A Six Sigma Series from Advance

Posted by Sten Westgard, MS

Last month, Advance for the Laboratory published a three part series on Six Sigma in the Clinical Laboratory, written by David Plaut, Nathalie Lepage, and Kim Przekop:

Part 1

Part 2

Part 3

While it's great that Advance has invested in in-depth coverage of the Six Sigma topic, unfortunately one of their examples in part 2 demonstrates a misunderstanding of the application.

See the mistake, after the jump...

In the example, three different analytes are assessed, and a lot of confusion might result. Either there's a mistake in the table or there's a mistake in their calculations.

With the first example, hemoglobin, the mean of a method is 12.4 g/dL while the group mean is 12.3 g/dL, and the lab SD is 0.7 g/dL. That means the CV is (0.7/12.4) = 5.6% and the bias is (12.4-12.3/12.3) = 0.8%. Given that the CLIA requirement for TEa is 7%, this is not a Six Sigma method. Instead, the Sigma-metric is (7 - 0.8)/5.6 = only 1.1.

Now, if we believe the SD is being expressed in table as a percentage and not units (i.e. that "g/dL" in the second row left column is a mistake), the Sigma metric is now (7 - 0.8) / 0.7 = 8.8 Sigma. Unfortunately, this is a number not listed in the table. So we can't be sure this is what they did.

The example attempts to derive through a laborious method an estimate of the SD necessary to achieve 6 Sigma. The easiest way to do that is simply divide the TEa by 6:  7/6 = 1.16% or about 1.2%. In units at a level of 12.4 g/dL that works out to about 0.2 g/dL. The table lists 1.2 g/dL, which is six times the ideal SD.

Whenever you mix units and percentages, there's always a danger you make miscalculations. That's why we strongly encourage labs to convert everything into percentages, and then make the Sigma-metric calculation only in those percentages.