Questions

Post-calibration verification, how do we do it? A thorny question with a guest expert answer.

Question and Answer: What to do for Post-Calibration Verification

June 2025

"My question is regarding post-calibration varification. we usually do 2 level control and one retained sample of patient to varify the calibration. but my question is which retained sample we should use? because if we are calibrating the analytes for the out of QC reason,  the same day patient samples might be on either higher or lower side and after calibartion, if we run same sample, the different would be too much to accept as per CLIA. So shall we choose the retained sample> from the time before the QC processes were out?"

For this Q and A, which focuses on the topic of recalibration, we are happy to be joined by Dr. Nils Person. In his long career, he encountered a lot of laboratory recalibration practices and provided practical advice to all of them.

Dr. Person:
"The question seems to focus on how best to use retained patient samples as part of calibration verification following a calibration performed as response to out of range QC. There are a couple of issues here that need to be addressed and some pragmatic thoughts based on experience.

"First, and foremost, recalibrating a method in response to QC results out of range is almost always a really bad idea. It is all too common an approach to troubleshooting QC and is generally useless, a waste of time and materials and chasing after a non-existent problem. On modern systems calibration drift between recommended calibration intervals is exceedingly rare.

"Most often, if a method is recalibrated in response to out-of-range QC, the QC results look very good and the assumption is the problem is solved. Almost never is that the case. The usual scenario is things appear great after recalibration - first because calibration was not the issue in the first place and, second, because the recalibration just put a band aid over whatever underlying problem might be there. Recalibration temporarily adjusted the method to compensate for the problem without actually addressing or fixing the problem. So the problem, if there is one, is still there and will return creating an endless cycle of QC out followed by recalibration. The only time recalibrating in response to out-of-range QC makes sense is if there is hard evidence that calibration is the root cause. A good example would be if the method is recalibrated because the recommended calibration interval is reached and QC immediately after calibration is out. That would strongly suggest calibration is the issue. There are other instances as well, but recalibrating a method as a initial step after QC failure is rarely a good idea.

"Next, the question is asked about the need to test previously analyzed patient samples following recalibration and how much change in those patient results might be expected. If you have established your QC performance and "rules" and have not changed the reagent or calibrator lot, there is no real reason to retest patient samples in this situation. Nothing has changed that would impact the ability of QC samples to detect significant change in method performance so QC results should be adequate to confirm calibration was acceptable.

"If you are calibrating a new reagent lot, then the CLSI EP26 approach works nicely. For each method, decide in advance what concentration(s) are clinically critical and determine, in advance, the medically significant difference at that concentration (following the guidance). Then, when a new lot is started, you collect a small number of patient samples at the predetermined concentration(s) and test with both lots. The difference is compared to the predetermined acceptance limit. The number of samples needed is statistically determined based on method performance and the predetermined significant clinical difference. Detailed guidance is given in EP26 which has recently been updated.

"Finally, if you are retesting patient samples to check for change in method performance as part of troubleshooting and can only practically spot check at a few concentrations, the best approach is to select patient samples with values near medical decision point(s). That is the concentration that matters. Trying to test high or low samples to "check the range" is not usually a practical exercise and can be limited to special situations like initial method validation or periodic range verification studies. First because finding samples at extreme concentrations can be difficult, and then because, if you are quick checking method performance following troubleshooting.... who cares if there is a little drift and the high or low end? It's not going to change any medical decisions. What matters is performance at the decision point(s)."

Special thanks to Dr. Person for his generous time and useful guidance.

Nils B. Person, Ph.D., FAAC is a board-certified clinical chemist with over 40 years’ experience in laboratory medicine. He spent 15 years directing hospital laboratories prior to joining Siemens and spent 22 years supporting Siemens technical staff and customers primarily with regards to method validation and quality control. Dr. Person has also been part of a number of Clinical and Laboratory Standards Institute (CLSI) Standards development teams and was involved in the development of EP23 Laboratory Quality Control Based on Risk Management, EP26 User Evaluation of Between Reagent Lot Variation, EP21 Total Analytical Error and C24 4ed Statistical Quality Control for Quantitative Measurement Procedures.