| Food Composition Data: A User's Perspective (1987) |
|Report and recommendations of the conference|
The papers presented can be divided into five general areas. The first three papers, each by a member of the INFOODS staff, deal with the background and general context of the meeting. Vernon Young (paper 1) discusses the forces that led to the formation of INFOODS, arguing that it is now time for the field of food composition data to be drawn together and, particularly, to consider agreed-upon, unified, and standardized approaches to its problems. These problems arise from a chaotic situation wherein everyone seems to be working independently. Much of the food composition data that exist are incomplete, incompatible, and inconsistent. William Rand (paper 2) presents some general ideas of food composition data as data, with emphasis on its variability and its inherent incompleteness for solving specific problems. This paper points out some of the consequences of the current situation of incompatibility and inconsistency between data bases, with special attention to the lack of confidence in the whole field of food composition data. John Klensin (paper 3) presents the role of INFOODS as a "broker" of food composition data, helping to increase the accessibility and interchange of data, and co-ordinating the development of guidelines. A major point made by this paper is that the users should define the systems they need, and that computers should be viewed as tools to assist the user rather than as constraints.
The second group of papers is concerned with specific uses of food composition data. Ann Sorenson of the US Department of Health and Human Services and her colleagues at the University of Pittsburgh (paper 4) report on a number of international epidemiological research programmes, the problems they have with existing food composition data, and the implications for how food composition data should be organized and managed. Lenore Arab of the University of Heidelberg (paper 5) discusses her experience working in the area of nutritional epidemiology in Europe, focusing on practical problems and potential solutions that range from the difficulties of determining dietary intake to those of dealing with data from several different food tables, each with its own conventions. Ritva Butrum of the US National Cancer Institute (paper 6) presents many of the Institute's programmes, both within the United States and internationally, which require food composition data, and the problems that these programmes face without globally complete, consistent, and compatible data. Betty Peterkin, of the US Department of Agriculture (paper 7), shows how the data that USDA collects are used to satisfy various programmes in research, information, education, regulation, and food assistance, with special attention given to the assessment of the quality of the US diet and preparation of dietary guidelines. This paper lists differing aspects of food composition data which individual users require: sets of nutrients, various stages of processing or preparation, levels of statistical analyses, degrees of precision and accuracy, units of measure, and different formats. Carol Windham and her colleagues at Utah State University (paper 8) discuss the needs that university research in the areas of nutrition guidance and nutrition education place on food composition data. The paper also describes the components of and problems with the system that Utah State University and the Human Nutrition Information Center of USDA developed to fulfil these needs. The need for more and better data, and the importance of documentation to characterize the data, are stressed. Fred Steinke of the Ralston Purina Company (paper 9) discusses the uses that food companies make of food composition data, surveys the data that are currently being used, and maintains that safety and acceptability by the consumer are the primary concerns of the food industry.
The third group of papers is authored by developers, managers, and users of large food composition data bases. Loretta Hoover of the University of Missouri-Columbia (paper 10) discusses the general status of nutrition data bases within the United States in the context of the requirements and problems that the user faces in dealing with these systems. She introduces the concept of two tiers of users: those experienced users who are primarily interested in issues of accessibility, installation, updating, availability, and computational concerns as well as being aware of the problems of accuracy and applicability of different software and data bases; and those inexperienced users who are primarily interested in utility, compatibility, and cost. Any efforts to improve the status of food composition data need to be directed primarily toward the first tier of users; however, the results of improvement will shortly effect those in the second tier. Frank Hepburn of USDA (paper 11) describes the management of the USDA data base, illustrating how the data has evolved in response to a changing food supply and the more sophisticated needs of users. Marilyn Buzzard and Diane Feskanich of the Nutrition Co-ordinating Center at the University of Minnesota (paper 12) describe their system and the procedures that they have initiated to satisfy user needs. The specific needs stressed include: standardized methods of obtaining data; up-to-date and complete nutrient profiles; adequate documentation of data sources; quality control for the data base itself; and stability, comparability, and flexibility. Grace Petot of Case Western Reserve University (paper 13) discusses the management of a large academic system, cataloguing user needs and expectations.
The fourth group of papers is concerned with food composition data outside the United States. Hernando Flores of the Federal University of Pernambuco, Brazil (paper 14) illustrates the very real problems which arise from using the outdated INCAP tables to estimate consumption of important nutrients. Aree Valyasevi of Mahidol University, Thailand (paper 15) describes the status of food composition data in Asia, and reports on progress in the establishment of ASIAFOODS. Ake Bruce and Lena Bergstrom (paper 16) catalogue the data bases within Sweden and other Nordic countries, showing the profusion of data bases that can exist in a relatively homogeneous part of the world. Pamela Verdier of Health and Welfare Canada (paper 17) describes the national data base of Canada, how it was put together and managed, and how it is integrated into the nutritional activities of Canada.
The fifth and last group of papers discusses food composition data from several points of view slightly removed from the field. Joanne Holden and her colleagues at the Beltsville Nutrient Composition Laboratory of USDA (paper 18) discuss the problems of maintaining the quality of the data in data bases and of assessing and communicating that quality. This paper presents a specific system for evaluating published data on selenium, and discusses the aspects such a system must consider. George Beaton of the University of Toronto (paper 19) examines the effect of uncertainty in food composition on the estimation of nutrient intake from food consumption surveys, concluding that this variability is often less important than previously assumed. Reliability of food composition data is most important in those foods that make the greatest contribution to total nutrient intake, and the effect of improving reliability will be most evident when examining diets which include few foods. This paper also notes that these considerations focus on variability and that the presence of bias in the data is often an important source of error. Catherine Woteki of the US National Center for Health Statistics (paper 20) discusses in detail the US National Health and Nutrition Examination Surveys, which are a major user of the USDA data base. John Klensin of INFOODS (paper 21) examines food composition data bases from the point of view of modern information systems theory and practice, showing that the field has changed considerably since most data bases were first implemented.