Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

Managing a nutrient data-base system: meeting users' needs and expectations

Introduction
The HVH-CWRU nutrient data-base system
Uses and users
Meeting users' needs and expectations
Conclusions
References

GRACE J. PETOT

Departments of Nutrition, Epidemiology and Biostatistics, School of Medicine, Case Western Reserve University, Cleveland, Ohio, USA

Introduction

Approximately 25 years ago, the first computerized food composition tables were used for dietary evaluation in epidemiological studies [8,21]. Computer processing of dietary data demonstrated improved comparability of data from different sources, easier application of new approaches to analysis and more efficient storage of data for future use [9]. Tremendous advances in computer hardware and software development, along with the burgeoning interest in food, diet, and health, have expanded the applications for food composition information far beyond those of 25 years ago.

The purpose of this paper is to describe a large nutrient data-base system and to document those activities involved in managing it to meet users' needs and expectations.

The HVH-CWRU nutrient data-base system

The Highland View Hospital-Case Western Reserve University (HVH-CWRU) Nutrient Data Base was developed in the early 1960s to study the diets of chronically ill people who were living at home. At that time, a suitable nutrient data base was not available and, because of the large number of individuals to be studied over time, it was decided to create a food composition table using the computer [11]. The multidisciplinary development team included a physician, a research dietitian, a biostatistician, and a systems analyst, each of whose interest was in the use of the data generated as well as the generation and processing of the data. The team approach to the continuing development and management of the system has remained until the present time. By the mid-1960s, it was recognized that there were other research efforts that would require more extensive food composition information; however, sources of funds were not found. In spite of this, the group decided to completely revise the original data base, with local support for development and research from the Division of Nutrition at Highland View Hospital and the Department of Biometry at Case Western Reserve University. Since 1978, with the dissolution of the Division of Nutrition at Highland

View Hospital, the data base has been maintained by the Departments of Biometry (at present the Department of Epidemiology and Biostatistics) and Nutrition in the School of Medicine, Case Western Reserve University.

The primary source of food composition data was, and remains, the United States Department of Agriculture (USDA) Handbook No. 8. Machine-readable forms of the data were not available during the early development years, and the data were therefore transferred by hand to computer cards. Since USDA data were not always adequate to meet the various research needs, other sources of information were used. Criteria for inclusion of data were established, the most important of which was that the data base reflect current knowledge of the nutrient composition of food. These criteria are, of course, fundamental to the validity of the entire process. For nutrients or for foods not included in Handbook No. 8, information is selected from research reports in refereed journals, directly from the food industry for brandname foods, and from calculations of recipes based on nutrient values for ingredients. Close communication is maintained with the Food Composition Group at USDA for advice and consultation regarding the reliability of all data.

The food composition table at present contains more than 3,000 food items and recipes. For each food item, there is storage space for the nutrient values per 100 grams of edible food product, shown in table 1. Information stored with each food item is illustrated in table 2. The food items are placed in one of 44 food groups and are arranged alphabetically into subgroups related to common attributes of form, preparation, processing, or nutrient content. The coding manual for the NVH-CWRU

Table 1. Nutrients included in the NVH-CWRU nutrient data base

Table 2. Information stored for each food item in the HVH-CWRU nutrient data base

Nutrient Data Base contains a listing of all food items with identification codes, volume or household measure codes. For all food items, nutrient values may be retrieved for any one of six or seven household measures. The variety of measure codes provides considerable flexibility in expressing household or volume measures of food items and reduces professional and clerical time in preparation of data for nutrient analysis. To obtain nutrient data for an item, the following information obtained from the coding manual must be provided to the computer: (a) the food item identification number; (b) a permissible measure code; and (c) a quantity or amount of the measure.

Two versions of the nutrient data-base software are maintained. A Fortran 77 version is transportable to a large number of mainframe or mini-computers with Fortran compilers. The C version is for computers running the UNIX operating system and features simplified coded data input, file storage, and manipulation of data output.

Standard system calculations include nutrient summaries for lists of foods, menus, dietary intakes, etc., averages for up to 99 summaries, percentage of recommended dietary allowances (RDAs) for each age, sex, pregnancy or lactation category, percentage distribution of energy for total protein, total fat, total carbohydrate, animal and plant protein, animal and plant fat, refined and natural carbohydrate, polyunsaturated and saturated fatty acids. Electrolytes are expressed in weight units and in milliequivalents.

The nutrient data-base system has been made available to a large number of users for the past ten years; it has been purchased by 15 institutions and one practicing dentist. Although close communication is maintained with active owners, it has not been possible to document all of their applications. Major categories of their uses are clinical practice and research, education, market research, food-product development, and nutrient-analysis services. The institutions owning the data-base system are university academic departments, university medical centres, hospitals, food manufacturers, and a food trade association.

Uses and users

Eight broad categories of uses of the data system have been identified (table 3). Clinical research utilizing dietary intake represents the area of greatest use for investigations of the relationships between diet and health.

Table 3. Uses and users of the HVH-CWRU nutrient data base

These include studies of dietary components and hypertension [22], weight reduction with drugs, anorexia, dialysis in renal disease, pulmonary disease, alcohol consumption [3, 4,14], aging [5, 6, 7], cervical dysplasia [23], multiple sclerosis [12, 26], mental retardation, dental caries [2], atherosclerosis, high-risk pregnancy, and lactation.

The majority of these studies have been conducted at university medical centres, clinical research centres, and academic departments; however, some have been done by pharmaceutical companies investigating the nutritional adequacy of dietary supplements or enteral feeding products and by food manufacturers for in-house research for new product development. A number of clinical research investigations have been completed, or are in progress, by doctoral and masters' candidates in nutrition and medical computer systems.

Epidemiological surveys by independent research contractors with government agencies have been completed or are in progress. The largest survey was the National Evaluation of School Nutrition Programs, contracted by the Food and Nutrition Service, USDA, with System Development Corporation, in which approximately 7,000 24-hour recalls were coded and analysed for 26 nutrients [25].

The primary requirement for clinical and epidemiologic dietary studies is for current, valid, reliable nutrient calculations for specific foods, beverages, or special dietary products, in specific quantities as consumed by persons of all ages living in any region of the United States. Foodintake information is usually received as 24-hour recalls, quantified diet histories, food-intake diaries or observed records of intakes. Portion sizes may be weighed or, more frequently, estimated. There is a notable lack of standardized protocols for obtaining this information, and therefore it represents many levels of completeness. The records may be obtained by trained individuals, but in many cases are self-reported. Therefore, depending on the quality of such records, it becomes the responsibility of the data-base management team to qualify the results in terms of the completeness of the collected information.

Evaluations of the nutritional quality of analysed records, other than comparisons with the RDAs, are not provided with nutrient-analysis services. It has been the policy of this management team to recommend consultation with dietitians or nutritionists about recording of data and for interpretation of results. In spite of this, some investigators continue to plan and conduct dietary studies without recognizing the need for such assistance.

Clinical applications of nutrient analysis have been implemented in a number of large hospitals. University Hospitals of Cleveland, a 1,000-bed teaching institution, accesses the data base daily for assessing the dietary intakes of about 100 patients [16,17]. The hospital dietary department and the data-base management team have developed a recipe file system for storing the coded ingredients of each recipe. This file provides access to current nutrient values for each recipe without actually storing pre-calculated nutrient values. Efficient procedures, using computer generated precoded forms for recording intakes, are used. Completed reports are then delivered to dietitians within hours. As costs of purchasing and maintaining hardware decline, and as software becomes more transportable, it is anticipated that many more institutions will include dietary nutrient analyses as a routine component of nutritional assessment.

A major use of the data-base system is for the nutrient analysis of menus and recipes for institutional food service. These analyses usually serve as documentation of the nutritional adequacy of food served in hospitals, nursing homes, correctional institutions, state-supported residential institutions, schools, and the armed services. Governing or accrediting agencies for these institutions have set nutritional standards and usually require a comparison with the RDAs as evidence of adequacy. Rarely do they request analyses for other nutrients or food constituents. Independent developers of software are incorporating nutrient analysis into computerized systems for hospitals, restaurants, and food-service contractors for management of food inventories, recipes, menus, production schedules, cost accounting, or clinical services. Many developers have realized that creating and maintaining a food composition table is costly and are purchasing food tables with or without programs for accessing and for formating reports.

Another user of a nutrient analysis system is the broker or distributor who offers menu and recipe analyses as a service to purchasers of food products. Food manufac turers have found that using a computer for examining and manipulating ingredient and foodproduct formulas for optimal nutrient content saves time and money in the development of new products [13]. Calculations of nutrient content per portion of many prepared, ready-to-heat-andeat food-service products are supplied to institutions with the products purchased. It is not necessary for these analyses to meet the Food and Drug Administration nutrition labelling regulations, which require laboratory analysis if nutrient content appears on the label. Several physicians have formulated special dietary products for use in their practices and are using nutrient calculations during the development and testing phases.

As the computer replaces printed food composition tables, dietetic, nutrition, and food science students have been introduced to a very effective method for learning what is in food. The fast response time for queries and manipulations of menus, recipes, and portion sizes provides more information with greater understanding of the interrelationships of foods, nutrients, portion sizes, and diets. This capability extends to the potential provision of nutrient information to practicing professionals through nationwide access to nutrient data-base systems. Professional and trade associations and health promotion organizations are exploring the feasibility of providing this access to members and clients.

At another, less sophisticated level, nutrient-analysis services are being made available to the public. Health-care professionals are using simplified, understandable nutrient summaries of patients' dietary records or of their favourite foods. Food exchange lists are easily developed to meet a variety of diet prescriptions. The American Heart Association uses nutrient analysis services in the development of educational materials for physicians, dietitians, patients, and the general public [1].

Recently, through an agreement with a local software developer, a nutrient-analysis system designed for a variety of personal computers has been made available at low cost to the home user. It incorporates a data base of 800 foods and recipe ingredients with 15 nutrients which were extracted from the HVH-CWRU Nutrient Data Base. It allows the home user to plan and evaluate the nutritional adequacy of recipes, menus, and dietary intakes. Other purchasers of this software include physicians, dentists, libraries, elementary schools, secondary schools, and universities. An interesting feature of this version is the expression of refined carbohydrate in teaspoons in the nutrient summaries. A professional package for use on microcomputers, with the same food table, is being marketed to small hospitals, nursing homes, consulting dietitians, physicians, and dentists.

Two federal agencies have used the HVH-CWRU Nutrient Data Base to examine the nutrient content of common or usual portions of food [18,19, 20]. Reports generated included nutrients per portion, percentage US RDA per portion, Index of Nutritional Quality, and nutrients per 100 kilocalories. This information was used by the Federal Trade Commission for proposing rules regulating the advertising of a "nutritious food" and by the Food and Nutrition Service of USDA for regulating the kinds of foods sold in competition with federally subsidized school meals.

Meeting users' needs and expectations

All users of the nutrient data-base system express the need for: (a) reliable and valid food composition data with (b) easy access in as short a time as possible, (c) reported in the desired form, and (d) at little cost. An unexpressed need for many is one of hand-holding, consultation, education, and encouragement.

Food Composition Table and Software

It has already been shown that food composition data are used for many purposes. It then becomes evident that the degree of precision and accuracy of nutrient data reports is dependent upon the accuracy and completeness of the quantitative and descriptive information provided for analysis. Users of data for clinical research and applications have the greatest need for complete and accurate analyses. Experience indicates that familiarity with the state of knowledge of food composition is necessary during the research planning stages. All too often, dietary intake records have been collected prior to determining the availability of the data for a food constituent. Research activities involving specific food components of current interest may occur far in advance of the acquisition of composition analyses, for example studies of dietary carotene intake. Constituents for which data have recently been requested are beta carotene, total carotenes, retinol, Vitamin K, choline, chromium, iodine, selenium, total sugars, total refined (or added) sugars, individual sugars, all individual fatty acids, and dietary fibre. Values for some of these constituents are included in USDA Handbook No. 8 revisions; however, for a large number of foods as consumed, data for these constituents are not available. These foods include many packaged, processed foods, fast foods, frozen ready-to-eat foods, and bakery products. Food items are added to the data base as requested by users; thus its composition does reflect its users' requirements. The data base at Case Western Reserve University is being updated continuously, with annual updates made available to owners.

In response to data-base owners' inquiries about updating the food table themselves, it is recommended that they use the recipe file software to accommodate those food items specifically for their use. This avoids the problem of overwriting their data in the food table as they receive updated versions.

It is important that users know how unknown nutrient values are accommodated in summaries. The summary reports a flagged nutrient sum if there is a missing nutrient value for one or more food items included for analysis. Additionally, users may be provided with an output format which indicates missing nutrient values for specific foods (fig. 1).

The degree of precision achieved with calculations of summaries is dependent on the precision of the nutrient values stored in the food table and the computer on which the programs are run. Guidelines for precision provided by USDA are used for inclusion of data from other sources. As smaller data bases are used on smaller computers, the degree of precision decreases. At the same time, as fewer food items are included in a data base, more substitutions must be made, thereby reducing the specificity of the reports for dietary intakes [10]. Such data bases may not be satisfactory for individual dietary records or as a generalpurpose tool. These limitations must be understood and used for appropriate applications.

Inquiries have been received about the use of retention factors. The HVH-CWRU Nutrient Data Base does not incorporate retention factors and the issue is handled somewhat ambiguously. For most purposes, it is possible to select edible portions of foods as consumed. The recipe file system allows for coding of the edible portion of each ingredient by using yield factors for preparation and cooking losses. However, another frequent request is for a data base of usual or common portion sizes. No attempts have been made to add these to the food table; however, this feature would be greatly appreciated for food-service and restaurant menu analysis. Planning menus or diets which meet specified nutrient requirements is an expressed need by a few users. The development of this capability is at present being explored as part of a clinical research study.

Ease of Access and Turn-around Time

The two requisites for ease of access and fast turn-around are a readily available computer and user-friendly, interactive processing and query procedures. The UNIX operating system affords an optimal time-sharing environment for direct access by telephone. Response time is fast and ideal for clinical and day-to-day use in practice or educational settings. Coding is still timeconsuming and the most costly activity associated with analysis; it also requires knowledge of foods and nutrient composition.

Smaller data bases with fewer food items require greater coding knowledge, as more judgements regarding substitutions must be made than for larger data bases containing more foods and more adequate descriptive attributes.

Interactive on-line coding is desirable and would be more error-free; this will be implemented as computer time becomes less expensive. Response time, however, must be very short to alleviate boredom and tedium while entering data. For large data sets, batch processing with offline data entry is more practical and less costly.

Report Forms

The most frequent reports requested are listings of foods with summaries of nutrient content and comparison with RDAs. Printed, easily readable formats are provided for verification and are most often used for clinical practice and patient education. Machine-readable formats must be used for computerized statistical analysis. Summaries of nutrient values or percentages of RDA or energy distribution may be aggregated by meal, time, place, food group, or any other set of descriptive features. Nutrients may be reported in ratios, the most common being nutrient/kilocalorie and polyunsaturated/saturated fatty acids. All reports may be produced in printed table or machine readable formats. Under development at the present time are procedures for directing nutrient analysis output into an existing data-base management system. This will facilitate the generation of almost any kind of desired report.

Many nutrient data users have requested that reports include intake evaluations with recommendations for dietary improvement. The provision of these evaluations is not in accordance with the present policy of providing analyses for professionals' interpretation and use.

Cost

From the data-base user's point of view the costs are high, while from the manager's point of view charges seem never to be high enough to support continuing maintenance. This includes all of the related activities involved in understanding users' needs, explaining how their expectations can or cannot be met, and the inevitable hand-holding throughout the process as users and the data-base team are educated. The costs to users of the data-base system comprised two major components: (a) the costs of development and maintenance of the food table and software and (b) the computer costs. As hardware has become less expensive and more efficient, development and maintenance activities represent the major portion of expenses.

Justification for the costs of using nutrient data-base systems for clinical research and applications has been difficult, owing to concern about the reliability of food composition data and the problems of obtaining food consumption information.

Conclusions

Managing a large data-base system is challenging and, at the same time, frustrating. External factors contributing to this ambivalence include the dynamic nature of the food supply and the proliferation of research findings implicating food, nutrient, diet, and health relationships. Within the relatively small community directly associated with accumulating and disseminating food composition information in a useful form, the issues are recognized and are being discussed [15]. Many user's needs, as now expressed, seem to require knowledge and capabilities beyond the present ability to supply complete and timely information. Maintaining close communication and continuing dialogue between users and providers of data is critical to achieving a consensus for producing reliable, accurate, and useful reports with consistency and comparability.

References

1. American Heart Association, Northeast Ohio Affiliate, Cooking Without Your Salt Shaker (American Heart Association, Dallas, Texas, 1978).

2. L. Bardach, J. Geduldig, and I. D. Mandel, "Dietary Patterns in Caries Resistant vs. Caries Susceptible Adults," J. Dent. Res., 59 (special issue A): 508 (1982), abst.

3. H. T. Bebb, H. B. Houser, J. C. Witschi, and A. S. Littel, "Nutritive Content of the Usual Diets of Eighty-two Men, J. Am. Diet. Assoc., 61: 407-415 (1972).

4. H. T. Bebb, H. B. Houser, J. C. Witschi, A. S. Littel, and R. K. Fuller, "Calorie and Nutrient Contribution of Alcoholic Beverages to the Usual Diets of 155 Adults, A.J.C.N., 24: 1042-1052 (1971).

5. P. J. Garry, J. S. Goodwin, and W. C. Hunt, "Nutritional Status in a Healthy Elderly Population: Riboflavin,"A.J.C.N., 36: 902-909 (1982).

6. P. J. Garry, J. S. Goodwin, W. C. Hunt, and B. A. Gilbert, "Nutritional Status in a Healthy Elderly Population: Vitamin C," A.J.C.N., 36: 332-339 (1982).

7. P. ]. Garry, J. S. Goodwin, W. C. Hunt, E. M. Hooper, and A. G. Leonard, "Nutritional Status in a Healthy Elderly Population: Dietary and Supplemental Intakes," A.J.C.N., 36: 319-331 (1982).

8. M. H. R. Goodloe, G. Waters, C. G. Quinlan, and J. G. Barrow, "Studies in Atherosclerosis. IV. Electronic Computers in the Processing of Dietary Data," A.J. C.N., 13: 304-310 (1963).

9. O. B. Hayes, S. Abraham, and C. A. Caceres, "Computers in Epidemiologic Studies," J. Am. Diet. Assoc., 44: 456- 460 (1964).

10. L. W. Hoover, "The Problems of Using Small Nutrient Data Bases," Proceedings of Eighth National Nutrient Data Bank Conference, July 25-27, 1983 (US Department of Commerce National Technical Information Service, Springfield, Va., 1983).

11. H. B. Houser, "Nutrient Data Banks- the State of the Art," Introduction in G. Petot and H. B. Houser, eds., Proceedings of Fourth National Nutrient Data Bank Conference April 23-24, Case Western Reserve University, Cleveland, Ohio, 1979 (Case Western Reserve University, Cleveland, Ohio, 1987).

12. H. B. Houser, A. 1. Sorensen, A. S. Littel, and J. C. Vandervoort, "I)ietary Intake of Non-hospitalized Persons with Multiple Sclerosis. 1. Food Diary and Coding Methods," J. Am. Diet. Assoc.. 54: 391397 (1969).

13. B. P. Klein, "Nutrient Analysis Information for Decision Making," Food Tech., December 1983: 37-54

14. L. J. Lillien, A. M. Huber, and M. M. Rajala, "Diet and Ethanol Intake during Pregnancy," J. Am. Diet. Assoc., 81: 252-257 (1982).

15. G. J. Petot, "Commonalities and Differences: Are We All Speaking the Same Language?" Proceedings of Eighth Nutrient Data Bank Conference, July 25-27, 1983 (US Department of Commerce National Technical Information Service, Springfield, Va., 1983).

16. G. J. Petot, "Nutrient Database Systems in Patient Care and Research: Clinical Applications," Proceedings of Seventh Annual Symposium on Computer Applications in Medical Care, October 23-26, 1983, Baltimore (IEEE Computer Society Press, Silver Springs, Md., 1983).

17. G. J. Petot. H. B. Houser, and R. V. Uhrich, "An On-line Nutrition Information System for the Clinical Dietitian," Proceedings of Fourth Annual Symposium on Computer Applications in Medical Care, November 2-5, 1980, Washington, D.C. (IEEE Computer Society, Long Beach, Cal, 1980).

18. G. J. Petot, A. Whitney and H. B. Houser, Index of Nutritional Quality of Foods, contract with Food and Nutrition Service, USDA. March 1979.

19. G. J. Petot, A. Whitney and H. B. Houser, Nutrient Composition of Foods Commonly Available in the Marketplace, contract with Federal Trade Commission. August 1978.

20. G. J. Petot, A. Whitney, and H. B. Houser, Percentage of U.S. RDA in Frequently Used Serving Sizes and in 100 Kilocalorie Portions of Foods Commonly Available, contract with Food and Nutrition Service, USDA. December 1979.

21. E. M. Thompson and H. Tucker, "Computers in Dietary Studies," J. Am. Diet. Assoc., 40:308 312(1962).

22. S. Wassertheil-Smoller, H. G. Langford, M. D. Beaufox, et al., "Effective Dietary Intervention in Hypertension: Sodium Restriction and Weight Reduction," J. Am. Diet. Assoc., 85: 423-430 (1985).

23. S. Wassertheil-Smoller, S. L. Romney, J. Wylie-Rosett, et al., "Dietary Vitamin C and Uterine Cervical Dysplasia," Am. J. Epidem., 114: 714-724 (1981).

24. B. K. Watt and A. L. Merrill, "Composition of Foods: Raw, Processed, Prepared," Agriculture Handbook No. 8 (Science and Education Administration, USDA, Washington, D.C., 1963 and revisions).

25. J. B. Wellisch, S. D. Hanes, L. A. Jordan, K. M. Maurer, and J. A. Vermeersch, The National Evaluation of School Nutrition Programs: Final Report, vols. 1 and 2 (System Development Corporation, Santa Monica, Cal, 1983).

26. J. C. Witschi, A. S. Littel, H. B. Houser, and A. 1. Sorensen, "Dietary Intake of Non hospitalized Persons with Multiple Sclerosis. 11. Nutrient Intake for One Year Compared with Recommended Allowances," J. Am. Diet. Assoc., 56: 203 211 (1970).

Contents - Previous - Next