Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
The main purpose of evaluating nutrient data is to eliminate poor-quality data, leaving only reliable information for the calculation of a mean value to be used in tables and data bases. A CC of a can mean that only three studies, two of them excellent, have been published or that a great deal of minimally acceptable data exists. In the case of Se data in general, a CC of a signifies the second situation. Most of the studies reporting the Se content of foods were assigned QIs between 1.0 and 2.0 out of a possible 3.0. A CC of c implies few data of minimally acceptable quality exist. A CC of b indicates data quality falling between a and c.
Combining and interpreting data from different studies presents some unique challenges to the nutrient data evaluator and statistician. Specifically, the biases of each study must be taken into account: biases based on different samples, analytical method, reagents, instrumentation, analysts' performance, and degree of accuracy and precision for each study. Usually these biases for a given study are not quantified or documented. Differences in the mean values for various studies cannot easily be evaluated when laboratories analyse samples obtained from different sources, and use different handling techniques, reagents, etc. The calculation of a mean nutrient value across studies can be performed in several ways. Weighting strategies were of particular interest. Weighting the mean towards the number of samples in the studies is one approach that was considered: data from studies that reported analyses of the largest numbers of samples would be weighted most heavily. However, this approach would attach greater significance to the number of samples category than to the others. Another approach might be to weight most heavily data from those studies with the smallest variance. However this is not always possible because a standard deviation or standard error is sometimes not reported. A third approach can weight most heavily data from the study with the highest QI. This was also rejected owing to the narrow range of QIs and the subsequent lack of resolution in the scale. In view of the limitations discussed, weighting was deemed undesirable, and it was agreed to calculate a simple mean Se value at this time. As documentation and data quality improve, a weighting strategy could be considered for calculation of the mean nutrient value.
The derivation of the QI and consequent CCs also can be approached in several ways. A conservative scheme would be based on the assumption that the quality of a study is only as great as its weakest aspect, as was the system in the iron table [X]. From this viewpoint, the QI for each study would be equal to the lowest of the five ratings. However, applying this method of scoring to the existing Se studies would have resulted in very few acceptable data, since 0 is a frequent rating, especially in analytical quality control. Also, making the QI equivalent to the lowest rating would weight the QI toward the category with the greatest number of zero ratings. To avoid these consequences, a less conservative approach was taken which considered that: (a) sometimes quality-control measures are taken during the course of research, but not reported; and (h) the actual values found in Se papers with no mention of quality control often fall within or close to the range of values reported in Se papers that report appropriate qualitycontrol measures. This holds true for the examples shown here. For the purpose of having enough acceptable data, standards have been adjusted. However. considering this compromise on the derivation of the Ql, one safety feature was added to the calculation of the mean Se value: the exclusion of values with an index smaller than 1.0. This feature requires a study to meet a minimum level of overall quality for its data to be included. The minimum acceptable Ql was set at 1.0 because that seemed to be a reasonable cut-off point, i.e. a higher cut-off point would eliminate the majority of studies.
Users of these data should be aware that the mean Se value for each food may not be representative of average levels found in the nation's food supply. Acceptable mean values were derived from available data from one or several studies. Individual criteria were not weighted, and even low ratings for sampling plan would not disqualify a study, depending on the other ratings. However, in each case the mean value represents the best present estimate of Se in a given food item.
One of the purposes of developing this system was to encourage investigators to consider all five categories of criteria when designing studies and reporting results. The system is a dynamic one and can be modified to respond to improvements in such areas as analytical methodology and availability of SRMs as well as to the reporting of new data. As additional research is done that incorporates the top levels of these criteria' we expect to upgrade our standards to allow more stringent evaluation of published data, and thereby increase users' confidence. For example, the quantitation limit of the method was rarely reported in the studies we evaluated. Without this information, it is difficult to assess the validity of low Se levels in foods. Rating studies on this aspect of analytical method only when a quantitation limit was reported was a compromise based on the level of existing data. In future evaluations, it is hoped that a rating of 0 will be assigned in analytical method to those studies that do not report the quantitation limit of the method as well as to studies which report results below the stated quantitation limit.
Although the criteria were developed using Se as the test case, they are applicable to data compilations for other food components. The evaluation system becomes nutrient-specific with the customizing of the criteria and the scheme for deriving the QI and mean nutrient value. Use of this evaluation system for any particular food component requires these steps: collection of relevant papers; delineation of nutrient specific criteria at the various rating levels; assignment of ratings; and selection of strategies for deriving the QI and mean nutrient value. The quality of a given set of data influences the fitting of the criteria into the rating scale and the scheme selected for deriving the QI and mean nutrient value. This process is analogous to the familiar statistical problem of balancing type I and type 11 errors. If the rating scale is too rigorous, most available data will be eliminated. On the other hand, if it is too lenient, many less reliable studies will be included.
The delineation of levels of data quality permits data users to evaluate the suitability of a specific mean nutrient value for their data bases. Furthermore, access to criteria ratings that were assigned in the data evaluation procedure would allow the user of the data to assess the specific decisions made in the evaluation. The number of commercially available data bases is increasing as are the number of users. Data users must take on the responsibility of selecting nutrient data of known quality.
The authors appreciate Dr Nancy Miller-lhli7s input on technical issues relative to analysis of foods, and gratefully acknowledge the statistical advice of Dr Estelle Russik and Dr Arnold Greenland.
Names of products are included for the benefit of the reader and do not imply endorsement or preferential treatment by USDA.
1. W. H. Allaway, J. Kubota, F. Losee, and M. Roth, "Selenium, Molybdenum, and Vanadium in Human Blood," Arch. Environ. Health, 16: 342-348 (1968).
2. M. A. Amer and G. J. Brisson, "Selenium in Human Food Stuffs Collected at the Ste-Foy (Quebec) Food Market," J. Inst. Can. Sci. Technol. Aliment., 6: 184-187 (1973).
3. D. Arthur, "Selenium Content of Canadian Foods," Can. Inst. Food Sci. Technol. J., 5: 165-169 (1972).
4. J. C. Bruhn and A. A. Franke, "Trace Metal and Protein Concentrations in California Market Milks," J. Food Protect., 40: 170-173 (1977).
5. C. J. Cappon and J. C. Smith, "Chemical Form and Distribution of Mercury and Selenium in Edible Seafood," J. Anal. Toxicol., 6: 10-21 (1982).
6. W. G. Cochran, Sampling Techniques (John Wiley & Sons, New York 1977).
7. Committee on Dietary Allowances, Food and Nutrition Board, Commission on Life Sciences, National Research Council, Recommended Dietary Allowances, 9th ed. (National Academy Press Washington D.C., 1980).
8. J. Exler, Iron Content of Food, Home Economics Research Report, no. 45 (Consumer Nutrition Division, Human Nutrition Information Service, USDA, Washington, D.C., 1983).
9. S. N. Ganapathy, B. T. Joyner, D. R. Sawyer, and K. M. Hafner, "Selenium Content of Selected Foods," Trace Elem. Metab. Man. Anim. Proc. Int. Symp. 3rd, 1978: 322 (abstr).
10. D. M. Hadjimarkos and C. W. Bonhorst, "The Selenium Content of Eggs, Milk, and Water in Relation to Dental Caries in Children," J. Pediatr., 59: 256-259 (1961).
11. L. H. Keith, W. Crummett, J. Deegan, Jr., R. A. Libby, J. K. Taylor, and G. Wentler, "Principles of Environmental Analysis," Anal. Chem., 55: 2210-2218 (1983).
12. W. R. Mabey, J. S. Winterle, T. Podoll, et al., "Elements of a Quality Data Base for Environmental Fate Assessment, Final Report," Environmental Protection Agency Contract no. 68-03-2981, SRI Project PYU 2073, Work Assignment no. 6, 5 July 1984.
13. V. C. Morris and O. A. Levander, "Selenium Content of Foods," J. Nutr.. 100: 1383 1388 (1970).
14. A. L. Moxon and D. L. Palmquist, "Selenium Content of Foods Grown or Sold in Ohio," Ohio Report,Report, 65: 13-14 (1980).
15. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1577a, Bovine Liver (National Bureau of Standards Washington D.C., 1982).
16. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1549, Non-Fat Milk Powder (National Bureau of Standards, Washington, D.C., 1984).
17. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1571, Orchard Leaves (National Bureau of Standards, Washington, D.C., 1977).
18. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1566, Oyster Tissue (National Bureau of Standards, Washington, D.C., 1979).
19. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1568, Rice Flour (National Bureau of Standards, Washington, D.C., 1978).
20. National Bureau of Standards Certificate of Analysis, Standard Reference Material 1567, Wheat Flour (National Bureau of Standards, Washington, D.C., 1978).
21. O. E. Olson and 1. S. Palmer, "Selenium in Foods Consumed by South Dakotans," Proc. S. D. Acad. Sci., 57: 113-121 (1978).
22. O. E. Olson, I. S. Palmer and M. Howe, Sr., "Selenium in Foods Purchased or Produced in South Dakota," J. Food Sci., 49: 446-452 (1984).
23. J. A. T. Pennington, B. E. Young, D. B. Wilson R. D. Johnson and J. E. Vanderveen, "Mineral Content of Foods and Total Diets: The Selected Minerals in Foods Survey, 19821984," J. Amer. Diet. Assoc. (in press).
24. H. A. Schroeder, D. V. Frost, and J. J. Balassa, "Essential Trace Metals in Man: Selenium," J. Chron. Dis., 23: 227-243 (1970).
25. K. K. Stewart, "Nutrient Analyses of Food: A Review and a Strategy for the Future," in G. R. Beecher, ea., Nutrition Research, BARC Symposium no. 4 (Allanheld, Osmun & Co. Publishers, Totowa, N.J., 1981), pp. 209-220.
26. K. K. Stewart, "The State of Food Composition Data: An Overview with Some Suggestions," Food Nutr. Bull., 5: 54 68 (1984).
27. S. Williams, ed., Official Methods of Analysis of the Association of Official Analytical Chemists, 14th ed. (Association of Official Analytical Chemists, Washington, D.C., 1984).
28. W. R. Wolf and M. Ihnat, "Evaluation of Available Certified Biological Reference Materials for Inorganic Nutrient Analysis, in W. R. Wolf, ea., Biological Reference Materials: A vailability, Uses, and Need for Validation of Nutrient Measurement (John Wiley & Sons, Inc., New York, 1985), pp. 89-105.
29. US Department of Agriculture, Agricultural Statistics (US Government Printing Office, Washington, D.C., 1983).
Magnitude of the reported variability of composition
Impact of composition variation on a one-day food intake
Additional impact of a random error in intake estimation
Some implications for data analyses
Validation of food intake data: implications of food composition variation
Systematic errors in food composition data
Relevance to priorities for food composition data
G. H. BEATON
Department of Nutritional Sciences, Faculty of Medicine,
University of Toronto,
In another setting the author has been concerned with approaches to the nutritional assessment and interpretation of population data . Two problems have been identified in this work: day-today variation in intake must be taken into account or there can be potentially serious errors in the estimation of the prevalence of either inadequate or excessive intakes; and the nutritional adequacy of intake (or the risk of excess associated with detrimental factors) must be approached on a probability basis. With replicated observations the first of these problems may be addressed by statistical adjustment of the distribution following ANOVA to estimate the partitioning of variance. The second can be addressed by generating probability statements based upon the distribution of requirements among individuals [ 1, 10, 6].
In the course of developing specific approaches applicable to large-scale surveys, consideration was directed to the question of food composition "errors." Current USDA tables provide estimates of the standard error associated with the average content figures for individual foods . Standard deviations can be derived. These appear very large, with coefficients of variation (CV) ranging from about 10 to 50 per cent depending upon the food and nutrient. At first it appeared that this error, whether real (i.e. methodologic) or simply due to the range of compositions that a particular sample of a class of food might have (biological variation), was so large that any approach to assessment might be in jeopardy. To examine this specific issue, some examinations of the predicted "error" of one-day intakes were undertaken. The results of these examinations and considerations of implications are presented below.
Table 1. Variability of food composition as empirically estimated from USDA composition tablesa
CV range assumed
|Below cut-off||Above cut-off|
|Calcium||20 mg/ 100 g||5-50||5-15|
|Iron||1 mg/ 100 g||5-65||10-30|
|Riboflavin||0.05 mg/100 g||5-50||10-30|
|Vitamin B6||0.1 mg/100g||5-50||10-30|
|Vitamin A||30 IU/100g||5-65||10-30|
a. CVs estimated from reported SE of the mean value presented in table and reported number of determinations for the food .
Table 1 presents some estimates of the standard deviation of food composition generated from the standard errors and number of assays presented in the new USDA data bases . This tabulation is impressionistic rather than systematic - that is, an examination of some sample foods rather than a thorough examination of all foods was carried out in order to generate the estimates. Empirically it was observed that for each of the nutrients there was a distinction between the CV at low concentrations and at higher concentrations. This was probably due to the impact of methodologic error when concentration was low. In table 1, estimates are presented above and below arbitrary cut-off points to illustrate this phenomenon.
It can be seen that the very high CVs are associated with the low concentrations of nutrients. Above the arbitrary cut-off points the nutrients fall into two classes as far as CVs are concerned. These probably represent the general magnitude of the biological variability of food composition.
To assess the impact of this variation on the estimate of intake of an adult for one day, two sample diets, identified in tables 2 and 3 as HW1 and HW2, were taken from recorded foodintake studies. (The compositions of these diets can be found in the National Academy of Sciences report .) For each food and nutrient in these diets the CV of composition was either generated from the food composition table or, if the table presented no SE estimates, it was imputed, selecting a random value from the ranges suggested in table 1. Thus, for every food and nutrient in the two records, a mean and CV of composition was generated. The SD of composition was thus available as well.
Table 2. Food composition and variability estimates associated with food record HWIa
|Food item (g)||Weight eaten (g)||Composition per 100 8 (percentages given in parentheses)|
|Protein (mg)||Calcium (mg)||Iron (mg)||Magnesium (mg)||Sodium (mg)||Zinc (mg)||Vitamin C (mg)||Thiamine (mg)||Riboflavin (mg)||Niacin (mg)||Vitamin B6 (mg)||Folate (mg)||Vitamin A (IU)|
|Orange juice||124.0||0.68||9.0||0 10||10.0||1.0||0.05||38.9||0.079||0.018||0.202||0.044||43.8||78.0|
|Scrambled egg||64.0||9.32||74.0||1.46||12.0||242.0||1 10||0.2||0.061||0.243||0.066||0.091||35.0||486.0|
|(13 4)||(-)||(15 4)||(18 8)||(7 5)||(14 3)||(-)||(13.7)||(11.7)||(9 2)||(29.4)||(-)||(-)|
|Shredded wheat||28 4||11.00||38.0||4.22||132.0||10.0||3.30||0.0||0.260||0.280||5.25||0.253||50.0||0.0|
|Milk, whole||122 0||3.29||119.0||0.05||13.0||49.0||0.38||0.94||0.038||0.162||0.084||0.042||5.0||126.0|
|(5.3)||(6 9)||(44.3)||(37.4)||(16.3)||(16.2)||(50.2)||(27 6)||(14.0)||(15.7)||(49.4)||(28 3)||(34.5)|
|Coffee whitener||15.0||1.00||9 0||0.03||0.0||79 0||0.02||0.0||0.0||0.0||0.0||0.0||0.0||89.0|
|(20.0)||(47 7)||(83 3)||(27 3)||(-)||(59.1)||(37 8)||(35.3)||(19.5)||(4.4)||(23.6)||(28.2)||(10.2)|
|(8.0)||(12.6)||(19.9)||(13 4)||(7 3)||(31. 5)||(-)||(33 9)*||(25.3)*||(22.0)*||(44.3)*||(55.1)*||(18.1)*|
|(10.8)*||(14.8)*||(38.4)*||(34.2)*||(5.4)*||(55 0)*||(16.1)*||(34 0)*||(21.0)*||(19.0)*||(14 7)*||(30.2)*||(36.6)*|
|Skim milk||245.0||3.41||123.0||0.04||11.0||52.0||0.40||0.98||0.036||0.;40||0.088||0.040||5.0||204 0|
|(4.3)||(13 3)||(38.2)||(17.0)||(27 1)||(36.9)|
|Number of items with imputed CV (*)||3||4||3||3||3||5||3||4||4||4||5||5||6|
a. CV values are shown below average composition.
Contents - Previous - Next