Combined EEG, aEEG and MRI biomarkers in the neonatal period to predict neurodevelopmental outcomes in infants with neonatal encephalopathy: a diagnostic test accuracy systematic review and Bayesian meta-analysis

Objective To assess the predictive accuracy of early neurophysiological and neuroimaging biomarkers, alone and in combination, for adverse neurodevelopmental disorders in term-born infants with neonatal encephalopathy (NE). Design Systematic review and meta-analysis conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy guidelines. Eligible studies included infants born at term with NE who underwent amplitude-integrated EEG (aEEG) or EEG and MRI of the brain within the first month of life. Adverse outcomes, assessed at 18–36 months of age, were defined as cerebral palsy, postneonatal epilepsy, severe hearing or visual impairment, moderate-to-severe developmental delay, or death attributable to NE. Searches were conducted in MEDLINE, CINAHL, Embase and Web of Science from database inception to 10 June 2025; risk of bias of included studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-Comparative (QUADAS-C) tool. Main outcome measures Sensitivity, specificity and diagnostic odds ratio (DOR) of abnormal aEEG background, EEG background, EEG seizures and MRI injury, individually and in combination, for predicting adverse outcomes, pooled using Bayesian bivariate random effects meta-analyses. Results 27 studies including 1843 infants were analysed. MRI injury was the individual predictor with higher DOR estimate (31.01, 95% CI 15.07 to 72.82), followed by abnormal EEG background (16.84, 95% CI 5.88 to 50.59), while abnormal aEEG background and EEG seizures performed less well (7.99, 95% CI 2.40 to 33.00; 4.46, 95% CI 1.86 to 11.42). Combining EEG background with MRI injury improved DOR (78.59, 95% CI 19.72 to 321.36) and specificity (93.8%, 95% CI 85.2% to 97.9%) compared with MRI alone. Conclusions MRI is a strong individual predictor of adverse outcomes in NE. Combining it with early EEG improves prognostic accuracy and may better support clinical decision-making. PROSPERO registration number CRD42024585816.