AOAC Methods in Codex STAN 234 (Preliminary Methods Review)

J. ASSOC. 01'1'. ANAt.. CHEM. (VOL. 65, NO. 4, 1982)

986

CAP/\R liT AL.:

R. Baetz, Food and Drug Administration (FDA), Dallas, TX P. Beavin, FDA, Baltimore, MD P. H. Davis, Ross Laboratories, Columbus, OH K. Faul, FDA, Kansas City, MO W. T. Gardner, Campbell Soup Co., Camden, NJ C. Hansen, FDA, Atlanta, GA R. Hopefl, Libby, McNeill & Libby, Inc., Chi· cago, IL J. 0. Hostetter, California Canners and Growers, Stockton, CA T. Klopp, Government of Canada Fisheries and Oceans Inspection Laboratory 301, Vancouver, British Columbia M. D. Loges, FDA, Buffalo, NY R. Moffitt, Carnation Research Laboratories, Van Nuys, CA C. B. Newton, Del Monte Research Center, Walnut Creek, CA R. T. Newton, FDA, Seattle, WA C. Paillard, Quebec Department of Agricul– ture, Sainte Foy, Quebec, Canada K. Panaro, FDA, Boston, MA R. D. Satzger, FDA, Elemental Analysis Re· search Center, Cincinnati, OH D. Schwartz, General Foods Corp., Tarrytown, NY P. Snider, FDA, Kansas City, MO R. F. Sudde.ndorf, FDA, Division of Nutrition, Washington, DC

um accuracy-. These differences appear to be caused by the slightly less accurate results for the green beans; however, these lower accuracies may be an artifact of low Level 1 concentra– tions. Figures I and 2 are linear regression plots of the collaborative results vs reference values for cadmium and lead, respectively. Prediction intervals at the 95% confidence level are also given. The slopes of the regression lines provide a good estimate of the overall accuracy, with a slope of 1.00 corresponding to perfect accuracy. For cadmium the slope is 0.93 and for lead it is 0.94. In comparison, the averagt> accuracies calculated from the data in Tables l and 2 are 96% for cadmium and 97% for lead. The prediction interval can be used to estimate the precision of a future analytical result. For example, if a lab· oratory analysis shows that a sample contains 0.14 ppm cadmium and if a horizontal line is drawn from that point on the y-axis in Figure I, the line will intercept the upper and lower boundaries of the prediction interval at 0.1 I and 0.21 ppm cadmium, respectively. On the basis of the collaborative study data, therefore, an analytical result of 0.14 ppm cadmium has a 95% probability of having the associated true cadmi– um value that is between 0.11 and 0.21 ppm. Recommendation It is recommended that this dry ash ASV method for the determination ,>f lead and cad– mium in foods be adopted official first action. Acknowledgments The authors thank John W. Jones for per• forming the AAS analysis of the collaborative samples and the following collaborators for their cooperation in the study:

N. Tepedino, FDA, Brooklyn, NY J. Williams, FDA, San Francisco, CA

RffERENC:ES (I) Gajan, R. J., Capar, S. G., Subjoc, C. A., & Sanders, M. (1982) f. Assoc. Off. Anlll. Chem. 65, 970-977 (2) Youden, W. J,, & Steiner, E. H . (1 975) Statistical Manual of the AOAC. AOAC, Arlington, VA

Codex Trial Method Review