"Our problem started with a 3000-L vessel equipped with a mixer that was situated above the vessel with a shaft penetrating the vessel lid and the mixing propeller at the bottom of the shaft," explained our GMP Agent-In-Place. "The seal needed to be replaced, so a new one was installed. Unfortunately, the new seal design was inadequate and allowed some lubricant to leak out. The lubricant was assumed to have fallen into the product, so we tested the product for contamination. At this stage, the product is relatively crude, so there are several purification steps downstream.
"We quarantined the 20 batches that were made after the seal replacement and started a deviation investigation. Design standards for bearings and seals were reviewed and instituted.... The seal-lubricant manufacturer was contacted to obtain the complete formulation of the lubricant ingredients, and the toxicity of each ingredient was evaluated. Assuming that a few milliliters dropped into one batch of 3000 L of product, we would have had a parts-per-million contamination. Thus, the downstream purification steps would mean parts per billion at most, which was determined to be too small for any test. As a result, only a theoretical safety assessment could be made based on a safety assessment of the ingredients in the lubricant. Ultimately the batches were released."Too (TWO?) cool
"We installed a new giant cooler in our distribution center, validated it, and filled it with product," said our GMP Agent-In-Place. "It had two cooling units, each capable of handling a full load. The design specification was for each cooling unit to carry the load half the time, alternately cycling on and off automatically. After using the new cooler for about a month, the engineer noted the units weren't sharing nicely. One unit was doing most of the work and he ordered a fix.
"The cooler began having low temperature excursions, and a deviation was initiated," continued our agent. "As part of the investigation, the validation group wanted to revalidate the cooler. But nothing had been said about fixing the problem or evaluating the temperature-measurement system calibration to see if it might be the source of the problem. Our very excitable quality representative was overheard saying, 'Fix it first, then revalidate. Don't start a validation study with the intent of changing the system throughout it. Fix it first!'"
"We were installing magnetic stirred mixing vessels in many of our newer facilities," explained our GMP Agent-In-Place. "For those who haven't heard of this type of vessel, think of your chemistry laboratory glass beaker with a small, white magnetic stirring rod in it and a magnetic stirrer underneath. Now scale this up to 3000 L. The stirring rod needs a bearing to sit on but works much the same way. In this case, the stirrer slowly stressed a nearby tank weld and caused leakage. We took it out of service and repaired it. Part of the deviation evaluation included looking at all the magnetic stirred vessels across the company, including at all our international sites."
"Transferring a product is always fun," our GMP Agent-In-Place chuckled. "We were moving a time-release bead product from a company in England to our US factory. The plan was to mix the beads so that they would target the time-release profile. The English company kept an inventory of 9 or 10 coating pans and mixed them into three pan-sized lots. We had ambitious plans, and wanted 40 pans in inventory with the goal of mixing nine at a time.
"This all sounds good," began our agent, "until I saw that the computer program they provided to calculate the best mix of blends from the inventory pans. The program calculated all possible combinations of mixes, choosing the set that would give the lot blends closest to the ideal target. Our manufacturing department's plan was to use an identical program and change the nine-pan inventory to 40, and the three-pan blend to nine. I laughed when I heard this and asked, 'What millennium would you like the response?' The number of combinations to be calculated was so large that it could take thousands of centuries of computer time to get an answer based on the original program and the computers in use at that time! The laugh was on me in the end. They asked me to come up with something workable. So we devised a semi-manual method of choosing pan mixes, with computer calculations to verify final blend targets."
Pharmaceutical Technology's monthly "Agent-in-Place" column distills true-life cautionary tales from the secret files of Control, a senior compliance officer. If you have a story of clueless operators, oblivious management, inopportune lapses of judgment, or Murphy's Law in action, please send it to Control at AgentinPlace@advanstar.com