Early detection of preterm intraventricular hemorrhage from clinical electroencephalography

Abstract

Objectives: Intraventricular hemorrhage is a common neurologic complication of extremely preterm birth and leads to lifelong neurodevelopmental disabilities. Early bedside detection of intraventricular hemorrhage is crucial to enabling timely interventions. We sought to detect early markers of brain activity that preempt the occurrence of intraventricular hemorrhage in extremely preterm infants during the first postnatal days. Design: Cross-sectional study. Setting: Level III neonatal ICU. Patients: Twenty-five extremely preterm infants (22-28 wk gestational age). Measurements and Main Results: We quantitatively assessed electroencephalography in the first 72 hours of postnatal life, focusing on the electrical burst activity of the preterm. Cranial ultrasound was performed on day 1 (0-24 hr) and day 3 (48-72 hr). Outcomes were categorized into three classes: 1) no intraventricular hemorrhage (grade 0); 2) mild-moderate intraventricular hemorrhage (grades 1-2, i.e., germinal matrix hemorrhages or intraventricular hemorrhage without ventricular dilatation, respectively); and 3) severe intraventricular hemorrhage (grades 3-4, i.e., intraventricular hemorrhage with ventricular dilatation or intraparenchymal involvement). Quantitative assessment of electroencephalography burst shapes was used to preempt the occurrence and severity of intraventricular hemorrhage as detected by ultrasound. The shapes of electroencephalography bursts found in the intraventricular hemorrhage infants were significantly sharper (F = 13.78; p < 0.0001) and less symmetric (F = 6.91; p < 0.015) than in preterm infants without intraventricular hemorrhage. Diagnostic discrimination of intraventricular hemorrhage infants using measures of burst symmetry and sharpness yielded high truepositive rates (82% and 88%, respectively) and low false-positive rates (19% and 8%). Conventional electroencephalography measures of interburst intervals and burst counts were not significantly associated with intraventricular hemorrhage. Conclusions: Detection of intraventricular hemorrhage during the first postnatal days is possible from bedside measures of brain activity prior to ultrasound confirmation of intraventricular hemorrhage. Significantly, our novel automated assessment of electroencephalography preempts the occurrence of intraventricular hemorrhage in the extremely preterm. Early bedside detection of intraventricular hemorrhage holds promise for advancing individual care, targeted therapeutic trials, and understanding mechanisms of brain injury in neonates.