Background: Extracorporeal membrane oxygenation (ECMO) is accepted therapy for cardiorespiratory failure. Even after a successful ECMO course, patient deterioration may occur and a second course of ECMO may be contemplated. Although data regarding second ECMO courses exist in neonates, there are no reports describing second ECMO courses in pediatric patients. We hypothesized that data from a national ECMO registry would be useful in identifying which pediatric patients would be optimal candidates for a second course of ECMO.
Methods: We obtained data from the national Extracorporeal Life Support Organization registry from 1981 to 2007 on all patients 1-18 years old who required single-run ECMO (SRE) or multiple-run ECMO (MRE). Primary outcome measures were complications and survival. Continuous variables were assessed for distribution normality by using a Shaprio-Wilk statistic to guide nonparametric testing. SRE and MRE patients were compared by using chi2 tests (Fisher's exact and McNemar's) to assess differences in categorical variables; continuous data were assessed by using Mann-Whitney U or Wilcoxon signed-rank testing. Two multivariate regression models were constructed to identify independent predictors of survival and complications in MRE patients. Statistical significance was assumed at P < 0.05.
Results: A total of 3937 pediatric patients received ECMO for cardiac or respiratory failure. Of them, 3810 (96.8%) children underwent a single course of ECMO, whereas 127 (3.2%) required multiple ECMO runs. Compared with SRE patients, the first ECMO course in MRE patients was notable for higher rates of cardiac ECMO (61% versus 44%, P < 0.001), venoarterial ECMO (88% versus 78%, P = 0.04), and central cannulation (28% versus 17%, P = 0.007). There was no survival difference between MRE and SRE patients (44% versus 49%, P = 0.28). Median time between MRE courses was 9.0 days (interquartile range = 5-20 days). The mean number of complications per MRE patient was higher in the second ECMO run compared with the first (3.93 versus 3.12, P = 0.008). Multivariate regression identified 2 variables as independent predictors of survival in MRE patients: (1) renal complications during first ECMO run (P = 0.04); and (2) total number of complications during second ECMO run (P = 0.005). A separate multivariate analysis identified 3 variables independently predictive of complications in MRE patients: (1) age (P < 0.001); (2) duration of second run (P < 0.001); and (3) total number of complications during first ECMO run (P < 0.001).
Conclusions: ECMO therapy achieves 49% survival in children 1-18 years of age. When a second ECMO course becomes necessary, survival rates comparable to the first ECMO course are possible. Patients developing renal complications during their first ECMO course have worse outcome with a second ECMO course. Patients are at greater risk for complications during a second ECMO course if they experience a high number of first-run complications, are >3 years old, or undergo a prolonged second ECMO course. These data are useful when deciding whether to offer a second ECMO course to an eligible pediatric patient.