Click for next page ( 66


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 65
6 RATIONALE AND RECOMMENDATIONS FOR EARLY ACTION INTRODUCTION As indicated in Chapter 5, improvement in the data collection, processing, and reporting methods of the Nationwide Food Consumption Surveys (NFCS) and the National Health and Nutrition Examination Survey (NHANES) can be advanta- geous both in the uses of the core (NFCS and NHANES3 survey data bases inde- pendently and in their conjoint use, e.g., in a wider system of nationwide food and nutrition monitoring. Implementation of the recommendations pre- sented here and in Chapter 7 will permit the surveys to be used more widely and more effectively. The Committee has evaluated sugges Lions from data us er s regarding possible modifications of the survey instruments. Survey design modifications that increase the specificity of the data obtained (e.g., incorporation of food- item brand names, identification of both physical and logistic impediments to food intake, and sub jects' -use of medications that might interact with nutrients ~ were suggested during workshop and symposium discussions. Although these suggestions would be advantageous to some users, the Committee's recom- mendations are based on giving highest priority to changes that will benefit the data user community at large . The Committee also cons idered potential cost constraints in developing its recommendations. Recommendations for early action are discussed in this chapter and those expected to require longer-term implementation are discussed in the next chapter. The recommendations of the Committee are related primarily to modifica- tions in survey design, food intake methods, and information on the compo- sition of foods, that is, information on what people are eating and infor- mation on who is eating that food. More specifically, the recommendations on food intake survey methods concern continuous collection, processing, and reporting of survey data, continued collection of replicated data, and similarity in methods of data collection. The recommendations on information on the composition of foods concern expans ion of the qualitative and quan- titative data on the composition of foods, use of standardized tables on the composition of foods, and use of standardized codes for reporting the food ingested. The recommendations on population data concern sampling similarity and the use of compatible sampling frames for selecting survey subjects. Implementation of these recommendations will be most helpful to those us ing the survey data to assess dietary intake of various food components of biologic importance. In addition, these changes can also increase the utility

OCR for page 65
66 of survey data in designing or modifying feeding programs for population groups; developing nutrition education~programs, nutrition guidance, and dietary recommendation materials; and developing new programs in agricultural production and food processing to yield products useful in food and dietary modifications . SURVEY PURPOSES AS GUIDES The purposes of the separate surveys provide a useful perspective for con- sidering agency or interagency coordination of Committee recommendations. The NFCS serves as the major national focus for collection of food con- sumption and dietary intake data that address the question of who is eating w _ . All other uses of the data are derived from the answers to these funda- mental questions . Furthermore, it is necessary that NFCS data and reports be provided to users promptly and in a format that fosters compatibility and com- parability with other data bases. The purposes of the NHANES are related to measurements of the prevalence of disease states and health conditions. The dietary component of the NHANES is complementary to its fundamental purposes. The role of the NHANES data is critical for emerging uses of survey data, particularly uses concerned with epidemiologic and public-health applications. These uses require that the NHANES data base be comparable and compatible with the NFCS and that infor- mation be reported in a timely fashion. Some of the Committee recommendations discussed in this chapter are related more to dietary intake data than to data on the prevalence of health conditions and disease states. Given the different purposes of the surveys, the Committee believes that the U.S. Department of Agriculture (USDA) is the likely agency to.lead in implementing several of these recommendations. The Committee recommends that USDA serve as the lead federal agency for survey standardizations dealing primarily with the food-related aspects of federally funded food consumption and dietary intake surveys. Where health conditions and disease states would be a major focus for survey modifications, the Com- mittee recommends thee the Department of Health and Human Services (0HHS) serve as the lead federal agency. Given the different purposes and data collection methods of the surveys, recommendations are not always the same for both surveys. The Committee's recommendations are based on consideration not only of both the independent ant the conjoint current uses of the survey data, but of emerging uses for data. Some recommendations on survey modifications that would foster more timely collection and reporting of the survey results are intended to serve future data users better without compromis ing currently essential uses .

OCR for page 65
RECOMMENDATIONS ON SURVEY DESIGN Data from the NFCS and the NHANES have many and varied uses. Given the variety of current and emerging uses for these data, it is difficult to de- velop a single set of recommendations that will serve the needs of all data us ers . CONTINUOUS EVALUATION PROCESS While accepting the realities of continuing independent but mutually com- patible surveys, the Committee recognized the merit of a continuing consider- ation of the desired features of a composite single survey. As discussed in Chapter 5, the Committee developed a systematic approach to examining issues of definition of data needed and of requisite precision and reliability of da ta ~ Append ix A) . The Committee found its approach useful and recommends that it be made an on-going process in the continuing rev few and evaluation of the separate surveys, their purposes and their design. CONTINUOUS SURVEYS A consistent need of all data users is the timely reporting of survey information. Thus, implementation of changes in food intake methods that would provide for timely reporting of data and other survey results would be beneficial to all users of food consumption and dietary intake surrey data. The core surveys have his torical ly been conducted at intervals . NHANES I and II were conducted in 1971-1974 and 1976-1980 and the two mos t recent NFCS more than 10 years apart (1964 and 1977-1978~. Data collection for the 1977-1978 NFCS took a year. NFCS methods re uire data collection from a large, representative sample of U.S. households and persons in those households. The large volume of data collected results in a large volume of data to be coded and processed. In the view of the Com- mittee, a fundamental barrier to timeliness in reporting survey data is the size of the data collection -- about 15,000 households and about 30,000 indi- vidual~. This collection phase has the potential for "input overload" with a backlog of data in the initial survey process . Redes ign of the Individual Dietary Intake component of the NFCS to collect and report data continuous ly would help to overcome this barrier to timely data collection and reporting. USDA has proposed a continuous system to monitor and report on the food consumption and dietary intake of women aged 20-50 and their children; other high-risk groups, such as the elderly, might be added to the panel. This

OCR for page 65
68 system of data collection would provide the data base for a series of annual reports . But, for ache same cos t, the sample of persons interviewed each year will be much increased if they are visited only once. Therefore, the Com- mittee does not believe that the panel approach is as likely to be as efficient as a continuous sampling to detect changes in food consumption pat- terns of subsections of the population (e.g., in consequence of recent unem- plo~rment) that cannot be well identified beforehand. The Committee recommends redes ign of the Individual Dietary Intake component of the NFCS. Instead of an intermittent 1-year survey of a large sample, the Committee sugges ts an annual survey that would distribute the total sample over a number of years. -A design that provides for the collection and processing of data from 20% of the full sample per year has been suggestedl3. Addition of data accumulated in the preceding 4 years to the data Cal lee ted during a current year would provide data on the total sample. This design would reduce the problem of input overload to a minimum. The data collected in the Committee's proposed continuous Individual Dietary Intake survey should include the detailed population and socioeconomic descriptors that adequately and precisely identify the survey subjects and their household characteristics, as well as their food consumption and dietary intake. Similar considerations about "input overload" pertain to the NHANES. Data for the 1976-1980 survey covered 20,000 individuals, and some of the data col- lected are still not available for analysis in 1984. The next NHANES is scheduled for 1987. The Committee recommends redesign of future NHANES. Instead of intermittent surveys, the Committee suggests a survey that would distribute the total sample over 5 years. Collection and processing of data from 20% of the full sample per year would permit addition of data accumulated in the preceding 4 years to be added to the data Cal lected during the current year, to provide data on the total 5-year ~ample. This design would reduce the problem of input overioat to a minimum ant would permit a moving 5-year average to be reported every year. CONTINUOUS DATA PROCESSING AND REPORTING A common user need identified by the Committee is for data to be made available as soon as possible after being collected. To be used effectively,

OCR for page 65
69 data on foot consumption and dietary intake should reflect information that is current, given the limitations of survey design. A continuous data processing and reporting system would probably involve calculations updated annually on the basis of tats accumulated in each given year ant the preceding 4 years (i.e., a 5-year moving average). Annual would also be possible, e.g., ~~ of necessity, analyses over a period of years would also De posezole, e.g., over a '-year period. This system of moving averages would, of necessity, involve some data several years old. However, the advantage of providing users with con- tinually updated data is that it largely removes concerns about the incor- poration of otter data in the moving average. The availability of this updated information, coupled with information on the foot consumption and dietary intake of subjects, would make such data more widely and effectively used. Along with implementation of continuous data collecting systems for the Individual Dietary Intake component of the NFCS and for the NHANES, the Committee recommento the implementation of continuous systems for processing data, releasing data tapes, ant reporting. Dmplementation of a continuous system would result in the more efficient and timely collection, processing, ant reporting of survey data on food consumption ant dietary intake. COLLECTION OF REPLICATED DATA The core surveys yield statistical data on food consumption and dietary intake of two general types: data related to group mean intake across popu- lation groups and data relates to intake ant distribution of intake within population groups. The NHANES collects a 24-hour recall of dietary intake and thus con- tributes population late on group means across population groups. With NHANES data, the estimated mean nutrient intake (e.g., iron, calcium, or fat) can be established for population groups defines by sex, age, race, income, or other selected variables. The Individual Dietary Intake component of the NFCS, however, collects 3-day dietary intake data ant thus can provide two types of information: data on group means (as above) and, more important, data that can be used to derive estimated of the distribution of intakes among individuals within groups. The 3 individual days of intake data collected by the NFCS probably provide a minimum of data points needed to estimate the partitioning of '~etween- subject" and "day-to-day" variances. The '~etween-subject" variance is a descriptor of the distribution of intakes within the population group (defined by sex, age, race, income, and other variables). This '~etween-subject" distribution is what is neetedl6~24 to address questions relates to the prevalence of high or low dietary intake of given food components. Thus, the

OCR for page 65
70 3 individual days of dietary intake information, rather than the 3-day mean, obtained from NFCS tata collection covit provide partic- ularly important information. However, the data should be collected by the same method for each of these three days. This is not the present practice where recall data in used for one day and diary data for the other two. Both types of information -- variability across population groups (NHANES and NECS) and variability within population groups (NFCS) -- are important for many uses of the separate survey data bases. They take on particular impor- tance when they can be linked to additional data sources for nationwide food and nutrition monitoring. The Committee recommends that USDA, at a minimum, continue to collect replicated data on food consumption and dietary intake of individuals, e.g., that USDA maintain, at a minimum, the collection of 3-day intake information used in the individual intake phase of the 1977-1978 NFCS. This replicate data, however, should be collected by a single method. The Committee notes that the minimal number of replicate days required to provide reliable estimates of the distribution of intakes within a population group depends on group size and the day-to-day variability of intake. It also recognizes that questions of comparability of dietary methods used in col- lecting information and of seasonal and other effects warrant consideration. Existing data bases are not adequate for this purpose. However, implemen- tation of a system of collection of additional replications, distributed across a year, coupled with careful statistical analyses could provide extremely important information for the refinement of survey design and improvement of interpretation. USDA already has plans for replications of data collection from the same individuals during a 1-year period. The Committee endorses this as a method- ologic study, even though not recommending it as a continuing survey system. Because of the need for statistical analysis of intraindividual variability, the Committee supports USDA's proposed plan for the repeated surveying of individuals over a 1-year period. The Committee believes that this will allow needed statistical analyses whose results will, over the longer term, permit determination of the best data collection procedures. The NHANES also needs replicate data, because the dietary data are used to investigate the effects of diet on health. Knowledge of the variability of diet in an individual is necessary for other analyses. Where the NHANES already does other replicate subsampling, addition of 24-hour recall replicate

OCR for page 65
71 sub~amples is necessaryi63~7~27328 to estimate intraindividual variability for interpretation of regression analyses that use dietary information as a determinant29 in health-related analyses. Expansion of the NHANES to permit collection of replicate dietary intake data in a subeample ant thus to address currently unresolved statistical concerns about intraindividual variability is ai80 encouraged. This replicate data collection can be done within the current survey design and time constraints imposed by current use of the mobile examination unite, inasmuch as the mobile examination units remain in a single locality for about 6 weeks. Within that Midweek period, some subjects can be reinterviewed. RECOMMENDATIONS ON FOOD INTAKE METHODS Usere identified other needs for data that have implications for food intake methods and information on the composition of foods. Many of these needs call for data that answer questions broadly related to what is being consumed. Table 4-2 identifies many of there questions. Several of the tats needs identified by users can be met by changing the methods for surveying food intake. Committee recommendations for modifying these methods are made in response to those user needs. In some instances, a modification suggested by the Committee involves only one of the core surveys; in other instances, the Committee has recommended joint action to effect common changes in food intake methods. CO=ON DIETARY RECALL METHODS If dietary data are to be interpreted adequately, reliable estimates of the distributions of usual intake within the population or population groups are needed. Such estimating requires planned replication of 1-say dietary intake data collection or, as recommended earlier, continuation of the USDA process of collecting multiple days of intake data. When couplet with appro- priate statistical analysis, such data sets can yield the required estimates of the dis tributions of usual intake. The combination of this survey des ign feature, an NFCS sample larger than that of the NHANES, its improved repre~en- tation of the population, and the increased opportunity for dividing the popu- lation into groups of particular interest has let the Committee to emphasize the role of the USDA data- base as the primary dietary data base. The I-day intake (24-hour recall) data bane of the NHANES can be uset to obtain reliable estimates of the average usual intake (but not distribution of intake) of population groups, provided that the groups are large enough. For

OCR for page 65
72 some analyses, such as those in which groups are classified on the basis of other criteria and group mean intakes are compared, either data bane could be used. Indeed, that is a current practice . For some applications, it is pose ible to link findings across the two surveys by using mean and variance of dietary intake of defined groups to confirm that the groups are comparable between the two survey data bases. Such confirmation lends greater confidence when data from the surveys that do not overlap are linked (e.g., linking NHANES health or biochemical data to NFCS household descriptors, economic tata, or other traits characterizing groups). In the context of this report, the dietary corroboration is impor- tant, because nutritional, and hence dietary, problems are the focus of the analyses discussed here. The Committee has received many indications that the approach discussed above may not yet be possible. Even when groups with similar general charac- teristics are selected in the NFCS and NHANES populations, there are strong reasons to believe that agreement of mean dietary intake would not be seen -- not that it toes not exist, but rather that the computed intakes are likely to differ. Some of the reasons are related to the differences in coding pro- cedures ant in data bases on the composition of foods . In addition, the two surveys differ in food intake data collection methods and that this might contribute to dif ference~ in es timated intakes . As discussed earlier, the 1977-1978 NFCS methods incorporated a 24-hour recall and a 2-day record to collect information on 3 days of individual dietary intake. The 1976-1980 NHANES methods included a 1-day recall and, for other reasons, a foot-frequency questionnaire technique. Thus, in part, the methods -- with inclusion of 1-day recall -- are similar, although the actual procedures us et for this common element were not standardized across the surveys. In the judgment of the Committee, standardization of some of the core elements with this common technique could foster comparability and facil- itate the cross-linking of information. Where feasible, Standardization of some of the methods used to collect the 24-hour dietary recall information would increase the utility of both surveys without contravening the internal purposes of either. The Committee recommend. that core elements of the methods used for the 24-hour recall in the two surveys be standardized. This would also include both the standardization of interviewing techniques and the format of interviewing subjects. As discussed later in this report, ~ tandardization should also include the coding of food intake data and the use of common data bases on the composition of foods. The Committee notes that USDA has previously commissioned studies to evaluate food intake methods. Results of these studies may lead to modifications of the 24-hour, as well as dietary record, data

OCR for page 65
73 collection procedures. In commending USDA for this continuing effort to improve food intake methods, the Committee urges that any changes in core elements of the 24-hour component be implemented uniformly across the two surveys. The Committee, using the survey purposes as guides, suggests that USDA serve a lead role in this regard, but emphasizes that the logistic implications of changes should be considered in the context of both surveys. INFORMATION ON D IETARY SUPPLEMENT USE Both the NHAl7ES and the NFCS obtained limited information on the use of dietary supplements (e.g., vitamin or vitamin~mineral preparations). However, neither survey collected such information in the detailed manner that was used to collect other food consumption and dietary intake data. A recent report on the use of dietary supplements30 indicated that about 43X of people surveyed were using vitamin or vitamin-mineral supplements. But such data are no substitute for data collected as an integral part of a food consumption and dietary intake survey. Thin telephone survey report serves to point out that future surveys should collect better information on this source of nutrients. More detailed data, collected within the context of a food con- sumption and dietary intake survey, can show how frequently supplements are being ingested ant when and will provide a more realistic perspective on the nutrients present in the dicta of survey subjects. In anticipation that core elements of dietary intake questionnaires will be standardized, the Committee recommence that the two surveys include questions to obtain information on the use of dietary supplements. There possible, brand names and quantities ingested or other qualitative and quantitative information (e.g., potency), should be ~ought. Because of the poss ible complexity and multiplicity of information that could be obtained, the Committee suggests that as interagency working group be convened to determine the nature ant wording of the core questions used to obtain information on the use of dietary supplements. With the survey purposes as guides, given that the thrust of this e f fort is dietary, the Committee recommends thee USDA serve as the lead agency in convening the sugges ted working group.

OCR for page 65
74 D ISCRET IONARY I) IETARY COMPONENTS . . Although water, alcoholic beverages, sugar, salt, ant fat are important dietary components, their intake is difficult to estimate. In part, this is related to their discretionary use in diets (e.g., sugar and salt used at the table ant trimming or not trimming fat) and the variability in their use -- both between persons and in one person's own behavior. Both the NFCS and the NHANES have obtained some information on discre- tionary dietary use of alcoholic beverages and salt. However, data users have expressed a need for increased quantitative information on the use of discre- tionary dietary components. Alcohol, sugar, and fat are discretionary sources of dietary energy. The estimation of their contribution to total dietary energy (i.e., what is con- sumed) is important for all data uses. Quantitative estimates of fat, sugar, ant salt consumption are particularly important for public-health uses of the data. Water is an important component of the diet and can be a source of nutritionally essential trace minerals. The Committee believes that it will be difficult to develop core com- ponent~ for the survey questionnaires that will result in improved estimates of discretionary dietary components. For some components, special surveys may be required. Nevertheless, the Committee expects that the addition of more detailed and reliable quantitative information on discretionary dietary com- ponents is an important modification for future surveys. The Committee recommends that USDA and DHHS convene a joint working group to develop improved techniques for determining the consumption of discretionary dietary components. The Committee suggests that this review include considering the utility of special surveys for determining the consumption of these components and, alternatively, the utility of adding core components to the existing surveys that better measure the consumption of discretionary dietary components. FOOD-FREQUENCY QUESTIONNAIRE TECHNIQUE A food-frequency (tietary-frequency) questionnaire technique is part of the design of the NHANES. These questionnaires are uses, often in epidemi- ologic studies, as a means of acquiring information on the usual pattern of general dietary intake of specific foots or nutrients over a long period. The information recorded includes whether foods in various groupings -- such as

OCR for page 65
75 milk, meat, fish, eggs, fats ant oils, legumes and nuts, cereals, fruits, vegetables, and alcoholic beverages ~ have been consumed in the past ant number of times each food was consumed daily or weekly. It has been suggested that this technique for collecting ant reporting information on a survey sub ject to ''usual" food intake be incorporated into the NFCS. However, the Committee is not convinced that the food-frequency tech- nique has been developed to its full potential, partly because adequate vali- dation studies have not been done with typical American cultures.3 More importantly, there has been inadequate recognition of the need to define end use of the data before design of the instrument. When this is done, the food- frequency technique can be an appropriate tool32~33 for specified purposes. The Committee recommends review of the use of a food-frequency ques tionnaire technique . Potent ial modifications to allow some quantification of individual intake over time should be explored. Adequate resources should be made available to test the validity and cost effectiveness of the food-frequency ques tionnaire technique . This review should be based on the experience of investigators in both the public and the private sectors. The Committee anticipates that this review will provide highly useful information. However, if, after appropriate investigation, no uses of these data for important survey purposes can be identified, the Committee would recommend that the technique then be deleted from the NHANES. In any event, the Committee sugges ts that the foot-frequency ques tionnaire technique not be added to the NFCS be fore the recommended review is complete. INFORMATION ON THE COMPOSITION OF FOODS The survey techniques used collect both qualitative and quantitative data on food intake. These data are needed to estimate food profiles (and, by der- ivation, nutrient profiles) of diets, food intake patterns (ant, by deri- vation, nutrient intake patterns), ant foot relationships (ant, by derivation, nutrient relationships) to socioeconomic factors. Qualitative and quanti- tative information is also used to estimate dietary exposure to nonnutritive food components that have biologic or physiologic importance. Thus, the qualitative ant quantitative data developed about and from food intake infor- mation are pivotal for many uses of national survey data on foot consumption and dietary intake. Some Committee recommendations regarding survey modifi- catione are directed at this fundamental part of the survey information system -- its information on the composition of foods.

OCR for page 65
- 76 EXPANSION OF EXISTING DATA BASES , Data on the compos ition of foods have a crucial role in the ef fective use of survey data on food consumption and dietary intake. A reliable qualitative and quantitative data base on the composition of foods is fundamental to the wiser and more effective use of these survey tata. Much can be done to expand the data on the composition of foods. The Com- mittee accepts the premise that sufficient knowledge and technology are avail- able in the field of analytical chemistry to make many needed improvements in data on the composition of foods at a reasonable cost. In some cases, existing analytic methods are adequate, ant direct analysis of foods will be sufficient to acquire the needed data. In other cases, new or improved methods will be needed. In any event, significant improvements can be made if improvements are approached sys tematical ly . The Committee believes that it is realistic to pro ject the success ful expansion of data bases on the nutrient composition of foods. For example, although the U. S. food system contains thousands of food items, there are some general patterns of food consumption. Findings suggest that a "core" of food items consumed by the population accounts for about 90134 o f the food consumed. Qualitative and quantitative analys is of core foods in the diet would provide maximal information with a minimal number of determinations. The Committee views the continuing expansion of the data on the composition of foods as an important contribution to the more effective and wider use of the results of food consumption and dietary intake surveys . The Committee, us ing the survey purposes as guides, recommends that USDA serve as lead agency in th is ef fort to develop a cos tee f fective approach to expanding the data base on the composition of foods, e.g., through the identification and chemical analysis of core foods in the diet of the U. S. population. The Committee suggests consideration of the fol lowing for data acquis i Lion: es tab 1 ish ing or i ter ia to evaluate the current ~ Late of data on the composition of foods, using these criteria to evaluate current methods for food analysis, and , where needed, developing ~ trategies for improving the data in data banes on the compos ition of foods . The Committee also Rugged ts that appropriate educational strategies on analytic methods be developed for those who provide analytic data for the data bases and for those who maintain and evaluate data base systems. A greater effort should be made to

OCR for page 65
educate those who use various data base systems, including the users of the food consumption and die tary intake surveys . DATA ON THE COMPOSITION OF FOODS A cornerstone of the core purveys is the availability of good qualitative ant quantitative data on the composition of foods. Such data are used to cal- culate es timates of dietary exposure to and inges Lion of food components that have biologic, physiologic, or nutritional significance. Current data banes on the compos ition of foots are generally des igned to provide values representative of the "normal" or "usual" amounts of nutrients, that is, the amounts of nutrients likely to be encountered by the consumer most of the time. Little data base information is available on the non- nutritive components of foods. Other analytic data on the composition of foods often have been limited to the values listed on commercial-product labels. In the United States, such "label values" reflect both regulatory and corporate constraints, as well as scientific data on food consumption. More- over, some values may underestimate or overestimate the nutrient content of foods, as prepared and consumed. An absolute requirement of any effective use of survey data on food consumption and dietary intake is a complete data base on the qualitative and quantitative information about the composition of foods . To improve use of future food consumption and dietary intake surveys, the Committee recommends that the data bases on the composition of foods be expanded and improved. Information for data bases on the composition of foods should be derived from direct chemical analysis of food products (as prepared and consumed) wherever analytic methods make such analysis possible . The Committee recommends that label values not be used in data bases on the composition of foods unless no other information is available. Incorporation of label value information into future data bases should be discontinued as information from direct analysis becomes available. IMPROVED ANALYTIC METHODS FOR FOOD ANALYSIS The current knowledge of the nutrient composition of UPS. foods has been reviewed by Stewart35. Table 6-1 summarizes that review. As indicated in the table, substantial data on the composition of foods are lacking. The absence of data may be due to the lack of reliable analytic methods. However,

OCR for page 65
78 this is not always the case; the analytic methods are sometimes considered reliable. Until foods are analyzed and these data are available, the benefits of the surveys cannot be realized. lrhe Committee recommends that, where appropriate analytic methods exist, analyses be undertaken to provide missing data on nutrients present in foods, giving pr iority to foods identified as core foods . Implementation of this recommendation may require commitment of resources on the part of USDA and many other government agencies. The Committee believes that USDA and agencies with particular needs for data on the composition of foots could develop cooperative agreements to ob tain such data. This would provide useful information both for a requesting agency and for incorporation into future data bases on the composition of foods. The Committee, with the survey purposes as guides, recommends that USDA serve an lead agency in this effort . STANDARDIZED DATA ON COMPOSITION OF FOODS , Many uses of food consumption and dietary intake survey data do not require information on the foods as they are consumed, but involve informs tion on the foot consumption and dietary intake of components that is derived from food consumption information. Thus, a necessary step in data processing is the calculation of the dietary intake of particular food components. Calculations of dietary exposure to or intake of food components for a given period involve knowledge of the quantitative consumption of a given food, qualitative and quantitative information on the composition of the food, and the Rum of the intakes of individual products containing that component. Thus, the estimates of food component intake are derived by combining data on the compos ition of foods with die tary intake data . The core surveys do not use a common, standardized data base on the compo- sition of foods to compute the dietary intake of given food components. Each agency develops its own data on the composition of foods, although some of the information comes from colon data. Given this, users will continue to have difficulty in determining whether differences reported in the calculated con- sumption of food components reflect differences among those surveyed or dif- ferences in the data bases on the composition of foods that are used to calculate intake. Colon, 8 tandardizet data on the compos ition o f foods are required for effective uses of the survey data and valid comparisons between surveys.

OCR for page 65
- 79 - Differences between surveys in intake estimates that are due to the use of different data bases on the composition of foods must be precluded. The Committee strongly recommento that all future food consumption and dietary intake surveys supported with federal funds use a common, standardized data base for estimating dietary intakes of food components. The Committee recommento that such a standardized data base on a given food contain the foot identifier (i.e., the food code), the mean content ant variance of the reported components of the food, the number of samples used to determine the reported measurements, and an estimate of the reliability of the analytic data. For reasons cited earlier having to do with the purposes of the core surveys, the Committee strongly recommends that USDA serve as the lead agency in joint agency efforts to standardize common data on the composition of foods that are used in federally funded food consumption and dietary intake surveys. Because special NHANES needs must be recognized, taken into account, and met, the Committee recommends that the NCHS survey planners be directly involved in these joint agency efforts. The resulting standardized data bare should be made readily available to the public and professionals. COMMON FOOD CODES Although the codes used are not now standardized between the surveys, the coding of dietary responses provided by survey subjects is a process common to the two surveys. In this process, the raw data on each food that is consumed are evaluated and assigned identification codes. lithe coded food response is normally among the first data entered into the computerized data collection syn teme and is necessary to provide an orderly data set . Coding is thus a crucial step for using food consumption and dietary intake survey data. The use of a common coding process is crucial for data uses that involve consumption estimates for the various food components of biologic, physio- logic, or nutritional importance. Colon food codes can increase the effec- tiveness of data uses that involve "pooling" information about food con- sumption and dietary intake, whereas differences in food codes uset by the core surveys may feat to dif ferences in es timates of tie tary exposure to food components, depending on the survey data used. Thus, differences in foot coding between the surveys would affect survey uses, survey users, ant the usefulness of the purveys.

OCR for page 65
- 80 Given the importance of foot codes in identifying foods consumed and in using survey data, the foot codes used in the surveys should be standardized. The Committee strongly recommence that future foot consumption and dietary intake surveys supported by federal funds use a compatible food coding system for the identification of foods. The Committee also recommends that a standard set of criteria be developed and used for determining how a food is to be coded when the food (e.g., a home-recipe food) is not readily identifiable with the usual nomenclature of commerce. The development of appropriate foot codes requires consideration of complex issues beyond the Committee's charge. However, the Committee recommends that USDA and DHHS jointly establish a mechanism for the development of common food codes. The Committee, using the survey purposes an guides, suggests that USDA serve as the lead agency in this e f fort . Academic and incus try scientis ts who use these data should be given an opportunity to contribute to this development. POPULATION INFORMAT ION A fundamental question that the surveys mus t answer is 'who is consuming the foods" The survey components that address this question must be com- parable and compatible across the core survey data bases. Application of com- mon population descriptors clearly will benefit data users. Although the different purposes and logistics of separate surveys will inevitably be reflected in the subroutines of the core survey instruments, use of identical sampling universes and population descriptors would benefit all data users. The common identifiers that address the question of ''Who is con- suming the food?" should be retained on the data tapes. Particularly for application of these data bases to nationwide food and nutrition monitoring, identical population descriptors across the surveys ensure that information from individual survey data bases would be consistent. LINKAGE BETWEEN NFCS AND NHANES Art important use of the data from the core surveys involves the ability of users to identify current socioeconomic and foot purchas ing factors associated with food consumption patterns, dietary intake of food components, and nutritional status. In addition, these data are used to ascertain the relation of program or policy changes to desired changes in food consumption patterns, dietary intake of foot components, and nutritional s tattle. Data users have expressed a need to facil itate their analyses through linkages

OCR for page 65
81 between the NFCS and the NHANES data bases, because the data on socioeconomic status ant food purchasing information in the NFCS data banes are more exten- sive than data in NHANES. Ideally, linkages between the surveys would be most increased if the sample subjects in the surveys came from the same households and were, in fact, the same subjects. In application, however, such sampling would increase respondents' burden and increase costs, because of the need for increased sample size. From the practical standpoint, with other types of improved linkages , inferences could be made across the two surveys by using data on the socio- economic status and foot purchasing determinants of foot consumption from the NFCS data bases ant then examining characteristic groups in the NHANES for differences in nutritional and health status. The Committee has previously discussed the need to have comparable and compatible food intake methods and standardized data on the composition of foods to make more effective conjoint use of the separate data bases. Linkages among foot consumption, dietary intake, and nutritional status data bases should also include Chose related to the selection and weighting of survey sub jects and their precise description. The Committee recommends that survey planners from USDA ant DHHS continue a joint effort to assese the current compatibility in the sampl ing frames and population descriptors of the two surveys. The goal of this joint effort should be to develop and implement sampling plans and descriptions of populations that ensure the greatest possible compatibility between the two surveys. When the above compatibility is achieved, it is important for reader e of repor to to be ab le to 1 ink the f indinge repot ted by the two surveys . Th is requires identical or at least compatible divisions within variables such as geographic area, age groups, income levels, and years of education. The Committee recommends that the joint working group, mentioned above, set the appropriate divisions of population descriptors so that they are compatible between reports from the two surveys. LINKAGE BEYOND NFCS AND NHANES There are potential uses that require that information from other sources be combined with information contained in the two surveys. For example, certain analyses require linkages of detailed household expenditure survey data (e .g., that obtained from the Consumer Expenditure Survey ~ and infor- ma t ion from NFCS .

OCR for page 65
a ~ - >~ The Committee recommence that such uses be identified ant considered and that appropriate linkage points, where necessary, be established to fulfill there uses. There is a specific use that the Committee is already aware of that requires speci fic linkage information. If adequate identi Eying information is collected for individuals, it would be possible to link survey data with other data nets, particularly those involving health outcomes. In particular, the National Death Index provides a mechanism to follow surveyed individuals for mortality. To be most effective, Social Security numbers should be collected in both surveys. Mechanisms to ensure confidentiality are possible. For example, individual identifying information can be stored with a separate agency, such as the Bureau of the Census or NCHS, which is covered by specific privacy acts. Linkage could be accomplished under its auspices without risk of releas ing individual data. The Committee recommends that such identi lying information be collected in both surveys. GENERAL USER NEEDS The data userel conferences were designed specifically to develop infor- mation on and recommendations for modifications in survey methods. However, the Committee has chosen to include recommendations on some general user needs that were identifies during its own information gathering process. The recom- mendations included here are related to data handling techniques and the implementation of a continuing Late users' information system. The implemen- tation of these additional but more general recommendations, although not Specifically a part of the Committees charge, would also serve the needs of future data users. The Committee therefore urges their consideration. DETAILED DESCRIPTION OF DATA HANDLING TECHNIQUES Data users suggested the incorporation of detailed information on data collection, processing, data reporting, ant other data handling techniques into the data tapes. Such descriptions would serve user needs for more detailed description about the data handling (data documentation). Some user needs relate to quantitative documentation of response rates and of standardization proceedures. Other specific user needs for data documen- tation that were identi f fed are factors involved in decis ions about data process ing, reporting, and rules and procedures for survey design and data processing (e.g., sample weighting, coding rules, and rules for imputed or estimated values). In addition, some data users suggested the incorporation of documentation about data developed from such procedures (e.g., distinctions

OCR for page 65
83 between absence of values and "zero" values that are as appropriate as any other values, estimates of measurement, precision and reliability36~. Some Committee recommendations in Chapter 6 are related to the standard- ization of methods and information. Were such recommendations are imple- mented, the Committee anticipates that there will also be standardization of data handing techniques. The Committee recommend. that, where fees ib le. information on the group of techniques necessary be incorporated into the data tapes and be added to published data reports to ensure a complete record of the reasons for changes in variables. Incorporation of data documentation would assist users, particularly users of data tapes, in undertaking more effective and appropriate data analyses. The Committee anticipates that some aspects of Rate documentation will be peculiar to the separate core survey data bases and will be incorporated into only one agency 'a data base tapes. Where methods are standardized in and common to the two core surveys, the Committee recommends that each agency ts data base tapes contain identical data documentation. IMPLEMENTATION OF A POSTRELEASE INFORMATION SYSTEM AS illu8 bated in Chapters 2, 3, 4, and 5, uses for data on the food con- sumption and dietary intake of the U.S. population are many and varied. There are only a few current efforts to ensure that postrelease information about the data is itself provided deco data users in a timely ant systematic way. For example, an NHANES users group meets informally three times a year. The users group address list provides a means of conveying important postrelease infor- mation about the NHANES data, but only to those on the list . USDA's Human Nutrition Information Service ha. address lists for those who have asked for data or particular pub fished reports . The National Technical Information Service maintains a list of those who have ordered the data tapes. However, those lists often do not provide the names of the data users, but only institutions ' mail-receiving addresses . The Committee suggests that means of providing continuing postreleane information to data users be explored and that a system be developed and implemented to provide this information to data users . The fees ib ility of us ing modern bib liograph ic information systems, an information feedback system, ant an educational sys tem for data users should all be reviewed .

OCR for page 65
84 The Committee suggests that the agencies review the use by scientific journals and literature search systems of a keyword notation system when publications cite or use NFCS ant NHANES survey data. The goal of this effort would be assurance that, when these data are cites or us et in scientific reports, the specific articles can be identifies and retrieved through the modern computer izet ~ cienti f ic 1 iterature s earch systems (e.g., Medline, Toxline, Biological Abstracts, and Chemical Abe tracts ~ . The Committee further suggests the development and implementation of an information feedback sys tem between data collectors and data users. Both groups would then be able to develop an effective knowledge of data uses, data users' concerns, and other information that can help to ensure both the appropriate use of data and meeting the needs of future data users. In addition, the Committee urges consideration of a continuing educational system, perhaps through the development of short courses for data users, to provide useful information to the broad community of users. An educational system can help to prevent unrealis tic or inappropriate interpretations o f and expectations for these data. In the view of the Committee, the variety of uses for these data makes it likely that an interagency effort in user education would be most useful and most cost~ffective. SUMMARY lathe foregoing recononendatione of. the Committee are related primarily to modifications in survey design, data processing, and reporting that respond to expressed user needs for data that are timely and that are comparable ant com- patible across the separate NFCS and NHA~S data bases. Others concern modi- fications in food intake methods and use of intake methods common to the separate surveys. The Committee also recommence expans ion of the Individual Dietary Intake component of the NFCS and the NHANES to incorporate continuous data collection and processing and reporting of accumulated data over a 5-year period (a 5-year moving average) while the Household Food Use component of the NFCS is continued on an intermittent basis. Committee recommendations related to food intake information are des igned to respond to users' needs for better and more compatible qualitative and quantitative information on the foods that are consumed. The Committee recom- mendations include expansion and standardization of current data bases on the compos ition of foods and standardization of food coding.

OCR for page 65
85 - Further recommendatione focus on user needs for more comparable and com- patible information about the subjects in the separate survey samples. Some general recommendatione are also included in the Committee's recommendations. Although not specifically within the Committee's charge, the general reco~en- dations are designed to respond to needs of both current and future data users for more detailed use of survey tats.

OCR for page 65
- 8~6 - Table 6-1. Foot composition knowledge as percent of applicable food categories containing substantial datea Nutr lent To tat protein Total fat Niac in Riboflavin (vitamin B2 Th iamin (vitamin B Calcium Iron Phos phones Sodium Pa tas ~ ium Chol es tero 1 Magnet ium Zinc Copper Fatty acids Vi tamin A Vi tamin B6 Simple sugars Percent of food categories with subs tantial data ~! . _ 82 54 53 53 53 46 44 44 44 44 42 40 40 37 34 27 20 10 a USDA has divided food supply into 42 broad categories. Because come nutrients are known to be absent from some foods and foot categories, not all nutrient analyses are desired or needed for all 42 categories. Percent of food categories with substantial data is calculated only for categories in which nutrient is suspected to be present in important amount. "Substantial data" means sufficient information to establish normal amount of nutrient in most foots in category. Calculated from Stewart.36