Objective: Although ECMO has decreased mortality in neonates with severe cardiorespiratory failure, neurologic sequelae occur in 10-20% of survivors. Cannulation of the right common carotid artery for VA ECMO may contribute to risk for cerebrovascular complications, while the VV procedure may pre-select infants with less compromised cardiac function. To determine whether neonatal clinical variables and neurologic sequelae were different in these two groups, we compared presentation, EEG recordings, CT or MRI scans, and neurodevelopmental outcome at 10-18 months in survivors of VA vs VV ECMO treated during a four year period.

Methods: From February 1992 through January 1996, 151 infants were treated non-randomly with VA or VV ECMO. Fifty-eight were excluded because of complex congenital anomalies (3), need for complex surgery (2), diaphragmatic hernia and lung hypoplasia (14), age >28 days at ECMO onset(7), or early death (32). EEG's before and/or during ECMO were classified as normal, or as mildly, moderately or markedly abnormal. Bayley Scales of Infant Development (BSID) were assessed at 10-18 months in 93 survivors (VA=47, VV=46). Of these, 72 had at least one EEG before or during ECMO, and 66 had neonatal cerebral imaging (MRI or CT).

Results: There were no significant differences in inborn/outborn rates, distribution of primary diagnoses, sex or race, or lowest PaO2 pre-ECMO between VA and VV treated neonates. Onset of ECMO was later (63 vs 39 hours, P<.05), and duration of ECMO longer(144 vs 120 hours, P<.05) for VA than for VV bypass. BSID scores were similar in the two groups: A mean mental development index of 96 in VA vs 102 VV (P=.071), and a mean psychomotor development index of 90 in VA vs 92 VV treated neonates (P=.663). ANOVA revealed no differences in mental or psychomotor scores predicted by EEG results prior to or during ECMO, or neonatal cerebral imaging studies for either the VA or VV groups.

Conclusions: Although VA ECMO is often considered optimal for supporting critically ill babies (eg those requiring cardiac inotropes pre-ECMO), VV subjects in this study met the same criteria for initiating bypass. No significant differences in neonatal EEG and neuroimaging abnormalities, or neurodevelopmental scores at 10-18 months were observed in VV compared to VA ECMO survivors. Venovenous ECMO may, therefore, be the preferred (least invasive) access for providing ECMO support in the majority of neonates with severe cardiopulmonary failure.