Disease Registries Voluntary agencies Public health agencies School Health Records Population Surveys National Health Survey Special studies Epidemic Intelligence Centers for Disease Control Public health agencies Appropriate safeguards against invasion of privacy and breach of confiden- tiality must be built into the system in such a way that linkage is not proscribed. At present there is a vast amount of information on mortality and morbid- ity in existing computer tapes. Most of the types of data shown in Table 3 are currently collected and stored in accessible form. There is a need for new ideas and methods to utilize these data sources in combination with food consumption data in defined populations. For example, fetal nutritional experience seems to offer an excellent prospect for studying chronic disease within a realistic time cycle. Surveillance of various indices of infant birth and mortality, linked to birth certificate data, and study of these data in a comprehensive way could shorten the long observation periods needed to collect mortality and morbidity perspectively in adult populations (Haw- thorne, 198 1~. For purposes of illustration, certain selected conditions considered at present to be strongly associated with food consumption are listed in Table 4. The types and sources of health data that could be utilized in population

OCR for page 29
36 c, c, ~ ~ a.~ ._ ~ c., c~ c, u' 1 0 =0 ~v v'= o _ c~ o o o 3 au 4 - ._ c~ o c~ c~ 4_ ct o ct .= cn ct v: - c~ x x x x x x x x x x x x u, c, ._ _ ~ c~ (L) t4 cn .= ~ c: ct c~ t4 <., 0 ~Q ~ _ c~ ct ~3 c., c~ c~ - ~ ~ x x .~ ce ~ u, x x x x x x x x x x x x x x x x x x x x x x x x s ~ ~ e E 0 ~ ~, D ~ _ e <0~ 0 ~ _ _ l 1 | |

OCR for page 29
Date Base Requirements studies are indicated in the table. While the sources vary in the degree of validity of the data contained in them, they form a rich pool of information for assessment of the health of populations. The many health data files that exist in the United States should be cataloged and those that may provide data for relating health status to food consumption patterns identified. It is recommended that a special work group be established to evaluate current health statistical information and recommend a system whereby these data can be collated in a manner that will permit identifying population segments showing unusual health pat- terns. Once identified, the integration of this information with the proposed system for monitoring food consumption patterns should be developed. In this manner studies can be targeted to specifically identified groups either through special studies or as a part of the ongoing monitoring program. If information on food consumption and nutritional status were obtained in a compatible format, this system could be used to explore associations between food consumption and health status, while taking into account the modifying effect of other factors. There is a current need in the United States for specific information on food consumption and health status (obtained at the same time and from the same individuals) in order to identify specific associations. There is an additional need to obtain this type of information in a consistent manner, on the same individual as often as possible, and at repeated time intervals in order to detect trends and changes over time. 37