Bacterial Endotoxin Testing: Representative Samples

enigl at enigl at
Thu Mar 20 18:21:49 EST 1997

In article <19970314030900.WAA11499 at>, panjablion at (PANJABLION) writes:

> How do you get representative samples
>from a batch of finished products once they have already been packaged and
>ther are no segregation of beginning, middle and end of the batch?  If
>anyone has any information, please e-mail me at panjablion at

IMO, At this late date, the FDA will _never_ be satisfied that you took enough samples and got representative samples.  Your SOP originally said sampling needed to be B-M-E to control the process.  Also, was B-M-E  justified in the first place by studies with more or less sampling?  I recommend time sequence or piece sequence blocked sampling.  At this point you can't claim random or blocked design. . .  But, you could try numbering the boxes and then re-randomizing them.  That might be ok but a major investigation will have to run more samples than normal to validate that design.

Look at the USP and PDA monographs for sampling schemes.  Be able to show the FDA a photocopy of the pages you are using to justify the representative sampling scheme from a valid source, (i.e., don't use an internet reference or "verbal" suggestion because they will not carry any weight with the FDA).    Add the pages to your investigation report and get high ranking sign-offs e.g., VPs of QA and Compliance.

Davin C. Enigl, (Sole Proprietorship) MEAS

Microbiology Consulting,  Hazard Analysis and
Critical Control Points (HACCP), CGMP, and Validations
for the Food, Cosmetic, Nutritional  Supplement, and Pharmaceutical Industry

enigl at

March 20, 1997
1:29 pm

More information about the Microbio mailing list