ERP Micro December 2019

Data for each contamination level was analyzed using the probability of detection (POD) statistical model (13) and conducted using the AOAC Binary Data Interlaboratory Study Workbook (Version 2.3; 14). The probability of detection (POD) was calculated as the number of positive outcomes divided by the total number of trials. LPOD values were calculated as the total POD values for all collaborators. The LPOD was calculated for the candidate presumptive results, LPOD CP, the candidate confirmatory results (including false negative results), LPOD CC , the difference in the candidate presumptive and confirmatory results, dLPOD CP, presumptive candidate results that confirmed positive (excluding false negative results), LPOD C, the reference method, LPOD R , and the difference in the confirmed candidate and reference methods, dLPOD C . A dLPOD C confidence interval not containing the point zero would indicate a statistically significant difference between the candidate method and the reference method at the 95 % confidence level. In addition to POD values, the repeatability standard deviation (s r ), the among laboratory repeatability standard deviation (s L ), the reproducibility standard deviation (s R ) and the homogeneity test value (P T ) were calculated and interpreted according to ISO 5725-1 (13, 15). The homogeneity test value can provide information on the homogeneity of the test portions sent to the laboratories or inconsistent performance of the method in certain labs.

7