| Food Composition Data: A User's Perspective (1987) |
|Experiences with food composition data: the context|
|INFOODS: Background and current status|
Growing need for and availability of information on components of foods
Information on the nutrient and non-nutrient composition of foods has been and continues to be used for a variety of purposes: assessment of the intake of and need for nutrient and nonnutrient constituents of foods by populations; meal planning and calculation of therapeutic diets; food production and nutrition policies; and the evaluation of food as a vehicle for transfer of toxic chemicals from the environment to man. To date, this information, particularly in relation to energy-yielding substrates and essential nutrients, has been obtained largely from food composition tables, a historical development of which has been presented by Hertzler and Hoover . However, the increasing interest in and concern for the relationships between diet and food habits and degenerative diseases (including coronary heart disease, diabetes, hypertension, stroke, and cancers) has stimulated the further gathering of composition data on foods. Simultaneously, this has called attention to major limitations in the available data to support a variety of research activities in this area. These are especially evident in epidemiologic studies, where attempts are being made to characterize dietary differences between various populations exhibiting markedly different incidences of the diseases thought to be associated with diet and nutritional factors. Such studies have characterized interand intra-national differences in diet by broad categorizations, but suffer from the lack of detailed information on the nutrient and non-nutrient content of the enormous variety of foodstuffs consumed by different populations and subgroups within specific countries or geographic areas. Additionally, the introduction of nutrient labelling in the US, for example, has promoted the accumulation of data by the food industry on the composition of foods and their products. These data offer a further opportunity to expand the food data base.
Most of the information on the components of foods is still disseminated in the conventional, printed-page format; this is likely to continue as an important mode of communication for many users. However, there can be little doubt that the use of modern computer techniques will expand in the future, for a number of reasons:
1. Certain functions can be served more effectively through organization of a computerbased data bank than by conventional methods. Thus, one advantage of data dissemination in computer-readable form is the relative ease of updating; another is the possibility of carrying out sophisticated search strategies for material that has some specified set of desired properties, a problem that can be tedious or perhaps impractical when the information is tabulated in the conventional, hardcopy form.
2. The distribution of data in computer-readable form offers considerable advantages for its subsequent manipulation and usage.
3. The proliferation of small machines or of personal computers and interest by consumers in diet and food habits will inevitably result in an increased demand for computer-aided food composition data systems.
The number of food and nutrient data bases is increasing rapidly and many new entrepreneurs are becoming active. This is made evident, in part, by the growing number of meetings being devoted specifically to various problems associated with the development and uses of nutrient data bases . Extensive nutrient data banks, such as the one established at USDA's Consumer Nutrition Center [6,10], and at other locations throughout the world, illustrate a clear trend in converting existing food component data to a computer format.
With increased use of computer techniques, both for archiving food component data and for application of these data to various purposes, it becomes absolutely critical to examine carefully the ways in which food composition data are handled and used. Although computerization can be very efficient, it does not guarantee optimum or effective service to the final user of the data. A study by Danford  is illustrative of this problem. It was found in this study that when a diet based on common, unambiguous foods and simple serving sizes was submitted for analysis using 11 commonly used computerized data bases, considerable variation existed between the output for various nutrients. This is a particularly disturbing finding when it is recognized that all of the data bases had evolved from a common data source, namely USDA Handbook No. 8. A similar, but more detailed, comparative study of a dietary record was considered by Hoover , and she also found that the values for nutrients varied considerably between the various computerized nutrient analysis systems included in the study, even after elimination of coding judgements relative to portion size and selection of food items. Dwyer and Suitor  have also pointed out the uncertainty about the reliability and validity of the data in many computer-aided diet analysis systems.
Because of the problem of the reliability of the available data in nutrient data banks and the disservice to the scientific and technical community that can easily arise from non-standardized, non-evaluated data storage and dissemination, it may not be unreasonable to suggest that consideration be given to the registration of food component or nutrient data banks by an appropriate authoritative body. This is especially important, as has been emphasized by Lide , when attempts are made to provide data directly to the user, because both quality assurance and the opportunity for users to assess the suitability of the data for meeting a particular need are key factors.