Saeid Ebrahim, Quality Systems & Statistical Techniques Manager, BioTeknica, is often asked about validating processes. Below he shares some important considerations.
Q: I can’t fully verify the output of each process and need to provide objective evidence that the process has been validated with a high degree of assurance. How do I validate my process when I’m not even sure what “high degree of assurance” means?
A: The 21 CFR820.75 (a) (Process Validation) states:
Where the results of a process cannot be fully verified by subsequent inspection and test, the process shall be validated with a high degree of assurance and approved according to established procedures. The validation activities and results, including the date and signature of the individual(s) approving the validation and where appropriate the major equipment validated, shall be documented.
The challenge that many life science manufacturers now face is to establish what is meant by a high degree of assurance. Since 100% verification is not possible in many cases, typically, statistical technique is a powerful tool to use in the determination of the sample plan required to provide a high degree of assurance. We need to define and select a Sampling Plan that is representative of the entire population of each lot being produced.
Before we get there though, there are a couple of steps that need to be taken.
First, we typically help our clients to establish the quality characteristic in question to be tested as an output of a process with a clear inspection/test method that has been validated. The characteristic could be either a variable or an attribute (pass/fail).
Second, a risk-based approach is taken to establish the severity of the failure associated with the characteristic in question. This is typically done in a pFMEA (Process Failure Mode Effect Analysis) document.
Finally, we establish the sampling plan required to validate the process. Knowing the severity will help us to define the Reject Quality Level (RQL), which in turn will help to establish the Sampling Plan. Passing the acceptance criteria for such a sampling plan provides objective evidence that the process has been validated with a high degree of confidence.
Over the years, on numerous occasions, BioTeknica has helped clients with process validation (Performance Qualification–PQ) for products such as guidewires. On several occasions, the clients were about to outsource the product, and our expertise was needed to validate the products produced by the external manufacturers. For example, let’s say a tip diameter couldn’t exceed 0.013 inches.
An initial step would be to help the client identify the characteristic and determine the severity level and Reject Quality Level (RQL). Per the client’s established procedure for statistical technique, RQL for a variable and nominal process output with a severity of 1 is defined as 20% with 95% confidence level. Based on the example’s established procedure, this sample size provides the following protection: AQL = 1.9%, RQL.05 = 20% with 95% confidence. (One can say with 95% confidence that the process is at least better than RQL of 20%, i.e., less than 20% defects.) The sampling plan associated with this RQL as per this example is n=15, Ppk≥0.49. (Ppk is the Process Performance index that tells how well a process is producing around the target specification.) Once the protocol is executed, the data is analyzed using Minitab or JMP or other available software to calculate the Ppk.
In this case, let’s say we would agree to produce three lots to capture all sources of variations and demonstrate reproducibility. If all three lots passed such acceptance criteria, one could draw the conclusion that the process has been validated with a high degree of confidence.
We have helped many clients successfully complete PQs in their projects. If you have any comments or questions for Saeid or if you’d like to submit a different question for any of our SMEs, please click on Ask a BioTeknica SME. Our BioTeknica Subject Matter Experts (SMEs) will gladly share their insights.