[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index

Re: Sampling raw materials

Andy Minkevich,

In general, there is no accepted standard
for sampling chemical raw materials for
the production of an API.

Each firm should establish (prove) that the
raw material inspection (sampling and sample
assessment) plans they use to inspect each
lot shipment provides a lot-shipment-
representative sample set with sufficient
granularity that the annual risk of a
production batch failure is, at the 95% (or
higher) confidence level, statistically less
than 1 in 100 to less than 1 in a 1,000,000,
with the statistical level chosen depending
directly on the number of annual batches or
lots produced, the safety risk associated with
the use of off-spec material, and the value of
each batch or lot.

For example, for liquids in tankers, a top
and bottom sample from each tank compartment
may be adequate.

Whereas, for a CRITICAL API precursor solid
delivered in 50-kg drums, multiple-level
(minimum, top, middle, and bottom) sampling
of each container may be needed.

For a lot of 50 consecutively filled and
time-stamped 25-kg bags of charcoal from a
single lot, a single sample from a random
bag on each layer on the pallet may be
sufficient.

Moreover, in general, the API should have a
hierarchical sampling plan that provides
multiple levels (typically, intensified,
normal, and reduced) of inspection intensity
and uses the outcomes of the results'
history from the testing of previous batches
of lots from a given source to change the
inspection level for a given raw material.
[Because of the risks and the time delays
associated with "resampling," in general,
it may be prudent to always sample at the
intensified level and only change the level
of assessment of the samples sampled. This
is especially true when your firm is being
CHARGED for the time you hold a bulk delivery
vehicle before allowing the lot in the tank
to be discharged into your receiving tanks or
silos, or rejecting the lot.]

Finally, the AUDITED level of documented
"quality built in" by the manufacturer of
each raw material should factor into the
inspection (sampling and assessment) plans
adopted by the manufacturer of the API.

As it is with the manufacturer of drug products,
it is up to each manufacturer to establish
(prove) that their inspection controls are
adequate to ensure that no problems are
missed.

Hopefully, this general answer has addressed
your question.

Respectfully,

Dr. King
http://www.dr-king.com

PS: As has been stated repeatedly, the
     sampling plan "1 + (integer {[square
     root of the number of containers] +
     0.5)" is not a valid statistical
     sampling plan and, unless the number
     of containers is 3 or less, does NOT
     A PRIORI guarantee that the number of
     containers sampled is representative
     of the lot or batch. [NOTE: To VALIDLY
     infer from the outcomes observed for
     those containers sampled and tested to
     the acceptability of the lot or batch
     of containers, a REPRESENTATIVE set of
     samples must be assessed. Moreover,
     as the % of the containers assessed
     decreases, one's level of confidence
     in the acceptability of the lot also
     decreases.]

     In general, the ISO standards for discrete
     units (the ISO 2859 series for attribute
     assessments and ISO 3951 for variable
     assessments) CANNOT validly be applied to
     raw material containers UNLESS the API
     manufacturer can establish that the observed
     variability within each container is
     SIGNIFICANTLY less than the variability
     between containers in a given lot or batch.


******************
Pharmaceutical Sci-Tech Discussion Group
To subscribe or unsubscribe go to:
http://www.pharmweb.net/pharmscitech.html
******************

[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index