Because therapy may be either instituted or withheld on the basis of information obtained from this type of monitor, it is essential to understand the performance characteristics of available systems. The overall performance of Oximetrix 3 and SAT-2 compared with CO-oximetry was nearly identical (Table 4). Differences in performance of oximetry PA catheter systems, therefore, do not appear to be caused by inherent limitations of two-wavelength systems as previously thought. Regression analysis and correlation coefficients for Oximetrix 3 and SAT-2 are similar. Bias and precision are also quite similar in magnitude; however, there is a directional difference in bias between the Oximetrix 3 group (negative bias) and the SAT-2 group (positive bias). While a similar correlation coefficient and bias were obtained in the HEM0PR02 group, precision appears worse. Both ambiguity and RMSE have been proposed as global indicators of performance relative to a criterion standard. The Oximetrix 3 group and SAT-2 group have virtually identical ambiguity (5.05 vs 5.29) and RMSE values (3.64 vs 3.92). The performance achieved by the HEM0PR02 group is apparently worse when compared by these two tests (ambiguity = 7.52, RMSE = 5.71).
In conclusion, if knowledge of the correct value for Sv02 is necessary for clinical decision-making, the differences in ambiguity, RMSE, and 95 percent confidence limits based on total variability are important. If, however, these devices are used primarily as trending monitors, where identification of directional changes in saturation is of paramount importance, then ESD and 95 percent confidence limits based on intrasubject variability are most relevant and the differences in performance are less clear.