191 / 328
guidelines under Appendix J, Annex F prescribe the analysis of the data set as a whole for the
POD analysis, is there a need to also consider the equivalency of the methods in each individual
lab. This may be outside of the realm of the guidelines, but I believe that this type of evaluation
could be useful. For example in this method for the cottage chese, overall the combined data
set produced 66 and 64 positive results for the presumptive and confirmed candidate results
respectively, while a total of 73 positives were detected using the reference method. As a
whole in this case the LCL and the UCL of the dPOD encompasses the 0 value showing no
statistical differences. This is however not true for two of the participating laboratories,
laboratory 7 and 9 both produced differences between the candidate and the reference method
which are statistically significant. In both cases the candidate method produced results that
were much lower than the reference method. Does this indicate that there was a particular
problem with these sample sets, the method or the analysis? The table below shows an
example of the results in lab 7 where the candidate method detected 4 confirmed positives
whereas the reference method detected 10.The deli turkey sample set does not appear to have
the same issue above. • It is interesting to see that sample 9 in the low sample set for cottage
cheese of the candidate method for labs 7-15 all produced negative results and overall only one
lab that in the included sample set produced a positive sample. To illustrate this point, 1 out of
11 labs had a positive sample giving a POD of 9%. This is much lower than the overall POD of
about 50% for the sample set and falls outside of the statistical (normal) distribution expected
with a 50 % fractional positive result. All other sample sets exhibit a more normal distribution
of fractional positive results as would be expected. The same pattern does not repeat
anywhere in the reference method data set. • The APC counts for the sample sets seem to have
a broad range of values where the range covers 3 logs for cottage cheese and 4 logs for the deli
turkey of counts. I was not expecting such a difference given the method that is used to
enumerate these results is a well validated and used quantitative method.
ER 6
Some of the numbers need to be checked and edited accordingly.
ER 7
No
ER 8
No
General Comments (2)
ER 1
None.
ER 2
Suggestions are given below for shortening the manuscript
ER 3
NA
ER 4
No additional comments
ER 5
ER 6
ER 7
ER 8
EDITORIAL EVALUATION CRITERIA
Is the Validation Study Manuscript in a format acceptable to AOAC?
ER 1
Yes
ER 2
Yes
ERP PROFILE SUMMARIES
155