From Hillary Thurman: Here is some additional information, in response to Bob Gosselin's 11-6-21 response:
The RIs currently in use are the manufacturer’s suggested ranges. Each site collected screened normal samples and EP Evaluator was utilized to ensure the patient population did not fall outside this range. Each site (there are seven) currently has one geomean for each site, ranging from 10.4–11.0 seconds, determined using the samples collected during the verification of the new instruments. The proposal is to combine data from each site to create the system-wide geomean. Each site will contribute 20 samples towards the combined mean. We will complete this each time we change reagent lots.
From a draft of our policy on verifying new reagent lots:
“In general, we will retain the current geometric mean. However, the geometric mean recovered on the new lot must match the actual geometric mean in use by ±0.3 seconds in order for the new calculated geometric mean to be valid. If this does not match within ±0.3 seconds, the geometric mean should be verified with 20 new fresh screened normal samples or by using samples with known values and comparing them with peer results or by using INR Validate.”
I am unclear how peer results or INR Validate verify the geometric mean.
As for instrument comparability, the current practice is to evaluate the instruments using the results from CAP CGL surveys (run on one analyzer, of course) and then evaluating the results based on CLIA allowable error. Some sites also use the Cross-Check surveys to compare their two instruments, but this is not standardized, nor is there a defined practice or policy for using pooled comparison samples on each individual instrument.
Thank you for your time,
Hillary Thurman, MS, MLS(CM)
Medical Technologist Coordinator
St. Michael Medical Center