Good day everyone and welcome back to another week of MedTech Compliance Chronicles. Today we will go over a topic that, for many, is shrouded in mystery: Process Validation. What is process validation? How do you do it? Do you have to do it for every process? How do you tell which processes need it and which do not? These are all very common questions to have and we will answer them all here!
In short, process validation is documented objective, scientific evidence that a process consistently produces outputs that meet their requirements. There are several factors in doing this. First you must decide which processes to validate. Next you must perform the validation, which will include making sure that all infrastructure used in the process is properly installed and maintained, the process can meet its requirements and that the process still meets its requirements under normal operating conditions. Finally, you will need to create a plan to monitor the status of validated processes and determine when they might need revalidation.
When and What to Validate
The first question on everyone’s mind when talking about process validation is when and what to validate. If you are strictly adhering to the regulations, you only validate processes which cannot be fully verified by subsequent measuring and monitoring. This sort of implies a requirement that you must fully verify the results of any process which can be fully verified. However, the wording in ISO 13485:2016 gives a little more wiggle room and, in practice, the FDA does as well. The wording in the standard is that a process must be validated when the output “cannot be or is not verified.” Those three little words “or is not” allow for the manufacturer in some instances to validate a process even though the output is capable of being verified.
That all sounds very nice but what does it mean? At this point we have established that all processes have inputs and outputs and that these inputs and outputs each have specific requirements which they must conform to. For example, if the input to a machining process is a metal and the output is a screw, some requirements for the inputs would be the material specifications and some requirements for the outputs may be the threading type, pitch and diameter of the screw, etc. The FDA expects that manufacturers of medical devices maintain objective evidence that every process in the manufacturing of a medical device produced outputs which met their requirements. For some product features/characteristics this is easily verified and when easily verified, verification is the expected route. These types of verifications would be your in-process inspections where you take measurements and verify that the items are within their allowable tolerances. Some features/characteristics of a device are not as easily verified or not verifiable at all without destructive testing. This is where process validation becomes a requirement. Validation is required in any process where the result is not capable of being verified. Destructive tests are one example, other very common examples in the medical device industry are sterilization processes and sealing processes for sterile barrier systems because verifying the outputs of these process would inherently involve introducing them to contamination or breaking the sterile barrier seal which would in turn preclude the product from use.
Per a strict interpretation of the FDA requirements and some very strict auditors, processes where the output is not capable of verification are the only processes that should be validated. However, in practice many organizations validate processes in order to reduce the burden of verification activities. This is allowable in certain instances but should be done with extreme caution. The first and foremost thought on anyone who is considering validating a process where verification is capable is, what are the risks of not fully verifying the output of this process. If the risk is medium to high, you probably will have a hard time justifying not verifying the outputs. If the risk is low you can begin to craft a justification to validate. Types of processes that are typically validated in this way are processes where the resulting outputs are technically capable of verification, but too numerous or too complex to feasibly do.
IQ, OQ, PQ
Now into the meat and potatoes of process validation. IQ, OQ and PQ stand for installation qualification, operational qualification and performance qualification. They comprise the actual activities which will be undertaken to perform process validation. The activities essentially come down to making sure everything used in the process is properly installed, then making sure that it all works together to produce an acceptable result and finally handing it off to the manufacturing team to verify that the process will still output an acceptable result in real manufacturing conditions.
Installation qualification is where you will verify that all process equipment, environmental and infrastructure requirements have been met and are in working order. The obvious first step, yet often the most overlooked step, in performing any activity is making sure that you both have everything needed to complete the activity and also that everything needed works as it is supposed to. The most common considerations will be the equipment and floorspace needed to perform the process. You should make sure that you assign the proper amount of space in your facility to reasonably allow the person to perform the process. Any equipment needed should be installed in accordance with the manufacturer’s instructions and evidence of such installation maintained. This will usually take the form of a checklist made from the installation instructions in the user manual which you will fill out as you perform the installation. Some heavier industrial equipment may require more thorough installations or may even require technicians from the manufacturer of the equipment to visit your facility and install it. If this is the case, you must still maintain the records of all installation activities. Some final considerations for IQ would be any environmental conditions necessary such as temperature or humidity control or cleanroom qualification(s).
Operational qualification is where you will verify that the process is capable of producing an output which conforms to its requirements and establish the process control limits. This is the stage of validation where you will challenge the process to the worst case conditions and establish the process control limits. Worst case conditions are simply what is hardest for the process to do or, in another way, what inputs are most challenging for the process to produce conforming outputs. For example, let's say you are validating a milling process utilizing some type of rolling mill. You may have established that it is hardest for the equipment to mill larger sized inputs. It is also typical of rolling mill processes that the speed of the mill has a significant impact on the efficiency of the milling. It is also good design of experiment methodology to minimize the number of trials performed (i.e., test as many aspects in each trial as you can without invalidating the results) Therefore, to establish process limits and challenge the process using a worst case condition for this process you might design an experiment where you only use your largest size input materials and adjust the speed up and down until your find the lower and upper limits of where the output is acceptable. From this data, you should also be able to find where the ‘sweet spot’ was, or where the process outputs varied the least from the specifications for the outputs. You will then do this same process for all of the controllable parameters that have an impact on product quality. At the end of operational qualification you should have established the upper and lower limit for all of these parameters under the worst case conditions. You should also have an idea of the optimum settings for these parameters. Another output of OQ should be your formal documented procedures for how to perform the process, or at least drafts of them for now.
Finally, there is performance qualification. The primary purpose here is to take that perfect, ideally developed and executed process and procedures and put it to the test in actual production settings. Many times OQ will be performed by the scientist or engineers who developed the product and sometimes a lot of OQ might rely on retrofitted process development data from R & D, so some outputs of OQ might not even have been performed on the equipment that will be used in production. Therefore, both the fundamental knowledge and understanding of process and equipment may be very different in actual day to day production. I will note here that while OQ may sometimes utilize process development data from unqualified equipment, it is the equipment which will be used in production that must be the subjects of IQ and PQ. Back to the point, PQ is where you will introduce all of the regular sources of variation that will exist during production. This must include all of the different shifts, a number of different operators representative of those who will perform the process, environmental conditions of the production facility, and production at the actual scale that will be regularly produced. PQ is also where you should put your procedures to the test and make sure they are clear and understandable by people who did not take part in the development of the process. The amount of production runs you will need to make must be based on statistically valid rationale and must be enough to provide objective evidence that this process, when run under these conditions with these settings, consistently and reliably produces an output that conforms to its specifications.
Monitoring and Control
Nothing in a QMS is ever truly finished and processes have their way of drifting over time. For this reason, it is required that all validated processes are continually monitored for ongoing effectiveness. Monitoring and control plans, as they are often called, are usually developed using the data from OQ and incorporating any lessons learned from PQ. In OQ you needed to develop upper and lower process limits. In monitoring and control you must develop methods to continually monitor the process parameters identified in OQ during production and keep them within the limits.
The plan should include either an alert for when a parameter is nearing one of its limits or if a limit has been exceeded, as well as actions to take to bring the process back within its limits. Some practical examples of this would be a machine with an alarm to alert the operator when its internal temperature has exceeded a certain point or designating one device out of every batch for destructive testing. Follow-up actions might include checking the machine’s coolant system or just waiting ‘X’ amount of time and, if a destructive test failed, perhaps draw a statistically reasonable sample from the lot for additional testing. You are not required to lay out in detail exactly what will be done in every situation but general actions should be identified that cover all of the reasonably foreseeable situations. Your plan should also identify when the process might require revalidation and to what extent revalidation needs to take place. For example, if you move a piece of equipment, you might need to perform IQ and PQ again but much of OQ should still apply regardless of the location.
Finally, you must keep very detailed records of the monitoring and control of the process. You should maintain all of the data from process monitoring and even document the equipment and operators used for each lot or batch. This information will be relied upon during investigations should an adverse event occur.
Conclusion
Process validation is a very complex topic that varies widely depending on what your specific process is. The goal of this post was to develop a fundamental understanding on what process validation generally is. You should now have an understanding that process validation is the provision of objective, scientific evidence that a process continually produces conforming outputs. You do this by ensuring that all infrastructural resources are adequately provided, identifying the proper settings and methods to use them and then introducing all of the natural sources of variation present in your production environment and ensuring that the process remains robust. You must then make sure you develop adequate plans to maintain the process in a state of control and develop actions to bring it back in control when it naturally drifts.
Comments