AOAC ISPAM Stakeholder Panel Meeting Book 9-24-17

Quantitative Microbiology Validation Acceptance Criteria

ƒ Original Statistics - ƒ Log 10 transform data • Can’t be performed for Uninoculated Control Level (0 Data Points) ƒ Perform ANOVA or t-test to determine if mean replicates at each level are different ƒ Calculate repeatability by determining the standard deviation (s r ) at each level ƒ Graph the mean results for each contamination level (Candidate – Y axis, reference – X axis) and determine the slope and linear correlation coefficient (r 2 )

ƒ Acceptance Criteria per matrix per level was established • p value (obtained from ANOVA or t-test) should be >0.05 to establish equivalence between methods

Quantitative Microbiology Validation Acceptance Criteria

ƒ Revised Appendix J Guidelines ƒ Log 10

transform data using updated equation

Log 10 [CFU/unit + (0.1)f] f is the reported CFU/unit corresponding to the smallest reportable result

ƒ Modification to Log 10

transformation is to normalize the data and can be

used to evaluate uninoculated control levels (data results with 0) ƒ Plot all data, candidate (Y axis) vs. reference (X axis) – Identify outliers ƒ Perform outlier tests (Cochran and Grubbs) - What does the outlier mean? If data is an outlier can you exclude? How to report them? ƒ Calculate repeatability/standard deviation (s r ) and reproducibility (Collaborative study only) ƒ Additional requirement for OMA methods is to construct a Youden Plot