DisplayTitle

Agricultural commodities are often inspected for mycotoxins and classified into acceptable and unacceptable categories depending on whether the mycotoxin level in a sample. Because of the large variability associated with the sampling, sample preparation, and analytical steps of a mycotoxin test procedure, the true mycotoxin level in a lot can’t be determined with 100% confidence and as a result some lots will be misclassified by the mycotoxin sampling plan. Some good lots will be rejected (seller’s risk) and some bad lots will be accepted by the sampling plan (buyer’s risk). These two risks can be described by an operating characteristic (OC) curve. A statistical method based upon the knowledge of the variability and distribution among replicated sample test results from the same lot was developed to calculate an OC curve related to a specific mycotoxin sampling plan design. The total variance associated with a mycotoxin test procedure was partitioned into the variance associated with sampling, sample preparation, and analysis. The sampling step was typically observed to be the major source of variability. The sampling, sample preparation, and analytical variances were shown to be a function of the mycotoxin concentration and regression equations were developed to describe the functional relationships. The observed mycotoxin distribution among replicated sample test results was positively skewed and could be adequately described by the negative binomial or the compound gamma distributions. Thes two distributions were chosen because they are skewed distributions and the distribution parameters match the observation that the variance is greater than the mean. The method to evaluate and design mycotoxin sampling plans has been used to manage mycotoxin levels in the food and feed industries regulatory agencies, and international organizations (Codex).