| Food Composition Data: A User's Perspective (1987) |
|The uses of food composition data|
|Using food composition data to communicate nutrition to the consumer|
Food composition data characteristics and limitations
Many factors influence the reliability and accuracy of nutrient composition data, and an understanding of these factors is essential before working with nutrient data bases. Recently, the strengths and limitations of nutrient data have been addressed in detail in conferences, speeches, and papers [31,51,62]. Of special interest are discussions by Perloff  and Hepburn  on USDA food composition data. Their papers discuss the uses, strengths, and limitations of food composition data in general. Four issues about food composition data are particularly important: (a) the representativeness of the data; (b) the sources of data; (c) the method used to derive nutrient values; and (d) the adequacy of analytical methodologies.
The representativeness of data depends on the number and quality of laboratories performing food composition analyses, the number and quality of samples used, and any weighting procedures used. Representative data reflect the nutrient composition of food products on a nationwide, year-round basis. Perloff  suggests that data are generally more reliable when they are based on analyses of a large number of samples from many locations and when they are compiled from several laboratories.
Often, a weighting scheme is used to average nutrient values from a number of samples. This makes nutrient values more representative of a national food supply and allows varieties of foods that are produced and consumed in larger quantities to be more accurately represented in the final value.
The reliability and usefulness of data also depend on its source. Analytical procedures and methods differ between laboratories and can influence the reliability and accuracy of nutrient information. This underlines the importance of using data compiled from several sources .
Published food composition values are derived either through direct analysis or by calculation. When data are obtained from the direct analysis of samples, natural variations in a food sample should be considered. Techniques for analysing samples also change with changes in prevailing cultivars or breeds, food products, and advances in food technology .
Values that cannot be averaged from actual analytical data can be calculated using analysed values. For example, protein is calculated from the nitrogen content of the food. Calculations for food mixtures are performed using analytical values for ingredients in the mixture. For cooked foods, calculations are based upon analytical values for the raw products, then adjusted for yield and retention factors.
Calculated values provide a different kind of data than those from direct analysis. The quality of calculated values depends on the quality of the original analytical values and the accuracy of calculation procedures. As advances are made in analytical methodologies, truer measurements of nutrient levels will result in better quality and greater reliability of data from both direct analysis and calculations .
The state of analytical methodology is an important factor influencing the availability and quality of food composition data. Hepburn  suggests that the status of analytical methodologies is often dependent on the interest expressed by experts in the field as well as on advancements in technology. Interest in certain nutrients shown by the professional community often sparks further research in the development of analytical methods for those nutrients.