My Mistake: Errata on Planning SQC strategies and adapting QC frequency for patient risk
A recent Westgard, Bayat, and Westgard paper published in Clin Chim Acta had a one digit error. But that error cascaded throughout the paper. Here are the corrections.
My Mistake: Errata Clin Chim Acta 2021 “Planning SQC strategies and adapting QC frequency for patient risk”
James O. Westgard, PhD
As a lifelong advocate of quality, it is embarrassing to discover a mistake. In a recently published paper, I made the mistake of entering 0.9964 in a table when the correct number should have been 0.9664. That seems like a small mistake, but because that number is the slope in a regression equation, it leads to several other errors in the paper. Below are the corrections that will be published in the journal.
[Editor's note: those of you who frequent westgard.com may note from time to time that there are also typographical errors on lessons and articles. Those errors, however, are not the fault of "the" Westgard, but instead they are mine, i.e. "a" Westgard (Sten). Whenever you come across one, let me know. I endeavour to continuous correction and improvement, and I sincerely hope you do not blame the father for the sins of the son.]
Westgard JO, Bayat H, Westgard SA. Planning SQC strategies and adapting QC frequency for patient risk. Clin Chim Acta 2021;523. https://doi.org/10.1016/j.caa.2021.08.028
Table 2. Row for 1:3s N=2, slope should be 0.9664 rather than 0.9964.
Table 3. Calculations for run sizes for 1:3s N=2 should be 84, 21, and 107, replacing the values 116, 28, 141. Run sizes from 85 to 100 patient samples can be accommodated by use of a 1:3s rule with a minimum of 2 control measurements per run.
Table 4. Calculations for run sizes for 1:3s N=2 should be 1000, 15, 21, 3, 99, 55, 64, and 65 replacing values of 1000, 20, 28, 4, 137, 75, 99, and 85. Run sizes of 50 could be justified for uniform use of 1:3s N=2 and run sizes from 100 to 125 patient samples with use of an N=2 multi-rule procedure.
It is, of course, embarrassing when something like this happens. I might excuse this by acknowledging that my nearly 80-year old eyes are not as sharp as they used to be. However, it is more likely a mental transposition of similar looking numbers, 9664 vs 9964, and it is harder to admit that my 80-year old mind is not as sharp as it used to be. Nonetheless, it shows that regardless of our good intentions, we humans make mistakes that reveal our limitations and that we need others to understand both our good intentions and our bad limitations.
For those of you who have made use of our QC Frequency calculator (http://tools.westgard.com/frequency_calculator.shtml), we have already made the correction in the calculations for the 1:3s N=2 QC procedure (as of 9/13/2021). The run sizes calculated for the other 9 QC procedures remain correct.
We encourage you to try the QC Frequency calculator. We are finding that “run size” is a very sensitive parameter for optimizing SQC strategies to fit the workloads and desired reporting intervals in service laboratories. We will continue to provide guidance and examples in the near future.