Neurological complications in patients on extracorporeal membrane oxygenation: predicators, outcomes, and implications for surgical management

Abstract

Objective: There is paucity of data in existing literature to guide management of neurological complications in extracorporeal membrane oxygenation (ECMO) patients. We studied the incidence of neurological complications, predictors of intracranial bleeding and 30-day mortality, and reviewed outcomes of management in this unique group of patients. Method: From August 2015 to July 2021, 349 consecutive patients had ECMO insertion at Queen Mary Hospital. Patients with ischemic or hemorrhagic intracranial events were included. Nested case-control with 1:2 nearest neighbor matching was performed (without replacement) to identify a group without neurological complications. Univariate and multivariate logistic regressions studied the predictors for intracranial bleeding (ICH) and 30-day mortality. Finally, a case series of ECMO patients who underwent neurosurgical interventions were reviewed. Result: The incidence of neurological complications was 11% (39/349). Extracranial bleeding (adjusted OR 8.25 [2.78-25.11], p<0.001), prior CPR (cardiopulmonary resuscitation) (adjusted OR 2.63 [1.16-6.25], p=0.021), lower platelet (adjusted OR 0.99 [0.99-1.00], p=0.049) and higher APTT levels (adjusted OR 1.15 [1.01-1.32], p=0.032) were independent predictors of intracranial haemorrhage. Presenting Glasgow coma scale (GCS) (adjusted OR 0.80 [0.68-0.95], p=0.012) and ICH volume (adjusted OR 1.05 [1.00-1.10], p=0.038) predicted 30-day mortality. The 30-day mortality was 42.9% for the 7 ECMO patients who underwent neurosurgical interventions, which had no significant difference compared with the non-operated group (13.6% vs 23.5%, p=0.350). Conclusion: Three were significantly more intracranial haemorrhage in our ECMO cohort compared to Caucasian data. Extracranial bleeding, prior CPR, lower platelet and higher APTT levels independently predicted incidence of intracranial bleeding in ECMO patients. Presenting GCS and ICH volume significantly predicted 30-day mortality. Neurosurgical interventions did not alter the rate of 30-day mortality.