Objective: To assess the ability of two different pulse oximeters to display continuous venous oxygen saturation through an extracorporeal bypass circuit with a degree of accuracy comparable to direct in-line oximetry.
Design: Prospective, comparison study of pulse oximeters (test oximeter 1 or test oximeter 2) and an in-line oximeter (test oximeter 3).
Setting: A tertiary care neonatal intensive care unit.
Patients: Sixty-five consecutive neonates with severe cardiorespiratory failure undergoing extracorporeal life support.
Interventions: The accuracy of the oximeters was determined by simultaneously comparing the saturation displayed by the pulse oximeters (test oximeters 1 and 2) and/or the in-line oximeter (test oximeter 3) with the measured fractional venous oxygen saturation obtained at regular intervals from the extracorporeal circuit.
Main outcome measures: Venous oxygen saturation was the criterion standard used to determine accuracy. Bias was defined as the mean difference between observed pulse oximeter or in-line oximeter values and the measured venous oxygen saturation. Mean biases were calculated for venous oxygen saturation measurements between 55% and 99% at intervals of 10%. Precision (the standard deviation of the bias) was calculated for low (55% to 75%), medium (76% to 81%), and high (82% to 99%) venous oxygen saturation values. A total of 983 venous oxygen saturation measurements were made and compared with simultaneous oximeter readings from test oximeter 1 (n = 600), test oximeter 2 (n = 478), and test oximeter 3 (n = 587).
Results: Test oximeter 1 was the most precise instrument at each level of venous oxygen saturation (SD, 4.0 to 4.8). Test oximeter 3 demonstrated the most consistent mean bias (range, 8), but was the most inaccurate oximeter across all levels of venous oxygen saturation.
Conclusions: In addition to its known clinical usefulness, pulse oximetry may serve as an adequate substitute for in-line oximetry during extracorporeal life support.