Evaluating CAP TRC - Statisticians needed :-)

[Cliff]

For those of you that perform the CAP TRC, or even those good with statistics, can you offer suggestions on how to evaluate a score?

 

Even though there are more than 10 participants, CAP grades this survey as an educational challenge.

 

This is our first time using the Adam cell counter and I think our results are good; however, one of our SDI's is a little high.  CAP offers very little info on what to do with an SDI, even though it seems like a useful comparison to the group.

 

This is from the Participant Summary:

 

 

This is an explanation I found of what an SDI is:

 

The target SDI is 0.0, which indicates there is not any difference between the laboratory mean and the consensus group mean. A SDI ±1 indicates a possible problem with the test.

The SDI expresses bias as increments of the standard deviation. A SDI of -1.8 indicates a negative bias of 1.8 standard deviations from the consensus group mean. This is not favorable.

Bias increases or decreases the percentage of patients outside the defined reference limit. For example, a positive bias decreases the percentage of patients normally outside the lower limit and increases the percentage of patients normally outside the upper reference limit. This creates an increase in false positive test results. Negative bias has an opposite effect and decreases true positives and creates false negatives.

Use the following guidelines to interpret the SDI:

 

Our result for TRC-26 was 11.79 with an SDI of +2.4.  The CAP discussion for this analyte indicates "...100% of participants using fluorescence microscopy obtained the intended result...", to my implying we "passed".

 

We evaluate everything we "fail" with a short writeup and present it to the medical director.  Historically we have always had an SDI less than 2, I can't very well say that this meets that requirement.

 

How would you evaluate this if this were your result?