Q: "You mentioned that there can be variation in performance even with different reagent lots. Are you recommending that we change QC rules based on the performance of each lot on each instrument?"
This is another question from our Cut Your QC webinar, now available as an on-demand recording. https://us02web.zoom.us/webinar/register/WN_F5IRlqySRR6eRcrm9Zz9mQ
I can see the worry already. For labs that are already beleaguered by their current QC practices, the contemplation of changing QC for every test on every instrument with every reagent lot sounds horrific. Constant chaos.
In practice, what we see is not never-ending churn of QC rules. Particularly on chemistry instruments, performance is stable not only within a reagent lot, but across reagent lots. Thus, while Sigma metrics might slightly fluctuate, they don't gyrate. I have seen instruments with tests that have the same general Sigma metric for years at a time.
That said, there are also methods and instruments that do have significant changes in performance from lot to lot, or even within lot. IA methods often have different performance with different reagent lots. Hematology methods often see a degradation in the performance of the control during its significantly short lifespan.
If we see that a method significantly changes the Sigma metric from lot to lot, should we ignore that and respond with a flat set of QC rules applied to the method, regardless of the performance? For the sake of a simpler QC approach, we would be sacrificing the ability to detect important errors (if we standardized our limits wider than needed), or wasting our time detecting changes that were not clinically significant (if we standardized our limits tighter than needed). In both cases, we are accepting an inefficiency in practice in order to sustain a simpler set up in design. If you want to make that bargain, at least make sure you know what that cost is. A simpler QC approach might be easy for your brain, but costly for your patients.
What most labs encounter with Sigma metrics is not one set of performance on all tests, but a mixture. The lab will see that some methods are Six Sigma, some are Five Sigma, Four, Three, and we hope only a few that are less than Three Sigma. Likewise, some methods will be very stable in performance, while a few might change significantly from lot to lot. The most practical approach is to take the savings and efficiencies where you can find them, and find compromises for the methods that don't meet your desired level of performance. If a few methods turn out to fluctuate lot to lot, and in a bad way (dropping to 3 or less than 3), it may be prudent to decide that a full set of Westgard Rules (but no 1:2s) is a useful baseline for that method. If the method ever jumps up above 3 Sigma, and you stick with the baseline set of Westgard Rules, you will find there is a cost of additional false rejection. Your convenience will be generating more false alarms than necessary, so you will spend more time in troubleshooting (or, if you haven't modernized your approach, repeating and recalibrating). That's the cost of convenience.
Similarly, a "standardized" QC approach across multiple instruments can generate extra false rejections on the one hand, or a weakened ability to detect important errors on the other hand.