Objective: To compare changes in regional cerebral or flank oxygen saturation measured by near-infrared spectroscopy with changes in central venous oxygen saturation (ScvO2) and to determine clinical variables associated with these changes.
Design: Prospective observational cohort study.
Setting: University tertiary care center, pediatric cardiac intensive care unit.
Patients: Seventy postoperative congenital cardiac surgical patients (median age 0.3 yrs; interquartile range 0.02-0.46 yrs).
Interventions: None.
Measurements and main results: We measured temporally correlated regional oxygen saturation (rSO2) with hematologic (hematocrit), biochemical (arterial blood gas, ScvO2, and lactate) and physiologic (temperature, heart rate, mean blood pressure, and pulse oximetry) variables in the first postoperative day. Cerebral and flank rSO2 were strongly correlated with ScvO2, in both cyanotic or acyanotic patients and single- or two-ventricle physiology with and without aortic arch obstruction (all p < .001). However, individual values had wide limits of agreement on Bland-Altman analysis. The correlations of change in these measurements were weaker but still significant (all p < .0001), again with wide limits of agreement. Similar direction of change in cerebral rSO2 and ScvO2 was present 64% (95% confidence interval, 55-73%) of the time. Change in arterial pressure of carbon dioxide (delta PaCO2) was associated with cerebral delta rSO2 (delta PaCO2 beta = 0.35, p < .0001) but not flank delta rSO2 or delta ScvO2. A pattern of relative cerebral desaturation (flank rSO2 > cerebral rSO2) was noted in a majority of patients (81%) with two-site monitoring regardless of bypass method or age.
Conclusions: Neither individual values nor changes in rSO2 are interchangeable measures of ScvO2 in postoperative pediatric cardiac patients. The unique relationship between changes in PaCO2 and cerebral rSO2 supports the hypothesis that cerebral near-infrared spectroscopy monitors regional cerebral oxygenation. Clinical application of this monitor must include recognition of the clinical variables that affect regional brain oxygenation.