Is near infrared spectroscopy an accurate marker of low cardiac output after stage 1 palliation in single ventricle patients?
Background: Balancing the systemic and pulmonary circulation during the early postoperative period is paramount in the management of children who have physiology compatible with single ventricle parallel circulation after stage 1 palliation. Near-infrared spectroscopy (NIRS) is a non-invasive, continuous method of evaluating real-time regional oximetry. Previous studies have shown a moderate correlation between cerebral near-infrared spectroscopy (cNIRS) and other markers of low cardiac output, although, there is a paucity of data in patients with single ventricle parallel physiology.
Methods: Retrospective chart review including children with single ventricle parallel physiology who underwent stage 1 palliation between January 2010 and December 2019 in our institution. IRB approval was obtained. Multiple perioperative variables were recorded during the first 24 postoperative hours, including demographics, vital signs, vasoactive inotropic score, cerebral and somatic NIRS values, and laboratory data. The main outcome variable was low cardiac output (CO) defined by the pediatric cardiac critical care consortium. A secondary analysis was performed using composite adverse outcome defined as one of the following: cardiac arrest, need for extracorporeal membrane oxygenation, or death.
Results: 93 patients with mean (SD) age of 29.9 (54.3) days were included in the study. Demographic and perioperative characteristics are illustrated in table 1. cNIRS values were lower in the low CO group compared to the normal CO group at 1 hour (40.75 ± 13.8 vs 49.23 ± 11.6; p=0.002 ), 6 hours, (43.75 ± 9.8 vs 51.96 ± 9.4; p<0.001) and 12 hours (51.95 ± 8.5 vs 56.04 ± 9.4; p=0.037). Similarly, the arithmetic difference between somatic and cerebral NIRS was higher in the low CO group compared to the normal CO group at 1hour, (27.26 ± 12.8 vs 18.61 ± 15.6; p=0.007) 6 hours, (22.45 ± 19 vs 15.49 ± 12.8; p <0.001) and 12 hours (16 ± 10.8 vs 10.42 ± 11.4; p=0.027). Table 2 summarizes perioperative variables between the low CO group and rest of the cohort. Moreover, lower cNIRS at 6 hours (p=0.027) and lower systolic blood pressure at 1hour (p=0.04) remained independently associated with low CO after logistic regression analysis. Receiver operating characteristic curve showed mean cNIRS of ≤55% in the first 24 hours had 81% sensitivity, and 62% specificity in predicting low cardiac output. Secondary analysis revealed cNIRS at 1hour and NIRS difference at 24 hours were associated with adverse outcome (p=0.024 and p=0.048; respectively), although did not remain significant by logistic regression.
Conclusion: Near-infrared spectroscopy was an accurate marker of low cardiac output in our cohort. NIRS monitoring is an important addition to the clinician's arsenal when managing critically ill single ventricle patients in the early postoperative period. Future studies should investigate the utility and accuracy of NIRS in other critical care scenarios.