A Comparison of Three Pulmonary Artery Oximetry Catheters in Intensive Care Unit Patients (20)

Thus, a decline in the oximeter system reading from 70 percent to 64 percent likely represents an actual change in the patients condition. It does not predict the actual value of the patients Sv02.
The assessment of agreement between clinical measurements may also be influenced by the passage of time. Previous studies have demonstrated that the performance of in vivo oximetry systems may deteriorate with the passage of time. During our 24-h study period, there was no statistically significant change in the bias observed in either Oximetrix 3 or SAT-2. The HEM0PR02 group did exhibit a significant (p<0.05) change in bias over the course of the study, paradoxically showing improved performance as the study progressed (Fig 2).
The differences in reproducibility (precision) among the three groups are less clear than the differences in offset (bias). The overall precision for the Oximetrix 3 and SAT-2 groups was remarkably similar (±3.07 percent and ±3.49 percent, respectively), while the HEM0PR02 group showed more variation (±5.24 percent); however, none of the three groups showed a statistically significant change in precision over the course of the study.
If one considers the variability between in vivo and in vitro determinations in individual patients, the differences in performance are less clear.