Case Study ; Pharmaceutical manufacturing for cause audit

I have been auditing for over 20 years, at first I enjoyed travelling to new places but soon I realised audits are hard work. Yes you get to travel but you don't really see much as you spend your time preparing for and conducting the audit then writing up the report.

When its a for cause audit its slightly different. Its hard to prepare for something where you don't know what the problem might be.

I had an issue in this case with inexplicable variation in the related substances test results for a product we had made by a contract manufacturing organisation (CMO), as its the case for many of us now. They had completed a full manufacturing review and OOS investigation but could not identify any more potential root causes. All the ones they had identified during their investigations had been tested and ruled out.

When I plotted the data I could see we didn't have a normal population, the distribution wasn't skewed though there seemed to be regular outliers. The frequency was too high for it to be analytical variation as had been suggested. The qualification of the method was robust and no one had any more ideas so it was down to me to audit the site over in Sweden.

Unfortunately it wasn't the best of weather going over, normally I planned my schedule for warmer weather but this time we had a problem to resolve so no choice. The helicopter ride across from Schiphol was unusual, we had thick fog and they navigated by beacon, I was certainly pleased to land.

At the site I started with manufacturing, it was a closed system with very little human intervention, automatic controls with quite tight tolerances and no deviations related to the problem under investigation only supply issues and a low yield. Nothing to be concerned about. All the in process checks were satisfactory, interesting though they were performed on a UV Vis spectrophotometer whereas the finished product testing was on HPLC. Something to consider. Time is more of an issue in process so this isn't unusual but can cause inconsistency in results. Made me wonder if we were comparing like with like.

A quick review and it was clear the in process data was normal distribution and the finished product testing, including stability, were variable. Both of which used the HPLC. The manufacturing process validation was robust as was the test method validation in process and finished product. The raw materials changed several times but plotting the batch numbers in a linear sequence to align with each event failed to show a trend.

What could it be I wondered, I studied Biology at university a fact the head of the analytical laboratory was well aware of. He seemed amused that I was auditing his analytical laboratory on a for cause audit when he, with his PhD in analytical chemistry, couldn't determine the nature of the problem.

I entered the laboratory and it hit me, the heat of the lab hit me in the face straight away, I had already spent 2 days investigating the process, reviewing the documentation. Looking at trends and the validation protocols and reports not to mention the meetings prior to the visit.

Could it really be that simple, surely the analysts knew the samples were temperature sensitive and the test itself had to be carried out at 20oC. I asked the simplest question. What temperature is it in here? No one knew, all the equipment was on an automated data capture system. You couldn't just check the temperature of the room, now the water batch yes to 2 decimal places, why I wondered? A simple room thermometer that's all I wanted they obtained one and what a surprise 27oC. My so obvious question is the equipment jacketed to maintain temperature? No it's held at ambient. Is it always this hot, not always in the morning its much cooler so we turn everything on to get it warm, it is Sweden after all its so cold outside. I wasn't amused, we were paying for the reject batches and it could be as simple as a lack of temperature management in the laboratory.

I've been accused of being too detailed, but not this time. The temperature issue did take time to fix an interim solution was adopted to keep the samples as the correct temperature during preparation and testing. I'm glad to say the problem simply went away, strangely the next time I came to visit I was welcomed with a request for advice on how to fix a new problem, but that's another story.

This is a real case study, it did actually happen I haven't changed anything for emphasis, sometimes it can be the simplest issue which everyone misses that solves a crisis.