Timing of intracranial hemorrhage during extracorporeal life support

This study was conducted to determine the timing of intracranial hemorrhage (ICH) in patients on extracorporeal life support (ECLS) to improve the use of the head ultrasound in the detection of ICH. A review was conducted of all neonatal ECLS patients at the neonatal intensive care nursery at Kosair Children's Hospital in Louisville, Kentucky, from May, 1985 through November, 1994 to establish a study group of infants in whom an ICH developed while on ECLS. Thirty infants who had an ICH (excluding subarachnoid hemorrhage and infarction) on ECLS were included in the study. Data were collected that included patients demographics, age at initiation of ECLS, duration of ECLS, type of ECLS support (venoarterial or venovenous), oxygenation index and last arterial blood gas before ECLS, hours of ECLS before ICH, and grade of ICH. ICH occurred in 9.9% of the neonatal patients requiring ECLS. These included 8 infants with a Grade I bleed, 1 infant with a Grade II, 4 infants with a Grade III, and 17 infants with a Grade IV. Ten of the 30 patients had sepsis as their primary diagnosis, and these infants were more likely to have an ICH while on ECLS compared to nonseptic infants (p < 0.02). The Kaplan-Meier curve showed that 50% of ICHs occurred in the first 24 hours of ECLS, 75% by 48 hours, and that 85% of ICHs occurred within 72 hours of initiation of bypass. There was no difference in timing of ICH in the septic infants compared to the nonseptic infants. The late occurring bleeds (> 72 hours) were all associated with significant neurologic changes or with multiorgan failure. It is concluded that daily head ultrasounds should be performed during the first 3 days of ECLS because most ICHs (85%) occur in the first 72 hours of cardiopulmonary bypass. In this era of cost containment, subsequent head ultrasounds should be obtained with changes in the infant's neurologic status or with the development of multiorgan failure.