Modeling severe functional impairment or death following ECPR in pediatric cardiac patients: Planning for an interventional trial

Aim: We aimed to characterize extracorporeal CPR (ECPR) outcomes in our center and to model prediction of severe functional impairment or death at discharge. Methods: All ECPR events between 2011 and 2019 were reviewed. The primary outcome measure was severe functional impairment or death at discharge (Functional Status Score [FSS] > 16). Organ dysfunction was graded using the Pediatric Logistic Organ Dysfunction Score-2, neuroimag-ing using the modified Alberta Stroke Program Early Computed Tomography Score. Multivariable logistic regression was used to model FSS > 16 at discharge. Results: Of the 214 patients who underwent ECPR, 182 (median age 148 days, IQR 14-827) had an in-hospital cardiac arrest and congenital heart disease and were included in the analysis. Of the 110 patients who underwent neuroimaging, 52 (47%) had hypoxic-ischemic injury and 45 (41%) had hemorrhage. In-hospital mortality was 52% at discharge. Of these, 87% died from the withdrawal of life-sustaining therapies; severe neurologic injury was a contributing factor in the decision to withdraw life-sustaining therapies in 50%. The median FSS among survivors was 8 (IQR 6-8), and only one survivor had severe functional impairment. At 6 months, mortality was 57%, and the median FSS among survivors was 6 (IQR 6-8, n = 79). Predictive models identified FSS at admission, single ventricle physiology, extracorporeal membrane oxygenation (ECMO) duration, mean PELOD-2, and worst mASPECTS (or DWI-ASPECTS) as independent predictors of FSS > 16 (AUC = 0.931) and at 6 months (AUC = 0.924). Conclusion: Mortality and functional impairment following ECPR in children remain high. It is possible to model severe functional impairment or death at discharge with high accuracy using daily post-ECPR data up to 28 days. This represents a prognostically valuable tool and may identify endpoints for future interventional trials.