LAUSR

486 www.ccmjournal.org March 2018 • Volume 46 • Number 3 The management of sick, term neonates with respiratory failure is a challenging but rewarding endeavor. For most conditions, the survival is excellent with medical management. In the rare circumstance when extracorporeal membrane oxygenation (ECMO) is required for refractory cases, survival rates approach 100% for the most common etiology, meconium aspiration syndrome (1, 2). However, neonates with congenital diaphragmatic hernia (CDH) represent a more challenging subgroup. Despite the use of ECMO, only 50–60% survive to hospital discharge and an additional 10–15% of these children may not survive to school age (2). Given the severity of illness at presentation and comorbidities that may have existed or been acquired during the initial treatment, it is not surprising that a fair number of the survivors of neonatal critical illness may have impaired neurodevelopment and school problems (3, 4). In long-term follow-up, few risk factors for identifying these problems have been consistently observed, and study of the range of neuropsychologic impairment has been limited. In order to further address the above issues, Leewen et al (5), in this issue of Critical Care Medicine, have followed a cohort of children born during a 39-month period who required neonatal ECMO and/or had CDH and performed neuropsychologic testing at school age. They used a variety of instruments spanning six domains including intelligence, attention, visuospatial memory, visuospatial processing, verbal memory, and executive functioning. Their original cohort included 127 neonates with approximately 80% survival to school age. Of the survivors, just over 80% of the children were eligible for the study, and nearly 80% were recruited, giving a study cohort of 65 children. Slightly more than half of the children were treated with ECMO, one third of whom had CDH. The authors grouped the patients into three main categories for analysis: CDH children not treated with ECMO (CDH group), the CDH children treated with ECMO (CDHECMO group), and the children treated with ECMO for other respiratory reasons (R-ECMO group). In the children treated with ECMO, venoarterial ECMO was used in all children in the CDH-ECMO group, and veno-venous ECMO was used in the majority of the R-ECMO group. Impairment in the domains of attention, visuospatial memory, and verbal memory was common: more than half of the children tested scored 1 z score or more below normal in one or more of the domains. Additionally, the degree of impairment in these domains was greater than would be expected based on intelligent quotient (IQ) scores: only the CDH-ECMO group had a mean IQ score that was significantly below normal. However, the other five tested domains did not demonstrate significant differences between the three groups of patients. In order to identify risk factors for neuropsychologic impairment, the authors assessed demographic variables, ECMO-related variables, hospital and severity of illness variables, and growth at 1-year follow-up. On univariable analysis of associations with intelligence, several variables were found to be significant such as baseline patient group (i.e., CDH-ECMO-R-ECMO, etc), hospital length of stay, maximum vasoactive inotropic score (VIS), and growth at 1-year follow-up. However, no significant independent predictors were found on multivariable analysis. For the other domains, only VIS was associated with components of verbal and visuospatial memory on multivariable analysis. In their discussion, the authors hypothesize that poor perfusion to vulnerable regions of the brain, such as the hippocampus, may explain their finding of VIS as a significant risk factor in these two memory domains. Consistent with previous studies, this work reinforces the need for regular follow-up and neuropsychologic assessment in this high-risk group of neonates. Interestingly, as seen for pediatric cardiac patients, the authors found some hospitalrelated variables to be associated with IQ on univariable analysis, such as length of stay (6). The authors’ novel finding that VIS was independently associated with verbal and visuospatial memory is important and should be further investigated in terms of both a neuropathophysiologic basis and its reliability. Their suggestion that poor perfusion may result in anatomic abnormalities of the hippocampus that have been reported in ECMO-treated patients (7, 8), a structure that is important for delayed memory, is intriguing and supports the need for further longitudinal neuroimaging studies. Identifying this link between abnormalities identified by neuroimaging and outcomes is key since others have questioned the predictive value of neuroimaging (9). The potential utility in using the VIS to identify the timing and need for ECMO was also suggested by *See also p. 401.