AOAC-RI ERP Book Micro Jan 19 2017

positive ‘screening stage’ and a positive ‘confirmation stage’), true negatives (samples with a negative ‘screening stage’ and a negative ‘confirmation stage’), false positive (samples with a positive ‘screening stage’ but a negative ‘confirmation stage’) and false negatives (samples with a negative ‘screening stage’ but a positive ‘confirmation stage’ ) for the candidate method testing of frozen blueberries is provided in Table 4. The AOAC fractional positive criterion (POD between 0.25 and 0.75) was met, providing evidence that the inoculation procedure was successful for the frozen blueberry matrix. Additionally, no false positives or false negatives were identified when comparing the ‘ screening stage ’ and ‘ confirmation stage ’ of the candidate method. As described in Section 2.7, the results presented here were statistically analyzed according to the POD methods of the 2012 version of the AOAC International Methods Committee Guidelines for Validation of Microbiological Methods for Food and Environmental Surfaces document (1). The POD CP , POD CC , dPOD CP , POD C , POD R , and dPOD C values calculated at each level of inoculation for the candidate method and the reference method testing of spinach, and their corresponding 95% confidence intervals, are provided below in Table 5. Review of the POD calculation outcomes for the frozen blueberry candidate method (25 g Frozen Blueberries Analyzed via the 3M TM MDA) and reference method (25 g Frozen Blueberries Analyzed via the FDA BAM Ch. 4A), shows that the 95% confidence intervals contained 0.000 at all inoculation levels for the dPOD CP , comparing the ‘screening stage’ and ‘confirmation stage’ of the candidate method, and the dPOD C , comparing the candidate method and reference method. Because of this, for the candidate method, the POD CP values were not significantly different from their corresponding POD CC values. Based on this outcome, it can be stated that the probability of detecting a positive result in the “screening stage” of the candidate method was not significantly different fro m the probability of detecting a positive result in its “confirmation stage” at any of the inoculation levels tested. Additionally, for the candidate method versus the reference method, the POD C values were not significantly different from their corresponding POD R values. Based on these outcomes, by fulfilling the criteria to contain 0.000 within the 95% confidence intervals, it can be stated that the probability of detecting a positive result in the candidate method was not significantly different from the probability of detecting a positive results in the reference method.

CONFIDENTIAL INFORMATION Page 6 of 13