46 M.-H. Lee and T. Rosenthal: J Extra Corpor Technol 2026, 58, 43 – 50
Figure
1. The error in the calculated in-line PaO 2 of Quantum System at the 1st blood gas series has a strong correlation with the patient weight, which allows to derive a formula to compute the predicted in-line PaO 2.( A) A scattered XY plot was drawn for patient weight( X-axis) and error in mmHg of the 1st blood gas series( Y-axis). The data was fitted into a linear regression line, showing a strong correlation. R value is shown in the upper right corner.( B) A scattered XY plot was drawn for the calculated in-line PaO 2 of Quantum System( X-axis) and predicted in-line PaO 2 computed with our formula( Y-axis). The data was fitted into a linear regression line, showing a very strong correlation. R value is shown in the upper right corner.( C) Bland-Altman analysis of the calculated and predicted in-line PaO 2 shows strong agreement with a bias of 1 mmHg( bold solid line) and LOA of 101 and �99 mmHg( bold dashed lines). The X-axis represents the average of the calculated and predicted in-line PaO 2, while the Y-axis represents the difference between the calculated and predicted in-line PaO 2.( D) A scattered XY plot was drawn for the calculated in-line PaO 2( X-axis) and FiO 2( Y-axis) of Quantum System. The data was fitted into a linear regression line, showing a very strong correlation. R value is shown in the upper right corner.
this correlation disappeared at the 2nd blood gas series( R = 0.07) and at the 3rd blood gas series( R = 0.19) after the second in vivo calibration( Table 2).
From the strong linear correlation at the 1st blood gas series, we derived a formula to compute a predicted in-line PaO 2 based on the measured PaO 2 at the 1st blood gas series:
Predicted inline PaO 2 ðmmHgÞ ¼ measured PaO 2 þ 13 Weight þ 40:
Using this formula, we determined the predicted in-line PaO 2 of all 81 patients with the measured PaO 2 and patient weight and compared it to the calculated in-line PaO 2 of the Quantum System. We found a very strong linear correlation between the calculated and predicted in-line PaO 2( R = 0.8, slope = 1.0, Y intercept = 2.3, see Figure 1B). The Bland-Altman analysis shows that the bias( average difference) is 1 mmHg and LOA( 1.96 SD, 95 % of data is within the LOA) is 101 and �99 mmHg between the calculated and predicted in-line PaO 2( Figure 1C)[ 19, 21 – 23 ].
Based on the very strong linear correlation between the calculated and predicted in-line PaO2, we can use our formula to calculate the predicted in-line PaO 2 to achieve a target PaO 2 at the 1st blood gas series on CPB. This can be done by substituting the measured PaO 2 with a target PaO 2 in the formula:
Predicted inline PaO 2 ¼ target PaO 2 þ 13 Weight þ 40:
Forexample, ifatargetPaO 2 is 250 mmHg with a 10 kg patient, the predicted in-line PaO 2 is 420 mmHg. This suggests that if we set FiO 2 to achieve the calculated in-line PaO 2 of 420 mmHg when we perform the 1st blood gas series on CPB, the actual PaO 2 is approximately 250 mmHg.
Calculated in-line PaO 2 is essentially determined by FiO 2 before the first in vivo calibration
The error observed at the 1st blood gas series increased as patient weight increased, suggesting that the impact of oxygen consumption on the in-line PaO 2 calculation is minimal prior to the first in vivo calibration. Therefore, we investigated how strongly the calculated in-line PaO 2 is dependent on FiO 2