Chapter 4

Selection of nutrients and other components

The aim of food composition databases should be to include all nutrients or other bioactive food components that are known or believed to be important in human nutrition. This ideal can rarely be achieved, especially where resources are scarce, and therefore decisions must be made on priorities. Some measure of selectivity is both desirable and practicable, particularly in respect of analytical work, which constitutes the major demand on resources.

The following considerations, in addition to the availability of resources, will govern the selection of nutrients and other food components: 

1. the basic need for information; 
2. health problems in the country concerned; 
3. the state of current thinking in the nutritional and toxicological sciences; 
4. the availability of existing data; 
5. the existence of adequate analytical methods; 
6. the feasibility of analytical work; 
7. national and international nutrition-labelling regulations.

The stages in this process are outlined schematically in Figure 4.1.

The basic need for information

In all countries, information on water, protein, fats, carbohydrates and energy will be required as a minimum base.

Health problems in the country concerned

In countries where deficiency diseases are a pressing problem, information on key vitamins (e.g. vitamin A) and minerals (e.g. iron) will be required. In industrialized countries, however, where problems such as cardiovascular disease, diabetes mellitus, hypertension and cancer are predominant, data on energy, fat, fatty acids, cholesterol, individual carbohydrates and sodium may be seen as top priority. All countries with long dark winters, or where sunlight is prevented from reaching the skin for cultural or other reasons (e.g. purdah, institutionalization), food levels of vitamin D will be required. This range of constituents will be required worldwide if a complete epidemiological evaluation of degenerative diseases is to be made and guidelines for preventive dietary practices are to be established (Rand and Young, 1983). In a country where toxicological problems have been identified, relevant data on food toxins (e.g. goitrogens) or contaminants (e.g. mycotoxins [Van Egmond, 1984; Van Egmond and Speijers, 1999], heavy metals), may need to receive high priority.

FIGURE 4.1 Schematic outline of the selecting nutrient values

The state of nutritional and toxicological sciences

Food components to be included should also reflect the general state of nutritional and toxicological thinking. A comprehensive database should include all nutrients for which recommended intakes have been established nationally and, where appropriate, internationally.

In addition, those involved with the preparation of databases should try to anticipate needs for data. Interest in “new” or “rediscovered” components of food can escalate rapidly (Southgate, 1985); thus those who are responsible for database programmes must be aware of current developments and the interests of nutritional and clinical scientists. There is now, for example, significant interest in values for the glycemic indices of foods (Brand-Miller et al., 1999). These give a measure of the rate at which carbohydrates are digested (see Chapters 6 and 7) and some tables have been produced (Foster-Powell and Miller, 1995). Caution may be necessary in interpreting responses to questionnaires, however. For example, when Paul and Southgate (1970) reviewed the requests of some users of the United Kingdom food composition tables, they discounted advice to exclude nutritionally unavailable carbohydrates, because they were aware of the growing interest in dietary fibre.

Although these guidelines are primarily concerned with the provision of nutritional information, there is growing recognition that a wider range of constituents play an important role in the relationship between diet and health (Ames, 1983). These include naturally occurring biologically active constituents such as a range of phytochemicals including phytates, oxalates, flavonoids, glucosinolates and phytosterols. Some of these components, such as goitrogens (Gaitan, 1990; Speijers and Van Egmond, 1999) alter the nutritional values of foods, through interactions in the food or gut, or during metabolism. There is also interest in including information on food additives and contaminants in databases (Louekari, 1990; Burlingame, 2001). The amounts of additives in foods are highly brand-sensitive and often subject to variation with time, so it is particularly important for these data to be date-marked. The distribution of contaminants is often more complex than the distribution of the naturally occurring constituents within foods and representative values may be difficult to establish. Furthermore, sampling procedures for contaminants are often designed to identify maximum likely exposure in a population, and it may be misleading to list contaminant values in the same record as nutrients. For these reasons, these guidelines make only limited reference to contaminants, although their importance is recognized (Young, 1984).

Availability of existing data

A great deal of information is available for certain nutrients or non-nutrient components that have been the focus of research or have been measured for regulatory purposes. These data should be employed, provided that they meet the programme's quality criteria. Where resources are limited and preclude inclusion of all components in the user database, it would still be useful to store all available data at the archival levels of the data system.

Existence of adequate analytical methods

The availability of reliable analytical methods is an essential determinant of components for inclusion (Stewart, 1980) (see Chapters 6 and 7). It will not be cost-effective to analyse foods for a particular nutrient, however high its priority, if methods are untried or yield conflicting values. When methods are in doubt, it may be appropriate for methodological studies to be implemented as part of the database programme.

The emergence of a reliable new or improved method for measuring a nutrient may create the need for analysis (or reanalysis) of foods that are important in the food supply or that are known or suspected to be good sources of the nutrient concerned.

Feasibility of analytical work

The commissioning of analyses for each nutrient must be governed by practical factors: the cost and time required, and the availability of equipment, trained personnel, chemicals, etc. These are major considerations, especially in some developing countries. Costs must always be weighed against the nutritional or clinical requirements for particular nutrients. Where resources are limited it may be useful to search out other laboratories, such as governmental regulatory laboratories or those working on soil chemistry, for collaboration. Borrowing or calculating values would be the final option.

National and international nutrition-labelling regulations

Nutrition labelling has emerged in recent years as one of the more important and demanding areas involving food composition. The key international body concerned is the Codex Alimentarius Commission (FAO/WHO, 2003), operated jointly by FAO and WHO. Complete food labelling text, with a section on nutrition labelling, is available in print and electronic form (FAO/WHO, 2001). Compliance with Codex Alimentarius is voluntary, and many countries have their own unique nutrition-labelling regulations (FDA, 2001; EC, 1990; FSANZ, 2001). It is useful for food composition programmes to include all the nutrients required in their national nutrition labelling as well as those required in the labelling regulations in countries within their region. For food exporting countries, the nutrients required in the regulations of major trading partners are also important for inclusion in the food composition database.

Coverage at different stages of data management

As noted earlier, ideally, a food composition database system should include values for as many nutrients and other components as possible, with technical provision for adding more information as it becomes available. However, because a comprehensive database system is a national reference resource, it is useful to list the values for individual forms of nutrients separately, where separate analytical values are available or can be obtained, particularly in a reference database. The factors used for converting the different forms of a nutrient to a single value to give an indication of its biological value may change as the state of nutritional science advances. If only the calculated (derived) value is recorded in the database management system, it will not be possible to recalculate the putative total biological activity; thus, it is desirable that the measured values appear in addition to calculated values. In any event, all conversion factors used should be listed in numeric data fields as equivalent to components, or in the documentation sections of the database.

Component data can be expressed on many different bases. For example, amino acids can be expressed as mg per g nitrogen (N) (or as g per 16 gN) and fatty acids as percentages of the total fatty acids, and this is the preferred format for entering such data, if this is the way in which they were obtained from the analytical laboratory. However, at the user level, it is often more useful to present all the data for a particular food as g per 100 g edible portion (or per 100 ml for some beverages, along with density values). User databases (or, more usually, printed tables) will vary in complexity and coverage; hence specific decisions must be made on each component for the different data outputs. Thus, data may be presented as “total” or “available” values for nutrients, for which several forms exist, calculated using appropriate factors and a documented algorithm.

Analogously, in simplified printed tables it may be desirable to regroup some components, such as fatty acids and cholesterol, into separate sections. This will almost certainly be the case when printing costs are a constraint.

In the case of special-purpose tables, many formats are possible. In tables for non­specialists, values may be grouped (e.g. fat <1 g, 1–5 g, 5–10 g, etc.), or foods may be listed according to their ranking as sources of nutrients (excellent, good, fair, poor) depending on the proportion of the recommended daily allowance present in an average serving.

Suggested coverage of nutrients for different levels of data management is given in Table 4.1, and Table 4.2 provides examples of data dissemination formats. Comments on some of these components follow, and further details can be found in Chapters 6 and 7.

Water

It is essential to give values for water content in published tables and papers on food composition and at all levels of data management, including the comprehensive user database. Variations in water content are important determinants of the levels of other components, and data on water content make it possible to compare nutrient values (e.g. for different foods or different analyses of the same food) on a similar moisture basis. This information is also essential when data from different sources are being compared or combined. Analyses for some nutrients are conveniently performed on the dry matter (DM) sample. Therefore laboratory data may be reported per 100 g DM, and recorded in the reference database in this way. However, each DM value must be related to the analysed water content of the same sample, so that nutrient values can be recalculated to their appropriate fresh-weight basis. In simplified printed tables it may be unnecessary to list water content, but it should only be omitted when space is a critical constraint.

Protein

Values for protein are required at all levels of the data system. Conventionally, they are based on total nitrogen values using a nitrogen conversion factor (FAO/WHO, 1973), with all factors being recorded at the food level in the database. Values can also be based on the total nitrogen minus the non-protein nitrogen multiplied by a specific factor related to the amino acid composition of the food, or as the sum of amino acids (see Chapters 6 and 7). New amino acid data used in conjunction with the ratio of total amino acid residues to amino acid nitrogen seem to suggest that the nitrogen conversion factor should be lowered. Sosulski and Imafidon (1990) suggest a global conversion factor of 5.7 and Salo-Väänänen and Koivistoinen (1996) of 5.33, both with individual factors for different foods and food groups. At this time no new international agreement on conversion factors had yet been reached.

Total fat

Values for total lipids vary considerably with analytical method (see Chapters 6 and 7) and may be of limited nutritional significance; nevertheless, they are widely used and should be included at all levels of the database.

Fat (triacylglycerols). Inclusion of this item is desirable in the reference database, primarily for use in the calculation of food energy value, and also because of the interest in triacylglycerols from animal and vegetable sources. The widespread and increasing use of mono- and acylglycerols in manufactured foods is an additional reason for its inclusion.

Phospholipids. Values for the different classes of these substances should be included at the reference database level because of their wide use as emulsifying agents, and because of their physiological properties.

Sterols. Although cholesterol was once considered the most important sterol from a nutritional viewpoint, the significance of the other sterols (e.g. sitosterol) is now recognized; they should be included at the user database level.

Fatty acids. Data for individual fatty acid stereoisomers should be included in the reference database. At this level, the most convenient mode for expressing fatty acid values is as g fatty acid per 100 g total fatty acids. In user databases, however, expression as g fatty acid per 100 g of food is more useful. In simplified user databases the fatty acids may be grouped into total saturated, total mono-unsaturated and total polyunsaturated acids, or the ratio between the groups may be cited together with the total fat value. Another grouping of major interest is as n-9, n-6 and n-3 families of unsaturated fatty acids (Gurr, Harwood and Frayn, 2002).

Carbohydrates

Values for available (glycemic) and unavailable (non-glycemic) carbohydrates derived by analysis are desirable throughout the database system. The earlier practice of including carbohydrate calculated “by difference” has proven to be scientifically unsound and should be phased out as soon as possible (FAO/WHO, 1998).

Available carbohydrates (glycemic). These include all the sugars (glucose, fructose, sucrose, lactose and maltose) known to be glucogenic in humans and the polysaccharides (starch and partially hydrolysed starches, and glycogen) hydrolysed by the endogenous secretions of the human digestive tract (Table 4.3).

Unavailable carbohydrates (non-glycemic). These include all the polysaccharides that are not hydrolysed by the endogenous secretions of the human digestive tract: components of the plant cell wall (cellulose, non-cellulosic polysaccharides, pectic substances and hemicelluloses) and a range of polysaccharides used as food ingredients or food additives. These together are the non-starch polysaccharides (NSPs), which are often used as a definition of dietary fibre. There are several other definitions of dietary fibre, each identified by a different methodology, and each measuring different amounts of the non-glycemic carbohydrates, and other non-carbohydrate material (e.g. lignin).

Oligosaccharides. There is growing recognition of the potential nutritional importance of this group and an associated need to start assembling values for these components. Oligo­saccharides include tri-, tetra- and pentasaccharides of the raffinose series, analogous malto­derivatives and a range of fructose polymers, including those at the lower end of the poly­saccharides. Individual oligosaccharides need to be recorded separately because they are metabolized differently.

Polyols (sugar alcohols). These comprise a group of polyhydric alcohols structurally related to the sugars where the reducing group has been reduced to a hydroxyl compound. Very small amounts of them occur naturally in foods, but they are widely used as food additives for their humectant properties or as a replacement for sugars in reduced-energy products, low cariogenic sweets and foods for diabetics. Under the labelling regulations of some countries, polyols are included in the carbohydrate declaration, but in a nutritional database it is preferable to list them separately under their specific trivial names. Table 4.3 indicates the more important polyols used in foods.

Organic acids

These are important in relatively few foodstuffs, and their inclusion in a user database should be selective. Values should be given for fruits, fruit products (including juices), a few vegetables (particularly those preserved in acetic acid), and other manufactured products, such as vinegar, salad dressings that have organic acids listed as major ingredients, soft drinks and yoghurt. In these cases, organic acids should be included in energy calculations.

Alcohol

Alcohol (ethyl alcohol) may be a significant energy contributor; levels must be determined and used in energy calculations for alcoholic beverages, and for confectionery and desserts

Inorganic constituents

Total ash. Values for ash are frequently given in data sources and the values should be entered into the database system primarily because they can be used in internal checks on the sum of all the proximate components, the calculation of total or available carbohydrate by difference and the mineral content. Because the values are not of nutritional significance, they need not appear in simplified tables.

Individual inorganic constituents. All the essential inorganic elements should be included. Current instrumental techniques provide information on a wide range of minor trace constituents with little extra cost, and it is desirable to include a comprehensive list. The forms in which some trace elements occur are important in relation to their bioavailability and should therefore be recorded when this information is available.

Vitamins

Many vitamins occur in several active forms called vitamers; if it is technically possible, the vitamers should be analysed separately and the values held separately in the database system, in some cases at the user database level. In simplified tables, it will usually be enough to provide a value for the total activity of the vitamin in question. It is, however, essential to document the algorithms used to calculate these estimates of total activity.

Non-nutrient constituents

Contaminants. Contaminants include mycotoxins, heavy metals and residues of pesticides, herbicides and animal growth promoters. The distribution of contaminants in foods is such that the concept of representative values for contaminants differs from that for nutrients. It may be misleading to list contaminant values in the same record as nutrients. Listing in archival and/or reference auxiliary data records is preferred.

Bioactive substances. There has been a growing interest in the range of dietary phytochemicals in recent years, particularly in view of their possible protective action against cardiovascular diseases and certain cancers. These include isothiocyanates, polyphenols, flavonoids, isoflavones, lignans, saponins and coumestrol (AICR, 1996; Pennington, 2002). Consequently, there is a parallel interest in the inclusion of phytochemicals in food composition databases (Ziegler, 2001). The collection of data from data sources is useful, although it may not be possible to find complete data sets.

Antinutrients and toxicants. Some constituents have undesirable physiological effects, for example, goitrogens, haemagglutinins, antivitamin factors, trypsin inhibitors, oxalic acid and phytic acid. Data for these components should be included for the relevant foods. Other important natural toxicants include solanine, cyanides, glucosinolates, lathyrogens, mimosine and nitrosamines. Ideally, data for these natural components should be incorporated in the reference database.

Additives. Many additives are measured, in whole or in part, during the course of nutrient analyses. Salts, for example, are included in analyses for various cations and anions; protein additives are determined in nitrogen analysis; and some emulsifiers and thickeners are included in analyses for nitrogen, starch and unavailable carbohydrates. Clearly, specific analyses are preferable. However, the need for data on additives and other non-nutrient components of foods may relate to priorities regarding food safety and not necessarily to nutritional priorities.

Miscellaneous. Where data exist for other compounds of interest, such as caffeine, theophylline, theobromine, tannins and other bioactive compounds (carnosine, carnitine, creatinine), they should be listed in the database at least up to the reference level.