Hello everyone!

Just looking for some advice - we are replacing older Echos with 3 new Echos.  The machine itself and testing methodology is the same - the only difference is the instrument is now white and purple.  The last validation I did when we received new Echos ... I think I went overboard with the number of samples I ran for each test.

Any advice on the number of samples for validation of instruments that are the same in all aspects of testing - just newer models? 

 

Thank you!!

Sara 

Lies, damned lies, and statistics.

You've asked the question that has plagued my small brain for almost three decades.  How many is enough - the question cannot be answered to everyone's satisfaction.  Some will tell you to do more, others will tell you your plan is overkill.

Everything is a risk; you need to use sound judgment and work with your CLIA director to make sure they are satisfied with your plan.

When we moved to a new computer system, we had millions of records.  I chose to use statistics.  We fully validated 384 randomly selected records - lots of sites to determine this number.  This took a tremendous effort to complete.  As the years went on and we "simply" upgraded to a newer version, we had a solid history of working with, and validating, the software so we eased up a bit on our plan.

These are brand new instruments, albeit with the same methodology.  You need to perform a validation that will ensure each new device performs as expected.

I always started with the vendor; they almost always offer a suggested plan.

I have no experience in determining what is appropriate so take the following opinion with salt. It seems reasonable to save and freeze samples tested on your current Echos that include negative and a variety of positives . Then thaw and test on the new instruments and compare results.  I mean this process is used to compare prenatal titers. I would suggest 40 samples as 30 is regarded as the minimum number that permits statistical analysis (if memory serves). 

We did this 3 years ago. We did a minimal validation. The reagents were all the same so you're only really validating the mechanical components of the new analyzer. Therefore, we did just enough to show that the new machines got the same results as the old in regards to blood types, antibody screens (no identifications), DAT, and crossmatches. I think we did 10 specimens of each representing each blood type.

Our medical director always insists on 100 samples. This is WAY overkill for things that are not done frequently, like eluates, but for a TYSC or ABID on the Echos it's a statistical sample size for our volumes. We did 100 TYSCs on each Lumena, and 100 antibody panels on each Lumena (we have 2 Lumenas, and upgraded from 2 Echos). We were done in about 3 days for each instrument, running 4 at a time. Not too bad.