Marleen C. Tjepkema-Cloostermans

Early EEG contributes to multimodal outcome prediction of postanoxic coma.

Objectives: Early identification of potential recovery of postanoxic coma is a major challenge. We studied the additional predictive value of EEG. Methods: Two hundred seventy-seven consecutive comatose patients after cardiac arrest were included in a prospective cohort study on 2 intensive care units. Continuous EEG was measured during the first 3 days. EEGs were classified as unfavorable (isoelectric, low-voltage, burst-suppression with identical bursts), intermediate, or favorable (continuous patterns), at 12, 24, 48, and 72 hours. Outcome was dichotomized as good or poor. Resuscitation, demographic, clinical, somatosensory evoked potential, and EEG measures were related to outcome at 6 months using logistic regression analysis. Analyses of diagnostic accuracy included receiver operating characteristics and calculation of predictive values. Results: Poor outcome occurred in 149 patients (54%). Single measures unequivocally predicting poor outcome were an unfavorable EEG pattern at 24 hours, absent pupillary light responses at 48 hours, and absent somatosensory evoked potentials at 72 hours. Together, these had a specificity of 100% and a sensitivity of 50%. For the remaining 203 patients, who were still in the “gray zone” at 72 hours, a predictive model including unfavorable EEG patterns at 12 hours, absent or extensor motor response to pain at 72 hours, and higher age had an area under the curve of 0.90 (95% confidence interval 0.84–0.96). Favorable EEG patterns at 12 hours were strongly associated with good outcome. EEG beyond 24 hours had no additional predictive value. Conclusions: EEG within 24 hours is a robust contributor to prediction of poor or good outcome of comatose patients after cardiac arrest.

Electroencephalogram Predicts Outcome in Patients With Postanoxic Coma During Mild Therapeutic Hypothermia

Objective:To assess the value of electroencephalogram for prediction of outcome of comatose patients after cardiac arrest treated with mild therapeutic hypothermia. Design:Prospective cohort study. Setting:Medical ICU. Patients:One hundred forty-two patients with postanoxic encephalopathy after cardiac arrest, who were treated with mild therapeutic hypothermia. Measurements and Main Results:Continuous electroencephalogram was recorded during the first 5 days of ICU admission. Visual classification of electroencephalogram patterns was performed in 5-minute epochs at 12 and 24 hours after cardiac arrest by two independent observers, blinded for patients’ conditions and outcomes. Patterns were classified as isoelectric, low voltage, epileptiform, burst-suppression, diffusely slowed, or normal. Burst-suppression was subdivided into patterns with and without identical bursts. Primary outcome measure was the neurologic outcome based on each patient’s best achieved Cerebral Performance Category score within 6 months after inclusion. 67 patients (47%) had favorable outcome (Cerebral Performance Category, 1–2). In patients with favorable outcome, electroencephalogram patterns improved within 24 hours after cardiac arrest, mostly toward diffusely slowed or normal. At 24 hours after cardiac arrest, the combined group of isoelectric, low voltage, and “burst-suppression with identical bursts” was associated with poor outcome with a sensitivity of 48% (95% CI, 35–61) and a specificity of 100% (95% CI, 94–100). At 12 hours, normal or diffusely slowed electroencephalogram patterns were associated with good outcome with a sensitivity of 56% (95% CI, 41–70) and a specificity of 96% (95% CI, 86–100). Conclusions:Electroencephalogram allows reliable prediction of both good and poor neurologic outcome of patients with postanoxic encephalopathy treated with mild therapeutic hypothermia within 24 hours after cardiac arrest.

Effect of Burst Stimulation Evaluated in Patients Familiar With Spinal Cord Stimulation.

Spinal cord stimulation (SCS) is used for treating intractable neuropathic pain. It has been suggested that burst SCS (five pulses at 500 Hz, delivered 40 times per second) suppresses neuropathic pain at least as well as conventional tonic SCS, but without evoking paraesthesia. The efficacy of paraesthesia‐free high and low amplitude burst SCS for the treatment of neuropathic pain in patients who are already familiar with tonic SCS was evaluated.

Cerebral Recovery Index: Reliable Help for Prediction of Neurologic Outcome After Cardiac Arrest

Objective: Early electroencephalography measures contribute to outcome prediction of comatose patients after cardiac arrest. We present predictive values of a new cerebral recovery index, based on a combination of quantitative electroencephalography measures, extracted every hour, and combined by the use of a random forest classifier. Design: Prospective observational cohort study. Setting: Medical ICU of two large teaching hospitals in the Netherlands. Patients: Two hundred eighty-three consecutive comatose patients after cardiac arrest. Interventions: None. Measurements and Main Results: Continuous electroencephalography was recorded during the first 3 days. Outcome at 6 months was dichotomized as good (Cerebral Performance Category 1–2, no or moderate disability) or poor (Cerebral Performance Category 3–5, severe disability, comatose, or death). Nine quantitative electroencephalography measures were extracted. Patients were randomly divided over a training and validation set. Within the training set, a random forest classifier was fitted for each hour after cardiac arrest. Diagnostic accuracy was evaluated in the validation set. The relative contributions of resuscitation parameters and patient characteristics were evaluated. The cerebral recovery index ranges from 0 (prediction of death) to 1 (prediction of full recovery). Poor outcome could be predicted at a threshold of 0.34 without false positives at a sensitivity of 56% at 12 hours after cardiac arrest. At 24 hours, sensitivity of 65% with a false positive rate of 6% was obtained. Good neurologic outcome could be predicted with sensitivities of 63% and 58% at a false positive rate of 6% and 7% at 12 and 24 hours, respectively. Adding patient characteristics was of limited additional predictive value. Conclusions: A cerebral recovery index based on a combination of intermittently extracted, optimally combined quantitative electroencephalography measures provides unequalled prognostic value for comatose patients after cardiac arrest and enables bedside EEG interpretation of unexperienced readers.

Somatosensory Evoked Potentials in Patients with Hypoxic-Ischemic Brain Injury

Abstract Predicting the future of patients with hypoxic‐ischemic encephalopathy after successful cardiopulmonary resuscitation is often difficult. Registration of the median nerve somatosensory evoked potential (SSEP) can assist in the neurologic evaluation in these patients. In this article, the authors discuss the principles, applications, and limitations of SSEP registration in the intensive care unit, with a focus on prognostication. Registration of the SSEP is a very reliable and reproducible method, if it is performed and interpreted correctly. During SSEP recordings, great care should be taken to improve the signal‐to‐noise ratio. If the noise level is too high, the peripheral responses are abnormal or the response is not reproducible in a second set of stimuli; therefore, interpretation of the SSEPs cannot be done reliably. A bilaterally absent cortical SSEP response is a very reliable predictor of poor neurologic outcome in patients with HIE. It has a high specificity, but a low sensitivity, indicating that present cortical responses are a weak predictor of a good recovery. Further research is being done to increase the sensitivity. Somatosensory evoked potentials can be used in a multimodal approach for prognostication of outcome.

Predicting outcome in postanoxic coma: are ten EEG electrodes enough?

Introduction: Increasing evidence supports that early EEG recordings reliably contribute to outcome prediction in comatose patients with postanoxic encephalopathy. As postanoxic encephalopathy typically results in generalized EEG abnormalities, spatial resolution of a small number of electrodes is likely sufficient, which will reduce set-up time. Here, the authors compare a reduced and a 21-channel EEG for outcome prediction. Methods: EEG recordings from 142 prospectively collected patients with postanoxic encephalopathy were reassessed by two independent reviewers using a reduced (10 electrodes) bipolar montage. Classification and prognostic accuracy were compared with the full (21 electrodes) montage. The full montage consensus was considered Gold Standard. Results: Sixty-seven patients (47%) had good outcome. The agreement between the individual reviewers using the reduced montage and the Gold Standard score was good (&kgr; = 0.75–0.79). The interobserver agreement was not affected by reducing the number of electrodes (&kgr; = 0.78 for the reduced montage vs. 0.71 for the full montage). An isoelectric, low-voltage, or burst-suppression with identical bursts pattern at 24 hours invariably predicted poor outcome in both montages, with similar prognostic accuracy. A diffusely slowed or normal EEG pattern at 12 hours was associated with good outcome in both montages. Conclusions: Reducing the number of electrodes from 21 to 10 does not affect EEG classification or prognostic accuracy in patients with postanoxic coma.

Predicting success of vagus nerve stimulation (VNS) from EEG symmetry

PURPOSE Vagus nerve stimulation (VNS) has shown to be an effective treatment for drug resistant epilepsy, with achieving more than 50% seizure reduction in one third of the treated patients. In order to predict which patients will profit from VNS, we previously found that a low pairwise derived Brain Symmetry Index (pdBSI) could potentially predict good responders to VNS treatment. These findings however have to be validated before they can be generalized. METHODS 39 patients (age 18-68 years) with medically intractable epilepsy who were referred for an implanted VNS system were included. Routine EEG registrations, recorded before implantation, were analyzed. Artefact-free epochs with eyes open and eyes closed were quantitatively analyzed. The pdBSI was tested for relation with VNS outcome one year after surgery. RESULTS Twenty-three patients (59%) obtained a reduction in seizure frequency, of whom ten (26%) had a reduction of at least 50% (good responders) and thirteen (33%) a reduction of less than 50% (moderate responders). Sixteen patients without seizure reduction are defined as non-responders. No significant differences were found in the pdBSI of good responders (mean 0.27), moderate responders (mean 0.26) and non-responders (mean 0.25) (p>0.05). Besides seizure reduction, many patients (56%) reported additional positive effects of VNS in terms of seizure duration, seizure intensity and/or postictal recovery. CONCLUSION EEG features that correlate with VNS therapy outcome may enable better patient selection and prevent unnecessary VNS surgery. Contrary to earlier findings, this validation study suggests that pdBSI might not be helpful to predict VNS therapy outcome.

Detecting cortical spreading depolarization with full band scalp electroencephalography : An illusion?

Introduction There is strong evidence suggesting detrimental effects of cortical spreading depolarization (CSD) in patients with acute ischemic stroke and severe traumatic brain injury. Previous studies implicated scalp electroencephalography (EEG) features to be correlates of CSD based on retrospective analysis of EEG epochs after having detected “CSD” in time aligned electrocorticography. We studied the feasibility of CSD detection in a prospective cohort study with continuous EEG in 18 patients with acute ischemic stroke and 18 with acute severe traumatic brain injury. Methods Full band EEG with 21 silver/silver chloride electrodes was started within 48 h since symptom onset. Five additional electrodes were used above the infarct. We visually analyzed all raw EEG data in epochs of 1 h. Inspection was directed at detection of the typical combination of CSD characteristics, i.e., (i) a large slow potential change (SPC) accompanied by a simultaneous amplitude depression of >1Hz activity, (ii) focal presentation, and (iii) spread reflected as appearance on neighboring electrodes with a delay. Results In 3,035 one-hour EEG epochs, infraslow activity (ISA) was present in half to three quarters of the registration time. Typically, activity was intermittent with amplitudes of 40–220 µV, approximately half was oscillatory. There was no specific spatial distribution. Relevant changes of ISA were always visible in multiple electrodes, and not focal, as expected in CSD. ISA appearing as “SPC” was mostly associated with an amplitude increase of faster activities, and never with suppression. In all patients, depressions of spontaneous brain activity occurred. However, these were not accompanied by simultaneous SPC, occurred simultaneously on all channels, and were not focal, let alone spread, as expected in CSD. Conclusion With full band scalp EEG in patients with cortical ischemic stroke or traumatic brain injury, we observed various ISA, probably modulating cortical excitability. However, we were unable to identify unambiguous characteristics of CSD.

Deep Learning for outcome prediction of postanoxic coma

Electroencephalography (EEG) is increasingly used to assist in outcome prediction for patients with a postanoxic coma after cardiac arrest. Current literature shows that neurological outcome is invariably poor if the EEG remains iso-electric or low-voltage at 24 h after cardiac arrest or if it shows burst-suppression with identical bursts; such patterns are observed in approximately 30-50% of patients. Return of continuous EEG rhythms within 12 h after cardiac arrest predicts good neurological outcome with sensitivities in the range of 30 to 50% at specificities near 100%. In previous work, we reported on the Cerebral Recovery Index to assist in the visual assessment of the EEG. In this paper, we explore a deep learning approach, using a convolutional neural network for outcome prediction in patients with a postanoxic encephalopathy. Using EEGs from 287 patients at 12 h after cardiac arrest and 399 patients at 24 h after cardiac arrest, we trained and validated a convolutional neural network with raw EEG data (18 channels, longitudinal bipolar montage). As the outcome measure, we used the Cerebral Performance Category scale (CPC), dichotomized between good (CPC score 1-2) and poor outcome (CPC score 3-5). Using 5 minute artifact-free epochs from the continuous EEG recordings partitioned into 10 s snippets, we trained the convolutional neural network using 80% of the patients. Validation was performed with EEGs from the remaining 20% of patients. Outcome prediction was most accurate at 12 h after cardiac arrest, with a sensitivity of 58% at a specificity of 100% for the prediction of poor outcome. Good neurological outcome could be predicted at 12 h after cardiac arrest with a sensitivity of 58% at a specificity of 97%. In conclusion, we present a classifier for the prediction of neurological outcome after cardiac arrest, based on a convolutional neural network, providing reliable and objective prognostic information.

Early electroencephalography dynamics after cardiac arrest

To the Editor:With interest we read the article published in a recent issue of Critical Care Medicine by Rossetti et al (1), in which they confirm that early electroencephalography (EEG) contributes to outcome prediction of postanoxic coma. We concede the importance of external validation of previou