Contents - Previous - Next

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

TABLE 5. State of Knowledge of Nutrient Composition-Vitamins

Niacin Folacin Vita-
Biotin Chioline Panto
Baby foods S S S I I S S I NA I
Baked foods, bread NA S S I NA NA S I NA I I
cookies, crackers I I I NA I I NA I
sweet goods I I I NA I I NA I
Beverages I I I I NA I I I NA NA NA NA I
Breakfast cereals I I I I I I I I I I
Candies NA I I I NA I I I NA I
Cereal grains, whole I I I I NA I I I NA I S I
flours I I I NA NA I I NA I I I
pasta I I I NA NA I I NA
Dairy products S S S S S S S I S I I S
Eggs, egg products I S S S S NA S I I I S S
Fats and oils I
Fish and shellfish, raw I I I I I NA I I I I I
cooked I I
Fruits, raw S S S S NA S S I NA I
cooked, dried I I I I NA I I I NA I
frozen, canned S S S I NA S S I NA I
Legumes, raw S S I NA NA S I NA S I I
cooked S S I NA NA S I NA I I
processed I I I NA I I NA I
Meat, raw and cooked, beef S S S S S NA S S I I I I
lamb I S S S S NA S S I I I
pork, fresh and cured I S S S S NA S I I I I I
veal I S S S S NA S S I I I
sausage and lunch meat I S S S S S S I I I
Nuts and seeds I I I I NA I I I NA I I I I
Poultry, raw and cooked I S S I I NA S I I I I I
Snack foods
Soups S S S I I S S NA
Vegetables, raw S S S I NA S S I NA I I
cooked I I I I NA I I I NA I
frozen S S S NA S S I NA I
canned S S S NA S S I NA I
Fast foods S S S I I I S I
Frozen dinners I S S I S
Institutional food
Mixed dishes, commercial I S S I I I S I I
home prepared I I I I
Restaurant food

Nutrient Data Research Branch Consumer Nutrition Division Human Nutrition Information Service
US Department of Agriculture Hyattsville, Maryland 20782 January 1983


- Little or no data
S Substantial data
I Inadequate data
NA Not applicable

TABLE 6. State of Development of Methods for Nutrients in Foods

  State of Methodology *
Nutrient Category Adequate Substantial Conflicting Lacking
Carbohydrates, fibre and sugars   Individual sugars Fibre  
Energy     Food energy  
Lipids   Cholesterol Sterols  
Fat (total) Trans-fatty acids  
Fatty acids (common)    
Minerals/inorganic nutrients Calcium Iron (total) Arsenic Cobalt
Copper Selenium Chromium Haem-iron
Magnesium   Fluorine Molybdenum
Phosphorus   iodine Non-haem iron
Potassium   Manganese Silicon
Sodium     Tin
Zinc     Vanadium
Proteins and amino acids Nitrogen (total) Amino acids (most) Amino acids (some)  
    Protein (total)  
Vitamins   Niacin Vitamin A Biotin
  Riboflavin Carotenes Choline
  Thiamin Vitamin B12 V itamin K
  Vitamin B6 Vitamin C  
    Vitamin D  
    Vitamin E  
    Pantothenic acid  

* Description of methodology states:

Factors Adequate Substantial Conflicting Lacking
Accuracy Excellent Good Fair Poor
Speed of analysis Fast Moderate Slow Slow
Cost per analysis Modest (< $100) Modest to high High ?
    Method modif. Method develop. modif. Method develop.
Development needs   Extraction   Extraction process
    Extraction process  
  Applications Applications Applications

Source: G. R. Beecher, 1983.


There is increasing concern about the levels of chemical contaminants that find their way, directly and indirectly, into our food system. There are examples of chemical spills directly contaminating individual packages of food and there are numerous examples of compounds entering the system through water and soil contamination. For example, there are serious concerns with the levels of heavy metals and aflatoxins in the food system. Few data are available on the "normal" levels of these compounds found in foods.

Most of our knowledge of these compounds lies in the areas covered by the FDA and USDA recalls of contaminated foods. These data are not very useful for data banks, as they are compiled for foods that were recalled and therefore not eaten. Many of the assays are designed to determine whether or not the levels of these compounds are above some specified level, not to give good data on levels below the legal recall levels. A basic problem is that of sampling. One mouldy peanut can contaminate a large batch of peanuts with aflatoxin. One piece of solder can contaminate an entire can of fruit. Experience has taught us that, to get accurate data, the entire unit must be assayed. Thus, it is often the case that it is literally impossible to assay accurately any significant part of a lot of food that is eaten.

I have been unable to determine whether there is a constant lower level of contamination modulated by intermittent batches of very high contamination, or whether there are varying levels between lots, or both. Such questions are important if data bases are to contain data on inadvertent additives and contaminants.

The sensitivity of the issues surrounding chance contamination make it extremely difficult to get access to the private data bases that contain information pertaining to these matters. Our current procedures and legalistic operations and the highly competitive nature of the food marketplace make it most difficult for companies to release this information. The food, nutrition, and public health communities would benefit if some way could be found to resolve these difficult problems so that these data could be made available to the scientific community.


It is important in this evaluation not to leave readers with the idea that we are faced with an impossible task. This is not the case! There are many things that can be done to improve the situation and we have the technology to make these improvements at a reasonable cost. Probably most important is not to try to do everything at once. Things will proceed much more smoothly, and the programme will be much easier to develop, if the big problem is broken into its component parts.

Goals for INFOODS

The long-range goal of INFOODS is to make accessible essential data on the chemical content of human foods as consumed. To accomplish this goal there is a need to:

(i) assemble and/or develop the appropriate, validated analytical method for determining the composition of foods;
(ii) assemble, develop, and utilize sound sampling techniques of the food supply to ensure that representative samples are analyzed for their chemical content;
(iii) utilize validated methodologies for the de termination of the chemical composition of foods with the appropriate quality control systems to ensure the production of validated composition data;
(iv) develop and use the appropriate data systems and data delivery systems to provide the data user with the needed information in a timely, efficient, and cost-effective manner.

Uses for INFOODS Data - Practical Expectations

Such a system could be used for a number of different purposes. For example:

(i) the assessment of the levels of various chemical components in the diets of population groups;
(ii) the design of feeding programmes for population groups;
(iii) the development of nutrition and public health education programmes in the areas of recommended diets for population groups;
(iv) the development of new programmes for the modification of the diets of population groups through new programmes in agricultural production and food processing;
(v) epidemiological research studies on the causes of various disease states associated with food intake and diet;
(vi) short-term nutrition research programmes;
(vii) appropriate medical treatment of disease states;
(viii) nutritional and other labelling of various food products.

Over the next few years it is realistic to project the successful use of food data bases for items i to iv on this list. The data are available for many compounds of interest and can be obtained for many others. Currently, the data are not of sufficient quality and quantity for good epidemiological studies, although there are probably some exceptions. In general, data bases cannot be appropriately used for most short-term nutrition studies or for crisis medical treatment of illness. In these latter cases, individual assays of the compounds of interest will probably be required. However, the available data may well be adequate for long-term medical treatment and long-term nutrition research programmes.

Labelling is a special case. There are some foods for which the data are appropriate for label claims and there are others that are not. In general, it appears that the data bases could be used for labelling of commodities if the appropriate studies were done. On the other hand, it would appear unlikely that public data bases should be used for foods whose chemical compositions are recipe- and process-controlled. This means that a great deal could be done with the appropriate data systems.


There are too many tasks for any one data system. Some of our problems today come about because different users want different things from a data base. Trying to please everyone will actually please no one. The development of multiple data bases would permit a logical development of data bases suited for the individual user without many of the inherent conflicts that exist with today's monolithic approach. The development of a number of data base systems, each with its own purpose, should be considered. These would include: (i) a scientific data base; (ii) a "best value" data base; (iii) a labelling data base, and (iv) a nutrition education data base Each system would be designed for the appropriate user groups.

Scientific Data Base

The foundation for the entire system would be the scientific data base system for use by the food composition research community. It should contain all the information needed to permit professionals to evaluate critically the data and their usefulness. At a minimum, there would be a description of the methods used for sampling, determination, calculation, and validation for each data set. There should be some type of quality index (see below) attached to each piece of data. The basic criteria for acceptance into this data base should be a scientific peer review system. With such a system, an audit trail can be established for all composition data; without such a system and the lack of an audit trail, there is considerable potential for mistakes. Such a data base should be made available to all of the scientific research community.

"Best Value" Data Base

The "best value" data base would be derived from the scientific data base system and would be the primary user-oriented system. The values in this data base would be the best estimate of the composition level for the population of the food item of interest. It would be useful if the user could get such information as the median, range, and distribution function. The data should include the weighted mean, the standard error of the mean, the number of samples analyzed, and the confidence code (see below) for the data value

Labelling Data Base

The current legal requirements for labelling suggest that a special data base be set up for labelling. The selection of the data would be governed by the special requirements of this particular area.

Education Data Base

Finally, there should be some type of data base used for education purposes. The data would be derived from the "best value" data base, but severely collapsed. Most likely, only several hundred entries of foodstuffs and only a few dozen key components would be listed.


There is currently a considerable amount of composition data available from a variety of sources. However, the data are uneven in quality for individual compounds and food items. Because many of the public health problems cannot wait until we have perfect data, some tactics need to be developed to utilize what is currently available in an appropriate manner. The new procedure described in the recent iron table by Exler (3) is one that should be seriously considered.

Quality Indices for Food Composition Data Sets

The basic concept described by Exler (3) is that each separate data set is evaluated according to fixed standards. The data sets for each compound are scored on the quality of the documentation, the appropriateness of the method, and the quality control procedures used. Each category is scored separately and given a ranking of 0 to 3 according to the principles listed in table 7. The limiting quality index is the lowest score given and is the quality index assigned the data set for that compound.

Confidence Codes for Food Composition Data Bases

The concept of confidence codes was devised to make the best use of currently available data and at the same time inform the user that not all values could be used with equal confidence. In the development of "best value" data bases, the developers generally use means, or weighted means, of individual data sets to get the "best" value. The confidence codes are assigned by summing the quality indices of the individual data sets used in the calculation of the reported value, and then using the rules shown in table 8. An example of their use is shown in an excerpt from the iron table (see table 9).

The concepts inherent in the use of quality indices and confidence codes could be extremely useful in the utilization of current composition data and in the development of future new data. While much more work is needed in this are, the outcome should be well worth the effort.

TABLE 7. Criteria for Quality Indices

Evaluation Documentation of Analytical Method Sample Handling and Appropriateness of Analytical Method Quality Control
0 None Totally incorrect handling No duplicates
1 Unpublished, but written No documentation Duplicate aliquots
2 Published, but
Reasonable, documented common
Duplicate samples
3 Complete published write-up Extensive documented testing and appropriate method was used Standard reference materials, spikes, recoveries, or blind duplicates

Source: J. Exler (Ref. 3).

TABLE 8. Selection and Meaning of Confidence Codes

Sum of Quality Indices Confidence Code Meaning of Confidence Code
6 a The user can have confidence in the mean value.
3-5 b The user can have some confidence in the mean value; however, some questions have been raised about the value or the way it was obtained.
1-2 c There have been some serious questions raised about this value.
It should be considered only as a best estimate of the level of this nutrient in this food.

Source: J. Exler (Ref. 3).


There are so many foods and so many compounds that some care needs to be taken in the selection of the appropriate strategies for producing the necessary information. Filling in all of the blanks would cost too much and take forever. The developers of such data bases should concentrate on compounds associated with public health or potential public health problems for their country or region

Which Compounds

One strategy is demonstrated in figure 1. Efforts would be focused upon those compounds associated with public health problems for which there are inadequate data. If the assay methods are not adequate, efforts should be focused upon developing new methods. If the assay methods are adequate, efforts should be to assay the appropriate foods (see below). It should be noted that some effort may be required to get the analysts to use the appropriate methods. Often analysts do not because they lack knowledge of the method and how to use it, the cost of the method and its related equipment is too high, peer pressure is exerted against it, and local laws and regulations may intervene. Those who develop food composition data bases must address these and similar problems if INFOODS data bases are to be developed.

Which Foods?

Once a given compound has been selected for assay, it is necessary to ask, 'Which foods first?" Most countries have hundreds of different food items, and developed countries may well have thousands of them. Obviously, some strategy needs to be developed to ensure that the order of the foods selected for analysis yields the most important information at an early stage, leaving the less important information for a later stage. One approach that should yield important information first is shown in Figure 2. Several concepts are incorporated into this diagram. One is that further data are acquired only when the existing data are not adequate. While this may seem obvious, history has shown us that there has been a great deal of effort in duplicating analyses where quite adequate data already existed.

TABLE 9. Iron Content of Edible Portion of Food

Item No. Food Amount of Iron in 100 Grams Confidence
AH-8 Item No. (1963)
Mean Standard Number of
mg Error Samples
1 Cracked wheat 2.6 0.42 4 b** 444
2 French, enriched 2.8 0.12 38 a 446
3 Mixed grain 3.2 0.09 136 a** -
4 Raisin 2.9 0.29 11 b 452
5 Pumpernickel 2.9 0.19 4 b 456
6 Regular 2.7 0.10 43 b 454
7 Wheat 3.5 0.05 140 b  
8 White, enriched 3.0 0.02 445 a 461
9 Whole wheat 3.2 0.15 27 a 471
10 Danish pastry 1.8 0.10 9 b 1,899
11 English muffins, plain 2.8 0.09 25 a** -
12 Dinner, enriched 3.1 0.07 110 a 1,902
13 Frankfurter or hamburger, 3.0 0.03 250 a 1,902
14 Rye 2.8 ( ) 2 b  
15 Tortillas, corn 1.9 0.06 6 c*  
16 Hamburger, lean, cooked 2.7 0.16 4 b* 368
17 Lean meat, cooked 2.7 0.08 79 b*  
18 Liver, fried 5.7 1.20 5 b* 1,267

a The values reported in the table are the means of the data from two or more sources of data in which the mean values from each source do not differ from the overall mean by more then 30 per cent of the overall mean. Other data are designated by an asterisk. A single asterisk (*) denotes that the cats are from a single source. Two asterisks (**) denote that the data are from two or more sources, but the means differ from the overall means by more than 30 per cent of the overall mean.

The data presented in this table, and in food composition tables in general, are intended to represent values of the nutrient content of food on a nation-wide, year-round basis. The information on the reliability of each value in this table should be used to assess the confidence in how closely the iron content of a food sample is represented by that value.

Source: J. Exler (Ref.3).

A second concept is that the information that is really needed is the concentration of the item in food as eaten. Much of today's analytical data were obtained by analysis of raw foods and commodities, not of cooked and processed foods. However, a large proportion of today's foods are processed for direct consumption. Thus, data on foods as eaten become a scientifically feasible operation, although there are several technical difficulties. Since obtaining the data for ready-to eat food is now possible, we should get them.

The third concept is that of core foods. While there are about 6,000 food items in the United States food system, studies have shown striking patterns of food consumption. MacDonald's accounts for about 1 per cent of the foods purchased in the United States. The work of Wolf (4) indicated that 204 (out of a possible 5,439) food items accounted for 91 per cent of the food selected by his study group. Similar results are being found by others.



These findings strongly suggest that there is a small "core" of food items consumed by population groups that accounts for the bulk of the food consumed by that group. If this is so, then the problem of getting the critical information of what and how much of a given compound is being consumed by a population group may well become a manageable problem. The existence of a "core food group" would decrease the number of foods that need to be assayed from the thousands to the hundreds or from the hundreds to the dozens. The assay of several hundred food items is a manageable task, given the availability of automated assay techniques and computers. The establishment of the existence of core food groups and their identification would be very useful in establishing the priorities for the order in which the foods should be assayed.

The final concept in figure 2 is that most chemical compounds are not uniformly distributed throughout the food system; they are usually concentrated in a few items and occur in very low concentrations in others. A few screening studies could permit the analyst to concentrate on the acquisition of good data on the important sources of the compound in core foods and thus in the food supply. A strategy of studying those foods for which there are inadequate data, in the form in which they are eaten, moving down the core food list, and concentrating on the major sources of the compound, will reveal any given chemical component in the food supply of any country.

The state of the current methods has been briefly mentioned above and is addressed by Southgate elsewhere in this issue. It will be necessary to develop and validate numerous new methods for the determination of the components of foods. These methods must be accurate, reasonably precise (probably between 1 and 5 per cent coefficient of variation), rapid, sensitive, cost-effective, and appropriate for the technology available to the analyst.


IN FOODS should establish one or more working groups to address the following tasks:

(i) Establish at least two sets of written criteria (minimum and adequate) for analytical methods used in food analyses for food data systems, to include sampling procedures, method sensitivity, precision, selectivity (accuracy), and quality control criteria.

(ii) Evaluate the current state of method for food analyses using the criteria listed in (i), as well as cost factors, operator training programmes, and working conditions.

(iii) Suggest strategies for establishing acceptable assay techniques for the chemical analysis of foods.

(iv) Establish criteria for the evaluation of the current state of food composition data to include some form of quality indices for individual data sets and confidence codes for "best value" data sets.

(v) Suggest strategies for the improvement of the data in food composition data base systems.

(vi) Develop the concepts necessary for the appropriate education of those who:

a. provide analytical data for the data bases
b. evaluate and maintain data base systems
c. use various data base systems


Most of the scientific studies of the effect of ingestion of given compounds on the health of individuals and populations have focused on human biochemistry and physiology. Most of the efforts with regard to foods have been regulatory in nature. Scientific studies of the chemical composition of foods have been rare. I think it is now time to expand our efforts in the understanding of the effects of ingestion of various compounds, including the study of chemical composition of foods and food-stuffs. It is well that this topic was included in this planning conference. Our attempts to understand the relationship of diet to health and to develop the appropriate forces of action will benefit from the addition of the scientific study of the chemical composition of foods.


The statements in this paper were developed over the past eight years in discussion with a large number of people. While I am unable to list all those who have contributed to my thinking, I wish to thank them for their ideas and penetrating comments, especially G.R. Beecher, R. Butrum, F. Hepburn, W. Mertz, H. Slover, M. Stewart, J.T. Vanderslice, and W. Wolf for the many long hours of discussion on this topic.


1. National Academy of Sciencies-National Research Council, Chemicals Used in food Processing (NAS/NRC) Publication 1274, Washington, D.C., 1965).

2. National Academy of Sciences-National Research Council, Food Additives, Summarized Data for NRC food Additives Surveys (NAS/NRC National Technical Information Service, Washington, D.C., 1981).

3. J. Exler, Iron Content of food, Home Economics Research Report No. 45 (U.S. Department of Agriculture, Washington, D.C. 1983).

4. W.R. Wolf, "Assessment of Inorganic Nutrient intake from Self-Selected Diets", in G.R. Beacher (Ed.). Beltsville Symposts on Agricultural Research No. 4, Human Nutrition Research (U.S. Department of Agriculture, Beltsville, Maryland, 1981), pp. 175-196.

Contents - Previous - Next