Not all it's cracked up to be
"Microbiological tests take time," began our GMP Agent-In-Place. "Once you have a deviation, there are a bunch of already-made
batches involved. We had a product that had a bioburden action limit of 50 CFU/mL postcentrifugation. The batches started
running well, about that limit, and before we could investigate and find a root cause, 20 batches were affected. The deviation
was uniquely associated with one of two centrifuges, and that the bad centrifuge had a capillary fissure in the lid. After
consulting with the local health authority (not FDA) we agreed that a second, additional, and different pyrogen test would
be used to qualify these batches for sale. Ultimately the batches were sold, the centrifuge lid was repaired, and the current
trend is well below the bioburden action limit."
"We are a raw material-constrained manufacturing business," explaned our GMP Agent-In-Place. "And we were selling what we
were making with no extras. Normally that's a really good thing, but here we were with a new and improved version of our product
to launch that used the exact same raw material. We ramped up production of the new material on what we thought was the right
(overly aggressive) schedule based on FDA approval cycles so that we would have enough inventory soon after approval to launch
the new product. However, the inevitable delays set in, and we ended up with a product shortage—and numerous complaints."
Always ask more than one question
"Sometimes it's only one question," our GMP Agent-In-Place noted. "I was performing an audit of a contract laboratory. One
page of data showed a correction. I would have passed along, but the correction showed a reason code 'TE,' meaning transcription
error. In this case, there was no source for the data, so how could it be a transcription error? There was nothing to transcribe
it from and this entry was the result of a calculation (error 1).
"As an auditor, I asked about the documentation standards policy for laboratory records like this," continued our Agent. 'Did
you expect to see the entire equation written out?'
'No, just one example, per SOP,' they responded.
"There was no such example in this record (error 2). The only signature on the page was the correction notation," exclaimed
our Agent. "There was no other signature showing who was responsible for the work (error 3). It turned out, the contract laboratory
had no requirement that each page of laboratory work be signed, and there was no designated space on the form for such a signature
(error 4). Delving deeper into this record, I checked the calculations, and they seemed correct. It was a percent error of
observed versus standard calculation, not difficult. But then I noticed that the observed result that was written down from
the instrument was wrong (error 5). This led directly to the wrong initial calculation, beacuse the analyst was comparing
the result to the 10-8 standard rather than the 10-6 standard. If for the corrected result, the analyst had used the proper standard and the observed number as written, the
corrected result should have been 99.9%, but even the corrected result was wrong (error 6). Note that the auditor wouldn't
have noticed, except that he was looking at the correction."
Pharmaceutical Technology's monthly "Agent-in-Place" column distills true-life cautionary tales from the files of Control, a senior compliance officer.
If you have a story to share, please email it to Control at AgentinPlace@advanstar.com
. We won't use any names, but if we do use your experience in the column, you'll receive a Pharmaceutical Technology t-shirt.