Historical controls for extracorporeal membrane oxygenation in neonates.

A retrospective review of all patients cared for who met Loe's criteria for extracorporeal membrane oxygenation (ECMO) was conducted covering the 3-yr period 1983 through 1985. There were five out of 3,726 admissions who met criteria for ECMO (three of 127 outborn admissions). All infants were greater than 2 kg birth weight and met criteria based on alveolar-arterial oxygen pressure difference (P[A-a]O2), barotrauma criteria, or both P(A-a)O2 and barotrauma criteria. All infants had persistent pulmonary hypertension. Two patients also had hyaline membrane disease and one also had asphyxia and meconium aspiration. All patients were treated with conventional therapy and all survived. There were no patients who met criteria for ECMO and died and there were no patients referred for ECMO during this period. Published criteria for ECMO estimate a control group mortality rate of 80% to 94%. Mortality for this series was 0%. (Ninety-five percent confidence interval for mortality in a group of five survivors is 0% to 45%.) Controlled trials of ECMO were not done initially because it was considered unethical. This series shows that historical mortality rates are no longer valid and that controlled trials must be done.