Prognostic Value of Electroencephalography in Hypothermia-Treated Neonates With Hypoxic-Ischemic Encephalopathy: A Meta-Analysis

Background: Electroencephalography (EEG) background activity is associated with neurological outcome in neonates with hypoxic-ischemic encephalopathy. There is uncertainty about the prognostic value of EEG background activity after hypothermia was introduced. Methods: Searches were made on Pubmed, Embase, and the Cochrane Library, from inception to March 1, 2018. Pooled sensitivities and specificities were calculated to assess the diagnostic power of burst suppression, low voltage, and flat trace background activities in the prediction of an adverse neurological outcome in the follow-up period in hypothermia-treated neonates with hypoxic-ischemic encephalopathy. 12 was used to assess heterogeneity, and meta-regression was done to explore the source of heterogeneity. Results: Eighteen studies with 940 neonates were included. Pooled sensitivities and specificities in predicting the combination of death and neurodevelopmental impairment were burst suppression (sensitivity 0.87 [95% confidence interval (CI) 0.79 to 0.93], specificity 0.60 [95% CI 0.44 to 0.741), low voltage (sensitivity 0.84 [0.75 to 0.90], specificity 0.80 [0.58 to 0.921), and flat trace (sensitivity 0.85 [0.75 to 0.92], specificity 0.94 [0.77 to 0.99]). Subgroup analysis revealed the sensitivities of background patterns obtained after 24 hours of life were higher than those within age 24 hours, whereas the specificities were just the reverse. Flat trace performed best on sensitivity 0.93 (0.60 to 0.99) and specificity 0.90 (0.64 to 0.98) in predicting death. Burst suppression demonstrated the highest sensitivity 0.87 (0.58 to 0.97) and flat trace performed best on specificity 0.85 (0.60 to 0.96) in predicting neurodevelopmental impairment. Conclusions: EEG background activity is predictive of long-term neurological outcome in hypothermia treated neonates with hypoxic-ischemic encephalopathy. Burst suppression, low voltage, and flat trace are potential predictors of death or neurodevelopmental impairment. (C) 2018 Elsevier Inc. All rights reserved.