Oxygenation, a concept of multi-system integration and coordination in the intake, delivery, and use of oxygen for energy metabolism, should be measured using methods that provide continuous real-time data, allowing for simultaneous evaluation of intake, delivery, and use of oxygen. Dual oximetry (simultaneous use of arterial [Spa02 ] and venous oximetry [Spv02 ]) derived parameters meet these criteria. How well dual oximetry parameters agree with blood gas derived parameters is not known.
The purpose of this study was to evaluate dual oximetry derived parameters as continuous real-time indicators of oxygenation. Data sets (N = 190) were retrieved from the medical records of 40 adult critically ill, mechanically ventilated, surgical patients and used to calculate blood gas and dual oximetry derived equations for intrapulmonary shunt (Qsp/Qt, VQI), oxygen delivery (D02 , Dp02 ), oxygen extraction (02 ER, 0 2 EI), and oxygen consumption (V02 , Vp02 ). The data were analyzed using Bland and Altman statistical methods to examine the nature and extent of agreement between the measurement methods.
Spa02 tended to overestimate Sa02 by 2%; the precision estimate was 2.1%. The limits of agreement were -2.2% and +6.2%. Spv0 2 was found to be minimally biased with Sv02 (+0.10%) and moderately imprecise (3.5%) with limits of agreement of -6.8%, +7.0%. Large bias (-19%) and precision estimates (9.1%) for intrapulmonary shunt led to clinically unacceptable limits of agreement (-0.8%, -37%). Dp02 was judged to be interchangeable with D02 (+0.10% bias, 3% precision, and limits of agreement -5.5%, +53%). Vp02 was found to underestimate V02 (-3.6%), was highly imprecise (21.2%) and had wide limits of agreement (-45.85%, +38.6%). 02 EI had a -0.2% bias, 4% precision, and clinically unacceptable limits of agreement (-8.2%, +7.85%).
Measurement of arterial and mixed venous oxygen saturation by oximetry were found to be interchangeable with blood gas measurements. However, limits of agreement may be wider than previously thought. Although limits of agreement were clinically acceptable, measurement error may have propagated through the calculations to produce clinically unacceptable limits of agreement for dual oximetry derived parameters of oxygenation. Additional studies must use prospective designs to control bias and precision. Investigators can more accurately evaluate dual oximetry derived parameters as continuous real-time indicators of oxygenation.