Prognostic Tests in Term Neonates With Hypoxic-Ischemic Encephalopathy: A Systematic Review

Abstract

BACKGROUND AND OBJECTIVE: Hypoxic-ischemic encephalopathy (HIE) after perinatal asphyxia in term neonates causes long-term neurologic sequelae or death. A reliable evidence-based prognosis is essential. The study goal was to investigate the prognostic value of currently used clinical tests in neonatal patients with perinatal asphyxia and HIE. METHODS: Searches were made on MEDLINE, Embase, Central, and CINAHL for studies occurring between January 1980 and November 2011. Studies were included if they (1) evaluated outcome in term infants with perinatal asphyxia and HIE, (2) evaluated prognostic tests, and (3) reported outcome at a minimal follow-up age of 18 months. Study selection, assessment of methodologic quality, and data extraction were performed by 3 independent reviewers. Pooled sensitivities and specificities of investigated tests were calculated when possible. RESULTS: Of the 259 relevant studies, 29 were included describing 13 prognostic tests conducted 1631 times in 1306 term neonates. A considerable heterogeneity was noted in test performance, cut-off values, and outcome measures. The most promising tests were amplitude-integrated electroencephalography (sensitivity 0.93, [95% confidence interval 0.78–0.98]; specificity 0.90 [0.60–0.98]), EEG (sensitivity 0.92 [0.66–0.99]; specificity 0.83 [0.64–0.93]), and visual evoked potentials (sensitivity 0.90 [0.74–0.97]; specificity 0.92 [0.68–0.98]). In imaging, diffusion weighted MRI performed best on specificity (0.89 [0.62–0.98]) and T1/T2-weighted MRI performed best on sensitivity (0.98 [0.80–1.00]). Magnetic resonance spectroscopy demonstrated a sensitivity of 0.75 (0.26–0.96) with poor specificity (0.58 [0.23–0.87]). CONCLUSIONS: This evidence suggests an important role for amplitude-integrated electroencephalography, EEG, visual evoked potentials, and diffusion weighted and conventional MRI. Given the heterogeneity in the tests’ performance and outcomes studied, well-designed large prospective studies are needed.