At this point, it’s probably a good idea to talk a little about the types of validation.
If you think about it, there are devices that have been in production WELL before the regulations were established.
The ‘classic’ example is tongue depressors.
These have been around forever.
Do you think they did any validation when they first started – when validations were not required?
As mentioned, ‘typical’ validations, primarily PQ, require a substantial number of products built to complete validation.
If you have expensive product, are you just going to dump all that was used during validation?
Again, probably not.
Ideally, validations are completed prior to distributable product being created.
But as mentioned, this is not always feasible or possible.
In the US, there’s the notion of “cGMP” – where the (little) ‘c’ stands for CURRENT.
This means, that if the regulations are updated and you have a device on the market, you can’t always necessarily say that your product is ok because it met the regulations when cleared; you have to continually meet the CURRENT regulations to stay in the market.
Thus, there are times when you cannot always complete validation prior to manufacturing distributable product (of course, you should complete validation prior to actually distributing any more product).
So what do you do?
There are ‘defined’ approaches.
Prospective is the most conservative.
This is where you validate prior to distribution.
None of the product developed during validation is eligible for distribution.
Obviously, this is costly but is, by far the preferred approach.
Concurrent validation is, essentially, doing production runs during the validation.
This is done when you have costly product that is cost-prohibitive to trash just to support production.
In these cases, you put the process outputs in quarantine until you have successfully demonstrated that the process is validated.
Once validated, presuming there are no issues, the product can be released for distribution. This can be a reasonable balance between cost and risk.
The highest risk is retrospective validation.
This is done AFTER product has already been distributed.
For new products, of course, this should NOT be the selected approach.
If you already have product in the field but somehow failed to validate the process either due to manufacturing preceding the regulations in the case of the tongue depressors cited or in cases where, for some reason, it was not thought necessary, this is the only recourse.
What if, for example, you found that during a retrospective validation, you were able to pass nonconforming product into distribution? It is entirely possible you may need to recall ALL distributed product.
Consider those costs!
So, if issues are found in a retrospective validation, it’s necessary to determine the impact on existing product and take appropriate actions.