
Abstract
Objective
Neurologic injury is common in pediatric patients with extracorporeal membrane oxygenation (ECMO). Impaired cerebral autoregulation (CAR) may increase the risk of neurologic injury. We aimed to measure CAR in different phases of ECMO and analyze for association with radiographic neurologic injury.
Methods
We performed a retrospective, single-center, observational cohort study of pediatric (0–18 years) ECMO patients with no prior neurologic injury, cardiopulmonary bypass, ECMO, or congenital heart disease who underwent brain imaging. Correlation of time-matched cerebral oximetry and mean arterial blood pressure (MAP) data determined the cerebral oximetry index (COx) a noninvasive estimate of CAR. Impaired CAR was tested for association with neurologic injury score (NIS) as measured by blinded neuroradiologists. Secondarily, we calculated the area under the curve (AUC) of MAP above or below age-based 50th percentile MAP in temporal windows with impaired CAR. Univariate analysis and a multivariable linear regression model adjusting for type of ECMO, ECMO indication, and age determined the association between impaired CAR and the AUC metric with NIS.
Results
Data from 72 children were analyzed. The median age was 11.7 months, 47 children (65.3%) received VA ECMO, and severe neurologic injury (SNI) occurred in 9 (12.5%). There was no association between percent time with impaired CAR and SNI (16.6% versus 12.2%, P = 0.38). Impaired CAR burden was significantly higher pre-ECMO as compared with peri-cannulation and intra-ECMO time periods (26.5% versus 14.8% and 16.6%, P < 0.01). The multivariable linear regression model discerned a significant association between pre-ECMO hypotension during impaired CAR and NIS (coefficient = –0.008; –0.015, –0.001; P = 0.026; R2 = 0.19).
Conclusions
Total time with impaired CAR measured by the cerebral oximetry index was not associated with neurologic injury in pediatric patients with ECMO. However, impaired CAR was more common pre-ECMO, and pre-ECMO AUC incorporating magnitude of blood pressure derangement was significantly associated with neurologic injury, implicating pre-ECMO clinical derangements as imparting risk.