A-378 Acceptance and Harmonization of i-STAT Cartridges. A Practical Example with G3+ cartridges
Valerio M. Genta, Sara Shumate, Dina Radi, Carl Aston · Clinical Chemistry · 2025
Abstract Background i-STAT cartridges offer a rich menu of overlapping assays performed at the point of care which optimizes a rapid, flexible and interchangeable use for patient care. Acceptance and harmonization studies insure interchangeability and reliability of cartridges. We are presenting a practical example of the use of multivariate statistical techniques for acceptance and harmonization studies performed for i-STAT G3+ cartridges in sixteen facilities of Sentara Healthcare. Methods i-STAT™ cartridges: G3+®, CG4+® and CG8+® (Abbott Laboratories). Quality Control material Cue See® VeriSTAT QC Level 1 and Level 3. AMR material: Cue See® VeriSTAT LVM (Eurotrol), 5 Levels. Patient specimens: 212 (Arterial:82, Cord arterial:14, Cord venous:10, Venous:106). The two levels of QC material were assayed five times per facility. The five levels of AMR material were assayed in triplicate for each location of each facility. The data were collected and transferred electronically to Minitab® (Version 22, Minitab Inc.) and analyzed with multivariable statistical techniques for equality of means, variances and regression lines between facilities. The patient specimens were assayed in parallel and within fifteen minutes with CG3+ and, as references, CG4+ and CG8+ cartridges. Criterion for acceptance of absolute differences obtained between values obtained with CG3+ and the reference cartridges: target value +/- 0.04 pH Units. Results The multiple comparisons between facilities with ANOVA showed probabilistically significant mean differences between facilities (Level 1 and Level 3:P <0.001) ; however, these differences were due to very small MSE (Level 1=0.000039, level3=0.000013) which was not significant for either QC or clinical practices. The parallel box plots for each level of control clearly showed that the differences were within half of the total allowable error (0.04 pH Units).The weighted polynomial regression model detected probabilistically significant differences between AMR regression lines by facility, however this was due to very small MSE (=0.000009 pH Units) which was not significant for either QC or clinical practices. The normal probability of the standardized deleted residuals for the regression model showed a quasi-normal distribution, and the plot of the standardized deleted residuals by the fitted value showed a very similar, quasi-linear locally weighted scatterplot smoother (lowess) for each facility. The absolute differences between the assayed and the reference values was within target value +/- 0.04 pH units for all the facilities. The orthogonal regression model showed that the patient values obtained with G3+ cartridges had a quasi-linear relationship with the values obtained with the reference cartridges. Additionally, the absolute differences between the assayed and the reference values was within target value +/- 0.04 pH units for all the facilities. Conclusion The performance of G3+ cartridges was acceptable and interchangeable with that of CG4+ and CG8+ cartridges. This practical example showed that the use of multivariate statistical techniques and their graphic representations allowed immediate, parsimonious evaluation of studies for the acceptance and harmonization of i-STAT G3+ cartridges performed in several facilities of a Health system. For these studies the electronic transfer of data and the analysis with statistical software such as Minitab was critical.