There has been, is and always will be a strong focus on data in the pharmaceutical industry. When there is a manufacturing problem, data analysis can often indicate where to look for the root cause but profound process knowledge is usually required to find it.
In this article, we’ll look at 3 scenarios where pharmaceutical processing problems required detailed understanding of equipment / processes to identify (and subsequently address) the root causes and some of the lessons learnt that apply to all manufacturing operations.
Scenario 1
With increasing demand for a capsule product, a company moved to a new site. The new site included new equipment and processing lines to increase manufacturing capacity. The older equipment was shifted from the old site to the new site and as is common-practice in the industry, the newly procured equipment for the second line was the same as or equivalent to existing equipment to streamline the implementation.
Included in the new equipment was a fluid bed dryer (FBD2) used for spray coating of granules which was specified and installed as equivalent to the existing fluid bed dryer (FBD1) used for spray coating of the same product and load size. Despite the equivalence, it was found during commissioning and qualification that the same processing parameters yielded different outcomes. Despite an investigation at the time, the variation was not explained and with pressure to get the new processing line up and running alternative processing parameters were qualified/validated for FBD2 that did produce the same outcome as FBD1.
With a lean manufacturing review to optimise production sometime later, a new Production Manager took great interest in the variability which created the requirement for two sets of procedures and batch records.
The Production Manager engaged the Technical Services Officer to investigate. After following the equipment and process on the shop floor without observing anything significant, the Technical Officer found that both equipment had calibrated gauges for temperatures, equivalent spray coating peristaltic pumps flow rates, calibrated air flow gauges and confirmed fan speeds. The equipment supplier’s engineers investigated and declared that there were no differences in specification between the two FBD machines.
On further investigation, the Technical Services Officer entered the roof space above the manufacturing rooms where he observed a slight difference between the two FBDs. A pressure differential diaphragm sensor used to measure the air flow of the air intake duct for FBD1 was installed vertically, while the sensor for the air intake of FBD2 was installed horizontally. FBD1 had been installed by local engineers during the move from the old facility to the new factory, while the sensor for FBD2 had been installed by the engineers from the equipment manufacturer. While these appeared equivalent from a design perspective, the diaphragm sensor should be installed horizontally as gravity effects the diaphragm, giving a biased output reading.
After changing the orientation of the diaphragm sensor for FBD1 to horizontal, the FBDs were equivalent with the same processing parameters used to produce the same outcomes.
Scenario 2
The same liquid product was produced on different shifts by different operators in different but identically sized tanks with identically sized impellers and to the same level of quality, but there was a significant difference in the processing times from batch to batch.
The time difference was specifically observed at a raw material dissolution step where there seemed to be two distinctly different dissolution times with some batches taking about 15 minutes and others more than an hour.
It initially appeared to be random as to which batch would take longer. Batch records were analysed for raw material suppliers, production lines, shifts and even room temperature/humidity but no correlations were noted. A closer analysis of the batch records revealed a correlation between operators adding the powder to the tanks and the time taken for the powder to mix.
The person responsible for the investigation then went into the production rooms during the manufacture of various batches and observed that the operators that correlated with long mixing times carried box cutters and used them to slice the raw material plastic bags down the side to empty the powder into the tank. The operators that correlated with short mixing times did not have box cutters and they would untie the bags and use a stainless-steel scoop to add the powder. Further, observations noted that the powder was highly hydrophobic and would not wet properly when added quickly in the tank (as was the case with “blocks” from cut bags), forming a clump with a “fish eye” appearance with powder floating on the surface. However, powder that was added more slowly (as was the case with small scoops) would form smaller clumps that would get caught in the vortex and drawn down into the impeller facilitating mixing and dissolution.
After updating the processing instructions to define adding the powder slowly using a scoop, all batches took roughly 15 minutes to dissolve.
Scenario 3
A freeze dried finished product with an expensive, light-sensitive raw material, had been made by a company for many years. Some batches showed a significant colour-change in the finished product that didn’t comply with the specification and those batches were subsequently rejected at enormous cost to the business.
Much effort had been invested in protecting the raw material from light throughout the supply chain, on-site storage and manufacturing. No root cause for the colour-change of failed batches had been found despite investigations with the supplier and reviews of the manufacturing process.
A Process Improvement Manager was assigned to investigate the problem and started looking at the batch data which included hundreds of batches over the years and many failed batches due to colour-change. Experienced in manufacturing with sensitive materials, she found a correlation between the failed batches and the time of the year. Most of the failed batches occurred during the summer months and none occurred during the winter months. Specifically, the failed batches were all processed during the day and on those days the maximum outside temperatures were over 35 degrees Celsius. A trial of small scale batches in the R&D laboratory found that while the powdered raw material was not particularly temperature sensitive, once the powder was in solution, it was highly temperature sensitive.
Observing the process from raw material picking in a temperature and humidity controlled warehouse through to bulk manufacture and filling under red light in a temperature controlled production area, she noticed a critical anomaly. While the process was contained to GMP environmentally controlled environments, the pipework for the transfer from bulk mix to the filling room passed out of one room and into another but was not fully visible from the production area. Further investigation revealed that part of the closed-system transfer pipework, for only a few metres, was routed into an uncontrolled roof-space of the facility on its way between rooms.
Despite the short exposure time, extreme temperatures in the roof-space on those days had adversely affected the product and caused the colour change leading the costly batch rejections.
A short-term process change was implemented, including cooling the solution before transfer and passing cold water through the pipe just prior to transfer, to ensure the solution arrived at the filling room without rising above the degradation temperature. A project to re-route the pipework in only temperature controlled environments was implemented during the next scheduled shut-down period. After these changes, no batches were affected by colour change and the company saved hundreds of thousands of dollars.
Take Away Messages
Thorough desk-based reviews of designs and data can be helpful but there is no substitute for intimate knowledge of the 3 key components to pharmaceutical manufacturing: materials, equipment and processes. This knowledge exists in almost every manufacturing company and the challenge for businesses is to extract it from the right personnel and record it in a way that makes it available to the right future personnel.
The science of pharmaceutical manufacturing doesn’t end at Research and Development and opportunities for continuous improvement, like our focus on data, will always exist. There is a lot to be learnt from within a facility so get down to the shop-floor and witness manufacturing operations, listen to the operators who are at the coal-face and full of knowledge and assume you don’t know everything.
With decades of industry experience, many SeerPharma clients have leveraged our Consulting services to help investigate and resolve problems in production, with contamination etc.
Contact us if you're looking for external support to address an issue in your facility or with implementing the right root cause analysis and CAPA quality processes.
