This is the old United Nations University website. Visit the new site at

Contents - Previous - Next

Where to find food composition data

Potential sources of nutrient data include published food composition tables and data bases, journal articles, books and book chapters, proceedings from meetings, project reports, unpublished laboratory reports, university theses, reports from food producers, and materials produced for consumers. Data are generated and compiled by international and national agencies, academic groups, clinical/medical groups, food industries, trade associations, and private groups, usually for specific studies relating to the nutritional properties of foods or as part of efforts to create food composition data bases, either reference or special-purpose. (See Schakel et al. [78].)

International agencies, such as the FAO,collect food composition data in order to judge the quality of international food trade, and the potential needs and resources of various areas of the world. Government agencies conduct food composition analyses for regulatory purposes (e.g., to determine the nutritional quality of foods and to determine compliance with label claims), to develop data bases for use in dietary surveys evaluating the dietary status of population groups, and frequently to provide food composition data to professional groups and consumers. Academic groups generate food composition data as part of research efforts in food science and nutrition, while clinical/medical groups generate food composition data for analyses of patient diets and for nutrition intervention studies. The food industry analyses its products to determine their nutritional quality and to develop nutritional labelling claims. Trade associations generate food composition data to provide information for constituents and consumers. Private laboratories generate food composition data under commission for other groups.

Schakel et al. [78] provide a detailed listing of various sources of food composition data. The Nutrient Data Bank Directory [37], Computer Programs and Databases in the Field of Nutrition [31], and the INFOODS International Directory of Food Composition Tables [28] list many of the major tables and data bases in existence, and provide first steps for finding already compiled data. Journal, book, and proceedings references may be researched through computer retrieval, using keywords for titles, to assist in locating appropriate references. Food industries and trade organizations can be contacted directly, by post or by telephone. Listings of these industries and organizations (e.g., in the United States: the Thomas Register [90], the Million Dollar Directory [18], and the Trade Names Dictionary [106]) may be found in some libraries. It is more difficult to obtain project reports and unpublished laboratory data. Contact with individuals known to be involved with such work, as well as with government agencies, may give access to some of this information. INFOODS and its various regional liaison groups are also good potential sources of such information.

In some countries certain information about nutrients can be found on food labels, provided by food packagers and manufacturers. Nutrition information on food labels is generally under the regulation of a government agency seeking to ensure the safety and quality of the food supply. National laws may indicate that some label information is mandatory and some is voluntary, with the form and content of the mandatory information often being quite rigidly defined. As a result, the specific information provided by a food label usually varies both between and within countries.

In dealing with label information, manufacturers should be contacted to determine the specifics of the information, for example, if the values are based on analyses done by the manufacturer or if they are derived from a data base. It should be noted that label data for multi-ingredient foods may be calculated by the manufacturer from data base values of ingredients, and that label values may represent upper or lower estimates or be variously manipulated to be in compliance with national laws. Original data should be requested from the manufacturer with as much documentation as possible. If original data cannot be obtained, label data may be used as a "last resort" and subsequently updated or replaced when "better" data become available.

Label information is especially useful for identifying foods, describing foods, determining weight and other measures of serving sizes, identifying ingredients of multi-ingredient foods, and estimating nutrient values. Additionally, label information for multi-ingredient foods can sometimes be used to reconstruct recipes, and from this information the levels of nutrients not included on the label can be estimated. (See chapter 6.)

Evaluation of data from various sources

Once sources of relevant data are found, these data must be evaluated for suitability to the purpose at hand, and for their fit with the other data being assembled. Based on the discussion in chapter 1, the data must be examined according to the criteria of food identification, nutrient identification, data manipulation, and data quality.

Food Identification

A difficult task facing the food composition data base compiler is to determine if the data found in other sources are representative of the foods that are needed. While the detailed information required for such a determination is generally not available, it is still necessary to attempt to match the food for which data are being borrowed as closely as possible to the food required for the data base.

As an example, one may want to present nutrients for the food "fried calf. A search through the available sources may find nutrient values for four foods described as:

cooked beef liver
fried liver
calf liver, cooked
beef, yearling liver, fried

It is possible that these descriptions pertain to the same food or that, even though the foods are somewhat different, the nutrient values may be similar enough to warrant "borrowing" the data. This is a decision that ultimately the compiler must make on the basis of the projected use of the data base, although it is recommended that advice be sought from experts in individual food areas.

The effort to match food names encounters a plethora of problems due to the cultural importance of food. Foods and their names change from place to place, both between and within countries. The name "tortilla" is applied to a variety of foods throughout Latin America, while a variety of names is applied to carbonated beverages in the United States (e.g., cola, pop, tonic, soda). Cuts of meat vary around the world, as do parts of foods that are considered edible (e.g., tops and bottoms of beets and the rind of cheese). As one seeks to incorporate data from specific foods into a table it is essential that one know precisely what foods and parts of foods were analysed.

Nutrient Identification

When borrowing data from others, one must first determine the basis of the measurement, whether it is per purchased weight, per hundred edible grams, per wet or dry weight, etc. Some of these bases can be easily converted into one another (e.g., per pound to per kilogram), while others need special information (e.g., amino acids reported as milligrams of nitrogen per gram of total nitrogen can be converted to milligrams of nitrogen per 100 g edible portion only if total nitrogen content is known).

Beyond the basis of measurement are the forms and units of the nutrients. There are many nutrients which informally share the same name, yet are quite different for the chemist and are measured in different ways. Of special concern are dietary fibre, carbohydrate, protein, vitamin A, vitamin E, and niacin. (See chapter 3 and, for additional discussion, the INFOODS manual on nutrient identification [44].) The compiler of a food composition data base must completely specify the forms and units of the nutrients desired for the data base, and then carefully examine data obtained from other sources to ensure that they are expressed in, or can be converted to, the desired units.

Data Manipulation

There are two levels of data representation of concern to the food composition data base compiler: what the numbers that are presented represent, in terms of the distribution of the values for the nutrients that were originally measured, and how food samples that were analysed relate to the food described. In borrowing data from other sources, one must be sure that the reported data were manipulated in the same way as the data desired; for example, if means are to be used in the projected table, then means must be sought in other data.

To judge what the data represent in terms of a food, it is essential to go beyond the description of the food and explore the sampling method. Some information about the type of sample (e.g., locale, colour) may be inherent in the food name (e.g., Maryland steamed crab, green chili); however, information about the sampling scheme, number of samples, etc. may not be available from some data bases. Information of this type is generally more available from journal articles, books, proceedings, and unpublished reports. Such information is essential to determine the comparability of a food from a primary source to the food in the data base.

Data Quality

Perhaps the most important aspect of data borrowing, and the most difficult to judge, is that of "quality". This is discussed in chapter 1 in the context of determining the general quality of a data point. In the specific question of how to judge the quality of a borrowed data point, there are three factors involved: documentation, consistency, and reputation.


Evaluation of food composition data relies on the information that accompanies those data; data unaccompanied by descriptive information (what they are and how they were gathered) are not, in general, useful. Documentation should include a description of the food, details of the nutrients and how they were determined, and description of the sampling and the data manipulation. (See chapter 1 for further discussion of these topics.) It is the presence and content of such documentation that provides the basis for evaluation of the data.


Data obtained from other sources must be examined to determine how "reasonable" they are, in terms of both absolute and relative consistency. (See the discussion of validity checks in chapter 1.) Consistency checks are of three types:

These consistency checks must be performed with an awareness of the inherent variability in foods, so unusual data are not automatically excluded.


An important component of quality, and the most difficult to define, is the reputation of a food composition data source for performing analyses well and preparing and providing good data. If a data source is well known for quality work, then the values that it produces may be used with more confidence, even if they differ from what is expected. However, mistakes are possible even by the most reputable data sources. Questionable data should be brought to the attention of those who generated the data to rule out possible reporting errors (e.g., transcription or other inadvertent errors that could occur in preparing the data for publication). Data that are not from a known source must be examined more carefully and documented more rigorously. Unfortunately, many compilers are not well acquainted with many of the sources of data, which makes the assessment of reputation difficult.

5. Estimation from Data on Similar Foods

How foods differ
Adjustments for differences between foods
Assumed zero: a special case

Data are often needed for a food or nutrient that cannot be generated or found in the literature. In some of these cases it is possible to estimate the needed data on the basis of data from a "similar" food. It may be a biologically similar food (e.g., a different variety of apple, or cabbage for brussels sprouts), or a different form of the same food (e.g., raw for cooked). The basic problem is the choice of a food which is "close" to the food of interest; this requires an intimate knowledge of the foods and nutrients, and experience in matching them.

Occasionally, a food is found which can be considered so similar to the food needed that data can be substituted directly, with no adjustment necessary for the nutrient values. More often, however, the two foods differ enough that some adjustment, or calculation, is involved in the substitution process. These adjustments attempt to correct the data for the differences between the substituted food and the food of interest, and it must be realized that the resultant data are approximate.

Estimating food composition data from data on similar foods must be done with care, and only when other options are not feasible. No hard and fast rules can be given for these estimating procedures, only general guidelines. Documentation must be provided to indicate which values have been estimated and how they were estimated. This chapter discusses the major factors involved in judging the similarities and differences between foods, the specific adjustments that can be made to correct for estimated differences, and the special situation in which it is appropriate to assume the total lack of a nutrient in a given food.

How foods differ

Knowledge of how foods differ is important for judging a food's suitability for substitution. The personnel involved in these decisions must have a detailed knowledge of food composition, and, in particular, the effect of various factors on the nutrients of interest. The following sections discuss some of the more important factors that can affect the nutrient composition of foods. Further discussion of this important topic, and additional references, can be found in the work of Harris and Karmas [42], Pennington[61], Bender [9], and Breen [10].

Biological Relationships

The genetic similarity of food items is perhaps most important in determining nutrient similarity, although genetic similarity does not guarantee nutrient similarity. Since common names of foods vary from place to place, many food composition tables list the scientific names of the food items. This is not a universal solution, since the relationship between common name and scientific name is not entirely consistent. Not only does the consumer sometimes use the same name for several different foods, but scientific names may not be helpful for complex dishes. Polacchi [67] discusses the complex situation with regard to fish. Moreover, scientific names are not universally unique. For example, the 1981/82 German tables [79] place pears and apples in the same genus while the British and US tables separate them. There is the additional problem that scientific names give little assistance with mixed dishes, which are a major component of diets worldwide. Finally, it is not a general truth that closely related foods are similar in nutrient content. There are extreme variabilities among cultivars (e.g., of rice and corn). (See also Meyer et al. [52].) Often a nutrient data base developer must rely on verbal descriptions of a new food item and make assumptions regarding its genetic similarity to known food items.

EXAMPLES: USDA [96] uses amino acid data from standard yams for the related Hawaiian mountain yam.

Goddard and Matthews [24] have reported on five sweet potato cultivars in North Carolina: the carotene content varied from 2,175 IU to 3,950 IU per 100 g, almost a twofold difference.

Crowing Conditions

For genetically similar foods, much of the observed nutrient variability is related to the conditions under which the foods are grown or raised. The condition of the soil, the amount of rainfall and sunlight, the altitude, and the average and extreme temperatures all affect the nutrient content of plant crops. Similarly, the environmental conditions of animals during their growth, in addition to what they are fed and how they are handled, affect the nutrient content of the products derived from them. While it is sometimes possible to infer information about foods from the same geographic area, determination of the similarity of foods raised in different parts of the world is a very difficult procedure.

EXAMPLES: Paul and Southgate [59] show great differences in the carotene and vitamin D content of milk depending on whether it was obtained in the summer or winter.

Holden et al. [33] show that selenium values for homegrown foods in the United States are two to ten times those for foods bought commercially.

Maturity and Portion or Part

The nutrient composition of plants and animals changes during their lifetime, and therefore the maturity of a plant crop at harvest or the age and status of an animal at slaughter may greatly affect its composition. In addition, the part of the food item consumed is an important factor; generally, different parts of the plant or animal differ considerably in nutrient content. Cuts of meat vary tremendously, as do parts of plants (e.g., roots vs. leaves). Furthermore, between and even within different cultures, the portions considered edible may vary (e.g., apple peels and cores, cheese rinds). Thus, direct substitution may not be appropriate unless all these factors agree. Confusing the issue is the fact that the descriptions included in food composition tables are often culturally biased and may not be detailed enough to allow careful evaluation of similarity.

EXAMPLES: The vitamin A activity in tomatoes more than doubles (from 64 RE to 133 RE per 100 g) as they ripen from the green to the red-ripe stage of maturity [96], while the distribution of carotenoids changes from primarily beta-carotene in the green and pink stages to primarily lycopene in the red-ripe stage [77].

"Turnips (Brassica rapa)" may refer to either the root or the leaves. The former contains only a trace of vitamin A, while the latter are a concentrated source of this vitamin.

The fat content of meat depends very much on its maturity at slaughter. Furthermore, while fat is a prized component of meat products in many countries, in others it is routinely trimmed and discarded. In general, it is rarely appropriate to directly substitute prepared meat products across cultures, even though the items have the same apparent name.

Processing and Preparation

Processing and preparation are general terms used here to refer to any changes made to a food between harvest and consumption. Once the plant is harvested or the animal is slaughtered, changes in nutrient values begin to occur immediately. If consumption is delayed, a variety of processes can be used to preserve the food until it can be consumed, to make it more appetizing, or both. Almost all of these processes affect the nutrient content of a food, with different effects depending on the food, nutrient, and process. Finally, the preparation of a food prior to consumption is a further source of nutrient modification and variation. Some of these processes are of great commercial interest and have been examined, but more work remains to be done. Food processing and preparation procedures include heating or cooking, canning or bottling, freezing, drying, fermenting, irradiating, packaging, storing, separating mechanically, adding chemicals, cutting, grinding, and mixing.

These techniques have various effects on the nutrient content of foods, primarily as a result of mechanical separation, moisture addition or removal, heat, oxidation, acidity change, light exposure, and the addition of food additives. The effects depend on both the specific foods and specific nutrients involved. See Rechcigl [74], Karmas and Harris [42], Tannenbaum [88], and Bender [9]

Mechanical Separation

Mechanical separation includes milling, husking, peeling, trimming, and other mechanical activities that separate out a portion of the foods for discarding or repackaging. It is well known that nutrients are not usually uniformly distributed throughout a food, and differentially removing a portion of the food may change its nutrient composition.

EXAMPLE: The Japanese tables [75] show that in the processing of brown rice to "wet/milled" rice the iron content drops from 1.1 mg to 0.5 mg per 100 g while the thiamin level drops from 0.54 mg to 0.12 mg per 100 g.

Moisture Addition or Removal

Water is used to prepare and cook various foods such as rice, oatmeal, legumes, pasta, and vegetables. Water may also be added to rehydrate foods such as condensed soup or fruit juice concentrates. The addition of water dilutes the nutrient content of foods on a weight basis. Nutrient loss may occur if water-soluble nutrients migrate from the food and are discarded when excess water is drained.

Dehydration and the concentration of foods is a major way that foods are preserved (both commercially and in the home) since dry foods are less vulnerable to microbial contamination. The specific procedure by which the moisture is removed can be critical to changes in nutrient levels, with potential losses due to heat, as well as to removal of some of the water-soluble vitamins with the water. A primary effect of moisture removal, from the point of view of the table compiler, is the increase in the concentration of most nutrients on a weight basis.

EXAMPLE: In the USDA data base [96] (all per 100 g edible portion), raw apricots have 86.4 g water, 14 mg calcium, and 10.0 mg ascorbic acid while dehydrated apricots have 7.5 g water, 61 mg calcium, and 9.5 mg ascorbic acid. Therefore, drying concentrates the calcium but destroys much of the ascorbic acid.


Heat is intrinsic to the preparation of many foods (roasting, frying, baking, boiling, etc.) as well as to many commercial processes of food preservation (blanching, pasteurization, sterilization, etc.). It induces changes in many nutrients, both degrading their utility to humans (e.g., destroying vitamins directly as well as increasing rates of oxidation) and enhancing it (e.g., by denaturing some destructive enzymes). Most important in the prediction of the results of heat on nutrient values are the temperature and time exposed, facts often not recorded in the descriptions of foods.

EXAMPLE: In the British tables [59], spaghetti has an almost sevenfold increase in moisture content (10.5 to 71.7 g per 100 g) when it is boiled. This cuts to a third the concentration of dry solids (from 89.5 to 283 g per 100 g) and of all nutrients in the dry solids. However, due to degradation and cooking losses, there are disproportionate decreases in some nutrients. For example, thiamin decreases to 7% of its original value (0.14 to 0.01 mg) and riboflavin to 17% (0.06 to 0.01 mg), considerably less than the 33% predicted by dilution alone.


Oxidation, resulting basically from exposure to air, is an important phenomenon of many food preservation and preparation procedures (e.g., storage and shredding). Oxidation results in the destruction of some vitamins (e.g., vitamin A, vitamin C, folic acid).

EXAMPLE: In the USDA tables [96] there are data for frozen onions, both whole and chopped. While most of the nutrients are virtually unchanged by chopping, ascorbic acid levels differ by a factor of two (whole onions contain 8.0 mg per 100 g edible portion; chopped onions contain 3.3 mg per 100 g edible portion).

Acidity Change

A change in the pH of a food can trigger a number of changes in the level of its nutrients by enhancing or inhibiting chemical reactions. Alkaline conditions are used to extract proteins, and also accompany the use of alkaline baking powders. Particularly important with high pH (alkaline) processes are the increased destruction of certain vitamins (thiamin, ascorbic acid, and pantothenic acid) which are protected by acidity.

EXAMPLE: Tannenbaum et al. [89] note that the rate of ascorbic acid oxidation reaches a maximum at pH 4, declines to a minimum at pH 2 and then increases with increasing acidity.

Light Exposure

Exposure of food to light has the effect of accelerating the oxidation process and specifically degrades riboflavin and vitamin C.

EXAMPLE: Bender [9] notes that, in two hours of exposure to bright sunlight, bottled milk lost as much as 50% of its riboflavin.

Food Additives

Food preservation and preparation often involve adding substances to modify (e.g., enzymes), ferment (e.g., yeast), preserve (e.g., antioxidants), or enhance (e.g., spices) the food product or to do some combination of these. Additionally, metallic contamination may be an inadvertent side effect of food processing and preparation. Some of these additions increase nutrient levels (e.g., fermentation may initiate synthesis of vitamins); some protect nutrients (e.g., sequestering agents bind to minerals and prevent their enhancement of oxidation); and others degrade or destroy nutrients (e.g., bleaching of flour can promote oxidation of nutrients). While much research has gone into the effects of these additions, it is such a large area that relatively little is known about additive interactions and the quantitative effects of these additions.

EXAMPLES: Tannenbaum et al. [89] point out that sulfite, commonly used to prevent browning fruits and vegetables, protects ascorbic acid but is detrimental to thiamin.

Due to its commercial importance, the fermentation of cheese has been studied extensively. Dworschak [19] notes that during ripening of cheese some of the B vitamins increase while others deteriorate, in part depending on the specific fermentation process.

Many countries require fortification of various products, and both the types of additives and levels of fortification must be considered when substituting nutrient values for food items from different countries. Many countries iodize salt; some add vitamins A and D to milk.

Similarity of Ingredients

For multi-ingredient foods, the type and proportions of the ingredients affect nutrient content. For example, the amount of salt added to a product will obviously affect the sodium and chloride content of the food, and, similarly, the amount and type of fat added will affect the energy content and lipid profiles of a food. The amount of water added will affect the concentration of all nutrients. When many ingredients are involved, it becomes very complex to estimate the similarity of mixed dishes unless exact recipes are known. In most cases, it is more accurate to develop recipes for multi-ingredient dishes than to assume that dishes are similar enough to directly substitute one for another. (See chapter 6.)

The similarity between foods cannot be judged without careful assessment of the above factors. However, it must be emphasized that close agreement on all of the above does not ensure that the nutrient values for two apparently "similar" foods are similar.

Contents - Previous - Next