D.A.T. Southgate
5.1 Introduction
The definition of compatibility in the Oxford English Dictionary includes the phrase ‘able to be used in combination’, and this phrase seems very apt in relation to the objectives of this meeting: to reach the position where we can each use one another's data for the nutrient content of foods. Compatibility does not imply that the same format needs to be adopted or that there is a need to evolve one system to meet all perceived present and future needs, but merely that data can be used in combination.
One important requirement when using data from another worker or published source is confidence. Thus, confidence is required that the data are applicable to the particular situation and that the criteria of quality applied by another worker in accepting the data are equally acceptable in this situation. The range of criteria that need to be applied to data before the data become part of a nutrient data system either in the form of conventional food composition tables or a computerized nutrient data bank, is discussed. While, in general, the criteria are equally applicable to the two systems, the ease with which a computerized system can be revised permits some relaxation of the criteria.
In compiling data on the nutrient composition of foods there are two types of approach: the direct approach in which samples are collected, analyzed, and the values incorporated into the data system and the indirect approach in which values are drawn from the literature or unpublished reports. In practice, most compilations are a mixture of both approaches [1], although the original tables of McCance and Widdowson [2] were entirely based on analyses in their own laboratory. In the direct method, the compilers have control over the identity of the food, the sampling scheme of the analytical method used, and the execution of the analysis itself. Those who have been responsible for this type of operation are well aware of the limitations of their data and can take cognizance of these when presenting and using the data.
The indirect method is often seen as an easy and effective way of preparing data for composition tables, but in fact this method requires painstaking effort, a deep understanding of foods, and familiarity with nutritional analysis. It requires the establishment of criteria [1, 3] for the acceptance of data. These criteria are required to establish confidence in the compositional values and are applicable to the acceptance of data into nutrient data banks. They represent a ‘counsel of perfection’, but in practice all these criteria may not be met for a variety of reasons. However, if compatibility of nutrient data banks is to be achieved, it is vital that these criteria are used as guidelines.
5.2 Name of Food
The first essential requirement for compatibility of nutrient data is the unequivocal identification of a food item, and thus the choice of the identifying name has very high priority. In addition to the common name, the local synonyms need to be listed, and for international use of data, a thesaurus will be required to relate the locally used name to the names by which the same food is known in other countries. For unprocessed plants and animal foods the scientific taxonomic name will assist in identification, and for plants the variety should also be given. Prepared foods are more difficult to name correctly because some proprietary names have restricted use, and a few are used in different countries for different product formulations. Cooked dishes present even more problems because the composition of a named dish varies within households, between households, regionally, and internationally, and the name alone is often inadequate for unequivocal identification.
5.3 Origin of Food
Description of the origins of a food item will further assist in its identification. Such a description will need to include definition of the locality where a plant food was grown, the soil type, the method of husbandry, fertilizer treatment, harvesting, and post-harvest treatment. For animal products, locality and method of husbandry and slaughter should be defined. A great deal of food composition data originates from experimental studies in connection with plant and animal husbandry. The control groups in these studies can also be a source of data. However, the husbandry used must correspond to that used commercially.
5.4 Nature of Sample
The description of the sampling process, the place and time of sampling, the number of samples and their origin if purchased, by type of retail outlet, and finally the state in which the food was purchased also need to be considered. On the basis of these four criteria, it is possible to identify the food with reasonable certainty and to judge whether the sample is representative of the food as a whole or of a subset of the food, or whether data merely apply to the food actually analyzed.
5.5 Treatment of Samples before Analysis
The criteria required here are the details and conditions of storage and the preparative methods used to bring the food to a state suitable for analysis. At this stage it is usual to remove inedible material, and it is important to have a description of the inedible material removed and consequently the nature of the edible portion taken for analysis. Some compilations are based on analyses on pooled samples, and it is clearly important to know whether or not pooling was adopted. With cooked foods, descriptions of the cooking methods and recipe are also appropriate.
5.6 Analysis
The choice of the analytical method is frequently seen as a major cause of incompatibility of data. While there is some truth in this, it is not usually a major cause of discrepancies between published values, which are more usually due to the different samples analyzed and the natural variation in the composition of foods. However, it is clearly necessary for wider use of data to be aware of the methods used to obtain the values cited. In many cases reference to the method used is all that is required, and a description of any modification used since such a modification may make a significant difference to the values obtained.
For many nutrients there is a range of methods which give comparable results, and frequently the choice lies between one that uses complex instrumentation, for example, atomic absorption spectroscopy or high-performance liquid chromatography, and a time-consuming method based on simple techniques such as colorimetry or titration. The instrumental procedure is not necessarily better than the simple method that it has replaced, because many laboratories have invested in capital equipment and speed of analysis rather than staff. In some cases, the instrumentation enables determinations that were vitually impossible by manual procedures, such as amino acid analysis, while in other, complex and simpler methods give very similar results. Compatibility does not demand that the same analytical methods are used, but at least analyses based on the same principles should be used. Where the chosen methods are known to give different values, these differences should not be interpreted as genuine indications of variations in the composition of foods.
In order to have confidence in data it is essential that the standards of quality control of analytical work need to be agreed. These are as follows: (1) replicate analyses carried out as a matter of course; (2) analysis of reference materials; (3) regular use of standards; (4) recovery of added standards; (5) analysis of concealed replicates; (6) exchange of samples between laboratories; (7) collaborative tests of methodological protocols, and (8) comparisons of values obtained with literature values.
They are in fact the essential elements of good laboratory practice. It is unfortunate that published papers rarely contain any mention of these checks.
5.7 Factors to be Considered in the Choice of Methods
It is not possible here to discuss the choice of analytical methods for all nutrients. The basic principle should be that the method provides information that is nutritionally appropriate. This can be illustrated with some comments on carbohydrates.
One conventional approach to the measurement of carbohydrate is by difference: that is, by directly measuring the percentage of protein, fat, ash, and water, and deducting these from 100 to give the percentage of carbohydrate. This method is clearly inadequate for virtually all nutritional purposes because it combines in one value all the different carbohydrate species: sugars, starch, and components of dietary fibre, together with all the errors in the other determinations. Nutritional users require much more detail, and, therefore, the methods used must provide this detail. The metabolic effects of different sugars are not the same, and there is a need to have some knowledge of the different sugars present-either a range of specific procedures is required, e.g., enzymatic methods [4], or separation and analysis by a chromatographic procedure [5]. The higher oligosaccharides such as stachyose and verbascose are only poorly absorbed and are associated with gas production in the large intestine and should be measured separately. Analogously starch is associated with specific nutritional properties and values for it are also needed; the choice of method here must avoid interference from β-glucans and other non-cellulosic polysaccharides, and it is probable that enzymatic procedures give more relevant nutritional values.
In the case of dietary fibre, the progress in physiological studies of the mode of action is not sufficient to define the level of detail required, but it is clear that estimates of cellulosic and non-cellulosic polysaccharides are required [1] and that measurement of the monosaccharide and uronic acid components of the non-cellulosic fraction can give information which can be used to predict probable modes of action [6]. Consideration of the carbohydrates in foods leads to the remaining criterion that must be applied to data from other sources.
5.8 Mode of Expression
The use of different modes of expression can be a major cause of incompatibility, and it is important that either an internationally recognized system is followed or that the mode of expression used is stated clearly. For many nutrients there are no difficulties, although the basis of expression, for example, the units chosen or expression on ‘dry matter’ basis as opposed to a fresh or ‘as received’ basis, can produce ambiguities. The major difficulties arise where direct analytical data are converted into the quoted values for nutrients. Thus, the long-standing convention of using protein values derived by applying a factor to measured total N values and the calculation of energy values using the energy conversion factors [7] can produce problems of incompatibility. However, it must be emphasized that the magnitude of the discrepancies arising from these conventions is usually small.
The use of vitamin equivalents, for example in the expression of values for vitamin A activity which involves the conversion of carotene values into units of retinol activity, is another source of ambiguity because two systems are in use [8]. However again, with most foods the differences are small [1]. In principle, it seems preferable for the values in a data bank to include both the direct (analytical) values and the derived conventional values. The use of some analytical conventions is another cause of ambiguity, for example fatty acid data are frequently given in grams per 100 g total fatty acid, and amino acids are often expressed on a N basis: either milligrams per gram N or g/16 g N. The primary need in this connection is that the compilers of the data bank establish a uniform convention in their modes of expression and make certain that the convention they use is stated clearly. In some cases, the choice of one convention is a matter of national preference (or because of national legislation), but if the conventions are defined clearly, the necessary conversions can be made when using data from another source.
5.9 Development of Criteria and Their Adoption
Real compatibility ultimately depends on the acceptance of a series of criteria by those involved in the preparation and use of food composition data systems. In this paper only the main outlines have been sketched and so doing I am continuing the process begun by the Group of European Nutritionists [3] who saw the opportunities for a European system of food composition tables in the 1970s. The next stage in the development of these criteria will be carried out under the auspices of Infoods and lead to the production of a series of guidelines describing in detail the points outlined here.
At present, there is a large volume of compositional data to which the criteria suggested here with regard to the identification of the foods, the collection of the samples, and the analysis of the nutrients have not been applied. Most of these data are sound, and it is not suggested that they be discarded and that a mammoth programme of sampling and reanalysis be started. As new analytical work is started, it should be based on these criteria, and the primary need is that individual data banks are constructed and amplified, so that the existing numerical data are accompanied by information that will enable any user to be certain of (1) the identity of the food, (2) the origin and nature of the samples analyzed, (3) the analytical methods used, and (4) the modes of expression used. With this information the user will be able to assess whether he can use the data in combination with his own.
References
4 Bergmeyer, H.-U.: Methods of enzymatic analysis (Academic Press, New York 1974).
