outlier test and USP<1010>, Number of retests in OOS

08111832187

[ 本帖最后由 08111832187 于 2019-5-10 10:38 编辑 ]\n\n[ 本帖最后由 08111832187 于 2019-5-10 10:37 编辑 ]\n\nQuestion 2:
There is a new General Chapter <1010> in the USP. Will the statisticalrecommendations in <1010> be acceptable for the agency if they were part ofan internal procedure?
Answer:It would be OK for a firm to incorporate the procedures of <1010> except inthe cases where they are in conflict with the recommendations of our (FDA)guidances. For instance, FDA guidance recommends that outlier testing has avery limited role for chemical analysis, while USP <1010> gives quite anextensive discussion of the use of outlier tests.
The principle in our guidance remains the same: a result that is identifiedas a statistical outlier should not in itself lead to excluding it orinvalidating the result. It can be used as supporting evidence, along withother evidence developed in the investigation, that the result is invalid.

Question 3:
What's the agency's position regarding the methodologies of <1010>?
Answer:
FDA guidance makes no recommendations about the choice of statisticaloutlier tests to be used. It only says that the use of outlier tests shouldbe specified in an SOP and should be scientifically sound.
Q&As are from ECA news: https://www.GMP-compliance.org/gmp-news/fda-representatives-conduct-q-a-session-on-oos-results

"The maximum number of retests to be performed on a sample shouldbe specified in advance in a written standard operating procedure (SOP).The number may vary depending upon the variability of the particulartest method employed, but should be based on scientifically soundprinciples. The number of retests should not be adjusted depending onthe results obtained."It is to be welcomed that the document does not require a fixed numberof retests. However, an example given at the end of the guidance showshow, in the case of an inexplicable OOS result, seven retests help toprove that the first result did not reflect the true quality of the batch.This passage clearly refers back to the footnote of the Barr case of 1993!
number of retest is from ECA news: https://www.gmp-compliance.org/gmp-news/new-oos-guidance-by-fda

beiwei5du

"It is often important for the predefined retesting plan to include retests performed by an analyst other that the one who performed the original test. A second analyst performing a retest should be at least as experienced and qualified in the method as the original analyst."

Obviously, the FDA has taken into account industry comments saying that retests cannot always be performed by a second analyst, even if the second analyst is still considered very important.

请问这句话想表述的是什么意思呢？

https://www.gmp-compliance.org/gmp-news/new-oos-guidance-by-fda

beiwei5du

Question 2:
There is a new General Chapter <1010> in the USP. Will the statistical recommendations in <1010> be acceptable for the agency if they were part of an internal procedure?

Answer:
It would be OK for a firm to incorporate the procedures of <1010> except in the cases where they are in conflict with the recommendations of our (FDA) guidances. For instance, FDA guidance recommends that outlier testing has a very limited role for chemical analysis, while USP <1010> gives quite an extensive discussion of the use of outlier tests.
The principle in our guidance remains the same: a result that is identified as a statistical outlier should not in itself lead to excluding it or invalidating the result. It can be used as supporting evidence, along with other evidence developed in the investigation, that the result is invalid.

Question 3:
What's the agency's position regarding the methodologies of <1010>?

Answer:
FDA guidance makes no recommendations about the choice of statistical outlier tests to be used. It only says that the use of outlier tests should be specified in an SOP and should be scientifically sound.

其实outlier testing适用范围有限，可以用，但是担心的很多企业真的能做好相关统计学的支持嘛？