A matched pairs analysis of venoarterial and venovenous extracorporeal life support in neonatal respiratory failure.

It has been suggested that venovenous (VV) extracorporeal life support (ECLS) confers a survival advantage over venoarterial (VA) ECLS. These results have been confounded by differences in patient populations. In this study, a matched pairs comparison of survival and complication rates in neonatal respiratory failure patients managed with VA or VV ECLS was performed. Retrospective matching of 643 VA and VV patient pairs from the Extracorporeal Life Support Organization Registry was performed. Pairs were matched by same year, same diagnosis, gestational age +/- 1 week, birth weight +/- 0.3 kg, and oxygenation index +/- 5. Further matching for hemodynamic status was possible for 272 pairs and included pre ECLS CPR, use of epinephrine, and arterial pH +/- 0.1. Statistical significance was defined for outcome and selected complication rates using McNemar's chi-square analysis with correction for multiple comparisons. A survival advantage for VV was significant when matching for respiratory failure (83.8% VA versus 91.5% VV), but was not significant when matching for hemodynamic failure (90.4% VA versus 94.5% VV). In the latter match, hemolysis (10.7% VA versus 23.5% VV) and cannula kinking (0.4% VA versus 10.6% VV) were more common with VV ECLS. The incidence of intracranial hemorrhage did not significantly differ between groups (6.3% VA versus 7.4% VV). Survival is not significantly greater with VV ECLS when patients are matched for degree of respiratory and hemodynamic failure. Hemolysis and cannula kinking are more common with VV ECLS. There is no identified difference in the incidence of intracranial hemorrhage.