Extracorporeal membrane oxygenation (ECMO) has been used since the 1970s to rescue the sickest of children with cardiopulmonary failure that are refractory to conventional intensive care therapies [1, 2]. As was shown in the UK randomized trial of ECMO for neonatal respiratory failure, patients requiring ECMO are at high risk of neurological injury [3]. In the current era, with improving survival rates for critically ill children and an expansion in the role of ECMO to include patients in cardiac arrest (ECPR, extracorporeal cardiopulmonary resuscitation) [4], there is growing awareness that it is inadequate to focus solely on short-term survival. Due to the neurologic morbidities associated with ECMO support, longer term outcomes including physical, neurodevelopmental, and psychosocial outcomes need to be monitored and carefully evaluated to optimize quality of life [5].

In a recent issue of Intensive Care Medicine, two studies were published that deepen our understanding of longer term outcomes after ECMO.

Polito and colleagues present a report of risk factors for neurological injury in neonates undergoing ECMO based on the ECMO registry of the Extracorporeal Life Support Organization (ELSO) [6]. Of 7,190 neonates included in the Polito study, 1,412 (20 %) experienced a neurological complication, these being more likely in the smallest and sickest babies. Given that the Polito study is based on ELSO data captured within the intensive care unit where evaluation of neurodevelopmental outcome has inevitable limitations related to systemic illness and the use of sedatives, and that 62 % of patients with neurological complications died, this is likely to represent the worst end of the spectrum. Birth weight <3 kg, gestational age <36 weeks, need for cardiopulmonary resuscitation prior to ECMO, pre-ECMO blood pH ≤ 7.11, pre-ECMO bicarbonate use, prior ECMO exposure, and use of veno-arterial ECMO were associated with a higher risk of neurologic complications. While some of these predictors of worse neurologic outcome are not amenable to intervention (e.g., birth weight and gestational age), those that are modifiable should be targeted for future intervention to minimize future neurologic injury in the high-risk neonatal ECMO population.

Madderom and a multidisciplinary team of colleagues [7] from the neurodevelopmental follow-up clinic at Erasmus University report their experience with neonatal respiratory ECMO survivors to age 8 years. Of 135 children evaluated, 91 % were in mainstream school, 39 % of the cohort required educational support, and 9 % were in special education. Although intelligence fell within the normal range for the population, children treated with ECMO in the past were more likely than their peers to experience problems executing their schoolwork and to have emotional and/or behavioral issues. Specifically, the Dutch cohort by parent and teacher report demonstrated both internalizing (i.e., somatic complaints) and externalizing (i.e., aggression) behaviors, as well as issues with attention, thought, total problems, and social cognition. This phenotype is similar to that seen in child and adolescent complex congenital heart disease survivors who have undergone cardiopulmonary bypass as neonates in both the US and UK, which may significantly affect quality of life [8, 9]. This study offers some unique insights into the types of impairment children of school age who have survived ECMO may experience and therefore the support that they may require in future life.

A registry such as ELSO [10], which has wide coverage and a large number of patients, can offer objective data on risk factors for short-term outcome measures and is invaluable for answering research questions or for generating hypotheses for future mechanistic or intervention studies. It also offers unique and vital opportunities to benchmark outcomes, which is particularly important for such a high-risk and resource-intense treatment as ECMO. However, registry data have inherent limitations, and it is vital to complement these short-term data with detailed longitudinal assessment of outcomes including the kind of prospective follow-up of survivors reported in the Madderom study and others [11, 12].

Mechanisms of injury to the developing brain in children undergoing treatment with ECMO are both primary (including congenital syndromes or vascular malformations) and secondary, related to both the underlying disease process and aspects of ECMO support itself (including hypoxic-ischaemic encephalopathy, reperfusion injury, raised intracranial pressure, low cerebral blood flow states including cardiac arrest, cerebral bleeding, disordered cerebral autoregulation, thromboembolism and instrumentation of the cerebral vessels). Polito and colleagues allude to disorders of cerebral autoregulation being implicated in the evolution of severe neurological complications during ECMO in neonates. Observational data support the theory that cerebral autoregulation may be disordered during neonatal ECMO support [13]. Greater understanding of the mechanisms of neurological injury in the intensive care unit may lead to beneficial changes in practice—for example, cooling for perinatal asphyxia [14]—although the benefits of cooling to reduce neurological injury in ECMO-treated neonates have not been well established [15]. Further exploration of cerebral autoregulation and how it is affected by both critical illness and ECMO may represent a fruitful avenue of future research.

Retrospective evaluations of various single-center cohorts from the USA [1618], Canada [19], Australia [20], and the UK [21] shed further light on longer term outcomes in regards to late mortality events [21], neurodevelopmental deficits [12, 17, 19], and quality of life [18, 20] in neonates and children who survive to hospital discharge following ECMO. All these various data show it is essential for ECMO clinicians to invest in the follow-up of their patients so that patient-specific interventions can be put in place to optimize functional survival. Indeed, this has been recommended by several professional societies including the American Heart Association (AHA) [22], the Pediatric Cardiac Intensive Care Society (PCICS) [23, 24] and ELSO [25].

In conclusion, Polito, Madderom and co-workers have contributed important studies that take us forward toward a greater understanding of neurodevelopmental outcomes in ECMO patients. This represents an area that intensive care clinicians and tertiary hospitals managing children on ECMO will be required to attend to more closely in coming years. The critically ill children that undergo ECMO, their families and the local services caring for them all deserve a renewed focus on this aspect of their care. Future goals must include a greater mechanistic understanding leading to reduced levels of neurodevelopmental injury in ECMO patients and the initiation of innovative, scalable, long-term intervention protocols to meet the needs of this high-risk population.