Process Validation ISO13485 - Confusion on what processes need to be validated

[Tidge]

For a 13485-following company facing a variant of the "OMG that is a LOT of VALIDATION!" issue, there is a silver lining... the risk management processes required by 14971 can do the important job of discriminating between which production processes need to be validated (and which do not) and also (critically important IMO) at what level the processes requiring validation need to be validated. This question of level is important for test methods and whatever is intended to be segregated by application of the test method.

There are several sorts of "no, duh" manufacturing processes that will always need to be validated like those relating to sterilization, cleanliness, and other "invisible" outcomes (welding/sealing, homogeneous mixing, etc.). A Time/Pressure bag sealing process is one of these sorts of processes.

Additional comment: It is typical (in my experience) that most 'in-process' tests/checks during manufacturing assembly are testing for specific sorts of defects that could be introduced during the assembly... more so than "verifying" some requirement is being satisfied by the DMR. These types of tests are looking to find specific sorts of non-conformances rather than verifying conformity. Any test method validation typically accounts for this.

 

[d_addams]

For a 13485-following company facing a variant of the "OMG that is a LOT of VALIDATION!" issue, there is a silver lining... the risk management processes required by 14971 can do the important job of discriminating between which production processes need to be validated (and which do not) and also (critically important IMO) at what level the processes requiring validation need to be validated. This question of level is important for test methods and whatever is intended to be segregated by application of the test method.

There are several sorts of "no, duh" manufacturing processes that will always need to be validated like those relating to sterilization, cleanliness, and other "invisible" outcomes (welding/sealing, homogeneous mixing, etc.). A Time/Pressure bag sealing process is one of these sorts of processes.

Additional comment: It is typical (in my experience) that most 'in-process' tests/checks during manufacturing assembly are testing for specific sorts of defects that could be introduced during the assembly... more so than "verifying" some requirement is being satisfied by the DMR. These types of tests are looking to find specific sorts of non-conformances rather than verifying conformity. Any test method validation typically accounts for this.

One element that can help discern how to parse out these assessments is to stop asking 'does this process need to be validated' and only ask 'which critical process OUTPUTS need to be validated'.

From the original ask about a welding process and the critical OUTPUT of hermiticity as measured by leak testing. The process OUTPUT of hermiticity can be 100% verified via a non-destructive helium leak test. However, welds typically also have a critical OUTPUT of weld strength. One cannot 100% verify weld strength since it is a destructive test, therefore the OUTPUT of weld strength needs to be validated. In this case is the process 'validated'? The answer is yes and no and is why the question of 'does the process need to be validated' while appearing 'simple' leads to so much confusion. So one needs to be more specific about the OUTPUTS and not just discuss processes as monolithic.

CFR 820 as Bev quoted is very clear on this point so it isn't really open to interpretation.