QUESTION: Is it required that each and every container of API be identified before the release of the same for Production. Would it be acceptable if the batch release is based on statistical sampling and the identification of each and every container is done before the container is taken for dispensing.
The answer to your question depends on the country whose GMPs you are trying to meet (and you need to meet the GMPs of the country where manufacture takes place and the country where you are selling the finished dosage form).
For the US market you only need to test a representative sample of the the drums of the incoming API, for every batch - so if you receive 10 drums of a batch of API, you only need to perform identification testing on say 4 drums (sq. root of n+1) - prior to release of the raw material to production.
If you are in, or selling to, a country that follows WHO, or EU, or PIC/S GMPs, then you need to perform identification testing on each and every drum you receive in the shipment (and this also applies to excipients) - prior to release of the raw material to production.
I hope this helps.
---------- FOLLOW-UP ----------
QUESTION: Thanks Michael, If we Consider that each and every drum has to be identified, (which is practically difficult), would it be fine that the release is based on sq. root of n+1. The identification of the remaining drums to be done prior to dispensing. Would this be in accordance to WHO, EU and PIC GMP's
ANSWER: Good try, but no.
All GMPs world-wide require that your raw materials need to be released (i.e. tested) prior to release to production (i.e. dispensing).
Unless you intend to have a humungeous "quarantine area" in your warehouse this is not practical. And even if you do have a humungeous quarantine warehouse - why would you want to wait for the morning that you need to dispense product to find out that the material failed specification?
Incidentally - why is this such an issue for you?
---------- FOLLOW-UP ----------
QUESTION: Hello Michael,
We recieve tons of raw material for our consumption. Many a times we get about 40 to 50 tons of material as one consignment (Multiple batches) and most of them in 25 or 50 Kg packs. We follow the sq. root of n +1 to release the batch, but off late it was picked on by an auditor and they have insisted that it should be 100% as per EU guidelines. We went ahead with an FTNIR and also a Raman Spectrophotometer to cater to this issue, but the sheer volume of incoming material makes it very difficult to check the identity of each and every drum, Physically sample sq. root of n +1, make a composite for analysis and release the batch. So I am looking into ways wherein we do achieve this task practically. So i was thinking out loud to actually base the release on the statistical sample and then do the identification of each drum as it gets transferred to Dispensing. This will actually make the task less stressful for the staff and more manageable.
As I see it you are only shifting the pain (in time and location); and the trouble is that one day production will be in a super rush and will not wait for all the drums to be ID-checked; at which point your system has broken down.
Sorry I am not providing an answer that you like.