Factors associated with survival in pediatric extracorporeal membrane oxygenation--a single-center experience

We aimed to examine outcomes of extracorporeal membrane oxygenation (ECMO) therapy in the pediatric population and identify pre-ECMO and on-ECMO characteristics that are associated with survival.

We retrospectively reviewed the ECMO records at our institution between 1999 and 2008 and selected pediatric patients who were cannulated for respiratory failure or hemodynamic instability resistant to conventional interventions. We recorded details of pre-ECMO clinical characteristics, including blood gas variables and mechanical ventilatory support, and details of ECMO therapy including survival off ECMO and to hospital discharge. Predictors of survival were analyzed using logistic regression modeling and a prediction algorithm was developed.

Of the 445 ECMO runs, data from 58 consecutive patients were analyzed: 57% were successfully decannulated, and 48% survived to discharge from the hospital. The cohort included 32 (55%) female patients, 22 postoperative patients (38%), and 15 (26%) with an immunosuppressive condition, with a median age of 5 years and weight 19.5 kg, The mean duration of pre-ECMO respiratory support was 3 days, in the form of high-frequency oscillatory ventilation (n = 28, 48%) and conventional mechanical ventilation (n = 13, 22%). The median duration (interquartile range) of ECMO support was 142 hours (60, 321) or 5.9 days. Pre-ECMO pH above 7.2 (P < .001) and oxygenation index below 35 (P = .021) were associated with the highest survival rates. Pre-ECMO PaCO(2) and duration of mechanical ventilation were not associated with survival.

Based on our results, ECMO therapy should be considered early in children with oxygenation index greater than 35 with worsening metabolic status. The restriction of ECMO based on ventilator days alone needs to be revisited in this era of lung protective ventilation.