ID 171 – The revised cerebral recovery index: A tool for prognostication after cardiac arrest

Abstract

Objective Continuous EEG monitoring seems a promising tool for prognostication after cardiac arrest. We aimed to improve the cerebral recovery index (CRI), an outcome prediction algorithm, by adding automatic detection of burst-suppression with identical bursts and evaluating the relevance of features in the current CRI. Methods Automatic detection of burst-suppression with identical bursts was developed on a test set of 46 patients with burst suppression after cardiac arrest, evaluated in a test set of 19 EEGs and implemented in the CRI. The CRI was optimized in a training set of 88 patients and evaluated in a test set of another 88 patients. Results Burst-suppression with identical bursts was detected with 100% specificity and 67% sensitivity. Adding burst identity and removing two of the four features optimized the CRI. Higher sensitivities were achieved for good (14% vs. 5% at 12–24 h post resuscitation) but especially poor outcome (30% vs. 12% at 12–24 h, 38% vs. 33% at 24–36 h and 63% vs. 54% at 36–48 h) compared to the old CRI, all with a specificity of 100%. Conclusion The revised CRI predicts especially poor outcome with a high sensitivity and specificity. The improvements are an important step towards clinical implementation.