I am validating a new Hematology analyzer for our pre/post pheresis platelet counts for a back up. We also use it for QC on pheresis platelets. I had a few questions for anyone else who does similar testing.

Do you do linearity studies every six months? The manufacturer is saying this is what CLIA is now requiring. The kits are expensive to do if you have more than one instrument.

That brings me to the next question. We have an older instrument that would be back up to the new one but the new instrument runs higher than the old one on the platelet counts. Sometimes more than 30,000 difference. Supposedly they correlate but I can not see how that is accurate if they can be so different. Anybody have experience?

My last question is, most cell counting machines require EDTA samples for testing. But the pheresis is collected in ACD-A. The manufacturer recommended going to the literature. I was just wondering what instruments others are using and does anyone know any references about how ACD-A affects platelet counts.

Thanks!

It is not a CLIA requirement to perform linearity every 6 months. Calibration yes but not linearity. Most manufacturers require linearity upon start up and thats it.

Results can correlate but have a bias--which is probably what you are seeing between the analyzers. Every time we have changed analyzers we have seen a bias between machines. Most of the time it has to do with the threshold that the manufacturer has set to start counting platelets. The lower the threshold the more fragmented platelets get included in that platelet count.

We use the Sysmex XE2100D and have not seen any appreciable difference in platelet counts when collected in EDTA or ACD-A. I know that Sysmex has done an extensive study and has submitted to the FDA to get approval for the different anticoagulants for this analyzer.

donnawac is correct about the threshholds probably being the cause of biases. You can determine this from the mean platelet volume on correlation samples. There is probably no need to have the manufacturers adjust the threshhold to improve the correlation, a diffeence of 30,000 is not significant.

I disagree about linearities; they must be done periodically, but six months is too short an interval. I would suggest biannual. Be sure to put it in writing.