Accuracy, Precision, and Trending Ability of Perioperative Central Venous Oxygen Saturation Compared to Mixed Venous Oxygen Saturation in Unselected Cardiac Surgical Patients

Objective: To determine whether central venous oxygen saturation (ScvO₂) measurements could be used interchangeably with mixed venous oxygen saturation (SvO₂) measurements in adult cardiac surgery patients.

Design: A single-center prospective observational study.

Setting: A university hospital.

Participants: Eighty-five adult patients undergoing cardiac surgery.

Interventions: The study authors compared the oxygen saturations in 590 pairs of venous blood samples drawn from the pulmonary artery catheter (PAC) at three different time points during surgery and four different time points in the intensive care unit. They compared samples obtained from the distal pulmonary artery line (SvO₂) to those drawn from the proximal central venous line of the PAC (ScvO₂) with the Bland-Altman test and the four-quadrant method.

Measurements and main results: The mean bias between SvO₂ and ScvO₂ was -1.9 (95% confidence interval [CI], -2.3 to -1.5) and the limits of agreement (LOA) were -11.5 to 7.6 (95% CI, -12.5 to -10.7 and 6.8-8.5, respectively). The percentage error (PE) was 13.2%. Based on the four-quadrant plot, only 50% of the measurement pairs were in agreement, indicating deficient trending ability.

Conclusion: ScvO₂ values showed acceptable accuracy as the mean bias was low. The precision was inadequate; although the PE was acceptable, the LOA were wide. Trending ability was inadequate. The authors cannot recommend the use of ScvO₂ values interchangeably with SvO₂ measurements in the management of adult cardiac surgery patients.