Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
European Journal of Clinical Nutrition (1999) 53, 60±79 ß 1999 Stockton Press. All rights reserved 0954±3007/99 $12.00 http://www.stockton-press.co.uk/ejcn Comparison of nutrients in the food composition tables available in the nine European countries participating in EPIC G Deharveng1, UR CharrondieÁre1, N Slimani1, DAT Southgate2 and E Riboli*1 1 International Agency for Research on Cancer, 150 cours Albert Thomas, 69372 Lyon CeÂdex 08, France; 28 Penryn Close, Norwich, Norfolk NR4 7LY, UK Food composition tables were studied from nine European countries participating in the European Prospective Investigation into Cancer and Nutrition (EPIC): Denmark, France, Germany, Greece, Great Britain, Italy, The Netherlands, Spain and Sweden. They were compared from the point view of availability, de®nition, analytical methods, and mode of expression of the nutrients of interest for EPIC, and it was seen that most of the nutrients in the tables are analysed and expressed in a compatible way. For some nutrients, however, common methods and de®nitions (folate, dietary ®bre), or modes of expression (energy, protein, carbohydrates, carotenes, vitamin A and E) have not yet been agreed upon, so values are not comparable. For vitamin C a wide range of values are found due to the high natural variation in foods. For compiled tables, an additional problem is the use of several sources which may mean that the nutritional values are not comparable within the same table; and these values cannot be converted if the source is not stated. In addition, some tables were compiled using food composition values produced over 20 years ago with outdated analytical methods. In view of the inconsistent values for some nutrients and due to the large amount of foods reported within EPIC, it was concluded that standardised food composition tables have to be developed for the nine European countries involved in EPIC in order to provide comparable nutrient intake data. Descriptors: comparison; Europe; dietary; food analysis; food composition Introduction The assessment of dietary exposure is critical for the interpretation of the relationship between nutrition and health. Food intake may be measured with several dietary methods, which assess either recent diet (for example, 24 h recall, 7-d-diary), or long-term intake (for example, food frequency questionnaire). All dietary assessment methods have potential sources of errors which need to be identi®ed and controlled in order to obtain dietary intake estimates as close as possible to the true food intake (Bingham, 1987). Once data have been collected on the foods consumed, these are in general aggregated to match the items available in food composition tables, which introduces additional imprecision in nutrient intake estimation. Further potential error lies in the tables which use different systems to name, group and describe foods, and different de®nitions and chemical analytical methods for nutrients, which do not always provide comparable nutrient values. Furthermore the nutrient values are not always expressed consistently, and little or no information is available on the sampling of foods and on the extent of variation of the analysed values (Slimani, 1993). The sum of the above errors may result in an incorrect estimation of nutrient intake, especially at the individual level though possibly less at the group level. In studies on diet and health this measurement error may *Correspondence: Elio Riboli, IARC, 150 cours Albert Thomas, 69372 Lyon Cedex 08, France. Received 4 February 1998; revised 1 August 1998; accepted 18 August 1998 In this paper the term `table(s)' will be used both for tables of food composition and for nutritional databases. affect the observed relationship with disease risk, which may appear arti®cially strengthened or weakened. In order to obtain comparable nutrient intake values, especially in an international, multi-centre study, it is important that all methods and tools be standardised between each participating centre. The standardization has to be done at every level of the study such as the data collection, aggregation and coding of foods, and when applying the tables. This is one of the aims of the European Prospective Investigation into Cancer and Nutrition (EPIC) with its 460 000 subjects from nine European countries (Denmark, France, Germany, Greece, UK, Italy, The Netherlands, Spain, Sweden). Standardised protocols have been developed to collect data on previous illness, current medication, several lifestyle factors, anthropometry, diet, and collection of blood samples. For the dietary assessment, study subjects are requested to complete a centre-speci®c dietary questionnaire on their usual eating pattern. Additionally, a representative subgroup is interviewed by trained interviewers using EPIC-SOFT, a software programme specially designed to collect standardised 24 h diet recalls of foods consumed during the preceding day (Riboli & Kaaks, 1997; Slimani et al, 1998). Once simpli®ed, reported foods will ®rst be analysed at the food group level, comparing the quantities of foods consumed per food group. A second stage will consist in carrying out analyses at the nutrient level. Within EPIC, a large volume of nutritional data will soon be ready for analysis, and the question arises whether national tables can be used as such, or whether standardised EPIC tables need to be developed. This paper describes a detailed comparison Ð carried out at the International Agency for Research on Cancer Comparison of nutrients in nine European countries G Deharveng et al (IARC) in 1997 Ð of the nutrients present in the tables available in the nine EPIC countries. It focuses on the de®nitions and analytical methods of nutrients after a short description of the tables. This should be of assistance to compilers of nutritional values and users of food composition tables who wish to have an overview of the availability and comparability of nutrient values in the European tables studied. Materials and methods The tables included in this comparison are those commonly used in the nine European countries participating in EPIC. Additional tables do exist but they are either inadequate because of limited coverage or have been replaced by more recent data. The information presented in this review is Table 1 mostly taken from the introductory texts of the tables and from additional information provided by EPIC collaborators and by the local nutritionists in charge of compiling the tables. We shall ®rst describe the individual tables in general terms (Table 1) and secondly the coverage of major nutrients (Table 2). Thereafter, we shall present the results of the comparability and the reliability of nutritional values in terms of de®nitions and analytical methods. The comparison is focused on a selected number of nutrients of interest for the EPIC study with respect to a possible etiological relationship between cancer and nutrition. Additional tables describing the methods of analysis, de®nitions, modes of expression and units used to gather the information are not included in this article and can be obtained from the authors upon request. List of food composition tables as reported by the EPIC countries Country=year Denmark 1996 ± 4th ed. England 1991 ± 5th ed 9 supplements: 1988 ± 4th ed 1989 ± 4th ed 1991 ± 4th ed 1992 ± 5th ed 1992 ± 5th ed 1993 ± 5th ed 1994 ± 5th ed 1995 ± 5th ed 1996 ± 5th ed France 1995 ± 2nd ed 3 supplements: 1987 ± 1st ed 1987 ± 1st ed 1993 ± 1st ed Germany 1979 1992 1993 1994 ± 5th ed 1996 Greece 1992 Italy 1994 1998 The Netherlands 1996 Spain 1988 1990 1992 1993 1996 Name of the food composition tables Shortname used in the tables Availability: Presence of Range of P printed Number of cooked nutrient D disk foods=recipes foods values Levnedsmiddeltabeller DK96 PD 826 McCance and Widdowson's `The composition of foods' GB 91 PD 1188 Cereals and Cereal products, third supplement Milk products and Eggs, fourth supplement Vegetables, Herbs and Spices, ®fth supplement Fruits and Nuts, ®rst supplement Vegetable dishes, second supplement Fish and Fish products, third supplement Miscellaneous Foods, fourth supplement Meat, Poultry and Game, ®fth supplement Meat products and dishes, supplement, sixth supplement GB cereal GB milk GB veg GB fruit GB vdish GB ®sh GB foods GB meat GB mdish PD PD PD PD PD PD PD PD PD 360 335 463 340 347 308 418 429 286 ReÂpertoire geÂneÂral des aliments F 95 PD 800 ReÂpertoire geÂneÂral des aliments - Tome 1 Table de composition des corps gras ReÂpertoire geÂneÂral des aliments - Tome 2 Table de composition des produits laitiers ReÂpertoire geÂneÂral des aliments - Tome 3 Table de composition des `Fruits exotiques, fruits de cueillette d'Afrique' F fat P 83 F milk P 110 F fruit P 186 Energie und NaÈhstoffgehalt von Lebensmitteln Nutrition composition tables of milk and dairy products NaÈhstoffe in Lebensmittel Food composition and Nutrition tables BLS - Bundeslebensmittel-schluÈssel II-2 D 79 D milk D 93 D SFK D BLS P P P P D 841 930 1450 741 11 000 Greek Food Composition Table GK 92 PD 764 Tabelle di composizione degli alimenti Banca dati italiana di composizione degli alimenti per studi epidemiologici IT 94 IT 98 P PD ca. 400 ca. 800 NEVO Tabel - Nederlands voedingsstoffenbestand NL 96 PD 1476 Tablas de composicioÂn de alimentos Tablas de composicioÂn de alimentos La composicioÂn de los alimentos Tablas de composicioÂn de alimentos espanÄoles Contenido de carotenoides en verduras y frutas de mayor consumo en EspanÄa Tablas de composicioÂn de alimentos espanÄoles E 88 E 90 E 92 E 93 E carot P P P P P 543 294 234 413 60 E 97 P 68 S 94 S 96 D PD 1900 1100 1997 Sweden 1994 being updated DBNY94: recipe and food database 1996 Food composition table ± Nutrients and Energy 61 Comparison of nutrients in nine European countries G Deharveng et al 62 National food composition tables included in the review In all countries involved in EPIC, at least one table is available. Table 1 gives the title and the year of publication of the tables studied together with the abbreviations used in this article and some general information on the tables. The Appendix gives the addresses of the responsible institutes and the contact persons for queries on the data and for orders. Seven countries have one national table, some with socalled supplements on speci®c food groups. In this case one national institute is responsible for the analysis, compilation, update and publication of the table (Denmark, UK, France, Greece, Italy, The Netherlands and Sweden). In Denmark, the national table contains 826 raw and cooked foods, of which most have been analysed. The sources of the data are well documented, so values can easily be traced. In the UK, the complete National Nutritional Database is provided by the nine supplements on 3286 foods. The main table, GB 91, is in fact only an interim table prepared when the 4th edition was out of print. The majority of values are based on local chemical analysis or are calculated in a standardised way; some values have been carried on from the 4th edition using incompatible analytical methods; other values have been compiled from other sources, especially in two supplements (on ®sh and on herbs and spices). The British tables cover the widest range of cooking methods, including those using fat, such as frying. For the purpose of this review, the supplements are regarded as the national database, and GB 91 is used as the reference for the de®nitions and analytical methods. In France, the main table (F 95) covers about 800 raw and cooked foods; it includes the main data of the three previously published supplements (fat, milk, and exotic fruits). When comparing the nutritional values, care must be taken: in the fat and milk supplements the values are expressed per kg of food, whereas all the other tables express them per 100g of food. These two supplements are under revision and should be released by the end of 1998. In Greece, the of®cial printed table GK 92 is based on the 4th edition of the British table (1978) and includes only a selected number of nutrients from this table. GK 92 contains 764 foods and recipes; Greek recipes have been calculated using the British nutritional values. The national School of Public Health in Athens has developed a computerised food database based on GB 91 and on ®ve British supplements, it is extended and updated regularly. In Italy, a well-documented compilation was published mid 1998 (IT 98) with about 800 raw foods. About 32% of the nutritional values are taken from the national table published in 1989 (IT 89) which contains 400 raw foods and about 3% from the national table revised in 1994 (IT 94). As IT 98 covers more nutrients and foods than IT 89, some values are derived from other sources (British, American and German data) which are identi®ed at the nutrient level. In the Netherlands, the data on most of the 1476 raw and cooked foods have been analysed locally. The sources of data are coded at the nutrient level and are available on request. The complete computerised database is available to a very limited public; it contains varying numbers of nutrients depending on the food. In Sweden, the printed and PC-version of the national table do not cover exactly the same nutrients. They contain analytical and compiled data from local and foreign sources, but do not indicate the sources of values at the food or nutrient level. The MalmoÈ Diet and Cancer Study has developed a table (S 94) containing 1900 raw and cooked foods and recipes based on the 1993 PC-version of the national table, with some recipes added in MalmoÈ. Since then the national table containing some 1100 raw and cooked foods has been updated and published in 1996 (S 96 refers to the printed version of the national table). Germany and Spain have several independent tables. In Germany, the Bundeslebensmittelschluessel (D BLS) is widely used in epidemiological studies; it includes about 11 000 raw and cooked foods and recipes for which nutritional values have been mostly calculated. About 90% of the analysed nutritional data of the D BLS derive from the D SFK table, which contains 741 analysed raw foods. The D SFK is dif®cult to use because of its open-ended format and the fact that it is only available on paper. In Spain, most tables are based on rather old data. Sources of data and de®nitions of nutrients are not well documented, which means that the data are neither really reliable nor comparable. In E 90, for example, nutritional values are copied as such from different sources and are not compatible within the table itself because they do not follow common criteria in terms of mode of expression or method of analysis=mode of calculation. In 1997, a table (E 97) was published including 68 raw foods for which all nutrients were analysed locally, and de®nitions and methods of analysis are indicated. This is a ®rst attempt to produce up-todate and reliable Spanish data. All tables are available as printed versions, except two databases (S 94 and D BLS). A large number of tables are available on diskette and, generally speaking, these versions are more complete and up-to-date than the printed ones. The Spanish, most German tables (except D BLS) and IT 94 only exist on paper. All tables are published in the national language and most have an English translation; a few also have a French one (D milk and D SFK) or indicate Latin names (British tables, D SFK, IT 94, DK 96, E carot and S 96). The LANGUAL description system (Pennington et al, 1991) is not yet extensively used: F 95 gives the description of each food according to this system, while DK 96 uses it to describe only certain foods. Most computerised tables as well as the French supplements and D SFK have an open-ended format, that is, a varying list of nutrients depending on the food. With this type of format it is not obvious whether values are missing because the nutrients have not been analysed or because they are not present in the food item (zero value). The other printed tables use a ®xed format, that is, the same list of nutrients for all foods. The printed British tables indicate X X X X 6 X X 2,3 X X X X X X 9 X X X X X X X X X X X X X 6 ± X X X X X P X ± X X X X X ± X ± X X X X X X X 4 4 4 X X X X X X X X X X X X ± X X X X X ± X X X T, P ± X X X X X X X GB X DK 96 X X ± X X X X X P ± ± X X X X ± ± X X X X X X ± ± ± ± ± ± X ± X X X X ± X X X X X X X X F 95 3 3,6 3 3 3 3 3 3 3 ± ± 3 3 3 3 3 ± 3 ± 3 3 3 3 3 3 ± ± ± ± X ± X X ± ± ± ± X 4 4 4 3 ± X F fat5 3 3 3 3 3 3 3 3 3 (P) ± 3,7 3 3 3 3 3 ± 3 3 3 3 3 3 3 3 ± 3 3 3 X ± X 6 ± ± 3 ± X 4 4 4 X ± X F milk5 3 3 ± ± 3 ± 3 3 3 (P) ± 3 3 3 3 3 3 ± 3 3 3 3 3 3 3 3 ± 3 ± 3 X ± X X ± 3 ± X X 3,4 3,4 3,4 ± ± X COUNTRY=TABLE F fruit X X ± X X X X X X ± ± X X X ± ± ± X X X X X X X X ± X X X X ± X 6 ± X ± Xcrude X X ± X X ± X D 79 X X X ± X X X X X ± X X X X X ± ± X X X X X X X X ± X X X X ± X 6 ± ± X X X ± ± ± X ± X D milk Availability of food nutrients in the food composition tables. (The list is not exhaustive for some countries.) Unless indicated, nutrient values stand for 100 g food Energy Macro nutrients Water Total nitrogen Protein Total CHO1 Sugars1 Starch1 Lactose Dietary ®bre Fat Total saturated FA Total monounsat. FA Total polyunsat. FA Cholesterol Alcohol Minerals Na K Fe Ca P Mg Cu Zn Se I Cl Mn Vitamins Vitamin C Vitamin E, total Vitamin A, total Retinol fraction1 Carotene fraction1 Vitamin D Thiamin (B1) Ribo¯avin (B2) Niacin, total (T), preformed (P) Tryptophan=60 Tryptophan Vitamin B6 Vitamin B12 Folates (B9) Pantothenates Biotin Vitamin K1 Table 2 X X ± X X X X X T ± ± X X X ± ± ± X X X X X X X X X X ± ± X ± X X X X ± X X X X X X 9 X D 93 3 3,6 3 3 3 3 3 3 3 (P) ± 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 X ± 3 3 3,6 3 3 3 3 3 ± ± 3 3 X D SFK Table 2 continued X X X X ± X X X T, P ± X X X 6 X X X X X X X X X X X ± X X X X ± X 6 ± X X X X X X X X X X D BLS Comparison of nutrients in nine European countries G Deharveng et al 63 X X ± X X X X ± X X 2 2 2 2 3,9 X X X X X ± ± ± ± ± ± ± X ± X ± ± ± X X P ± ± ± ± ± ± ± ± X ± X X X X ± X X X X X X 9 X X X X X X ± X ± ± ± ± X ± ± X X ± X X T ± ± X ± ± ± ± ± IT 94 X GK 92 X X X X X X X X P ± ± X ± X ± ± ± X X X X X ± ± X ± ± ± ± X ± X X X X ± X X X X X X X X IT 98 X X X 3,8 ± X X X P ± ± X 3,8 3,8 ± ± ± X X X X X X X X X 3,8 ± ± X 3,8 X 6 X X 3,8 X X X 6 X X X X NL 96 X ± X ± ± ± X X X ± ± ± ± ± ± ± ± X X X X X X X ± ± ± X ± X ± X X ± ± ± ± X ± ± ± ± ± X E 88 X ± X ± ± X X X P ± ± X X X ± ± ± X X X X X X ± X ± X ± ± ± ± X X ± ± ± X X X X X X X X COUNTRY=TABLE E 90 X ± X X ± X X X T ± ± X X X ± ± ± X X X X ± X ± X ± X ± ± X ± X X ± ± ± X X 3 3 3 3 X X E 92 X X X ± ± X X X T ± ± X X X ± ± ± X X X X X X ± X ± X ± ± ± ± X X ± ± ± X X X X X X ± X E 93 X X X ± ± X X X X ± ± X X X ± ± ± ± ± X X ± X ± X ± X ± ± X X ± X ± ± ± X X X X X X ± X E 97 X X X X ± X X X T,P ± X X X X ± ± ± X X X X X X ± X X ± ± ± X ± X 6 ± ± ± X X X X X X X X S 94 X X X X ± X X X T,P ± ± X X X ± ± ± X X X X X X ± X X ± ± ± X ± X X X ± ± X X X X X X X X S 96 64 X nutrient present in the table; ± nutrient not present in the table.1 refer to detailed table for de®nition and=or term used. 2 given in appendix for some foods. 3 only for some foods. 4 also in % total fatty acids. 5 nutrient values are for 1 kg of food in this table (instead of 100 g of food). 6 refer to detailed table for additional fractions. 7 maybe in mg=gN. 8 not in the printed version, available only in the database. 9 expressed in g=100ml. Energy Macro nutrients Water Total nitrogen Protein Total CHO1 Sugars1 Starch1 Lactose Dietary ®bre Fat Total saturated FA Total monounsat. FA Total polyunsat. FA Cholesterol Alcohol Minerals Na K Fe Ca P Mg Cu Zn Se I Cl Mn Vitamins Vitamin C Vitamin E, total Vitamin A, total Retinol fraction1 Carotene fraction1 Vitamin D Thiamin (B1) Ribo¯avin (B2) Niacin, total (T), preformed (P) Tryptophan=60 Tryptophan Vitamin B6 Vitamin B12 Folates (B9) Pantothenates Biotin Vitamin K1 Table 2 (continued) Comparison of nutrients in nine European countries G Deharveng et al Comparison of nutrients in nine European countries G Deharveng et al for all nutrients either a value, or that no value is available, or mark `in trace amounts'. Nutrients included in the review The nutrients studied in this review are those of interest from a scienti®c point of view for research on the relationship between cancer and nutrition. This includes water, total N, protein, fat, total carbohydrates and some fractions (lactose and starch), energy, FAs (fractions and individual), cholesterol, alcohol, dietary ®bre, calcium, iron, potassium, free retinol, b-carotene, vitamin D, total vitamin E, individual tocopherols and tocotrienols, thiamin (B1), ribo¯avin (B2), vitamin B6, folates=vitamin B9, vitamin B12 and vitamin C. Other nutrients may be added to this list in the future. As shown in Table 1, a few tables systematically state the range of the analysed nutritional values, which is a guide for evaluating the natural variation of a given nutrient, and the number of analysed samples (French supplements, DK 96, D SFK); some do so occasionally (GB, IT 94). As the values depend on the sampling of foods and on what is regarded as the edible part of the food, it is helpful that all tables indicate the edible part. Sampling is extremely important in the process of producing nutritional values, and at the moment only the British tables provide some information on the sampling of foods. The source of the nutritional values (laboratory, table, literature, etc.) is indicated at the nutrient level in DK 96, IT 98 and NL 96; at the food level in the British tables and only in broad terms in the other tables. The source at the nutrient level is the only way to identify correctly the food being analysed and to verify the de®nition, analytical method, and mode of expression originally used. It enables the user to trace whether nutrient values are derived from different samples, for a given food, and=or from different methods which may not be compatible within a given table, for a given nutrient; and to correct them if possible and necessary. When all values have been analysed locally according to common criteria (for example, IT 94, E 97), information on sampling should be given in addition to that on analytical methods and modes of expression. For commercial foods, some tables utilize information from the manufacturers, either given directly to the compiler or taken from labels. As information on de®nitions and methods of analysis is not always provided, it is dif®cult to make the nutritional values compatible with the values from other sources, whenever necessary, before they are included in the food composition table. Results of the comparison of food composition tables for terminology, methods of analysis, availability and compatibility of nutrients Table 2 summarizes the coverage of the selected nutrients in the European tables studied. It shows that most tables state energy, water and all macronutrients (carbohydrates, protein, fat, alcohol, dietary ®bre), most vitamins and the important minerals such as sodium (Na), potassium (K), calcium (Ca), iron (Fe), magnesium (Mg) and phosphorus (P). 1. Water The water content of foods is not included in some Spanish tables (Table 2). Although water has no nutritional value, it is needed to identify and compare foods correctly. For example, when, for the same food, different tables state incomparable values for a given nutrient it is necessary to check whether the foods are in fact different, based on different water and=or fat contents, or whether the values as such are different because of natural variation. It is also useful to know the water content in order to estimate water loss due to cooking. Water content is calculated by drying. Several drying methods are in use, which give comparable results. 2. Macronutrients and energy 2.1 Protein Protein values are usually calculated based on the total-nitrogen (N) content of the food, which is measured by the Kjeldahl-type methods (HAOAC, 1980). Total-N values are given in GB, NL 96 (PC-version) and in E 97. In fact, E 97 is the only table giving both total- and protein-N values for all foods without calculating protein values; however protein is accounted for in the energy value by multiplying protein-N values by 6.25 for all foods. Calculation: Total-N values are multiplied by fooddependent conversion factors to obtain the protein content. The reference protein conversion factors given in Table 3 are those published by the FAO=WHO in 1973. They are mostly used as such (GB 91, GK 92, IT 98) or slightly modi®ed for some foods that are rarely consumed (F 95, S 94, S 96, IT 94, DK 96, D 93, D BLS and D SFK). For example, in some tables, the N value of cereal products is multiplied by 5.80, instead of 5.83, 5.70, 5.95 and 6.31 for speci®c cereal items. For foods that contain a signi®cant amount of non-proteinN (5 ± 10%), such as mushrooms, cartilaginous ®sh and molluscs, the British tables have applied protein conversion factors to protein-N values, rather than to total-N values. The German tables adjust for non-protein-N by multiplying the total-N of mushrooms by 4.17. Compatibility: The use of the FAO=WHO conversion factors, or slight modi®cations of them, has only a minor impact on the ®nal values for protein and on the global protein intake. This means that protein contents are comparable between the European tables that have stated the protein conversion factors, except in the Dutch table where a bigger impact on protein content is expected, as simpli®ed factors are used (6.38 for milk products and 6.25 for all other foods). The Dutch protein values should therefore be higher for cereals, cereal products and nuts. Table 3 Protein conversion factors FAO=WHO (1973): Milk and milk products 6.38 Nuts Almonds 5.18 Peanuts, brazil nuts 5.41 Other nuts, Seeds 5.30 Cereals Wholemeal wheat 5.83 Wheat bran 6.31 Other wheat ¯ours, pasta 5.70 Barley, oat, rye 5.83 Rice 5.95 Soya 5.70 Gelatin 5.55 Other foods 6.25 65 Comparison of nutrients in nine European countries G Deharveng et al 66 On the other hand, some German and Spanish tables do not mention the protein conversion factors used. Thus, for the latter tables, it is not possible to judge whether the protein values are exactly comparable with those of the other countries and they cannot be made comparable either. 2.2 Fat All tables mention the total fat content of foods in g except D BLS which expresses it in mg. De®nition: Although not mentioned in European tables, total fat is usually considered to be a mixture of triglycerides, phospholipids, sterols and related compounds. Methods: The result of the lipid analysis depends largely on the extraction medium and the method used for hydrolysis. Speci®c analytical methods have been developed for some food groups. The standard methods used in Europe are those of WernerSchmidt, Weibuhl-Stoldt described in Egan et al, 1981 and the AOAC (1980), which are broadly similar; they are utilised by GB, Italy and E 97. They use acid hydrolysis and an extraction with mixed solvent (ether and petroleum spirit) and do not allow further FA analysis. These methods can be used for most food groups but dairy products require alkaline hydrolysis instead of acid hydrolysis. The Folch-type methods (Folch et al, 1957) mentioned by NL 96 and Italy use extraction with chloroform and methanol. Compared to the standard methods they usually lead to higher but comparable values, and they directly provide the fat fraction for analysis of FAs. Soxhlet-type methods which use extraction with ether give lower values for cereals, because of incomplete extraction, but give comparable values for meat (excluding liver) and many meat products (Kinsella et al, 1975). Compatibility: Almost half of the tables do not mention the method used for fat and=or FA analysis. However, total fat values provided by recent analyses should be comparable, as agreements on extraction and hydrolysis procedures have been reached in Europe in recent years. 2.3 Carbohydrates (CHO) (Table 4) The carbohydrate values in the tables show possibly the most heterogeneity in nomenclature, de®nition, mode of expression and methods used, in addition to the level of detail with regard to the coverage of the different CHO fractions (Table 4). In all tables, total CHO and ®bre are listed as separate nutrients and expressed in g, except in D BLS. De®nitions: (a) Total CHO: Many tables use the term `available carbohydrates' which is usually de®ned as the sum of the free sugars and complex carbohydrates (starch, dextrins and glycogen)(Cummings et al, 1997). D BLS also includes sugar alcohols in the total CHO values. This leaves the data compatible with those of the other tables because sugar alcohols are very minor components of foods. The British tables use the term `carbohydrate' although the introductory text de®nes this as `available carbohydrate' which includes oligosaccharides. The German tables use `carbohydrate' in a similar way. Several countries use the term `carbohydrates' in the classical proximate analysis sense, that is, determined `by difference', to different extents. (b) Starch: Some tables explicitly include dextrins and glycogen within the starch values (DK 96, GB, GK 92, IT 94) whereas others (D 93 and D BLS) disregard or exclude these minor components although technically they are measured together with starch. In D 79, D 93, D BLS, NL 96, the starch fraction is called `polysaccharides'. No de®nition of `starch' is given in F 95. Starch values are not included in the Swedish and the Spanish tables. (c) Sugars: This term refers either to the sum of monosaccharides (called `soluble sugar' in IT 98), or to the sum of mono- and disaccharides (GB, GK 92, IT 94), or to the sum of mono-, di- and trisaccharides (DK 96). No de®nition can be found for F 95, NL 96. Some tables state monoand disaccharides separately (D 79, D BLS and S 94). The Spanish tables do not mention this CHO fraction. (d) Lactose: This is a disaccharide present in milk and milk products (Passmore et al, 1986). Values for lactose are available in g or mg in most tables, if relevant, except in the Spanish, Swedish, and Italian tables, and in D 79, D 93, GK 92 and F 95. Modes of expression: Two modes of expression are in use: as monosaccharides and in the anhydrous form. Values expressed as monosaccharides come straight from the analysis of the CHO. Values expressed in the anhydrous form are usually determined `by difference', which is the mode of expression chosen for labeling purposes in the European Community. Carbohydrate weights Ð in the anhydrous form Ð may be converted to monosaccharide equivalents using conversion factors depending on the length of the carbohydrate chain. Carbohydrate values expressed as monosaccharides are 1.05 times higher for disaccharides and 1.10 times for starch (GB 91). (a) Total CHO: Total CHO are expressed as monosaccharides in the British, French, Italian, Greek and all Spanish tables except E 88 and E 90. F 95 has chosen to express total CHO in both forms. The other tables express CHO in the anhydrous form. (b) CHO fractions: The British, Greek and IT 98 tables express `sugars' and `starch' as monosaccharides. The other tables express them in the anhydrous form. The sum of `sugars' and `starch' values is therefore lower than the value for `total CHO, in monosaccharides' in F 95 and in IT 94 because of the difference in modes of expression. Analytical methods: (a) Total CHO: There are two methods of obtaining values for total CHO. Either they are obtained `by difference' [100 7 (fat protein ash water)] which accumulates errors of the analysis of the other food components; dietary ®bre is further deducted in Sweden, but never in Denmark and not systematically in the Netherlands. Or the preferred approach is to add up the sum of the analysed CHO fractions. This is the reference method in England, France, Italy, the Netherlands and Germany (Table 4). Most of the tables do not describe the methods used for carbohydrates. The method `by difference' is used systematically in the Scandinavian countries and, to a lesser extent, in the French, German and Dutch tables. Available carbohydrates Available carbohydrates Available carbohydrates Available carbohydrates Carbohydrates ± Carbohydrates Available carbohydrates Available carbohydrates Idem GB Available carbohydrates Available carbohydrates F 95 F fat F milk F fruit D 79 D milk D 93 D SFK D BLS GK 92 IT 94 IT 98 DK 96 Carbohydrate Carbohydrates, total Country GB Terms used for total carbohydrates `Soluble sugars'#: sum of monosaccharides `Starch' #: starch and dextrins Idem GB ± ± ± ± ± (g=kg) (g=kg) and 7 ± Expressed as monosaccharides Yes, 7 No Other fractions available Sum of main fractions By difference No method mentioned No method mentioned No method mentioned Mostly sum of analysed fractions, sometimes by difference Mostly sum of analysed fractions, sometimes by difference Sum of main fractions No method mentioned. No method mentioned. ± ± No method mentioned Starch: Dean, 1978: enzymatic hydrolysis and measurement of glucose; Egan et al., 1981: polarimetry Sugars: Egan et al., 1981: Boehringer enzyme kit; Southgate et al, 1978a; Dean, 1978: HPLC; Southgate, 1976: colorimetry No method mentioned. No method mentioned. Methods used for carbohydrate fractions ± Idem GB Table 4 continued. Refer to 4th ed of GB Starch: Dean, 1978: enzymatic or acid hydrolysis and measurement of glucose; Sugars: Dean, 1978: HPLC Sugars: Southgate, 1976: colorimetry Starch: enzymatic hydrolysis Some individual carbohydratesy Sum of main fractions No method mentioned. By difference Many individual carbohydrates Sum of all carbohydrates and sugar No method mentioned. `Starch' (alone) alcohol fractions By difference ± Glucosey Fructosey Saccharosey Starchy ± Lactosey ± Sum of analysed fractions By difference (®bre included) Methods used for total carbohydrate `Soluble sugars': sum of mono- and ± Sum of analysed fractions di-saccharides `Starch': starch, dextrins and glycogen ± Usually, sum of main fractions Idem IT 94 `Monosaccharides' `Disaccharides' `Polysaccharides Lactose Other carbohydrates `Mono-=disaccharides': sum of glucose, fructose, saccharose and maltose `Polysaccharides': starch, but not glycogen, nor maltodextrins `Sugars'y `Starch'y `Monosaccharides' , `Disaccharides' `Oligosaccharides' (3 ± 10) `Polysaccharides' ( > 10) `Sugar alcohols' Idem GB ± (too much variability) `Starch' `Sugars' ± ± `Starch': starch, dextrins and glycogen dSugar added (label info) `Sugars': sum of mono-, di- and trisaccharides `Starch' #: starch, dextrins (if in very The list depends on the small quantity) and glycogen supplement `Total sugars' #: sum of monoand disaccharides Main fractions ± De®nitions # if expressed as monosaccharides Table 4 Total carbohydrates: Terms, de®nitions and methods used Comparison of nutrients in nine European countries G Deharveng et al 67 Comparison of nutrients in nine European countries G Deharveng et al By difference ± GLC No method mentioned. No method mentioned. No method mentioned. No method mentioned. Padmore, AOAC, 1990 By difference ± ± ± ± ± Sucrose ± ± ± ± ± `Monosaccharides': glucose, fructose and galactose `Disaccharides': sucrose, lactose, maltose `Mono-, disaccharides'z Glucose, fructose, maltose, lactose, saccharose. Given in appendix (that is, for some foods only). y for some foods only. z for the PC-version, see S 94. x S 96 d Carbohydrates, total E 88 E 90 E 92 E 93 E 97 S 94 Compatibility: CHO fractions are stated in various degrees Ð from none in the Spanish tables to many fractions in certain tables (Table 4). They are not always grouped in the same way Ð for example, simple sugars Ð and under the same term Ð `starch' vs `polysaccharides'. The fact that different groupings, de®nitions and modes of expression are in use makes it dif®cult to compare values for total CHO and CHO fractions precisely between countries (sometimes even within the same table). In a pragmatic sense the methods in use are probably compatible. The major incompatibilities arise from the modes of expression, the methods used to get the values (analysis vs `by difference') and whether or not `dietary ®bre' values are included in `total CHO' values. 2.4 Alcohol Alcohol is included in all tables except in some Spanish tables (see Table 2). It is most frequently expressed in g=100g except in GB 91, D 93, GK 92, IT 94 and IT 98, where it is expressed in g=100 ml, and in S 96 where two modes of expression are used: % vol and g=100g. As the density of spirits and liqueurs is 0.95 and 1.10, respectively, the values need to be converted to a common mode of expression before they can be compared. Distillation is the standard method, although most tables do not mention the method used. ± Glucids Carbohydrates Available carbohydrates Available carbohydrates Carbohydrates Carbohydrates, total NL 96 ? ? ± GLC Sugars: Enzymatic, HPLC and Luff-Schoorl =Kruisheer method Polysaccharides: Acid hydrolysis and optical rotation. Mostly by difference, sometimes sum of analysed fractions In the databasey: d and galactose `Mono- and disaccharides, total' `Polysaccharides, total' Total carbohydrates Country ± Methods used for carbohydrate fractions Methods used for total carbohydrate (b) CHO fractions: The carbohydrate fractions are analysed in three main steps (except for mono- and some disaccharides which do not require hydrolysis): extraction into aqueous alcohol, then hydrolysis (acid and=or enzymatic) followed by the measurement of glucose (or saccharose) by colorimetry, reductiometry, polarimetry, or High-Pressure Liquid Chromatography (HPLC), depending on the food. If the hydrolysis is done solely with acid, some of the nonstarch polysaccharides (NSP) are included in the starch value (NL 96). In the British tables a range of methods has been used for both sugars and starches, representing the development in carbohydrate analyses since the 1970s. Most tables giving starch values have used enzymatic hydrolysis followed either by a glucose-speci®c method (GB) or polarimetry (NL 96). Most recent values for sugars are obtained using HPLC (GB, NL 96). Terms used for total carbohydrates Table 4 continued Expressed as monosaccharides Yes, 7 No Main fractions ± De®nitions # if expressed as monosaccharides Other fractions available 68 2.5 Energy Energy may be expressed in kcal or in kJ (1 kcal 4.18 kJ). Most tables state both units, except the Danish and S 94 tables which express energy only in kJ, and some Spanish tables only in kcal. The energy content of foods in the tables has not been measured, but calculated. Calculation: Metabolisable energy may be obtained easily by adding up the relative energy contributions of CHO, fat, protein and alcohol, which are obtained by multiplying the respective weights of these nutrients by conversion factors. The traditionally applied conversion factors are the so-called `Atwater' factors, that is, 4, 9, 4, 7 kcal=g, or 17, 37, 17, 29 kJ=g (Atwater, 1910) applied to protein, fat, total CHO and energy contents, respectively. Whenever carbohydrates are expressed in monosaccharides, the so-called `Southgate and Durnin' factor of 3.75 kcal=g or 16 kJ=g is used, instead of 4 kcal=g or 17 kJ=g (Southgate & Durnin, 1970). F 95, D 93, D SFK, D BLS and NL 96 additionally include the Comparison of nutrients in nine European countries G Deharveng et al energy contribution of polyols and=or organic acids by applying a conversion factor of 13 kJ=g for organic acids (in general) and of 10 kJ=g for polyols. D 79, D milk and E 88 do not state how energy values are obtained. Compatibility: DK 96 has three particularities: the energy conversion factors applied are 38 kJ=g for fat and 30 kJ=g for alcohol instead of 37 and 29 kJ=g used in all other tables; in addition, ®bre is included in the calculation of energy whereas the other countries exclude it. Danish energy values should therefore be slightly higher than those in other tables. E 90 indicates that energy values have been copied as such from different sources (mainly USDA, GB 78, and French tables) which use different energy conversion factors. Energy values could therefore not be comparable within the table itself. Energy values are broadly similar whether obtained using the `Southgate and Durnin' or the `Atwater' coef®cients (F 95). Additionally, as most foods contain very small amounts of polyols and organic acids, their inclusion or exclusion does not have a major impact on the energy value. followed by some British tables, D SFK, and D BLS. The other tables give relatively low coverage, and it should be noted that in IT 94 more fractions are stated than in IT 98. Although analytical methods for FAs are not always stated, they are comparable. 2.7 Cholesterol For foods containing cholesterol, this nutrient is reported in all tables except E 88. The mode of expression is mg or g. Analytical methods: Cholesterol is analysed using GC (E 97) and GLC (GB 91, Italy, and NL 96). The British tables also state that some values have been calculated but do not provide the algorithm nor indicate which values were calculated and which analysed. In DK 96, cholesterol values are calculated as follows: (protein a)+fat where a 2.65 for pork and beef; a 3.25 lamb None of the other tables state the analytical method or whether the values were calculated. Compatibility: Cholesterol values, either obtained through analysis, or by calculation, may be used with reasonable con®dence. 2.6 Fatty acids (FA) Terminology: Fatty acids are named after their Latin names, their chemical names or chemical formulas for speci®c isomers of fatty acids (for example, C18:1, n-9). Modes of expression The usual mode of expression of FAs is in g=100 g food. In D BLS it is in mg=100 g; in F milk and F fat in g=kg food. In addition, the French supplements, DK 96 and the PC-version of NL 96 state them as a percentage of total FAs. The latter mode of expression is needed to correlate the percentage of FAs seen in blood with the pro®le of consumed foods. Analytical methods: After extraction by Folch-type methods or the Bligh and Dyer (1959) method (using chloroform and methanol), FAs are converted to their methyl esters, separated by GLC and their amounts calculated from GLC peak areas. They are expressed as quantitative percentages of the total FA area. This percentage can then be converted into g=100 g food by applying FA conversion factors to the total amount of fat in the food. These conversion factors are sometimes stated in the tables (for example, GB 91, S 96) and are comparable between each other. Compatibility: Almost all tables indicate the sums of saturated, monounsaturated and polyunsaturated FAs, whenever applicable. E 88 does not mention any of these, and D 79, D milk and D SFK do not cover all fractions. The individual FAs are found for some foods only, often in appendices. The most frequently reported are the saturated FAs C12 : 0 to C18 : 0, the monounsaturated FAs C14 : 1 to C22 : 1, and the polyunsaturated FAs C18 : 2 to C22 : 6. The oleic acid C18 : 1 and docosenoic acid C22 : 1 are sometimes stated as such, or by their isomers (for example, C18:1, n-9 alone) with or without the indication of the location of the double bond. The most complete coverage seems to be present in the French fat and milk supplements, 2.8 Dietary ®bre The study of the European tables illustrates the problems that have arisen in agreeing on a de®nition of `dietary ®bre' and translating this de®nition into an analytical method. As a consequence there are profound incompatibilities between the values in the different tables, which are virtually impossible to solve. Therefore `®bre' shows varying relations to colorectal cancer depending on the de®nition=method used (Kaaks & Riboli, 1995). Fibre is expressed in g, except in D BLS (mg), and reported, whenever relevant, in all tables except in E 88. De®nitions: `Crude ®bre' is the sum of plant substances resistant to hydrolysis by acids and subsequently by alkali. It is not relevant to human nutrition and values are substantially lower than for dietary ®bre but it is still used in D 79 and partly in F fruit. It is measured by the Weende method (AOAC, 1984). The de®nitions adopted for `dietary ®bre' are determined by the methods used. Most tables de®ne it as `the indigestible components of the plant cell wall' (Trowell, 1972). It can be divided into water-insoluble (cellulose, hemicellulose and lignins) and water-soluble (pectins, gums and mucilages). It is often described as unavailable CHOs plus lignin (which is not a CHO), whereas Trowell et al (1976) advocated de®ning it as the polysaccharides that are not digested by the endogenous secretions of the gastrointestinal tract. Analytical methods: There are four types of methods for obtaining values for total ®bre and=or ®bre fractions: (a) Total dietary ®bre (TDF) measured by the AOAC methods (based on the indigestibility of the components) is widely found. The AOAC reference method is Prosky et al, 1985, other related methods are also in use (Candlish et 69 Comparison of nutrients in nine European countries G Deharveng et al 70 al, 1987; Anderson et al, 1988, Mongeau & Brassard, 1989, Schweizer & Wuersch, 1979, Meuser et al, 1985) which give similar but not strictly identical results. They measure non-starch polysaccharides, lignin and resistant starch (Southgate, 1995). AOAC-type methods are used in DK 96, NL 96, E 97, F 95 and in the Italian, Swedish, German tables. (b) Dietary ®bre, de®ned as non-starch polysaccharides (NSP), is measured by the Englyst-type methods (Englyst et al., 1982; Englyst and Cummings, 1988). Lignin, waxes, cutins and resistant starch are excluded from this food fraction. The NSP de®nition is used in the British tables for all foods, and partially in F fruit, D SFK, and IT 98. (c) `By difference': total dietary ®bre 100 7 (water protein fat ash available CHO). It includes resistant starch. This method is used only in D SFK for some foods. (d) Dietary ®bre measured by Southgate-type methods is used in GK 92 (where values are expressed as monosaccharides); it is still mentioned in the recent UK tables and in IT 98 to some extent even though the method is no longer used nowadays. It is a colorimetric method measuring NSP, lignins and some starch as described by Southgate (1969), Southgate et al (1978b) and Wenlock et al (1985). (e) Dietary ®bre fractions. The dietary ®bre fractions depend on the chemical=analytical method used. The AOAC method provides values for the soluble and insoluble fractions, and the NSP method for more detailed fractions (for example, cellulose). Other methods not mentioned above also give dietary ®bre fractions, which are used in D BLS (Green®eld & Southgate, 1992). The British tables, D BLS and D SFK state `dietary ®bre' and its fractions. In the English tables the sum of the fractions is not equal to the value indicated for `dietary ®bre' for the same food, as these fractions have been assayed in different samples. This is not the case in D BLS where `dietary ®bre' is calculated as the sum of the fractions. Compatibility: E 88 does not give ®bre values and most Spanish tables do not state the method or de®nition used. In compiled tables it may not always be possible to determine which method was used to measure the values unless a source code is given at the nutrient level. The use of several incompatible methods makes it extremely dif®cult to compare ®bre intake between countries and even within a given country using the same table (for example, F 95, IT 98). The values obtained by Southgate-type methods should in principle be similar to those measured by AOAC-type methods. The AOAC and related methods give higher values than the Englyst method because ®bre values obtained by AOAC-type methods include lignin and resistant starch. Values of TDF depend on food processing (Englyst et al, 1988). In foods where lignin levels are low and resistant starch has not been formed (fruits, vegetables, and low extraction cereal ¯ours) the NSP and AOAC ®bre values should be similar. In high extraction cereal ¯ours, most processed cereal products and potatoes, the AOAC values may be several g=100 g higher than the NSP values (Mongeau et al, 1989). 3. Vitamins 3.1 Vitamin A=Retinol=Carotene (Table 5) Vitamin A activity derives from retinol and some carotenoids, of which b-carotene is the most active. In epidemiological studies, there is an increasing interest in the 600 or so identi®ed carotenoids because of their suspected cancer-preventing effects, which are independent of their vitamin A activity. The cancer protection is related to their capacity to trap singlet oxygen and free radicals. Carotenoids are also used increasingly as markers of dietary intake, especially of vitamins. It should be remembered, however, that carotenoid serum levels depend strongly on individual differences in terms of intestinal absorption and enzymatic transformation within the body (Riboli et al, 1987). De®nitions=Modes of expression: Vitamin A, retinol and carotenoids are usually expressed in mg=100 g. In F milk and F fat the unit is mg=kg and in D 79, D milk, E 88, E 90 and the Swedish tables it is mg or UI=100 g. The following are the de®nitions given of total vitamin A, retinol and carotenes. Indices of the de®nitions are used in Table 5. (a) Total vitamin A=retinol equivalents (RE): Total vitamin A activity is stated in retinol equivalents accounting for the different vitamin activities of retinol and carotenes (Table 5). a : 1 retinol equivalent 1 mg retinol (DK 96, D BLS, D SFK, 6 mg b-carotene IT 98, E 97, S 94, S 96) 12 mg other pro-vitamin A carotenoids 6 mg b-carotene equivalents b : 1 retinol equivalent 1 mg retinol (GB cereal, NL 96) 6 mg b-carotene (i.e., do not include other carotenoids) c : 1 retinol equivalent 1 mg retinol (F fat, F milk, E 92, E 93) 6 mg non-dairy b-carotene (exclude other carotenoids) 2 mg dairy b-carotene d : 1 retinol equivalent 1 mg retinol (IT 94) 6 mg non-dairy b-carotene 2 mg dairy carotenoids (b) Retinol: k : all-trans retinol equivalents all-trans retinol 0.75 13-cis retinol 0.90 retinaldehyde l : all-trans retinol equivalents all-trans retinol 0.75 13-cis retinol 0.90 retinaldehyde 0.40 dehydroretinol m : retinol trans-retinol cis-retinol (c) Carotenes: r : b-carotene alone, in principle (depends on the source and the date of analysis) s : 1 b-carotene equivalent b-carotene 2 b-carotene 2 -cryptoxanthin 2 b-cryptoxanthin s1 : 1 b-carotene equivalent 1 b-carotene 2 -carotene 2 b-cryptoxanthin t : total carotene: no precise de®nition (presumably sum of different carotenoids as analysed by HPLC or result of ± Retinol (vitamin A) (mg) ± Retinol (mg) Vitamin A (retinol)x (mg) Retinol (mg) Retinol (mg) ± Retinolk (mg) Retinolx, z (mg) ± ± Retinol (mg) ± ± ± ± ± ± ± Vitamin A activityc, x (mg=kg) Vitamin A activityc, x (mg=kg) ± ± Vit A (mg) ± Retinol eq.a, x (mg) Retinol eq.a (mg) ± Vitamin Ad (mg) Retinol eq.a (mg) Retinol eq. b (mg) Vitamin A (UI) Vitamin A (mg) Vitamin A: retinol eq.c (mg) GB fruit GB vdish GB ®sh GB foods GB meat GB mdish F 95 F fat D BLS GK 92 IT 94 IT 98 NL 96 E 88 E 90 E 92 F fruit D 79 D milk D 93 D SFK F milk x (mg=kg) ± ± ± ± b-carotenes (mg) ± ± carotener (mg) b-carotene eq.s, x (mg) Retinol (b-carotene) (mg) carotene (mg) Pro-vitamin A carotenoidst (mg) Total carotenoidst, x (mg) Total carotenet, carotenes (mg) b-carotene eq.s (mg) Total carotenet, x (m=mg=kg) Retinoll (mg) Retinol (mg) Retinolx (mg=kg) Retinolx (mg=kg) carotenes (mg) carotenes (mg) carotenes (mg) carotenet (mg) Retinol (mg) Retinol (mg) Retinol (mg) Retinol (mg) carotenes (mg) carotenes (mg) Retinolk (mg) ± GB veg Retinol (mg) ± carotener (mg) ± Retinolk (mg) DK 96 GB 91 Retinol equivb (mg) ± carotenes (mg) Carotene fraction GB cereal GB milk m Retinol fraction Retinol (mg) Retinolk (mg) a Term used for total vitamin A (unit) Vitamin A: Total activity and fractions ± availability, terms, de®nitions and methods used vitamin A (RE) ± Country Table 5 x, z b-carotene b-carotene Other carotenoids x -carotene b-carotene cryptoxanthin g-carotene mutatochrome b-carotene b-carotenex b-carotenex `retinol fractions': all-trans retinol retinaldehyde 13-cis retinol dehydroretinol y `retinol fractions': all-trans retinol dehydroretinol 13-cis retinol total retinol retinaldehyde y y `retinol fractions': all-trans retinol 13-cis retinol y : `carotenoid fractions': -carotene carotene equivalent b-carotene retinol equivalent b-cryptoxanthin y `carotenoid fractions': -carotene carotene equivalent b-carotene retinol equivalent cryptoxanthins b-carotene Fractions available mentioned mentioned mentioned mentioned mentioned Not mentioned Extraction, chromatographic separation and spectrophotometry Chromatography: Carr & Price, 1926, HPLC Chromatography: Carr & Price, 1926, HPLC Reverse phase HPLC Not mentioned Not mentioned Not mentioned Not Not Not Not Not Not mentioned Idem GB 91 Not mentioned Not mentioned Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 Not mentioned Reverse phase HPLC Chromatographic separation and absorption spectrophotometry Method used Comparison of nutrients in nine European countries G Deharveng et al 71 Analytical methods: Nowadays, two methods are used to analyse retinol and carotenoids: colorimetry and HPLC (Green®eld & Southgate, 1992). Hart & Scott (1995) have found that recent improvements in the HPLC technique lead to an increase in carotenoid recovery ranging from 60% to 90% in foods. Colorimetry can measure retinol and only total carotenoids, the detected fraction is then erroneously called `b-carotene'. In particular, values derived from the old method of `Carr & Price, 1926' give incompatible results compared to the newer method. Only the Italian tables state this old method. for some foods only y given in appendix (that is, for some foods only) z in the PC-version only a tot refer to main text. x ± Retinol (mg) (all-trans) Retinol eq.a (mg) S 96 ± Retinol (mg) (all-trans) S 94 ± ± Vitamin A: retinol eq.a (g) Retinol eq.a (mg) E 97 Retinol eq. E carot. Vitamin A: retinol eq.c (g) E 93 the analysis of `total carotene'). This expression gives the weight of carotenes, but does not adjust for relative vitamin A activities. b-carotene b-carotene (g) ± ± s1 Carotene fraction Retinol fraction Term used for total vitamin A (unit) Country Table 5 continued HPLC: Granado et al, 1992, Olmedilla et al., 1993 lutein -carotene zeaxanthin b-carotene licopene b-cryptoxanthin Not mentioned Fractions available Method used 72 retinol: HPLC: Brubacher, 1986 carotenoids: HPLC: Shearer, 1986 retinol: HPLC, normal phase carotene: HPLC, reverse phase retinol: HPLC, normal phase carotene: HPLC, reverse phase Comparison of nutrients in nine European countries G Deharveng et al Compatibility: The tables generally report total vitamin A, except most French and British tables, D 79, D 93 and GK 92 which distinguish the retinol and carotene fractions. All four de®nitions of RE are similar and should give comparable values. The two major differences are: ®rst, whether they include the vitamin activity of carotenes other than b-carotene; second, whether the better availability of dairy carotenoids is taken into account. D milk, E 90 and E 88 do not state whether it is expressed in RE, so values may not be comparable. All tables except IT 94 and most Spanish tables state retinol. Some British supplements (GB milk, GB ®sh and GB meat) provide values for cis- and trans-retinol for certain foods. Most tables do not state the de®nition of retinol. The values of the ®rst and second de®nitions of retinol cited above should be similar, as the amount of dehydroretinol is minor. The third de®nition should give slightly higher, but still comparable values. Values for carotene are available as carotene equivalents in the British tables, F 95, F fruit and IT 98. In F fat, F milk, D 93 and D SFK, the carotene fraction refers to total carotenoids, that is, not adjusted for the relative activities of the different carotenoids. E carot expresses the carotenes in RE even though fruits and vegetables contain no retinol. So the different means of calculations for the carotene fraction lead to very different values. Values for single carotenoids are available either for bcarotene alone (F fat, F milk, D BLS, E 92, S 94 and S 96), or also for other carotenoids (E carot, D SFK, and as appendixes in GB veg and GB fruit). Care should be taken when comparing vitamin A values between tables. For dairy products, total vitamin A activity differs depending on the factors used to express it in retinol equivalents and, for foods rich in carotenes, total carotene values depend on the analytical method and=or the mode of expression. Moreover the handling of food samples before analysis is extremely important (Green®eld & Southgate, 1992) and the natural variation in b-carotene is very high (ENG 91). 3.2 Vitamin D (Calciferol) In the tables, it is expressed in mg except in D 79 and E 90 (in mg). It is synthesised from cholesterol when UV rays from the sun are absorbed through the skin. De®nition: Vitamin D occurs naturally in animal foods as cholecalciferol (vitamin D3), while ergocalciferol Comparison of nutrients in nine European countries G Deharveng et al (vitamin D2) is manufactured. Both are added as forti®cation and have the same vitamin activity. Small amounts of 25-hydroxy-vitamin D, which has ®ve times more vitamin activity, are present in some animal foods (ENG 91). In the tables, the de®nition of vitamin D is generally not given. GB 91 and IT 98 de®ne vitamin D as the sum of vitamins D2 and D3; DK 96 and E 97 as vitamin D3. GB meat and GB ®sh are the only tables to state that vitamin D activity is the sum of the weight of cholecalciferol and ®ve times the weight of 25-hydroxy-vitamin D. Analytical methods: Most tables do not state the method used. In recent years, HPLC has mainly been used to measure vitamin D as it allows the three fractions to be separated (in GB 91, F 95, NL 96 and E 97). Older methods such as biological assay (GB 91) or colorimetry give unreliable data. Table 6 Country DK 96 GB 91 Total vit E (unit) 2 -TE (in mg) mg1 mg1 mg1 GB veg mg1 1 GB fruit mg GB vdish GB ®sh mg1 mg1 GB foods mg1 GB meat GB mdish F 95 F fat mg1 mg1 mg1 mg=kg4 F milk F fruit D 79 D milk D 93 D SFK mg=kg4 mg mg mg mg1 mg=mg1,4 D BLS GK 92 IT 94 IT 98 NL 96 mg TE1 ± ± mg1 (as -TE) mg3 1 De®nitions: Alpha-tocopherol equivalents (-TE) are de®ned in slightly different terms among tables giving comparable values. All recent tables except DK 96 and NL 96 apply the ®rst de®nition. The de®nition is however not stated in all tables. Indices of the de®nitions are used in Table 6. ± ± ± mg mg mg1 mg1 Fractions available Methods used -tocopherol ± No method mentioned Normal phase HPLC Reverse phase HPLC Colorimetry combined with GLC, Christie et al, 1973 Colorimetry combined with GLC, Christie et al, 1973 Colorimetry combined with GLC, Christie et al, 1973 HPLC HPLC ± For some foods: amounts of major tocopherols and tocotrienols on request -tocopherol d-tocopherol b-tocopherol g-tocopherol ± 5 -tocopherol d-tocopherol b-tocopherol g-tocopherol 5 -tocopherol d-tocopherol b-tocopherol g-tocopherol ± ± ± 4 Total tocopherols -tocopherol7 -tocotrienol7 b-tocopherol7 b-tocotrienol7 g-tocopherol7 g-tocotrienol7 d-tocopherol7 d-tocotrienol7 ± ± ± ± ± Total tocopherols4 Total tocotrienols4 -tocopherol4 -tocotrienol4 b-tocopherol4 b-tocotrienol4 g-tocopherol4 g-tocotrienol4 d-tocopherol4 -tocopherol 5 ± : -tocopherol g-tocopherol b-tocopherol d-tocopherol 6 D--tocopherol (mg) -tocopherol (mg) ± ± D--tocopherol (mg) D--tocopherol (mg)6 to3 refer to the main text. for some foods only. 5 : in appendix, that is, for some foods only. 6 in the PC-version. 7 expressed in % total tocopherols and in mg=kg. 4 3.3 Vitamin E (Table 6) Vitamin E activity is the sum of the activities of tocopherols and tocotrienols. Each fraction has a different vitamin E activity and -tocopherol is the component with the highest vitamin E activity (Passmore & Eastwood, 1986). Vitamin E: Total activity and fractions: Availability, de®nitions and methods used GB cereal GB milk E 88 E 90 E 92 E 93 E 97 S 94 S 96 Compatibility: Vitamin D activity is given in all tables except GK 92, IT 94 and E 88. It would be helpful to assert the link between vitamin D values and the method used in order to determine the relative quality of these values and to use only recently analysed data for dietary analysis. HPLC Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 Idem GB 91 HPLC HPLC No No No No No No method method method method method method mentioned mentioned mentioned mentioned mentioned mentioned No method mentioned Not applicable Not applicable No method mentioned HPLC Not applicable No method mentioned No method mentioned No method mentioned HPLC, Shearer, 1986 HPLC HPLC 73 Comparison of nutrients in nine European countries G Deharveng et al 74 1 Total activity in -TE -tocopherol 0.4 btocopherol 0.1 g-tocopherol 0.01 d-tocopherol 0.3 -tocotrienol 0.05 b-tocotrienol 0.01 g-tocotrienol 2 -TE as used in DK 96: d--tocopherol 0.5 btocopherol 0.1 g-tocopherol 0.3 mg -tocotrienol 3 -TE as used in NL 96: -tocopherol 0.4 btocopherol 0.1 g-tocopherol 0.01 d-tocopherol Analytical methods: The usual analytical method for vitamin E is HPLC. Only GB 91, GB cereal and GB milk state that colorimetry combined with gas chromatography (Christie et al, 1973) was used in addition to HPLC. Most tables do not state the method used. Compatibility: The Spanish and some German tables express vitamin E in mg=100 g without de®nition, and some French supplements in mg=kg, whereas the usual mode of expression is in -TE mg. All tables include vitamin E, except GK 92, IT 94 and most Spanish tables. E 90 and E 92 indicate values only for tocopherol. Single tocopherols are listed by about half of the tables, single tocotrienols only by F fat, D SFK, and in GB veg (on request). Most vitamin E values present in recent European tables were analysed using HPLC during the 1970's. However only vitamin E values obtained recently by HPLC are valuable. 3.4 Thiamin (B1), Ribo¯avin (B2), Vitamin B6 Thiamin is reported in all tables and is expressed in mg except in D BLS and E 88 (mg). Thiamin is heat unstable and damaged in alkaline conditions (Green®eld & Southgate, 1992). Ribo¯avin is reported in all tables and is expressed in mg except in D BLS and E 88 (mg). It is also destroyed by heat, and by light (Green®eld & Southgate, 1992). Vitamin B6 is the sum of pyridoxal, pyridoxamine, their phosphates and pyridoxine, which all have the same activity (Green®eld & Southgate, 1992). It is included in all tables except in IT 94 and E 88; it is expressed in mg except in D BLS (mg) and partially in D SFK (mg=mg). Analytical methods: Three methods are in use for vitamins B1 and B2 (microbiological, ¯uorimetry and HPLC) which give similar values. For vitamin B6, the major problem is not the method (HPLC and microbiological assay) but the extraction. However, few tables state the analytical methods used for these three vitamins. Compatibility: Vitamins B1, B2 and B6 values are globally comparable between countries. 3.5 Vitamin B12 (Cyanocobalamin) Vitamin B12 is present in small amounts in animal foods, especially in liver and egg yolk (Passmore & Eastwood, 1986). It is expressed in mg and reported in all tables except the Greek and Italian tables and E 88. Microbiological assay is the only method in use for measuring vitamin B12, and the same bacterium is used in all assays in Europe, resulting in a very standardised method. The problem with vitamin B12 values arises from the potentially high measurement error, as the concentrations in foods are so low. 3.6 Folates = Folic acid (B9) Folate activity in foods derives from a number of compounds, which are reduced forms of pteroyl glutamic acid (folic acid). Most naturally occurring folates are conjugates with 2 or more g-glutamyl residues (often called bound forms or polyglutamates). Folates are water-soluble and readily oxidised, and are therefore labile (Green®eld & Southgate, 1992). In order to be absorbed and metabolically active, the glutamyl residues must be deconjugated; the availability of folate varies between 25 ± 96% depending on the food (Passmore & Eastwood, 1986). De®nitions: There are considerable differences and lack of clarity in the terminology and the de®nitions used in the tables. Many tables use the term `folic acid' as a synonym for folates and de®ne it as the sum of the `free' and `bound' forms. D BLS applies the factor 0.20 to the `bound' forms to adjust for its lower bio-availability and gives total folate values in `free folic acid equivalent'. The amount of `bound' folate is calculated by difference between the amounts of `total' folic acid (measured after deconjugation) and of `free' folic acid (measured before deconjugation). Some tables (F fat, D BLS, E 90) give values for `free' folic acid. Analytical methods: Microbiological assay is currently the standard method for measuring folate activity in foods (Phillips & Wright, 1983). Several bacteria are in use, of which Lactobacillus caseii (rhamnosus) provides the most satisfactory values for total folate after deconjugation (GB, F 95, NL 96). The bacterium Streptococcus faecalis, used in the AOAC method, is only valid for measuring folic acid which is used for food forti®cation but which does not naturally occur in foods (Green®eld & Southgate, 1992). Unfortunately, in most tables the bacteria used are not stated. Additionally, a problem of quality control seems to occur between laboratories, especially mentioned by Denmark, where two laboratories use the same microbiological assay but obtain dramatically different values. HPLC methods (for example, Shearer, 1986) provide the possibility to measure the different fractions but their performance in collaborative trials has so far been unsatisfactory. HPLC is used by E 97 and partially by D SFK where its use is indicated in footnotes. The radio-immune assay used in Sweden, which derives from blood analysis, has not yet been validated for foods where different folate forms are present and the results are kit-dependent. Compatibility: All tables state `total folate' except GK 92, IT 94 and E 88 and the printed version of NL 96 (Table 2). It is expressed in mg except in D 79 (in mg). F fat, D BLS and E 90 also give values for `free folic acid'. According to Wright & Phillips (1985), it is desirable to obtain values for `total folates', and the concept of measuring `free folates' before deconjugation is invalid. It seems that the analytical methods used at present are neither inter-compatible nor consistent, resulting in incomparable values. In DK 96, folate values are generally higher than those of other countries using the same microbiological method; values obtained by HPLC are in general higher than those from microbiological assays; in Sweden the method has not yet been validated, and the AOAC method sometimes used in compiled tables gives lower responses than the microbiological assay used in Europe. In all cases, the sensitivity of folate compounds to temperature, pH Comparison of nutrients in nine European countries G Deharveng et al and oxidation causes problems in the analysis of this nutrient. Valuable data can be found in GB, F 95 and NL 96. This means that folate intake cannot be estimated con®dently at present using all European tables. 3.7 Vitamin C Vitamin C is vested in L-ascorbic (AA) and L-dehydroascorbic acids (DHAA), both of which are highly unstable. AA is degraded into DHAA during cooking, and DHAA, which is even less stable, is then degraded to an inactive form. In general there is over 10 times more AA than DHAA in fresh foods, but DHAA levels may be increased by food processing to 25 ± 50% of AA (Vanderslice et al, 1990). De®nitions: Vitamin C activity is usually de®ned as the sum of AA and DHAA. Although only six tables state this de®nition, in four tables it can be derived from the analytical method used. The Swedish tables explicitly report solely AA. This means that the values of raw foods should be comparable, but that values for processed or cooked foods should be lower than those in the other tables. Analytical methods: Three methods are in use to measure total vitamin C amounts (HPLC, ¯uorimetry Ð preferable method Ð and colorimetry). Titrimetry measures only AA (Green®eld & Southgate, 1992). Compatibility: All tables state vitamin C activity and use different de®nitions and=or methods. All analytical methods currently used are broadly comparable. However, titrimetry should give somewhat lower values for processed foods (used partially in GB 91, GK 92 and IT 98). The vitamin C values have to be regarded with caution because the handling of food samples before analysis is extremely important due to its low stability. Vitamin C content decreases rapidly with storage time (Vanderslice et al, 1990). The natural variation of vitamin C in foods is very high. 4 Minerals (Ca, Fe, K) Calcium (Ca), iron (Fe) and potassium (K) are included in all tables apart from K in E 97. They are usually expressed in mg. In D milk, Ca and K values are given in g, in D SFK and D BLS Fe is in mg and in E 92 K is in g. Analytical methods: All minerals are measured by atomic absorption spectrometry (Osborne & Voogt, 1978), ¯ame spectrophotometry (Dvorak et al, 1971), or emission spectrophotometry (Suddendorf & Cook, 1984) and controlled with standards. In Britain, colorimetry (Paul & Southgate, 1978) is also in use for iron, and titrimetry (Paul & Southgate, 1978) for calcium. Compatibility: All methods give similar results. However, care has to be taken for iron values as the samples can easily be contaminated by dust resulting in a high variability. Discussion Comparability of values Before comparing nutritional values, it is necessary to check that the food stated in the different tables is the same. It should refer to a similar edible part, water and fat contents. General knowledge about the foods consumed in the different countries is helpful, for example, some foods may have the same name in different countries but contain different ingredients. Concerning the comparability of values, it should be noted that the activities of the EUROFOODS-ENFANT programme, founded in 1982, resulted in more comparable modes of expression, analytical methods and de®nitions throughout European food composition tables. However values obtained by old analytical methods remain in newly published tables and usually cannot be identi®ed. Furthermore, the HPLC technique, used to analyse vitamin A and E components for example, has been improved considerably over the past 5 years or so, which means that HPLC values analysed before are incomparable. In the tables studied, the nutrients can be separated into three groups. The ®rst set of nutrients groups those with comparable values, although the de®nition and analytical methods may be slightly different (nitrogen, lactose, alcohol, water, cholesterol, fat, FAs, retinol, vitamin D, tocopherols, tocotrienols, thiamin, ribo¯avin, vitamins B6 and B12, calcium, iron and potassium). Work in progress on the American table means that it will soon indicate total fat contents as the sum of FAs and will no longer include phospholipids, sterols and related compounds. The inclusion of American fat values in European tables should therefore be avoided in future. The second set of nutrients concerns those which are not readily comparable due to the calculation or mode of expression but which may be easily converted to a comparable mode of expression (protein, carbohydrates, starch, sugars, energy, carotenes, vitamin A and vitamin E). For example, in order to eliminate arti®cial differences in protein contents, the total-nitrogen values of the foods should be collected or calculated back using the protein conversion factors indicated in the tables before applying a set of conversion factors common for all countries, to the total-nitrogen contents of foods. And for energy values, the ®rst step should be to express the total CHO values in the same way and then to calculate the energy values choosing one set of energy conversion factors for all tables, for example, 16, 37, 17, 29 kJ=g (3.75, 9, 4, 7 kcal=g) for CHO, fat, protein, and alcohol. An alternative approach for estimating vitamin E intake for most foods is to use the -tocopherol fraction alone as its marker. The third set of nutrients includes those which are not comparable between all the tables studied due to the methods or de®nition used (folate and ®bre). Use of the values for the last set of nutrients is not recommended at the moment unless the source of the value is documented and it meets the chosen criteria. For dietary ®bre, conversion is possible within a limited range of TDF values, which excludes extreme low and high values. Mongeau & Brassard (1989) give a regression equation (TDF 0.02)=1.28 NSP for linking values obtained by AOAC-type methods (TDF) with those obtained by Englyst-type methods (NSP). When constructing a standardised table, it may be useful to express `dietary ®bre' both as 75 Comparison of nutrients in nine European countries G Deharveng et al 76 NSP and as TDF so that local comparison can be carried out. Additional problems concern the comparability of values in compiled tables. Firstly, for one food, values might be copied from different sources which do not refer to the same food sample. Secondly, for a given nutrient, the values might be copied from several sources using incompatible methods, de®nitions and=or modes of expression. If they are copied without conversion or recalculation and without excluding incomparable values, the values become incomparable even within the same table. Conscientious compilers, however, should avoid these problems either by recalculating the nutritional values to meet common criteria, or by indicating the source of the values. A third problem is that, in several tables, old values produced with outdated analytical methods are included in recent versions without the source being indicated, and that compilers copy this `new' value into their database. Natural variability In some foods the natural variation of nutrients, for example, vitamin C and carotenes, may be higher than the variation due to the use of different de®nitions or methods. In the tables, no or almost no information is available on the sampling scheme. This makes it impossible to judge whether natural variation was accounted for when foods were selected for analysis and whether the sample represents the range of the local food available throughout the year. Availability of nutrients and foods In general, the northern and central European tables cover more nutrients and foods than the southern European tables. The most important macronutrients and minerals are well covered in all the food composition tables studied, but fractions of nutrients and vitamins are stated less extensively in the Southern European tables. The nutrients of interest for EPIC are not available in all tables (particularly in Spain and Greece), or only for raw foods (Italy). Additional problems arise in Spain and Greece where most values are old. In Spain the mode of compilation is doubtful, so it might be better to use more recent data from the other countries in addition to the newly analysed foods of E 97. Conclusions and recommendations The comparison of food composition tables shows that for some nutrients there are important differences in the de®nitions, methods of analysis and modes of expression between the tables studied. This applies to common foods, as well as less common foods. Thus, within a given country, local tables might be used, but they are not suitable as such for comparison of nutrient intake at an international level. Additionally, many foods reported within the EPIC study are not included in the European tables, so their nutritional values will have to be calculated. This is why standardised tables need to be developed for the nine European countries participating in EPIC. In Europe, there is a further need to agree on common de®nitions, analytical methods and=or modes of expression for those nutrients for which comparable values are not yet available such as carbohydrates, ®bre, folate, carotenes and vitamins A and E. While comparing de®nitions, methods and the resulting values, it becomes clear that more effort needs to be put into documenting the sources of data, the sampling, the methods and de®nitions, as well as the algorithms and recipes used to calculate missing values. In addition, it would be helpful if those values which have been calculated or copied were marked as such. The sources of values are best documented in the Dutch, Italian and Danish tables, while the British tables, which are often used as a source for other tables, give a short description of the food, but unfortunately do not document the source of data at the nutrient level. To compile nutritional values in a standardised way, strict guidelines should be followed, and this is possible only if the tables are well documented. The values to be copied need to be checked to determined: ®rst, whether the foods are really similar, especially in terms of edible part, water and fat contents; second, whether the de®nitions and methods used are compatible; third, whether the values are expressed in the same mode or if they need to be converted. Finally, the values of protein and energy always need to be re-calculated. References Anderson JW & Bridges SR (1988): Dietary content of selected foods. Am. J. Clin. Nutr. 47, 440 ± 447. AOAC (Association of Of®cial Analytical Chemists) (1980): Of®cial Methods of analysis of the Association of Of®cial Analytical Chemists, 13th Edition. AOAC Washington DC. AOAC (Association of Of®cial Analytical Chemists) (1984): Fibre (crude) in animal feed. In Of®cial Methods of analysis of the Association of Of®cial Analytical Chemists, 14th Edition. AOAC: Washington DC. 160 ± 162 Atwater WO (1910): Principles of nutrition and nutritive value of food. Fmrs. Bull. US Dep. Agric. No142 (2nd review). Bingham SA (1987): The dietary assessment of individuals; methods, accuracy, new techniques and recommendations. Nutrition abstracts and review (series A) 57, 705 ± 741. Bligh EG & Dyer WJ (1959): A rapid method of total lipid extraction and puri®cation. Can. J. Biochem. Physiol. 37, 911 ± 917. Brubacher G (1986): Vitamin A (retinol and retinyl esters) in food HPLC. In: Methods for determination of vitamins in food. Verlag: New York, pp. 26 ± 29. Candlish JK, Gourley L & Lee HP (1987): Dietary ®ber and starch in some Southeast Asian fruits. J. Food. Comp. Analysis. 1, 81 ± 84. Carr FH & Price EA (1926): Colour reactions attributed to vitamin A. Biochem. J. 20, 497 ± 501. Christie AA, Dean AC & Millburn BA (1973): The determination of vitamin E in food by colorimetry and gas ± liquid chromatography. Analyst 98, 161 ± 167. Cummings JH, Robertfroid MB, Andersson H, Barth C, Ferro-Luzzi A, Ghoos Y, Gibney M, Hermonsen K, James WPT, Korver O, Lairon D, Pascal G & Voragen AGS (1997): Review Ð A new look at dietary carbohydrate: chemistry, physiology and health. Eur. J. Clin. Nutr. 51, 417 ± 423 Dean AC (1978): Method for the estimation of available carbohydrate in foods. Food Chem. 3, 241 ± 250. Dvorak J, Rubeska I & Rezac Z (1971): Flame photometry: laboratory practice. Iliffe: London. Egan H, Kirk RS & Sawyer R (1981): Pearson's chemical analysis in foods, 8th Edition. Churchill Livingstone: Edinburgh. Englyst HN, Wiggins HS & Cummings JH (1982): Determination of nonstarch polysaccharides in plant foods by gas-liquid chromatography of constituent sugars as alditol acetates. Analyst 197, 307 ± 318. Englyst HN & Cummings JH (1988): Improved method for measurement of dietary ®ber as non-starch polysaccharides in plant foods. J. Assoc. Off. Anal. Chem. 71(4), 808 ± 814. Englyst HN, Bingham SA, Runswick SA, Collinson E & Cummings JH (1988): Dietary ®bre (non-starch polysaccharides) in fruits, vegetables and nuts. J Human Nut Diet 1, 247 ± 286. FAO=WHO Expert group (1973): Energy and protein requirements. In: Comparison of nutrients in nine European countries G Deharveng et al Report of a joint FAO=WHO ad hoc expert committee. FAO Nutrition Meeting Report Series No52. Rome:FAO=WHO. Folch J, Lees M & Stanley GHS (1957) A simple method for the isolation and puri®cation of total lipids from animal tissues. J. Biol. Chem. 226, 497 ± 509. Granado F, Olmedilla B, Blanco I & Rojas-Hidalgo E (1992): Carotenoid composition in raw and cooked Spanish vegetables. J. Agric. Food Chem. 40, 2135 ± 2140. Green®eld H & Southgate DAT (1992): Food composition data Ð Production, Management and Use. Elsevier Aplied Science: London. Hart DJ & Scott KJ (1995): Development and evaluation of an HPLC method for the analysis of carotenoids in foods, and the measurement of the carotenoid content of vegetables and fruits commonly consumed in the UK. Food Chem. 54, 101 ± 111. Kaaks R & Riboli E (1995): Colorectal cancer and intake of dietary ®bre. A summary of epidemiological evidence. Eur. J. Clin. Nutr. 49 (suppl 3), S10 ± S17. Kinsella JE, Posati L, Weihrauch J & Anderson B (1975): Lipids in foods: problems and procedures in collating data. CRC Crit. Rev. Food Technol. 5, 299 ± 324. Meuser F, Suckow P & Kulikowski W (1985): Methods for the determination of insoluble and soluble dietary ®ber in food stuff. Z. Lebens. Unters. Forsch. 181(2), 101 ± 106. Mongeau R & Brassard R (1989): A comparison of three methods for analysing dietary ®ber in 38 foods. J. Food Comp. Analysis. 2, 189 ± 199. Olmedilla (1993), Granado F, Blanco I & Rojas-Hidalgo E (1993): Quantitation of provitamin A and non-provitamin A carotenoids in the fruits most commonly consumed in Spain. In: Food and Cancer Prevention: Chemical and biological aspects (eds) Waldron KW, Johnson IT, Fenwich GR, RSC: Cambridge England, pp 141 ± 145. Osborne DR & Voogt P (1978): The analysis of nutrients in food. Academic Press: London. Padmore JM (1990): Of®cial Methods of analysis, 15th Edition. Association of Of®cial Analytical Chemists: Arlington Virginia USA, 1, 83. Paul AA & Southgate DAT (1978): McCance and Widdowson's The composition of foods. 4th ed. Her Majesty's Stationery Of®ce: London. Passmore R & Eastwood MA (1986): Davidson and Passmore Ð Human Nutrition and dietetics. 8th ed. Churchill Livingstone: Edinburgh. Pennington JAT & Butrum RR (1991): Food description using taxonomy and the `Langual' system. Trends Food Sci. Technol. November, 285 ± 288. Phillips DR & Wright AJA (1983): Studies on the response of Lactobacillus casei to folate vitamin in foods. Br. J. Nutr. 49, 181 ± 186. Prosky L, Asp NG, Furda I, de Vries JW, Schweizer TF & Harland BF (1985): Determination of total dietary ®ber in foods and food products: collaborative study. J. Assoc. Off. Anal. Chem. 68(4), 677 ± 679. Riboli E, Roennholm H & Saracci R (1987): Biological markers of diet. Cancer surveys 6(4), 685 ± 718. Riboli E & Kaaks R (1997): The EPIC project: rationale and study design. Int. J. Epidemiol. 26, (suppl.1), S6 ± S14. Schweizer T & Wuersch P (1979): Analyses of dietary ®bre. J. Sci. Food Agric. 30, 613. Shearer MJ (1986): Vitamins. In: HPLC Small molecules, IRL Press: Oxford, England. pp. 178, 214, 1624. Slimani N (1993): Comparability of European food composition tables with a view to their use in multi-centre epidemiological studies. In Report of Eurofoods Ð Enfant project meeting, 10 ± 12 Nov 1993, Portugal. Wageningen Agricultural University: The Netherlands, pp. 156 ± 160. Slimani N, Deharveng G, CharrondieÁre UR, van Kappel AL, Ocke MC, Welch A, Lagiou A, van Liere M, Agudo A, Pala V, Brandstetter B, AndreÂn C, Stripp C, van Staveren W & Riboli E (1998): Structure of the standardized computerized 24-hour diet recall interview used as reference method in the 22 centers participating in the EPIC project. Computer Methods and Programs in Biomedicine (in press). Southgate DAT (1969): Determination of carbohydrates in foods II Unavailable carbohydrates. J. Sci. Food Agric. 20, 331 ± 335. Southgate DAT & Durnin JVGA (1970): Calorie conversion factors: an experimental reassessment of the factors used in the calculations of the energy value of human diets. Brit. J. Nutr. 24, 517 ± 535. Southgate DAT (1976): Determination of food carbohydrates. Applied Science Publishers: London. Southgate DAT, Paul AA, Dean AC & Christie AA (1978a): Free sugars in foods. J. Hum. Nutr. 32, 335 ± 347. Southgate DAT, Hudson GJ & Englyst HN (1978b): The analysis of dietary ®bre Ð the choices for the analyst. J. Sci. Food Agric. 29 (11), 979 ± 988. Southgate DAT (1995): Dietary ®bre analysis. RSC: Cambridge, UK Suddendorf RF & Cook KK (1984): Inductively coupled plasma emission spectroscopic determination of nine elements in infant formula: Collaborative study. J. Assoc. Off. Anal. Chem. 67 (5), 985 ± 992. Trowell H (1972): Ischemic heart disease and dietary ®ber. Am. J. Clin. Nutr. 25, 926 ± 932. Trowell H, Southgate DAT, Wolever TMS, Leeds AR, Gassull MA & Jenkins DJA (1976): Dietary ®bre rede®ned. Lancet I, 967. Vanderslice JT, Higgs DL, Hayes JM & Block G (1990): Ascorbic acid and dehydroascorbic acid content of foods-as-eaten. J. Food Comp. Anal. 3, 105 ± 118. Wenlock RW, Sivell LM & Agater IB (1985): Dietary ®bre fractions in cereal and cereal-containing products in Britain. J. Sci. Food Agric. 36, 113 ± 121. Wright AJA & Phillips DR (1985): The threshold growth response of Lactobacillus casei to 5-methyl tetrahydrofolic acid: implications for folate assays. Br. J. Nutr. 53, 569 ± 573. 77 idem idem idem idem idem idem idem idem idem Centre Informatique sur la qualite des aliments (CIQUAL), Paris idem Cereals and Cereal products, third supplement Milk products and Eggs, fourth supplement Vegetables, Herbs and Spices, ®fth supplement Fruits and Nuts, ®rst supplement Vegetable dishes, second supplement Fish and Fish products, third supplement Miscellaneous Foods, fourth supplement Meat, Poultry and Game, ®fth supplement Meat products and dishes, sixth supplement ReÂpertoire geÂneÂral des aliments ReÂpertoire geÂneÂral des aliments ± Tome 1 ± Table de composition des corps gras ReÂpertoire geÂneÂral des aliments ± Tome 2 ± Table de composition des produits laitiers ReÂpertoire geÂneÂral des aliments ± Tome 3 ± Table de composition des `Fruits exotiques, fruits de cueillette d'Afrique' Istituto Nazionale della Nutrizione (INN), Roma Food composition and Nutrition tables BLS ± Bundeslebensmittel-schluÈssel II-2 Greek Food Composition Table Tabelle di composizione degli alimenti 1994 ± 5th ed 1996 Greece 1992 Italy 1994 Appendix continued. The Netherlands 1996 NEVO Tabel ± Nederlands voedingsstoffenbestand Banca dati italiana di composizione degli alimenti per studi epidemiologici National School of Public Health ± Department of Nutrition, Athens NaÈhstoffe in Lebensmittel 1993 1998 Obertiefenbach Nutrition composition tables of milk and dairy products 1992 Data: Stichting NEO ± TNO Voeding, AJ. ZEIST Print: NEVO- Voorlichtingsbureau voor de voeding, PO box 85700, 3508 CK's ± Gravenhage (La Haye) Istituto Europeo di Oncologia Div. epidemiologia e biostatistica, Milano Bundesministerium fuÈr ErnaÈhrung, Landwirtschaft und Forsten, Bonn und Deutsche Forschungsanstalt fuÈr Lebensmittelchemie, Garching bei MuÈnchen BGVV, Berlin Academie der Wissenshaften der DDR Zentral institut fuÈr ErnaÈhrung, Postdam ± RehbruÈcke Justus-Liebig-UniversitaÈt Giessen Fachgebiet Milchwissenschaft Energie und NaÈhstoffgehalt von Lebensmitteln idem Germany 1979 1993 ± 1st ed 1987 ± 1st ed idem Her Majesty's Stationery Of®ce, St Clements House, 2 ± 16 Colegate, Norwich NR3 1BQ McCance and Widdowson's (The composition of foods) England 1991 ± 5th ed 9 supplements: 1988 ± 4th ed 1989 ± 4th ed 1991 ± 4th ed 1992 ± 5th ed 1992 ± 5th ed 1993 ± 5th ed 1994 ± 5th ed 1995 ± 5th ed 1996 ± 5th ed France 1995 ± 2nd ed 3 supplements: 1987 ± 1st ed Data: Danish National Food Agency, Sùeborg Print: Gyldendals Forlagsekspedition, Lindgreens Alle 12, 2300 Copenhagen S Levnedsmiddeltabeller Denmark 1996 ± 4th ed Institute responsible Name of the FCT Country=year Appendix List of food composition tables as reported by the EPIC countries. General information Karin Hulshof E. Carnovale L. Marletta F. Miuccio Simonetta Salvini Prf. Dr Edmund Renner Dr Anna Renz-Schanen P.D. Dr Helmut Hescher Beate Hescher S.W. Souci W. Fachmann H. Kraut Dr Dehne Antonia Trichopoulou Prof Dr Habil Helmut Haenel idem idem idem Jayne Ireland-Rippert idem idem idem idem idem idem idem idem idem Susan Church Ian Quaterman Erling Saxholt Person responsible Comparison of nutrients in nine European countries G Deharveng et al 78 DBNY94: recipe and food database Food composition table ± Nutrients and Energy Sweden 1994 being updated 1996 1997 Irene Mattisson recipe: developed by the MalmoÈ Diet and Cancer study, MalmoÈ food nutrient database from the National Food Administration Swedish National Food Administration, Uppsala Helen Enghardt Luz Carretero Baeza Dolores Gomez Vasquez Wander ± Sandoz NutricioÂn Olga Moreiras Angeles Carbajal Maria Luisa Cabrera Jose Mataix Verdu Person responsible Ministerio de Salud y Consumo- SubdireccioÂn general de Higiene de los Alimentos Ciencias de la Salud, Universidad de Granada Insalud Tablas de composicioÂn de alimentos espanÄoles Contenido de carotenoides en verduras y frutas de mayor consumo en EspanÄa Tablas de composicioÂn de alimentos espanÄoles 1993 1996 Alter ± Division Dietetica NutricioÂn y salud EUDEMA Universidad Tablas de composicioÂn de alimentos Tablas de composicioÂn de alimentos La composicioÂn de los alimentos Spain 1988 1990 1992 Institute responsible Name of the FCT Country=year Appendix continued Comparison of nutrients in nine European countries G Deharveng et al 79