Abstract
PURPOSE: With the exception of neonatal respiratory failure, most centers are now using centrifugal over roller-type pumps for the delivery of extracorporeal membrane oxygenation (ECMO). Evidence supporting the use of centrifugal pumps specifically in infants with congenital diaphragmatic hernia (CDH) remains lacking. We hypothesized that the use of centrifugal pumps in infants with CDH would not affect mortality or rates of severe neurologic injury (SNI). METHODS: Infants with CDH were identified within the ELSO registry (2000-2016). Patients were then divided into those undergoing ECMO with rollertype pumps or centrifugal pumps. Patients were matched based on propensity score (PS) for the ECMO pump type based on pre-ECMO covariates. This was done for all infants and separately for each ECMO mode, venovenous (VV) and venoarterial (VA) ECMO. RESULTS: We identified 4,367 infants who were treated with either roller or centrifugal pumps from 2000-2016. There was no difference in mortality or SNI between the two pump types in any of the groups (all infants, VA-ECMO infants, VV-ECMO infants). However, there was at least a six-fold increase in the odds of hemolysis for centrifugal pumps in all groups: all infants (odds ratio [OR] 6.99, p<0.001), VA-ECMO infants (OR 8.11, p<0.001 and VV-ECMO infants (OR 9.66, p<0.001). CONCLUSION: For neonates with CDH requiring ECMO, there is no survival advantage or difference in severe neurologic injury between those receiving roller or centrifugal pump ECMO. However, there is a significant increase in red blood cell hemolysis associated with centrifugal ECMO support.