In case of analytical instruments, some companies use the term Out-Of-Calibration, others use Out-Of-Acceptance for failed recalibration or requalification.
Introduction - how to apply usual deviation elements
OOC is similar to all other deviations: what you need to go after are the Root cause analysis (RCA) and the Impact analysis.
Of course, technically you (or your service provider) may know what to do to correct the failure: do some special maintenance, finetune some instrument parameter, etc. After the correction, you are to decide which recalibration/requalification steps to repeat. This should be decided based on the type of the correction activities.
So this may be enough for the RCA for your OOC. But what about the impact analysis?
There are basically two types of recalibration/requalification:
As found: you go ahead and perform your recalibration without any maintenance before
As left: you perform your recalibration after maintenance
Currently, there are no written GMP rules which one to apply on which instruments, but there are some metrologically obvious choices how to differentiate.
Since these are to be handled in really different ways, let's discuss them one by one.
As found
So you don't do anyhing with your instrument or measuring device before recalibration, and you get an OOC.
This result describes your instrument's current 'behavior'. To have the worst-case approach, let's assume the instrument fell into this state 1 second after the last recalibration.
What we want to exclude, is the possibility of any Type 2 error in your release/stability tests: making the mistake of passing an OOS sample.
So you know the numerical deviation from the acceptance criteria, and you can get your result using your instrument. So you have every data to be able to recalculate your final results with the deviation determined as OOC.
But first let's always assess the criticality of your instrument.
Examples
If you use a balance to weigh sample for an LC-test using area percent calculation, you can quickly declare you have no impact on your results - you don't use the exact weighed amount to calculate your result(s).
In case of a burette used as a part of a KF-titrator, you simply evaluate the recalculated water content results - and if you determine an OOS during the recalculation, you should go ahead and test your retained sample to be sure. However, if you have a good OOT-limit structure, you should find no such case.
If it's about a temperature probe used to monitor a refrigerator, you can use Mean Kinetic Temperature calculation to determine if you have any real problem.
As left
So you got your OOC after maintenance. Okay, we need to be strong here: we have to declare we have no idea what was our instrument's status before the maintenance. All we can (and definitely should) take into consideration is any daily checks or any system suitability tests. If you have those, you can say the OOC has no impact at all on your QC-results, since your periodic checks prove your instrument kept the calibrated/qualified state until the first moment of your periodic maintenance.
Comments