For example, patient 2 of the Oximetrix 3 group had a bias for the study period of +2, while the average bias for all patients in that group was —1.98. Similarly, the individual precision for some patients was considerably better than the precision for the group as a whole. When similarity of catheter performance to manufacturers’ specifications was tested (Eq 10 to 12) (±2 percent saturation, ± 1SD), none of the systems rejected the null hypothesis to prove similarity at ±2 percent relative to CO-oximetry.
The performance characteristics for each of the systems are summarized in Table 4. The slope of the regression analysis for each group was similar; however, there were striking differences in the intercept values. The r2 values showed remarkable consistency for each of the three groups. As noted above, Oximetrix 3 and SAT-2 had comparable performance as measured by average bias and precision, while HEMOPROa showed poorer precision. Ambiguity and RMSE, both proposed as measures of overall performance, were similar for the Oximetrix 3 and SAT-2 groups. The HEM0PR02 group again demonstrated poorer performance by these measures. The ESD of repeated measurements within subjects, calculated as the square root of the average intrasubject variance (Eq 5), excludes intersubject variability.
Table 4—Overall Performance of Three Oximetry Pulmonary Artery Catheters Compared with CO-Oximetry
|Bias ± precision||— 1.98 ±3.07||1.86±3.49||-2.28 ±5.24|
|95% confidence limits (total variability)||±6.03||±6.86||±10.30|
|95% confidence limits (intrasubject variability)||±4.59||±5.66||±6.56|