EP Evaluator: Linear Regression vs. Multiple Instrument Module for Blood Gas Correlations
Hi everyone!
I'm a new POCT Coordinator trying to do my first blood gas correlation studies. In the past at my lab, EP Evaluator was used in method comparison mode to perform linear regression to compare each of our blood gas analyzers to the "primary" analyzer used by Respiratory Therapy. Data was not bounded with Total allowable error.
I'm trying to use EP evaluator in the multiple instrument comparison module instead, which compares all the analyzers together instead of only two at a time. Since I have to define total allowable error, this is creating more of an issue for me than the way it used to be done here.
I feel like the multiple instrument comparison module is the right way to do this but my lab manager (and the manager before that) is the one who had been doing the two-instrument-at-a-time linear regression method and I'm preparing for pushback on my reports being done differently.
Can anyone offer advice on which method is more valid? Is there a regulatory requirement somewhere stipulating that total allowable error must be used to bound data rather than only presenting linear regression analysis with no data bounds?
Thank you in advance from an overwhelmed new POCT Coordinator!
Ken
| Topic | Replies | Likes | Views | Participants | Last Reply |
|---|---|---|---|---|---|
| iSTAT in NICU -chem 8, CG4 | 5 | 0 | 252 | ||
| Rotem Sigma Validation Help | 1 | 0 | 102 | ||
| Hemochron Sig Elite use outside of manufacturer temperature range | 2 | 0 | 170 |
Ken,
I just completed this task for our I-STAT ABGs using the EP evaluator multiple instrument comparison module. The POCT All Common Checklist has standard COM 043000 which deals with this directly and states that there must be a written policy of how this task is accomplished and "acceptablility criteria are defined for comparability of non-waived instruments and methods used to test the same analyte, with records of action when criteria are not met." It also states that statistically defined acceptability limits should be used for quantitative assays. We utilize the CLIA acceptable performance criteria for each analyte. Where CLIA does not specify a limit (ie. iCa) we utilize the +/- 3 SD utilized by CAP to evaluate our PT survey results. In our recent CAP inspection the inspector actually looked to ensure that this procedure was written up and that the acceptability criteria were documented.
In EP evaluator, you will need to conduct a new experiment for each analyte and establish the boundaries, however once that is done the rest flows rather smoothly. Good luck.