Cerebral near-infrared spectroscopy during cardiopulmonary bypass predicts superior vena cava oxygen saturation

ObjectiveCerebral and flank near-infrared spectroscopy are used to monitor tissue oxygenation during cardiopulmonary bypass in pediatric patients. We sought to validate these noninvasive measurements as predictors of oxygen saturation in the superior and inferior venae cavae during cardiopulmonary bypass.MethodsEight patients underwent elective repair of congenital heart defects with bicaval cannulation. Ultrasonic flow probes and oximetric catheters were placed in the superior and inferior venae cavae limbs of the perfusion circuit. Cerebral and flank near-infrared spectroscopy and 12 additional variables were recorded each minute on cardiopulmonary bypass. Relationships between these variables and superior and inferior venae cavae oxygen saturation were analyzed by linear mixed modeling. The regression of superior vena cava oxygen saturation by current cerebral near-infrared spectroscopy and 1-minute lag cerebral near-infrared spectroscopy, which are equivalent to the regression of the superior vena cava saturation by the current cerebral near-infrared spectroscopy and the 1-minute change in cerebral near-infrared spectroscopy, were used to assess cerebral near-infrared spectroscopy as a trend monitor.ResultsThe mean number of observation time points per patient was 86 (median 72, range 34–194) for 690 total observations. The root mean square percentage error was 6.39% for the prediction model of superior vena cava saturation by single-factor cerebral near-infrared spectroscopy. The root mean square percentage error was 10.8% for the prediction model of inferior vena cava saturation by single-factor flank near-infrared spectroscopy.ConclusionsCerebral near-infrared spectroscopy accurately predicts superior vena cava oxygen saturation and changes in superior vena cava oxygen saturation on cardiopulmonary bypass. The relationship between flank near-infrared spectroscopy and inferior vena cava saturation is not as strong.