Contents - Previous - Next

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

User requirements for data bases and applications in nutrition research

Åke Bruce and Lena Bergstrom
Swedish National Food Administration, Uppsala, Sweden

During the last few years there has been an increasing demand for good, reliable food composition tables. At the same time it is becoming more and more expensive to carry out the extensive series of food composition analyses that form the basis for the tables. For this reason, international cooperation is starting. In the Nordic countries an official working group has just begun to co-ordinate the elaboration of food composition tables and data banks.

One important reason for the increased demand for reliable food composition data is the growing research on the connection between diet and cancer. In these epidemiologic surveys it is occasionally found that, not only are the traditional nutrients important, but also a number of other components in the diet have to be considered. There is also an increasing amount of ret search on interactions among different foods and the utilization of nutrients. In these studies it is sometimes necessary to calculate the intake of components other than traditional nutrients.

In this presentation we survey the interest of different groups of scientists regarding their needs for data on nutrients and other components in various foodstuffs. This paper deals only with the demands of human nutritionists. Most of our examples come from a developed country. We are aware that scientists in developing and tropical countries will have other requirements and we have also tried to consider these.

We found it difficult to separate the needs of epidemiologists from those of scientists working on other aspects of human nutrition. One difference is that the epidemiologists mainly work with hypotheses and test correlations between intake of different substances and diseases. In this paper, we consider only components in foodstuffs for which there are reliable data on a causal connection between intake and effect. However, the article, prepared shortly before Christmas, was written with the same feelings a child has when writing his list to Santa Claus of presents he would like. We want good, reliable data for all substances we discuss in this paper, but are realistic enough to believe that only a fraction of our wishes will come true.

We begin by defining the types of data different scientists need. Thereafter different groups of substances - nutrients, other components, and toxicants - are discussed. There are also some comments on the problems of inherent variation as well as on changes during the handling and processing of food. In closing, we request an international coding system.

THE REQUIREMENTS OF DIFFERENT SCIENTISTS

Traditional food composition tables have usually listed only nutrients in foodstuffs. Although these tables have existed for many years, they are far from complete. They usually contain sufficient information about vitamins in commodities and raw products, while data on vitamins in prepared foods are often limited. Usually there are sufficient data on macro-minerals, but far less information about trace elements is available. There are reliable fatty acid composition tables for most foodstuffs containing significant amounts of fats, while the quality of the data on amino acids varies, and when it comes to carbohydrates and dietary fibre we sometimes have surprisingly little information.

One of the major tasks of the clinical nutritionist is to plan diets for patients with different diseases. Some" times these diets must be very specialized, but usually the data in the traditional food composition tables are sufficient. Reliable data on fatty acid composition and cholesterol content are needed when diets for the different types of hyperlipoproteinemias are planned; mono- and disaccharide amounts must be known for planning diets for patients with hypertriglyceridemia, diabetes, and lactase deficiency, etc. The only diets for which the standard tables are not sufficient are those for persons with gluten intolerance, for which data on gluten and gliadins, particularly in complex foods and ready-to-eat foods are needed.

Within the large group of diseases caused by inborn error of metabolism, there are approximately 20 for which specific diets have been prescribed. The majority of these disorders are errors in the metabolism of amino acids (phenylketonuria, tyrosinemia, maple syrup urine disease, cystinosis, etc.), and for the preparation of diets for these patients good tables on amino acid composition in different foods are a necessity. Another group of diseases need data on individual mono- and disaccharides (fructose intolerance, galactosemia) or minerals (Wilson's disease, haemochromatosis).

Finally, there are a few disorders for which other data are needed. Diets for patients with gout or xanthinuria have to be low in purines; for patients with hyperoxaluria, oxalate has to be considered, while for the rare diseases beta-sitosterolemia and Refsum's syndrome, the content of plant sterols, physic acid, and phytol in the diet has to be known.

There seems to be a growing group of patients with allergies to various foods. Usually these patients are allergic to complex proteins in specific foods (for example, fish protein, soy or milk protein, etc.), but some react against rather simple compounds, such as benzoic acid, salicylic acid, some food additives, iodine, etc. Individuals who suffer from migraines are occasionally relieved by diets low in monoamines or similar substances.

There is a growing interest among pharmacologists for studying the effects of different diets on pharmacological actions. A number of naturally occurring components have pharmacological effects by themselves or can alter the metabolism of various drugs. A number of these substances are listed in Table 1. This is not a complete list, and should only be considered as an example of interesting substances.

TABLE 1. Natural Components with Pharmacological Actions

Vaso active amines: tyramine, dopamine, tryptamine, serotonin, hista mine
Psycho-active substances: caffeine, threobromine, theophylline
Goitrogens: glucosinolites, thiocyanate
Estrogenic factors

The toxicological aspect of diets is partly outside the scope of this paper. Usually, toxic substances occur only occasionally in the diet. However, there are a number of naturally occurring toxicants for which the intake can be calculated. Some of these affect the metabolism of nutrients. Table 2 shows a number of naturally occurring toxicants that are rather widespread in different foodstuffs (mainly plants). Sometimes there are reasons to calculate the intake of these substances from the whole diet.

SUBSTANCES IN FOODSTUFFS

Nutrients

Vitamins Modern food composition tables contain reliable data on most vitamins in individual foodstuffs. However, usually only the content of one chemical form of the vitamin is given. Quite often less active forms have been recalculated and expressed in terms of the most active components (e.g., carotenoids to retinol). Table 3 shows the vitamins essential for humans. It also contains data on the different chemical modifications or provitamins that generally occur in different foodstuffs. There seems to be an increasing demand for exact data about the latter components. Their importance as vitamins has to be reevaluated. There are also important pharmacological differences between some of them, e.g., nicotinamide and nicotinic acid. The epidemiological studies on chemo-prevention of certain cancers will probably also need better vitamin data (e.g., various retoinoids seem to have different protective effects against breast cancer).

TABLE 2. Some Naturally Occurring Toxicants

Enzyme inhibitors: trypsin inhibitors, amylase inhibitors, solanine, etc.
Anti-vitamin factors: thiaminase
Cyanogenic glycosides
Metal-binding constituents physic acid
Erucic acid

Human nutritionists have identified the organic substances that are called vitamins and thus essential for man. However, there is a group of compounds that are not essential for humans but occasionally are necessary for other mammals. The human body can usually benefit from an extra supply. They are sometimes called pseudovitamins, and reliable data on the content of the following compounds in different foods are sometimes of interest: inositol, choline, para-aminobenzoic acid, taurine, carnitine, and flavonoids.

Mineral and trace elements. In table 4, those minerals and trace elements regarded as essential for rodents, and most likely also for humans, are presented. There are often reliable data on the macro-elements in food tables. This precision is invalidated by the fact that food preparation can change the content considerably. Also, the data for the negative ions (Cl^-, HPO₃^2-, SO₄^2-, etc.) are of lower quality than the corresponding data for positive ions (Na⁺, K⁺, Mg²⁺, etc.). Among the trace elements, our knowledge about the iron content in different foodstuffs is good. However, on absolute value for the iron content in a diet says rather little about the amount of iron available. A number of factors are known to decrease (physic acid, tannins, etc.) or to stimulate (ascorbic acid, haem iron, meat protein, etc.) iron absorption. On the basis of these data, it is possible to predict the availability of iron in a composite diet. The situation is probably the same for other trace elements, although our current knowledge is insufficient to make similar calculations for the absorption of zinc, chromium, selenium, etc.

TABLE 3. Vitamins and Provitamins

TABLE 4. Essential Inorganic Elements

Protein and amino acids. Several good tables with data on the amino acid composition in different foodstuffs are available today. Occasionally, however, they lack data on those amino acids that are degraded with the usual hydrolytic procedures. Amino acid composition can also be changed during food preparation. Some amino acids react with sugars and form different Maillard reaction products or complexes with alanine (Iysinoalanine). The extent of these losses cannot usually be estimated from present tables.

In the ordinary diet, there are approximately 10 saturated fatty acids, 2-4 monoenes, and about 5 polyenes that ought to be calculated when the fatty acid composition is described. However, there is growing interest in some less abundant types of fatty acids, e 9., trans-fatty acids, longchain polyunsaturated fatty acids (e.g., eicosapentaenoic acid), and long-chained monoenes (e.g., erucic acid).

Carbohydrates and dietary fibre. The content of carbohydrates is traditionally calculated as the difference between dry weight and the weight of components that can be analysed. This is a very inexact method that ought not to be used. However, in many tables this is the source of the carbohydrate data. There are good reasons to have the carbohydrates divided up into starches, monosaccharides (glucose and fructose), and disaccharides (saccharose, lactose, and maltose). A special table containing a number of other, less abundant, saccharides and more or less complicated carbohydrates would also be a great value.

One major problem in research on dietary fibre has been to agree on a suitable analytical procedure that considers the chemical as well as the nutritional aspects of different types of dietary fibre. Different systems have been proposed, and it is still an open question how the data on dietary fibre should best be presented in food composition tables. Until it has been decided which procedure is most relevant for human nutrition, the final presentation of data on dietary fibre in food has to wait.

Other natural substances. The list of components in foodstuffs that are of nutritional interest is very comprehensive. Some of these components have pharmacological properties, as presented in table 1. Other can be metabolized by the human body (e.g., organic acids such as acetic acid, citric acid, etch A few inorganic ions have a widespread occurrence, such as nitrate and nitrite, ammonium, and sulphate ions. Particularly, the intake of nitrate and nitrite has been correlated with the incidence of gastric cancer. However, many of these substances only occur in a limited number of products and thus should not be included in a general food data bank system.

Substances Introduced by Man

Food additives. Most countries have a "positive" list of food additives in order to regulate intake. The Swedish National Food Administration has developed a data programme that makes it possible to calculate the per capita intake of food additives. It contains information on more than 3,000 different food additives and their mixtures in all types of edible foodstuffs available on the Swedish market. Such a system needs continuous maintenance and is labourious to keep up to date. Because regulations vary among countries, it is not likely that an international data bank would be easy to accomplish.

Pesticide residues. The levels of pesticide residues in different foodstuffs vary within wide ranges and are generally below detectable limits. In some products from specific regions the pesticide residues are rather high and constant, which facilitates calculations of the intake of such substances. However, this situation varies from country to country and it is doubtful whether an international data bank would be useful for such calculations.

VARIATIONS

The users of food composition tables always have to consider the reliability of the values. A number of factors, both systematic and unsystematic, influence the levels of nutrients and other compounds in different products. Some of these variations are, to a certain extent, predictable. The composition of bovine milk, and thus also the composition of butter, varies regularly between winter and summer seasons. The nutrient levels in imported vegetables vary, depending on the origin of the products. This is another source of seasonal variation. It is desirable that such variations be documented in the tables.

The different types of biological variations are illustrated in figure 1. Figure 1-a exemplifies fortified products. Technologically, there are no difficulties in keeping a stable level of iodine in table salt or vitamin A or D in margarines. Figure 1-b represents a product with a stable biological composition, e.g., milk or egg. In these products the levels of many nutrients vary within narrow limits.

Figure 1-c could represent meat. Muscle tissue has a rather constant composition, but meat is composed of several different tissues: skeletal muscle, connective, and adipose tissue. Depending on the proportion of these components, the content of nutrients in meat can vary within rather wide limits.

Figure 1-d represents, for example, minerals in many plant foods. Minerals, which are essential for the plants, can also show large variations. Substances that are not essential and usually present at low levels often vary within very wide limits.

There is probably no ideal method of describing the variations in nutrient contents. Often, the only value given is an estimated mean or median value. Sometimes the range is also stated. The best way to present the variation is probably the coefficient of variation. To a certain extent this value describes which of the curves 1-a to 1-d is relevant. Occasionally there is a skewed distribution of the value (1-e). This is, for example, the case for some of the toxic minerals. In these instances the variations are difficult to describe with a single value.

Another complicating factor is the variation in bigavailability for some nutrients, particularly minerals. At present, iron is one of the components for which the availability can be calculated. In future, similar calculations will probably be in use for other nutrients. It may then be an important task for an international system to supply relevant data in order to make these calculations feasible.

FIG. 1. Different Types of Biological Variations in Foods. Fig. 1-a exemplifies fortified products. Fig. 1-b represents stable biological composition as in milk or eggs. Fig. 1-c could represent meat, and Fig. 1-d, minerals in plant foods. Fig. 1-e shows skewed distribution.

VARIATIONS DURING HANDLING AND PROCESSING OF FOODS

All food composition tables contain data on nutrients in commodities and raw materials and in a number of processed products. Sometimes the latter values are based on industrially prepared foodstuffs for which detailed information is often available. Today, calculations on nutrient intake are based mostly on data from raw material and industrially produced products.

The changes in nutrient content during industrial processing have been studied intensively and good knowledge on systematic and unsystematic nutrient losses has been obtained. There are now existing computer systems for calculating nutrient contents in cooked dishes. These systems need continuous revision and improvement, and further cooking characteristics should be taken into account (e.g., preparation time, cooking temperature, number of portions in the recipe, size and material of cooking vessels, etc.).

Food preparation at home is a more unpredictable process and includes vitamin losses and loss of water, protein, and fat from evaporation and from dripping, as well as vitamin and mineral losses from soaking and extraction effects. The availability of some nutrients might also change in either a positive or a negative way.

IDENTIFICATION OF PRODUCTS

An important problem when consulting foreign food composition tables is the identification of individual products. When a Latin name is given it is easier to identify a product. Bovine milk and eggs are also easy to identify. As to meat products, however, the butchering process varies from country to country and consequently the distribution of muscle tissue, connective tissue, and adipose tissue. The recipes for ready-to-eat dishes and foods often vary so much among countries that data in foreign tables usually cannot be used.

An international coding system would increase the utilization of many food composition tables. An ideal system contains groups of numbers or letters representing different types of data. The first group of symbols could, for instance, state the biological source and anatomical location of the product. A second group of symbols could indicate the geographical origin, and, if possible, also the season for harvesting or production. In composite or processed products, the last group of symbols could give information about composition, handling, and preparation.

TABLE 5. Priority Lists

An ideal code could also include other information and would probably contain approximately 10 symbols, figures, or letters, in certain combinations.

PRIORITIES

In this paper we expressed our wishes without any restraints. In our opinion, however, the most important task is to settle the problem of variations that result from handling, processing, and preparation of foods. In the future, the different dishes in food composition tables should be analysed, but recipes for them should also be given to make it possible to compare dishes from one country to another. The recipes should also include the different cooking characteristics states above.

Three groups of nutrients are ranked in table 5. The first group, and that with the highest priority, includes those nutrients for which recommendations are usually given. The second list includes nutrients for which reliable data are urgently needed. The third list indicates the specialized tables that should be assembled in future, but that will be demanded by only a limited number of scientists.

Contents - Previous - Next