Food Authenticity Program Meeting Book (March 13, 2020)

This occurrence is factored into the probability distributions when the number of replicated were chosen. To allow some false positive or false negative results would require additional replicates to be performed. The chosen number of replicated with degree of proper identification was a compromise of the working group.

Sometimes there may be some replicate results that may be below the established limits for authenticity due to method or method+matrix variability. So that we don't unnecessarily reject one or more results that are outside the established range, we should look at the confidence interval generated by the authentic versus the suspect samples. If there is any overlap, then, we cannot conclude that the sample is adultered (using the binary result decision). Or should we let the method developer decide how confident the result is?

I propose to request the method developers to include a confidence interval (e.g at 95% level) of the 30 means is suspected versus authentic samples (also 30 reps) to verify this.

Complete documentation for the collection and use of authentic samples is must supplied by the method authors. The scope of authentic samples used to construct the “authentic” profile must be applicable to the defined scope of the NTT method. Expansion of the scope is possible with the inclusion of additional authentic samples into the baseline calibration, and validation using the protocol listed in this SMPR The FCC non-targeted guidance provides validation steps on how to develop and validate a non-targeted method, which are helpful information for stakeholders.

Please clarify the statement.

The SMPR leaves the authentic sample collection to the method developers who must supply full documentation on the process. The collected materials will be reviewed by the ERP to properly define the matrix scope of the method. Reference added to all 3 SMPRs.

Add: Appendix XVIII: Guidance on Developing and Validating Non- Targeted Methods for Adulteration Detection.