The electroencephalogram (EEG) contains information that is useful for the prediction of both poor and good neurological outcome in patients with postanoxic encephalopathy after cardiac arrest treated with mild hypothermia. The combined group of iso-electric, low voltage or burst-suppression patterns with identical bursts recorded at 24 hours after cardiac arrest reliably predicts poor neurological outcome with a sensitivity of 48% (CI: 35–60%) and a specificity of 100% (CI: 94–100%) (Chapters 2 and 4). In contrast, the sensitivity for bilateral SSEP absence was only SSEP 24% (CI: 10–44%) (Chapter 2). “Burst-suppression with identical bursts” is a distinct pathological EEG pattern characterized by bursts with a high similarity. Burst-suppression with identical bursts can only be seen after diffuse cerebral ischemia and is inevitably associated with poor neurological outcome (Chapter 3). In addition, normal or diffusely slowed EEG patterns at 12 hours after cardiac arrest are associated with a good neurological outcome with a sensitivity of 57% (CI: 42–71%) and a specificity of 96% (CI: 86–100%) (Chapters 2 and 4). The increased use of EEG monitoring leads to an increased detection of electrographic seizures and status epilepticus. However, it is currently unknown if and how aggressive patients with these patterns should be treated. In our retrospective study, moderate treatment with anti-epileptic drugs did not improve outcome of patients with electrographic status epilepticus after cardiac arrest (Chapter 5). Quantitative EEG analysis can assist in decreasing the time needed for visual interpretation of the long EEG recordings and in making the visual analysis more objective. We implemented two computer algorithms that can assist in the interpretation of long EEG recordings. The first system can be used for real-time classification of the EEG in critically ill patients. This system has an accuracy of 85–88% (Chapter 6). Secondly, we introduced the “Cerebral Recovery Index (CRI)”, which is a score ranging from 0 to 1, that can be used for the grading of EEGs in patients with postanoxic encephalopathy. At 24 hours after cardiac arrest, a CRI < 0.29 was always associated with poor neurological outcome, with a sensitivity of 55%(CI: 32–76%) and a specificity of 100% (CI: 86–100%). At the same time point a CRI > 0.69 predicted good neurological outcome, with a sensitivity of 25%(CI: 10–47%) and a specificity of 100% (CI: 85–100%) in the test set (Chapter 7). Finally, we showed by using a computational model that generalized periodic discharges, an EEG pattern that can be observed in patients with post-anoxic encephalopathy, can be explained as a reflection of selective ischemic damage of glutamatergic synapses (Chapter 8).
|Award date||10 Jan 2014|
|Place of Publication||Enschede|
|Publication status||Published - 10 Jan 2014|