Extracorporeal membrane oxygenation selection criteria: partial pressure of arterial oxygen versus alveolar-arterial oxygen gradient.

Extracorporeal membrane oxygenation (ECMO) has dramatically increased the survival rate of hypoxemic neonates who are unresponsive to maximum conventional medical therapy. Because ECMO involves multiple risks, including ligation of the right common carotid artery and right internal jugular vein, ECMO candidates should be neurologically intact neonates with a high probability of death despite maximum conventional ventilatory support. Currently, criteria based on the calculated alveolar-arterial oxygen gradient (A-aDO2) have replaced the neonatal pulmonary insufficiency index for predicting mortality and, thus, ECMO eligibility. A retrospective review of death prediction for the 26 months prior to the initiation of an ECMO program revealed a sensitivity of 67% and a specificity of 96% using the criterion of a PaO2 of less than 50 mm Hg for four hours. An equivalent A-aDO2 criterion of greater than or equal to 630 for four hours produced a sensitivity of 61% and a specificity of 96%. Prediction of mortality in neonates with sepsis was poor regardless of the criteria used. Excluding the deaths due to sepsis increased the sensitivity to 86% and 79% using criteria based on PaO2 and A-aDO2, respectively. It is concluded that the use of criteria based on PaO2 is equivalent to criteria based on A-aDO2 for predicting mortality. Criteria based on PaO2 may, however, decrease both the false-negative rate (patients with an elevated PCO2) and the false-positive rate (patients with intentionally induced hypocarbia secondary to hyperventilation alkalosis).