Calibration verification replicates

Sometimes we over-do out of habit or for security. I think you are doing enough. E.g Abbot does recommend running cal ver once for each level. Trouble shoot (include repeating) only if value obtained is not in range. Unless GEM states otherwise, I'd say once is OK -or 2 max like you are doing now. Basically there is no need to do 3 replicates of each level as I have seen some labs do for  calibration/verification of non-waived testing

We do 3 replicates for each level, which is the absolute minimum necessary to calculate accuracy and precision of the system. 

Looking at CAP's POC. 08300, it looks like we don't actually have to perform calibration verification ("Single use devices, and other test devices that do not allow user calibration, do not require calibration verification.").  So I guess we are really doing this as AMR verification.  In that case, does it matter if we are following the package insert instructions since we aren't actually using them to assess linearity?

From what I can tell, the i-Stat does not have specific guidance on the number of replicates, but the GEM PVP solution makes mention of running all 4 vials, 1 level at a time.  We also have other hospitals in our system that run RNA Medical CVC 223 on their Avox analyzers and its package insert states that you should run 3 replicates for each level.