Q & A: The Deleting Data Dilemma

Posted by Sten Westgard, MS

"We have a department supervisor that instructs the techs to delete (not omit) qc values outside of three s.d..Is there a specific CLIA rule against this? Where I come from, deleting qc values is wrong."

What do you think? The answer, after the jump...

There is a difference between the data that we record and document, and the data that we use to calculate our mean and standard deviation.

Regarding the recording of data, best practice is to retain all data, even out-of-control data, and even data that was screwed up by some factor or another -- just note in the log why the data should be considered unusual.

CLIA is pretty particular about retaining records on instruments. For example, even after you get rid of a method, typically CLIA asks you to retain records for two years. So deleting a QC point from your records, while not explicitly condemned in the regulations, is implicitly a no-no.

Regarding the calculations based on the data, deleting is a different issue. If data points outside of 3s are considered out-of-control (and usually this is the case), those points should not be included in any calculation of mean, SD, etc. Outliers are not representative of the standard operating performance of the method, so we don't include them in the calculations.

So... no and yes, can I give a more confusing answer than that?