Standardized EEG interpretation accurately predicts prognosis after cardiac arrest

Predicting Neurological Outcome From Electroencephalogram Dynamics in Comatose Patients After Cardiac Arrest With Deep Learning

OBJECTIVE

Most cardiac arrest patients who are successfully resuscitated are initially comatose due to hypoxic-ischemic brain injury. Quantitative electroencephalography (EEG) provides valuable prognostic information. However, prior approaches largely rely on snapshots of the EEG, without taking advantage of temporal information.

METHODS

We present a recurrent deep neural network with the goal of capturing temporal dynamics from longitudinal EEG data to predict long-term neurological outcomes. We utilized a large international dataset of continuous EEG recordings from 1,038 cardiac arrest patients from seven hospitals in Europe and the US. Poor outcome was defined as a Cerebral Performance Category (CPC) score of 3-5, and good outcome as CPC score 0-2 at 3 to 6-months after cardiac arrest. Model performance is evaluated using 5-fold cross validation.

RESULTS

The proposed approach provides predictions which improve over time, beginning from an area under the receiver operating characteristic curve (AUC-ROC) of 0.78 (95% CI: 0.72-0.81) at 12 hours, and reaching 0.88 (95% CI: 0.85-0.91) by 66 h after cardiac arrest. At 66 h, (sensitivity, specificity) points of interest on the ROC curve for predicting poor outcomes were (32,99)%, (55,95)%, and (62,90)%, (99,23)%, (95,47)%, and (90,62)%; whereas for predicting good outcome, the corresponding operating points were (17,99)%, (47,95)%, (62,90)%, (99,19)%, (95,48)%, (70,90)%. Moreover, the model provides predicted probabilities that closely match the observed frequencies of good and poor outcomes (calibration error 0.04).

CONCLUSIONS AND SIGNIFICANCE

These findings suggest that accounting for EEG trend information can substantially improve prediction of neurologic outcomes for patients with coma following cardiac arrest.

Preserved Electroencephalogram Power and Global Synchronization Predict Better Neurological Outcome in Sudden Cardiac Arrest Survivors

Quantitative EEG (qEEG) delineates complex brain activities. Global field synchronization (GFS) is one multichannel EEG analysis that measures global functional connectivity through quantification of synchronization between signals. We hypothesized that preservation of global functional connectivity of brain activity might be a surrogate marker for good outcome in sudden cardiac arrest (SCA) survivors. In addition, we examined the relation of phase coherence and GFS in a mathematical approach. We retrospectively collected EEG data of SCA survivors in one academic medical center. We included 75 comatose patients who were resuscitated following in-hospital or out-of-hospital nontraumatic cardiac arrest between 2013 and 2017 in the intensive care unit (ICU) of National Taiwan University Hospital (NTUH). Twelve patients (16%) were defined as good outcome (GO) (CPC 1-2). The mean age in the GO group was low (51.6 ± 15.7 vs. 68.1 ± 12.9, p < 0.001). We analyzed standard EEG power, computed EEG GFS, and assessed the cerebral performance category (CPC) score 3 months after discharge. The alpha band showed the highest discrimination ability (area under curve [AUC] = 0.78) to predict GO using power. The alpha band of GFS showed the highest AUC value (0.8) to predict GO in GFS. Furthermore, by combining EEG power + GFS, the alpha band showed the best prediction value (AUC 0.86) in predicting GO. The sensitivity of EEG power + GFS was 73%, specificity was 93%, PPV was 0.67%, and NPV was 0.94%. In conclusion, by combining GFS and EEG power analysis, the neurological outcome of the nontraumatic cardiac arrest survivor can be well-predicted. Furthermore, we proved from a mathematical point of view that although both amplitude and phase contribute to obtaining GFS, the interference in phase variation drastically changes the possibility of generating a good GFS score.

Neurological Prognostication After Hypoglycemic Coma: Role of Clinical and EEG Findings

BACKGROUND

Hypoglycemic coma (HC) is an uncommon but severe clinical condition associated with poor neurological outcome. There is a dearth of robust neurological prognostic factors after HC. On the other hand, there is an increasing body of literature on reliable prognostic markers in the postanoxic coma, a similar-albeit not identical-situation. The objective of this study was thus to investigate the use and predictive value of these markers in HC.

METHODS

We conducted a retrospective, multicenter, cohort study within five centers of the Critical Care EEG Monitoring Research Consortium. We queried our electroencephalography (EEG) databases to identify all patients undergoing continuous EEG monitoring after admission to an intensive care unit with HC (defined as Glasgow Coma Scale < 8 on admission and a first blood glucose level < 50 mg/dL or not documented but in an obvious clinical context) between 01/01/2010 and 12/31/2020. We studied the association of findings at neurological examination (Glasgow Coma Scale motor subscale, pupillary light and corneal reflexes) and at continuous EEG monitoring(highly malignant patterns, reactivity, periodic discharges, seizures) with best neurological outcome within 3 months after hospital discharge, defined by the Cerebral Performance Category as favorable (1-3: recovery of consciousness) versus unfavorable (4-5: lack of recovery of consciousness).

RESULTS

We identified 60 patients (30 [50%] women; age 62 [51-72] years). Thirty-one and 29 patients had a favorable and unfavorable outcome, respectively. The presence of pupillary reflexes (24 [100%] vs. 17 [81%]; p value 0.04) and a motor subscore > 2 (22 [92%] vs. 12 [63%]; p value 0.03) at 48-72 h were associated with a favorable outcome. A highly malignant EEG pattern was observed in 7 of 29 (24%) patients with unfavorable outcome versus 0 of 31 (0%) with favorable outcome, whereas the presence of EEG reactivity was observed in 28 of 31 (90%) patients with favorable outcome versus 13 of 29 (45%) with unfavorable outcome (p < 0.001 for comparison of all background categories).

CONCLUSIONS

This preliminary study suggests that highly malignant EEG patterns might be reliable prognostic markers of unfavorable outcome after HC. Other EEG findings, including lack of EEG reactivity and seizures and clinical findings appear less accurate. These findings should be replicated in a larger multicenter prospective study.

External validation of the 2020 ERC/ESICM prognostication strategy algorithm after cardiac arrest

Purpose

To assess the performance of the post-cardiac arrest (CA) prognostication strategy algorithm recommended by the European Resuscitation Council (ERC) and the European Society of Intensive Care Medicine (ESICM) in 2020.

Methods

This was a retrospective analysis of the Korean Hypothermia Network Prospective Registry 1.0. Unconscious patients without confounders at day 4 (72–96 h) after return of spontaneous circulation (ROSC) were included. The association between the prognostic factors included in the prognostication strategy algorithm, except status myoclonus and the neurological outcome, was investigated, and finally, the prognostic performance of the prognostication strategy algorithm was evaluated. Poor outcome was defined as cerebral performance categories 3–5 at 6 months after ROSC.

Results

A total of 660 patients were included in the final analysis. Of those, 108 (16.4%) patients had a good neurological outcome at 6 months after CA. The 2020 ERC/ESICM prognostication strategy algorithm identified patients with poor neurological outcome with 60.2% sensitivity (95% CI 55.9–64.4) and 100% specificity (95% CI 93.9–100) among patients who were unconscious or had a GCS_M score ≤ 3 and with 58.2% sensitivity (95% CI 53.9–62.3) and 100% specificity (95% CI 96.6–100) among unconscious patients. When two prognostic factors were combined, any combination of prognostic factors had a false positive rate (FPR) of 0 (95% CI 0–5.6 for combination of no PR/CR and poor CT, 0–30.8 for combination of No SSEP N20 and NSE 60).

Conclusion

The 2020 ERC/ESICM prognostication strategy algorithm predicted poor outcome without an FPR and with sensitivities of 58.2–60.2%. Any combinations of two predictors recommended by ERC/ESICM showed 0% of FPR.

Magnetic Resonance Spectroscopy of Hypoxic-Ischemic Encephalopathy After Cardiac Arrest

BACKGROUND AND OBJECTIVES

To correlate brain metabolites with clinical outcome using magnetic resonance spectroscopy (MRS) in patients undergoing targeted temperature management (TTM) after cardiac arrest and assess their relationships to MRI and EEG variables.

METHODS

A prospective cohort of 50 patients was studied. The primary outcome was coma recovery to follow commands. Comparison of MRS measures in the posterior cingulate gyrus, parietal white matter, basal ganglia, and brainstem were also made to 25 normative controls.

RESULTS

Fourteen of 50 patients achieved coma recovery before hospital discharge. There was a significant decrease in total N-acetylaspartate (NAA/Cr) and an increase in lactate/creatine (Lac/Cr) in patients who did not recover, with changes most prominent in the posterior cingulate gyrus. Patients who recovered had decrease in NAA/Cr as compared to controls. NAA/Cr had a strong monotonic relationship with MRI cortical apparent diffusion coefficient (ADC); Lac level exponentially increased with decreasing ADC. EEG suppression/burst suppression was strongly associated with Lac elevation.

DISCUSSION

NAA and Lac changes are associated with clinical/MRI/EEG changes consistent with hypoxic-ischemic encephalopathy (HIE) and are most prominent in the posterior cingulate gyrus. NAA/Cr decrease observed in patients with good outcomes suggests mild HIE in patients asymptomatic at hospital discharge. The appearance of cortical Lac represents a deterioration of aerobic energy metabolism and is associated with EEG background suppression, synaptic transmission failure, and severe, potentially irreversible HIE.

CLASSIFICATION OF EVIDENCE

This study provides Class IV evidence that in patients undergoing TTM after cardiac arrest, brain MRS-determined decrease in total NAA/Cr and an increase in Lac/Cr are associated with an increased risk of not recovering.

Regional Distribution of Brain Injury After Cardiac Arrest: Clinical and Electrographic Correlates

BACKGROUND AND OBJECTIVES

Disorders of consciousness, EEG background suppression, and epileptic seizures are associated with poor outcome after cardiac arrest. Our objective was to identify the distribution of diffusion MRI-measured anoxic brain injury after cardiac arrest and to define the regional correlates of disorders of consciousness, EEG background suppression, and seizures.

METHODS

We analyzed patients from a single-center database of unresponsive patients who underwent diffusion MRI after cardiac arrest (n = 204). We classified each patient according to recovery of consciousness (command following) before discharge, the most continuous EEG background (burst suppression vs continuous), and the presence or absence of seizures. Anoxic brain injury was measured with the apparent diffusion coefficient (ADC) signal. We identified ADC abnormalities relative to controls without cardiac arrest (n = 48) and used voxel lesion symptom mapping to identify regional associations with disorders of consciousness, EEG background suppression, and seizures. We then used a bootstrapped lasso regression procedure to identify robust, multivariate regional associations with each outcome variable. Last, using area under receiver operating characteristic curves, we then compared the classification ability of the strongest regional associations to that of brain-wide summary measures.

RESULTS

Compared to controls, patients with cardiac arrest demonstrated ADC signal reduction that was most significant in the occipital lobes. Disorders of consciousness were associated with reduced ADC most prominently in the occipital lobes but also in deep structures. Regional injury more accurately classified patients with disorders of consciousness than whole-brain injury. Background suppression mapped to a similar set of brain regions, but regional injury could no better classify patients than whole-brain measures. Seizures were less common in patients with more severe anoxic injury, particularly in those with injury to the lateral temporal white matter.

DISCUSSION

Anoxic brain injury was most prevalent in posterior cerebral regions, and this regional pattern of injury was a better predictor of disorders of consciousness than whole-brain injury measures. EEG background suppression lacked a specific regional association, but patients with injury to the temporal lobe were less likely to have seizures. Regional patterns of anoxic brain injury are relevant to the clinical and electrographic sequelae of cardiac arrest and may hold importance for prognosis.

CLASSIFICATION OF EVIDENCE

This study provides Class IV evidence that disorders of consciousness after cardiac arrest are associated with widely lower ADC values on diffusion MRI and are most strongly associated with reductions in occipital ADC.

SSEP N20 and P25 amplitudes predict poor and good neurologic outcomes after cardiac arrest

Background

To assess in comatose patients after cardiac arrest (CA) if amplitudes of two somatosensory evoked potentials (SSEP) responses, namely, N20-baseline (N20-b) and N20–P25, are predictive of neurological outcome.

Methods

Monocentric prospective study in a tertiary cardiac center between Nov 2019 and July-2021. All patients comatose at 72 h after CA with at least one SSEP recorded were included. The N20-b and N20–P25 amplitudes were automatically measured in microvolts (µV), along with other recommended prognostic markers (status myoclonus, neuron-specific enolase levels at 2 and 3 days, and EEG pattern). We assessed the predictive value of SSEP for neurologic outcome using the best Cerebral Performance Categories (CPC1 or 2 as good outcome) at 3 months (main endpoint) and 6 months (secondary endpoint). Specificity and sensitivity of different thresholds of SSEP amplitudes, alone or in combination with other prognostic markers, were calculated.

Results

Among 82 patients, a poor outcome (CPC 3–5) was observed in 78% of patients at 3 months. The median time to SSEP recording was 3(2–4) days after CA, with a pattern “bilaterally absent” in 19 patients, “unilaterally present” in 4, and “bilaterally present” in 59 patients. The median N20-b amplitudes were different between patients with poor and good outcomes, i.e., 0.93 [0–2.05]µV vs. 1.56 [1.24–2.75]µV, respectively (p < 0.0001), as the median N20–P25 amplitudes (0.57 [0–1.43]µV in poor outcome vs. 2.64 [1.39–3.80]µV in good outcome patients p < 0.0001). An N20-b > 2 µV predicted good outcome with a specificity of 73% and a moderate sensitivity of 39%, although an N20–P25 > 3.2 µV was 93% specific and only 30% sensitive. A low voltage N20-b < 0.88 µV and N20–P25 < 1 µV predicted poor outcome with a high specificity (sp = 94% and 93%, respectively) and a moderate sensitivity (se = 50% and 66%). Association of “bilaterally absent or low voltage SSEP” patterns increased the sensitivity significantly as compared to “bilaterally absent” SSEP alone (se = 58 vs. 30%, p = 0.002) for prediction of poor outcome.

Conclusion

In comatose patient after CA, both N20-b and N20–P25 amplitudes could predict both good and poor outcomes with high specificity but low to moderate sensitivity. Our results suggest that caution is needed regarding SSEP amplitudes in clinical routine, and that these indicators should be used in a multimodal approach for prognostication after cardiac arrest.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13613-022-00999-6.

Prediction of good neurological outcome in comatose survivors of cardiac arrest: a systematic review

PURPOSE

To assess the ability of clinical examination, blood biomarkers, electrophysiology or neuroimaging assessed within 7 days from return of spontaneous circulation (ROSC) to predict good neurological outcome, defined as no, mild, or moderate disability (CPC 1-2 or mRS 0-3) at discharge from intensive care unit or later, in comatose adult survivors from cardiac arrest (CA).

METHODS

PubMed, EMBASE, Web of Science and the Cochrane Database of Systematic Reviews were searched. Sensitivity and specificity for good outcome were calculated for each predictor. The risk of bias was assessed using the QUIPS tool.

RESULTS

A total of 37 studies were included. Due to heterogeneities in recording times, predictor thresholds, and definition of some predictors, meta-analysis was not performed. A withdrawal or localisation motor response to pain immediately or at 72-96 h after ROSC, normal blood values of neuron-specific enolase (NSE) at 24 h-72 h after ROSC, a short-latency somatosensory evoked potentials (SSEPs) N20 wave amplitude > 4 µV or a continuous background without discharges on electroencephalogram (EEG) within 72 h from ROSC, and absent diffusion restriction in the cortex or deep grey matter on MRI on days 2-7 after ROSC predicted good neurological outcome with more than 80% specificity and a sensitivity above 40% in most studies. Most studies had moderate or high risk of bias.

CONCLUSIONS

In comatose cardiac arrest survivors, clinical, biomarker, electrophysiology, and imaging studies identified patients destined to a good neurological outcome with high specificity within the first week after cardiac arrest (CA).

Outcome Prediction of Postanoxic Coma: A Comparison of Automated Electroencephalography Analysis Methods

Background

To compare three computer-assisted quantitative electroencephalography (EEG) prediction models for the outcome prediction of comatose patients after cardiac arrest regarding predictive performance and robustness to artifacts.

Methods

A total of 871 continuous EEGs recorded up to 3 days after cardiac arrest in intensive care units of five teaching hospitals in the Netherlands were retrospectively analyzed. Outcome at 6 months was dichotomized as “good” (Cerebral Performance Category 1–2) or “poor” (Cerebral Performance Category 3–5). Three prediction models were implemented: a logistic regression model using two quantitative features, a random forest model with nine features, and a deep learning model based on a convolutional neural network. Data from two centers were used for training and fivefold cross-validation (n = 663), and data from three other centers were used for external validation (n = 208). Model output was the probability of good outcome. Predictive performances were evaluated by using receiver operating characteristic analysis and the calculation of predictive values. Robustness to artifacts was evaluated by using an artifact rejection algorithm, manually added noise, and randomly flattened channels in the EEG.

Results

The deep learning network showed the best overall predictive performance. On the external test set, poor outcome could be predicted by the deep learning network at 24 h with a sensitivity of 54% (95% confidence interval [CI] 44–64%) at a false positive rate (FPR) of 0% (95% CI 0–2%), significantly higher than the logistic regression (sensitivity 33%, FPR 0%) and random forest models (sensitivity 13%, FPR, 0%) (p < 0.05). Good outcome at 12 h could be predicted by the deep learning network with a sensitivity of 78% (95% CI 52–100%) at a FPR of 12% (95% CI 0–24%) and by the logistic regression model with a sensitivity of 83% (95% CI 83–83%) at a FPR of 3% (95% CI 3–3%), both significantly higher than the random forest model (sensitivity 1%, FPR 0%) (p < 0.05). The results of the deep learning network were the least affected by the presence of artifacts, added white noise, and flat EEG channels.

Conclusions

A deep learning model outperformed logistic regression and random forest models for reliable, robust, EEG-based outcome prediction of comatose patients after cardiac arrest.

Deep learning of early brain imaging to predict post-arrest electroencephalography

INTRODUCTION

Guidelines recommend use of computerized tomography (CT) and electroencephalography (EEG) in post-arrest prognostication. Strong associations between CT and EEG might obviate the need to acquire both modalities. We quantified these associations via deep learning.

METHODS

We performed a single-center, retrospective study including comatose patients hospitalized after cardiac arrest. We extracted brain CT DICOMs, resized and registered each to a standard anatomical atlas, performed skull stripping and windowed images to optimize contrast of the gray-white junction. We classified initial EEG as generalized suppression, other highly pathological findings or benign activity. We extracted clinical information available on presentation from our prospective registry. We trained three machine learning (ML) models to predict EEG from clinical covariates. We used three state-of-the-art approaches to build multi-headed deep learning models using similar model architectures. Finally, we combined the best performing clinical and imaging models. We evaluated discrimination in test sets.

RESULTS

We included 500 patients, of whom 218 (44%) had benign EEG findings, 135 (27%) showed generalized suppression and 147 (29%) had other highly pathological findings that were most commonly (93%) burst suppression with identical bursts. Clinical ML models had moderate discrimination (test set AUCs 0.73-0.80). Image-based deep learning performed worse (test set AUCs 0.51-0.69), particularly discriminating benign from highly pathological findings. Adding image-based deep learning to clinical models improved prediction of generalized suppression due to accurate detection of severe cerebral edema.

DISCUSSION

CT and EEG provide complementary information about post-arrest brain injury. Our results do not support selective acquisition of only one of these modalities, except in the most severely injured patients.

Revisiting EEG as part of the multidisciplinary approach to post-cardiac arrest care and prognostication: A review

Since the 1960s, EEG has been used to assess the neurologic function of patients in the hours and days after cardiac arrest. Accurate and reliable prognostication after cardiac arrest is vital for tailoring aggressive patient care for those with a high likelihood of recovery and setting appropriate goals of care for those who have a high likelihood of a poor outcome. Attempts to define EEG's role in this process has evolved over the years. In this review, we provide important historical context about EEG's use, it's perceived unreliability in the post-cardiac arrest patient in the past and provide a detailed analysis of how this role has changed recently. A review of the 71 most recent and highest quality studies demonstrates that the introduction of a uniform classification and a timed approach to EEG analysis have positioned EEG as a complementary tool in the multimodal approach for prognostication. The review was created and intended for medical staff in the intensive care units and emphasizes EEG patterns and timing which portend both favorable and poor prognoses. Also, the review addresses the overall quality of the existing studies and discusses future directions to address the knowledge gaps in this field.

Critical care EEG standardized nomenclature in clinical practice: Strengths, limitations, and outlook on the example of prognostication after cardiac arrest

•

Optimal use of the ACNS nomenclature implies integration of clinical information.

•

Knowledge of pathophysiological mechanisms of EEG patterns may help interpretation.

•

Standardized therapeutic procedures for critical care patients are needed.

We discuss the achievements of the ACNS critical care EEG nomenclature proposed in 2013 and, from a clinical angle, outline some limitations regarding translation into treatment implications. While the recently proposed updated 2021 version of the nomenclature will probable improve some uncertainty areas, a refined understanding of the mechanisms at the origin of the EEG patterns, and a multimodal integration of the nomenclature to the clinical context may help improving the rationale supporting therapeutic procedures. We illustrate these aspects on prognostication after cardiac arrest.

Management of Status Epilepticus and Indications for Inpatient Electroencephalography Monitoring

Status epilepticus (SE) is a neurologic emergency requiring immediate time-sensitive treatment to minimize neuronal injury and systemic complications. Minimizing time to administration of first- and second-line therapy is necessary to optimize the chances of successful seizure termination in generalized convulsive SE (GCSE). The approach to refractory and superrefractory GCSE is less well defined. Multiple agents with differing complementary actions that facilitate seizure termination are recommended. Nonconvulsive SE (NCSE) has a wide range of presentations and approaches to treatment. Continuous electroencephalography is critical to the management of both GCSE and NCSE, while its use for patients without seizure continues to expand.

The accuracy of various neuro-prognostication algorithms and the added value of neurofilament light chain dosage for patients resuscitated from shockable cardiac arrest: An ancillary analysis of the ISOCRATE study

PURPOSE

In current guidelines, neurological prognostication after cardiopulmonary resuscitation is based on a multimodal approach bundled in algorithms. Biomarkers are of particular interest because they are unaffected by interpretation bias. We assessed the predictive value of serum neurofilament light chains (NF-L) in patients with a shockable rhythm who received cardiopulmonary resuscitation, and evaluated the predictive value of a modified algorithm where NF-L dosage is included.

METHODS

All patients who were included participated in the randomized ISOCRATE trial. NF-L values 48 h after ROSC were compared for patients with a good (Cerebral Performance Category (CPC) 1 or 2) and a poor prognosis (CPC 3 to 5 or death). The benefit of adding NF-L dosage to the current guideline algorithm was then assessed for NF-L thresholds of 500 and 1,200 pg/ml as previously described.

RESULTS

NF-L was assayed for 49 patients. In patients with good versus those with poor outcomes, median NF-L values at 48 h were 72 ± 78 and 7,755 ± 9,501 pg/ml respectively (P < 0.0001; AUC [95 %CI] = 0.87 [0.74;0.99]). The sensitivity of the modified ESICM/ERC 2021 algorithm after adding NF-L with thresholds of 500 and 1,200 pg/ml was 0.74 (CI 95% 0.51-0.88) and 0.68 (CI 95% 0.46-0.86), respectively, versus 0.53 (CI 95% 0.32-0.73) for the unmodified algorithm. In three instances the specificity was 1.

CONCLUSION

High NF-L plasma levels 48 h after cardiac arrest was significantly associated with a poor outcome. Adjunction to the current guideline algorithm of an NF-L assay with a 500 pg/ml threshold 48 h after cardiac arrest provided the best sensitivity compared to the algorithm alone, while specificity remained excellent.

Background frequency can enhance the prognostication power of EEG patterns categories in comatose cardiac arrest survivors: a prospective, multicenter, observational cohort study

Background

We assessed the prognostic accuracy of the standardized electroencephalography (EEG) patterns (“highly malignant,” “malignant,” and “benign”) according to the EEG timing (early vs. late) and investigated the EEG features to enhance the predictive power for poor neurologic outcome at 1 month after cardiac arrest.

Methods

This prospective, multicenter, observational, cohort study using data from Korean Hypothermia Network prospective registry included adult patients with out-of-hospital cardiac arrest (OHCA) treated with targeted temperature management (TTM) and underwent standard EEG within 7 days after cardiac arrest from 14 university-affiliated teaching hospitals in South Korea between October 2015 and December 2018. Early EEG was defined as EEG performed within 72 h after cardiac arrest. The primary outcome was poor neurological outcome (Cerebral Performance Category score 3–5) at 1 month.

Results

Among 489 comatose OHCA survivors with a median EEG time of 46.6 h, the “highly malignant” pattern (40.7%) was most prevalent, followed by the “benign” (33.9%) and “malignant” (25.4%) patterns. All patients with the highly malignant EEG pattern had poor neurologic outcomes, with 100% specificity in both groups but 59.3% and 56.1% sensitivity in the early and late EEG groups, respectively. However, for patients with “malignant” patterns, 84.8% sensitivity, 77.0% specificity, and 89.5% positive predictive value for poor neurologic outcome were observed. Only 3.5% (9/256) of patients with background EEG frequency of predominant delta waves or undetermined had good neurologic survival. The combination of “highly malignant” or “malignant” EEG pattern with background frequency of delta waves or undetermined increased specificity and positive predictive value, respectively, to up to 98.0% and 98.7%.

Conclusions

The “highly malignant” patterns predicted poor neurologic outcome with a high specificity regardless of EEG measurement time. The assessment of predominant background frequency in addition to EEG patterns can increase the prognostic value of OHCA survivors.

Trial registration KORHN-PRO, NCT02827422. Registered 11 September 2016—Retrospectively registered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-021-03823-y.

Neurophysiological Findings and Brain Injury Pattern in Patients on ECMO

Introduction. Brain injury is a major determinant of outcomes in extracorporeal membrane oxygenation (ECMO). Neurologic prognostication in ECMO has not been established. Absent electroencephalogram (EEG) reactivity and absent N20 on somatosensory evoked potential (SSEP) are associated with poor outcome in other types of brain injuries, especially following cardiopulmonary arrest. It is currently known if the same criteria are applicable in patients on ECMO. Methods. Continuous EEG (cEEG) was performed for patients with a Glasgow Coma Scale (GCS) <8 and SSEP data were performed for patients with a motor GCS < 4 in a prospective observational cohort undergoing ECMO at a tertiary center. EEG variables including reactivity were collected. SSEPs were categorized into absence, delay, or presence of N20. Poor outcome was defined as cerebral performance category 3 to 5 at discharge. Results. We present 13 consecutive patients who underwent both cEEG and SSEP. The median time from cannulation to EEG and SSEP were 3 (interquartile range [IQR] = 1-6) and 5 (IQR = 2-7) days, respectively. All patients were in coma and 12 (92%) had poor outcomes. Ten (77%) underwent brain computed tomography, the findings of which explained coma in only 2. Patients (n = 12) with poor outcome had poor variability, absent reactivity, and lack of sleep features with diffusely slow theta-delta background on the EEG. Despite poor outcomes, all had relatively preserved or normal N20 responses. One patient with preserved reactivity and sleep features on the EEG and intact SSEP had a good outcome. Conclusions. Absent EEG reactivity with the preservation of SSEP N20 was associated with poor outcome in comatose ECMO patients. We advise caution in interpreting electrophysiological tests in prognosticating ECMO patients until the patterns and outcomes are better understood.

Brain injury after cardiac arrest: pathophysiology, treatment, and prognosis

Post-cardiac arrest brain injury (PCABI) is caused by initial ischaemia and subsequent reperfusion of the brain following resuscitation. In those who are admitted to intensive care unit after cardiac arrest, PCABI manifests as coma, and is the main cause of mortality and long-term disability. This review describes the mechanisms of PCABI, its treatment options, its outcomes, and the suggested strategies for outcome prediction.

Complementary roles of neural synchrony and complexity for indexing consciousness and chances of surviving in acute coma

An open challenge in consciousness research is understanding how neural functions are altered by pathological loss of consciousness. To maintain consciousness, the brain needs synchronized communication of information across brain regions, and sufficient complexity in neural activity. Coordination of brain activity, typically indexed through measures of neural synchrony, has been shown to decrease when consciousness is lost and to reflect the clinical state of patients with disorders of consciousness. Moreover, when consciousness is lost, neural activity loses complexity, while the levels of neural noise, indexed by the slope of the electroencephalography (EEG) spectral exponent decrease. Although these properties have been well investigated in resting state activity, it remains unknown whether the sensory processing network, which has been shown to be preserved in coma, suffers from a loss of synchronization or information content. Here, we focused on acute coma and hypothesized that neural synchrony in response to auditory stimuli would reflect coma severity, while complexity, or neural noise, would reflect the presence or loss of consciousness. Results showed that neural synchrony of EEG signals was stronger for survivors than non-survivors and predictive of patients' outcome, but indistinguishable between survivors and healthy controls. Measures of neural complexity and neural noise were not informative of patients' outcome and had high or low values for patients compared to controls. Our results suggest different roles for neural synchrony and complexity in acute coma. Synchrony represents a precondition for consciousness, while complexity needs an equilibrium between high or low values to support conscious cognition.

Brain injury after cardiac arrest

As more people are surviving cardiac arrest, focus needs to shift towards improving neurological outcomes and quality of life in survivors. Brain injury after resuscitation, a common sequela following cardiac arrest, ranges in severity from mild impairment to devastating brain injury and brainstem death. Effective strategies to minimise brain injury after resuscitation include early intervention with cardiopulmonary resuscitation and defibrillation, restoration of normal physiology, and targeted temperature management. It is important to identify people who might have a poor outcome, to enable informed choices about continuation or withdrawal of life-sustaining treatments. Multimodal prediction guidelines seek to avoid premature withdrawal in those who might survive with a good neurological outcome, or prolonging treatment that might result in survival with severe disability. Approximately one in three admitted to intensive care will survive, many of whom will need intensive, tailored rehabilitation after discharge to have the best outcomes.

EEG patterns and their correlations with short- and long-term mortality in patients with hypoxic encephalopathy

OBJECTIVE

To analyze the association between electroencephalographic (EEG) patterns and overall, short- and long-term mortality in patients with hypoxic encephalopathy (HE).

METHODS

Retrospective, mono-center analysis of 199 patients using univariate log-rank tests (LR) and multivariate cox regression (MCR).

RESULTS

Short-term mortality, defined as death within 30-days post-discharge was 54.8%. Long-term mortality rates were 69.8%, 71.9%, and 72.9%, at 12-, 24-, and 36-months post-HE, respectively. LR revealed a significant association between EEG suppression (SUP) and short-term mortality, and identified low voltage EEG (LV), burst suppression (BSP), periodic discharges (PD) and post-hypoxic status epilepticus (PSE) as well as missing (aBA) or non-reactive background activity (nrBA) as predictors for overall, short- and long-term mortality. MCR indicated SUP, LV, BSP, PD, aBA and nrBA as significantly associated with overall and short-term mortality to varying extents. LV and BSP were significant predictors for long-term mortality in short-term survivors. Rhythmic delta activity, stimulus induced rhythmic, periodic or ictal discharges and sharp waves were not significantly associated with a higher mortality.

CONCLUSION

The presence of several specific EEG patterns can help to predict overall, short- and long-term mortality in HE patients.

SIGNIFICANCE

The present findings may help to improve the challenging prognosis estimation in HE patients.

Serum markers of brain injury can predict good neurological outcome after out-of-hospital cardiac arrest

PURPOSE

The majority of unconscious patients after cardiac arrest (CA) do not fulfill guideline criteria for a likely poor outcome, their prognosis is considered "indeterminate". We compared brain injury markers in blood for prediction of good outcome and for identifying false positive predictions of poor outcome as recommended by guidelines.

METHODS

Retrospective analysis of prospectively collected serum samples at 24, 48 and 72 h post arrest within the Target Temperature Management after out-of-hospital cardiac arrest (TTM)-trial. Clinically available markers neuron-specific enolase (NSE) and S100B, and novel markers neurofilament light chain (NFL), total tau, ubiquitin carboxy-terminal hydrolase L1 (UCH-L1) and glial fibrillary acidic protein (GFAP) were analysed. Normal levels with a priori cutoffs specified by reference laboratories or defined from literature were used to predict good outcome (no to moderate disability, Cerebral Performance Category scale 1-2) at 6 months.

RESULTS

Seven hundred and seventeen patients were included. Normal NFL, tau and GFAP had the highest sensitivities (97.2-98% of poor outcome patients had abnormal serum levels) and NPV (normal levels predicted good outcome in 87-95% of patients). Normal S100B and NSE predicted good outcome with NPV 76-82.2%. Normal NSE correctly identified 67/190 (35.3%) patients with good outcome among those classified as "indeterminate outcome" by guidelines. Five patients with single pathological prognostic findings despite normal biomarkers had good outcome.

CONCLUSION

Low levels of brain injury markers in blood are associated with good neurological outcome after CA. Incorporating biomarkers into neuroprognostication may help prevent premature withdrawal of life-sustaining therapy.

Association of Standard Electroencephalography Findings With Mortality and Command Following in Mechanically Ventilated Patients Remaining Unresponsive After Sedation Interruption

CONTEXT

Delayed awakening after sedation interruption is frequent in critically ill patients receiving mechanical ventilation.

OBJECTIVES

We aimed to investigate the association of standard electroencephalography with mortality and command following in this setting.

DESIGN, SETTING, AND PATIENTS

In a single-center study, we retrospectively analyzed standard electroencephalography performed in consecutive mechanically ventilated patients remaining unresponsive (comatose/stuporous or unable to follow commands) after sedation interruption. Standard electroencephalography parameters (background activity, continuity, and reactivity) were reassessed by neurophysiologists, blinded to patients' outcome. Patients were categorized during follow-up into three groups based on their best examination as: 1) command following, 2) unresponsive, or 3) deceased. Cause-specific models were used to identify independent standard electroencephalography parameters associated with main outcomes, that is, mortality and command following. Follow-up was right-censored 30 days after standard electroencephalography.

MEASUREMENTS AND MAIN RESULTS

Main standard electroencephalography parameters recorded in 121 unresponsive patients (median time between sedation interruption and standard electroencephalography: 2 d [interquartile range, 1-4 d]) consisted of a background frequency greater than 4 Hz in 71 (59%), a discontinuous background in 19 (16%), and a preserved reactivity in 98/120 (82%) patients. At 30 days, 66 patients (55%) were command following, nine (7%) were unresponsive, and 46 (38%) had died. In a multivariate analysis adjusted for nonneurologic organ failure, a reactive standard electroencephalography with a background frequency greater than 4 Hz was independently associated with a reduced risk of death (cause-specific hazard ratio, 0.38; CI 95%, 0.16-0.9). By contrast, none of the standard electroencephalography parameters were independently associated with command following. Sensitivity analyses conducted after exclusion of 29 patients with hypoxic brain injury revealed similar findings.

CONCLUSIONS

In patients remaining unresponsive after sedation interruption, a pattern consisting of a reactive standard electroencephalography with a background frequency greater than 4 Hz was associated with decreased odds of death. None of the standard electroencephalography parameters were independently associated with command following.

Stimulus-induced EEG-patterns and outcome after cardiac arrest

•

Presence of SIRPIDs on a late routine-EEG adds no reliable prognostic information.

•

SIRPIDs was rare among patients with a highly malignant EEG.

•

Whether specific subtypes of SIRPIDs have prognostic implications needs further investigation.

Objective

EEG is commonly used to predict prognosis in post anoxic coma. We investigated if stimulus-induced rhythmic, periodic or ictal discharges (SIRPIDs) add prognostic information after cardiac arrest.

Methods

In the multicenter Targeted Temperature Management trial, routine-EEGs were prospectively recorded after rewarming (≥36 h). Presence and subtype of SIRPIDs and main EEG-pattern (benign, malignant, highly malignant) were retrospectively reported according to a standardised classification. Patients were followed up after 180 days. Poor outcome was defined as severe neurological disability or death (Cerebral Performance Category 3–5).

Results

Of 142 patients, 71% had poor outcome and 14% had SIRPIDs. There was no significant difference in outcome between patients with and without SIRPIDs, even when subgrouped according to underlying main EEG-pattern. Comparing subtypes of SIRPIDs, 82% of patients with stimulus-induced periodic discharges had poor outcome compared to 44% of patients with stimulus-induced rhythmic delta activity, but the difference was not significant.

Conclusions

In EEGs performed ≥36 h after cardiac arrest, SIRPIDs cannot be used to reliably predict poor outcome. Whether certain subtypes of SIRPIDs indicate worse prognosis needs further investigation.

Significance

Categorising the main EEG-pattern has important prognostic implications, but assessment of late appearing SIRPIDs does not seem to add prognostic information.

Delayed awakening in neurocritical care

Delayed awakening is defined as a persistent disorder of arousal or consciousness 48 to 72h after sedation interruption in critically ill patients. Delayed awakening is either a component of coma or delirium. It results in longer hospital stays and increased mortality. It is therefore a diagnostic, therapeutic and prognostic emergency. In severe brain injured patients, delayed awakening may be related to the primary neurological injury or to secondary systemic insults related to organ failure associated with intensive care. In the present review, we propose diagnostic, therapeutic and prognostic algorithms for managing delayed awaking in neuro-ICU brain injured patients.

EEG Patterns and Outcomes After Hypoxic Brain Injury: A Systematic Review and Meta-analysis

Electroencephalography (EEG) is used to prognosticate recovery in comatose patients with hypoxic ischemic brain injury (HIBI) secondary to cardiac arrest. We sought to determine the prognostic use of specific EEG patterns for predicting disability and death following HIBI secondary to cardiac arrest. This systematic review searched Medline, Embase, and Cochrane Central up to January 2020. We included original research involving prospective and retrospective cohort studies relating specific EEG patterns to disability and death in comatose adult patients suffering HIBI post cardiac arrest requiring admission to an intensive care setting. We evaluated study quality using the Quality of Diagnostic Accuracy Studies 2 tool. Descriptive statistics were used to summarize study, patient, and EEG characteristics. We pooled study-level estimates of sensitivity and specificity for EEG patterns defined a priori using a random effect bivariate and univariate meta-analysis when appropriate. Funnel plots were used to assess publication bias. Of 5191 abstracts, 333 were reviewed in full text, of which 57 were included in the systematic review and 32 in meta-analyses. No reported EEG pattern was found to be invariably associated with death or disability across all studies. Pooled specificities of status epilepticus, burst suppression, and electrocerebral silence were high (92-99%), but sensitivities were low (6-39%) when predicting a composite outcome of disability and death. Study quality varied depending on domain; patient flow and timing performed was well conducted in all, whereas EEG interpretation was retrospective in 17 of 39 studies. Accounting for variable study quality, EEG demonstrates high specificity with a low risk of false negative outcome attribution for disability and death when status epilepticus, burst suppression, or electrocerebral silence is detected. Increased use of standardized cross-study protocols and definitions of EEG patterns are required to better evaluate the prognostic use of EEG for comatose patients with HIBI following cardiac arrest.

Bedside microdialysis for detection of early brain injury after out-of-hospital cardiac arrest

Bedside detection and early treatment of lasting cerebral ischemia may improve outcome after out-of-hospital cardiac arrest (OHCA). This feasibility study explores the possibilities to use microdialysis (MD) for continuous monitoring of cerebral energy metabolism by analyzing the draining cerebral venous blood. Eighteen comatose patients were continuously monitored with jugular bulb and radial artery (reference) MD following resuscitation. Median time from cardiac arrest to MD was 300 min (IQR 230–390) with median monitoring time 60 h (IQR 40–81). The lactate/pyruvate ratio in cerebral venous blood was increased during the first 20 h after OHCA, and significant differences in time-averaged mean MD metabolites between jugular venous and artery measurements, were documented (p < 0.02). In patients with unfavorable outcome (72%), cerebral venous lactate and pyruvate levels remained elevated during the study period. In conclusion, the study indicates that jugular bulb microdialysis (JBM) is feasible and safe. Biochemical signs of lasting ischemia and mitochondrial dysfunction are frequent and associated with unfavorable outcome. The technique may be used in comatose OHCA patients to monitor biochemical variables reflecting ongoing brain damage and support individualized treatment early after resuscitation.

Predicting early recovery of consciousness after cardiac arrest supported by quantitative electroencephalography

OBJECTIVE

To determine the ability of quantitative electroencephalography (QEEG) to improve the accuracy of predicting recovery of consciousness by post-cardiac arrest day 10.

METHODS

Unconscious survivors of cardiac arrest undergoing daily clinical and EEG assessments through post-cardiac arrest day 10 were studied in a prospective observational cohort study. Power spectral density, local coherence, and permutation entropy were calculated from daily EEG clips following a painful stimulus. Recovery of consciousness was defined as following at least simple commands by day 10. We determined the impact of EEG metrics to predict recovery when analyzed with established predictors of recovery using partial least squares regression models. Explained variance analysis identified which features contributed most to the predictive model.

RESULTS

367 EEG epochs from 98 subjects were analyzed in conjunction with clinical measures. Highest prediction accuracy was achieved when adding QEEG features from post-arrest days 4-6 to established predictors (area under the receiver operating curve improved from 0.81 ± 0.04 to 0.86 ± 0.05). Prediction accuracy decreased from 0.84 ± 0.04 to 0.79 ± 0.04 when adding QEEG features from post-arrest days 1-3. Patients with recovery of command-following by day 10 showed higher coherence across the frequency spectrum and higher centro-occipital delta-frequency spectral power by days 4-6, and globally-higher theta range permutation entropy by days 7-10.

CONCLUSIONS

Adding quantitative EEG metrics to established predictors of recovery allows modest improvement of prediction accuracy for recovery of consciousness, when obtained within a week of cardiac arrest. Further research is needed to determine the best strategy for integration of QEEG data into prognostic models in this patient population.

Prognostic Effects of Vasomotor Reactivity during Targeted Temperature Management in Post-Cardiac Arrest Patients: A Retrospective Observational Study

Early and precise neurological prognostication without self-fulfilling prophecy is challenging in post-cardiac arrest syndrome (PCAS), particularly during the targeted temperature management (TTM) period. This study aimed to investigate the feasibility of vasomotor reactivity (VMR) using transcranial Doppler (TCD) to determine whether final outcomes of patients with comatose PCAS are predicted. This study included patients who had out-of-hospital cardiac arrest in a tertiary referral hospital over 4 years. The eligible criteria included age ≥18 years, successful return of spontaneous circulation, TTM application, and bedside TCD examination within 72 h. Baseline demographics and multimodal prognostic parameters, including imaging findings, electrophysiological studies, and TCD-VMR parameters, were assessed. The final outcome parameter was cerebral performance category scale (CPC) at 1 month. Potential determinants were compared between good (CPC 1-2) and poor (CPC 3-5) outcome groups. The good outcome group (n = 41) (vs. poor (n = 117)) showed a higher VMR value (54.4% ± 33.0% vs. 25.1% ± 35.8%, p < 0.001). The addition of VMR to conventional prognostic parameters significantly improved the prediction power of good outcomes. This study suggests that TCD-VMR is a useful tool at the bedside to evaluate outcomes of patients with comatose PCAS during the TTM.

Chasing Certainty After Cardiac Arrest: Can a Technological Innovation Solve a Moral Dilemma?

When information on a coma patient’s expected outcome is uncertain, a moral dilemma arises in clinical practice: if life-sustaining treatment is continued, the patient may survive with unacceptably poor neurological prospects, but if withdrawn a patient who could have recovered may die. Continuous electroencephalogram-monitoring (cEEG) is expected to substantially improve neuroprognostication for patients in coma after cardiac arrest. This raises expectations that decisions whether or not to withdraw will become easier. This paper investigates that expectation, exploring cEEG’s impacts when it becomes part of a socio-technical network in an Intensive Care Unit (ICU). Based on observations in two ICUs in the Netherlands and one in the USA that had cEEG implemented for research, we interviewed 25 family members, healthcare professionals, and surviving patients. The analysis focuses on (a) the way patient outcomes are constructed, (b) the kind of decision support these outcomes provide, and (c) how cEEG affects communication between professionals and relatives. We argue that cEEG can take away or decrease the intensity of the dilemma in some cases, while increasing uncertainty for others. It also raises new concerns. Since its actual impacts furthermore hinge on how cEEG is designed and implemented, we end with recommendations for ensuring responsible development and implementation.

Automated Assessment of Brain CT After Cardiac Arrest-An Observational Derivation/Validation Cohort Study

Objectives

Prognostication of outcome is an essential step in defining therapeutic goals after cardiac arrest. Gray-white-matter ratio obtained from brain CT can predict poor outcome. However, manual placement of regions of interest is a potential source of error and interrater variability. Our objective was to assess the performance of poor outcome prediction by automated quantification of changes in brain CTs after cardiac arrest.

Design

Observational, derivation/validation cohort study design. Outcome was determined using the Cerebral Performance Category upon hospital discharge. Poor outcome was defined as death or unresponsive wakefulness syndrome/coma. CTs were automatically decomposed using coregistration with a brain atlas.

Setting

ICUs at a large, academic hospital with circulatory arrest center.

Patients

We identified 433 cardiac arrest patients from a large previously established database with brain CTs within 10 days after cardiac arrest.

Interventions

None.

Measurements and Main Results

Five hundred sixteen brain CTs were evaluated (derivation cohort n = 309, validation cohort n = 207). Patients with poor outcome had significantly lower radiodensities in gray matter regions. Automated GWR_si (putamen/posterior limb of internal capsule) was performed with an area under the curve of 0.86 (95%-CI: 0.80-0.93) for CTs taken later than 24 hours after cardiac arrest (similar performance in the validation cohort). Poor outcome (Cerebral Performance Category 4-5) was predicted with a specificity of 100% (95% CI, 87-100%, derivation; 88-100%, validation) at a threshold of less than 1.10 and a sensitivity of 49% (95% CI, 36-58%, derivation) and 38% (95% CI, 27-50%, validation) for CTs later than 24 hours after cardiac arrest. Sensitivity and area under the curve were lower for CTs performed within 24 hours after cardiac arrest.

Conclusions

Automated gray-white-matter ratio from brain CT is a promising tool for prediction of poor neurologic outcome after cardiac arrest with high specificity and low-to-moderate sensitivity. Prediction by gray-white-matter ratio at the basal ganglia level performed best. Sensitivity increased considerably for CTs performed later than 24 hours after cardiac arrest.

Quantitative analysis of EEG reactivity for neurological prognostication after cardiac arrest

OBJECTIVE

To test whether 1) quantitative analysis of EEG reactivity (EEG-R) using machine learning (ML) is superior to visual analysis, and 2) combining quantitative analyses of EEG-R and EEG background pattern increases prognostic value for prediction of poor outcome after cardiac arrest (CA).

METHODS

Several types of ML models were trained with twelve quantitative features derived from EEG-R and EEG background data of 134 adult CA patients. Poor outcome was a Cerebral Performance Category score of 3-5 within 6 months.

RESULTS

The Random Forest (RF) trained on EEG-R showed the highest AUC of 83% (95-CI 80-86) of tested ML classifiers, predicting poor outcome with 46% sensitivity (95%-CI 40-51) and 89% specificity (95%-CI 86-92). Visual analysis of EEG-R had 80% sensitivity and 65% specificity. The RF was also the best classifier for EEG background (AUC 85%, 95%-CI 83-88) at 24 h after CA, with 62% sensitivity (95%-CI 57-67) and 84% specificity (95%-CI 79-88). Combining EEG-R and EEG background RF classifiers reduced the number of false positives.

CONCLUSIONS

Quantitative EEG-R using ML predicts poor outcome with higher specificity, but lower sensitivity compared to visual analysis of EEG-R, and is of some additional value to ML on EEG background data.

SIGNIFICANCE

Quantitative EEG-R using ML is a promising alternative to visual analysis and of some added value to ML on EEG background data.

Multimodal Approach to Predict Neurological Outcome after Cardiac Arrest: A Single-Center Experience

Introduction: The aims of this study were to assess the concordance of different tools and to describe the accuracy of a multimodal approach to predict unfavorable neurological outcome (UO) in cardiac arrest patients. Methods: Retrospective study of adult (>18 years) cardiac arrest patients who underwent multimodal monitoring; UO was defined as cerebral performance category 3–5 at 3 months. Predictors of UO were neurological pupillary index (NPi) ≤ 2 at 24 h; highly malignant patterns on EEG (HMp) within 48 h; bilateral absence of N20 waves on somato-sensory evoked potentials; and neuron-specific enolase (NSE) > 75 μg/L. Time-dependent decisional tree (i.e., NPi on day 1; HMp on day 1–2; absent N20 on day 2–3; highest NSE) and classification and regression tree (CART) analysis were used to assess the prediction of UO. Results: Of 137 patients, 104 (73%) had UO. Abnormal NPi, HMp on day 1 or 2, the bilateral absence of N20 or NSE >75 mcg/L had a specificity of 100% to predict UO. The presence of abnormal NPi was highly concordant with HMp and high NSE, and absence of N20 or high NSE with HMp. However, HMp had weak to moderate concordance with other predictors. The time-dependent decisional tree approach identified 73/103 patients (70%) with UO, showing a sensitivity of 71% and a specificity of 100%. Using the CART approach, HMp on EEG was the only variable significantly associated with UO. Conclusions: This study suggests that patients with UO had often at least two predictors of UO, except for HMp. A multimodal time-dependent approach may be helpful in the prediction of UO after CA. EEG should be included in all multimodal prognostic models.

Clinical Correlates of Electroencephalographic Patterns in Critically Ill Patients

OBJECTIVE

The objective of this study was to determine the frequency and clinical correlates of different electroencephalographic patterns and their association with clinical outcomes in critically ill patients.

SUBJECTS AND METHODS

This retrospective cross-sectional study was performed in the Neurology Department of King Fahd Hospital of the University, Kingdom of Saudi Arabia and involved a review and analysis of medical records pertaining to 179 intensive care unit patients who underwent electroencephalography (EEG) in the June to November 2018 period.

RESULTS

Among the different etiologies, presence of spike and wave or sharp wave (SWs) was associated with encephalitis (P = .01) and large artery stroke (P = .01), whereas markedly attenuated EEG activity (p = .04) and burst suppression (P = .01) were associated with large artery stroke and hypoxic ischemic encephalopathy (HIE), respectively. Generalized theta activity (P = .01) was significantly found in patients of septic encephalopathy, while generalized delta activity (P = .02) and asymmetrical background (P = .04) were significantly associated with traumatic brain injury. Presence of periodic discharges in EEG was significantly associated with more adverse clinical outcomes (P = .001), whereas rhythmic delta activity (RDA) (P = .03), persistent focal slow wave activity (P = .01), and asymmetric background (P = .002) were found in patients who were discharged from hospital with sequelae of current illness.

CONCLUSION

Certain EEG patterns are associated with particular underlying etiologies like SWs for encephalitis, markedly attenuated EEG activity and burst suppression with large artery stroke and HIE, respectively. Whereas few EEG patterns, including periodic discharges, RDA, persistent focal slow wave activity have some prognostic value in critically ill patients. However, they cannot be used as markers for prognostic assessment of patients without considering other clinical and diagnostic variables. Further larger prospective studies with continuous EEG monitoring with control of confounding factors are needed.

[Long-term EEG monitoring in patients with acute traumatic brain injury]

OBJECTIVE

To investigate the informativeness of long-term scalp EEG monitoring in patients with acute traumatic brain injury (TBI).

MATERIAL AND METHODS

The informativity of long-term EEG monitoring (LTM) was performed in 60 patients with acute severe TBI. Odd ratios (OR) of unfavorable outcome and non-convulsive status epilepticus (NCSE) among clinical, neurophysiological and radiological features were calculated.

RESULTS

EEG features of the unfavorable outcome are: slowing of the dominant background rhythm below q range (OR 3.5, CI 1.2-10.7), absence of frontal-occipital gradient (OR 10.2, CI 1.89-10.12), absence of reactivity (OR 8.75, CI 2.14-35.7), absence of variability (OR 6.25, CI 1.72-22.6) and absence of NREM sleep, stage 2 (OR 5.8, CI 1.79-18.91). Clinical features associated with the unfavorable outcome are: a decrease in GCS score (OR 1.25, CI 1.07-1.47), TBI severity (OR 2.46, CI 1.16-5.18), axial dislocation (OR 4.45, CI 1.08-18.29). ORs for NCSE are significant for the following EEG features: presence of rhythmic and periodic patterns (RPP) (OR 11.92, CI 1.37-103.39), stimulus induced RPP (OR 23.14, CI 2.56-209.34), "plus" modifier (OR 4.11, CI 1.13-14.91) and electrographic evolution (OR 13.05, CI 3.59-47.39). Background rhythm slowing below q range reduces NCSE probability (OR 3.33, CI 1.09-10).

CONCLUSION

Long-term EEG monitoring is an informative tool for prognosis of outcome and diagnosis of NCSE in patients with severe TBI. The risk of NCSE increases with Marshall score but NCSE is not associated with poor outcome that requires an individual selection of intensive care.

Hypoxic-Ischemic Encephalopathy Evaluated by Brain Autopsy and Neuroprognostication After Cardiac Arrest

Importance

Neuroprognostication studies are potentially susceptible to a self-fulfilling prophecy as investigated prognostic parameters may affect withdrawal of life-sustaining therapy.

Objective

To compare the results of prognostic parameters after cardiac arrest (CA) with the histopathologically determined severity of hypoxic-ischemic encephalopathy (HIE) obtained from autopsy results.

Design, Setting, and Participants

In a retrospective, 3-center cohort study of all patients who died following cardiac arrest during their intensive care unit stay and underwent autopsy between 2003 and 2015, postmortem brain histopathologic findings were compared with post-CA brain computed tomographic imaging, electroencephalographic (EEG) findings, somatosensory-evoked potentials, and serum neuron-specific enolase levels obtained during the intensive care unit stay. Data analysis was conducted from 2015 to 2020.

Main Outcomes and Measures

The severity of HIE was evaluated according to the selective eosinophilic neuronal death (SEND) classification and patients were dichotomized into categories of histopathologically severe and no/mild HIE.

Results

Of 187 included patients, 117 were men (63%) and median age was 65 (interquartile range, 58-74) years. Severe HIE was found in 114 patients (61%) and no/mild HIE was identified in 73 patients (39%). Severe HIE was found in all 21 patients with bilaterally absent somatosensory-evoked potentials, all 15 patients with gray-white matter ratio less than 1.10 on brain computed tomographic imaging, all 9 patients with suppressed EEG, 15 of 16 patients with burst-suppression EEG, and all 29 patients with neuron-specific enolase levels greater than 67 μg/L more than 48 hours after CA without confounders. Three of 7 patients with generalized periodic discharges on suppressed background and 1 patient with burst-suppression EEG had a SEND 1 score (<30% dead neurons) in the cerebral cortex, but higher SEND scores (>30% dead neurons) in other oxygen-sensitive brain regions.

Conclusions and Relevance

In this study, histopathologic findings suggested severe HIE after cardiac arrest in patients with bilaterally absent cortical somatosensory-evoked potentials, gray-white matter ratio less than 1.10, highly malignant EEG, and serum neuron-specific enolase concentration greater than 67 μg/L.

Application of a standardized EEG pattern classification in the assessment of neurological prognosis after cardiac arrest: A retrospective analysis

INTRODUCTION

Electroencephalogram (EEG) is used in the neurological prognostication after cardiac arrest. "Highly malignant" EEG patterns classified according to Westhall have a high specificity for poor neurological outcome when applied within protocols of recent studies. However, their predictive performance when applied in everyday clinical practice has not been investigated. We studied the prognostic accuracy and the interrater agreement when standardized EEG patterns were analysed and compared to neurological outcome in a patient cohort at a tertiary centre not involved in the original study of the standardized EEG pattern classification.

METHODS

Comatose patients treated for out-of-hospital cardiac arrest were included. Poor outcome was defined as Cerebral Performance Category 3-5. Two senior consultants and one resident in clinical neurophysiology, blinded to clinical data and outcome, independently reviewed their EEG registrations and categorised the pattern as "highly malignant", "malignant" or "benign". These categories were compared to neurological outcome at hospital discharge. Interrater agreement was assessed using Cohen's Kappa.

RESULTS

In total, 62 patients were included. The median (IQR) time to EEG was 59 (42-91) h after return of spontaneous circulation. Poor outcome was found in 52 (84%) patients. In 21 patients at least one of the raters considered the EEG to contain a "highly malignant" pattern, all with poor outcome (42% sensitivity, 100% specificity). The interrater agreement varied from kappa 0.62 to 0.29.

CONCLUSION

"Highly malignant" patterns predict poor neurological outcome with a high specificity in everyday practice. However, interrater agreement may vary substantially even between experienced EEG interpreters.

Early recovery of frontal EEG slow wave activity during propofol sedation predicts outcome after cardiac arrest

AIM OF THE STUDY

EEG slow wave activity (SWA) has shown prognostic potential in post-resuscitation care. In this prospective study, we investigated the accuracy of continuously measured early SWA for prediction of the outcome in comatose cardiac arrest (CA) survivors.

METHODS

We recorded EEG with a disposable self-adhesive frontal electrode and wireless device continuously starting from ICU admission until 48 h from return of spontaneous circulation (ROSC) in comatose CA survivors sedated with propofol. We determined SWA by offline calculation of C-Trend® Index describing SWA as a score ranging from 0 to 100. The functional outcome was defined based on Cerebral Performance Category (CPC) at 6 months after the CA to either good (CPC 1-2) or poor (CPC 3-5).

RESULTS

Outcome at six months was good in 67 of the 93 patients. During the first 12 h after ROSC, the median C-Trend Index value was 38.8 (interquartile range 28.0-56.1) in patients with good outcome and 6.49 (3.01-18.2) in those with poor outcome showing significant difference (p < 0.001) at every hour between the groups. The index values of the first 12 h predicted poor outcome with an area under curve of 0.86 (95% CI 0.61-0.99). With a cutoff value of 20, the sensitivity was 83.3% (69.6%-92.3%) and specificity 94.7% (83.4%-99.7%) for categorization of outcome.

CONCLUSION

EEG SWA measured with C-Trend Index during propofol sedation offers a promising practical approach for early bedside evaluation of recovery of brain function and prediction of outcome after CA.

Somatosensory Evoked Potentials and Neuroprognostication After Cardiac Arrest

Improved understanding of post-cardiac arrest syndrome and clinical practices such as targeted temperature management have led to improved mortality in this cohort. Attention has now been placed on development of tools to aid in predicting functional outcome in comatose cardiac arrest survivors. Current practice uses a multimodal approach including physical examination, neuroimaging, and electrophysiologic data, with a primary utility in predicting poor functional outcome. These modalities remain confounded by self-fulfilling prophecy and the withdrawal of life-sustaining therapies. To date, a reliable measure to predict good functional outcome has not been established or validated, but the use of quantitative somatosensory evoked potential (SSEP) shows potential for this use. MEDLINE and EMBASE search using words "Cardiac Arrest" and "SSEP," "Somato sensory evoked potentials," "qSSEP," "quantitative SSEP," "targeted temperature management in cardiac arrest" was conducted. Relevant recent studies on targeted temperature management in cardiac arrest, plus studies on SSEP in cardiac arrest in the setting of hypothermia and without hypothermia, were included. In addition, animal studies evaluating the role of different components of SSEP in cardiac arrest were reviewed. SSEP is a specific indicator of poor outcomes in post-cardiac arrest patients but lacks sensitivity and has not clinically been established to foresee good outcomes. Novel methods of analyzing quantitative SSEP (qSSEP) signals have shown potential to predict good outcomes in animal and human studies. In addition, qSSEP has potential to track cerebral recovery and guide treatment strategy in post-cardiac arrest patients. Lying beyond the current clinical practice of dichotomized absent/present N20 peaks, qSSEP has the potential to emerge as one of the earliest predictors of good outcome in comatose post-cardiac arrest patients. Validation of qSSEP markers in prospective studies to predict good and poor outcomes in the cardiac arrest population in the setting of hypothermia could advance care in cardiac arrest. It has the prospect to guide allocation of health care resources and reduce self-fulfilling prophecy.

Early neuro-prognostication with the Patient State Index and suppression ratio in post-cardiac arrest patients

PURPOSE

Cardiopulmonary resuscitation guidelines recommend multimodal neuro-prognostication after cardiac arrest using neurological examination, electroencephalography, biomarkers, and brain imaging. The Patient State Index (PSI) and suppression ratio (SR) represent the depth and degree of sedation, respectively. We evaluated the predictive ability of PSI and SR for neuro-prognostication of post-cardiac arrest patients who underwent targeted temperature management.

METHODS

This prospective observational study was conducted between January 2017 and August 2020 and enrolled adult patients in an intensive care unit (ICU) with non-traumatic out-of-hospital cardiac arrest with return of spontaneous circulation (ROSC). PSI and SR were monitored continuously during ICU stay, and their maximum, mean, and minimum cutoff values 24 h after ROSC were analyzed to predict poor neurologic outcome and long-term survival.

RESULTS

The final analysis included 103 patients. A mean PSI ≤ 14.53 and mean SR > 36.6 showed high diagnostic accuracy as single prognostic factors. Multimodal prediction using the mean PSI and mean SR showed the highest area-under-the-curve value of 0.965 (95% confidence interval 0.909-0.991). Patients with mean PSI ≤ 14.53 and mean SR > 36.6 had relatively higher long-term mortality rates than those of patients with values >14.53 and ≤ 36.6, respectively.

CONCLUSIONS

The PSI and SR are good predictors for early neuro-prognostication in post-cardiac arrest patients.

Targeted temperature management and early neuro-prognostication after cardiac arrest

Targeted temperature management (TTM) is a recommended neuroprotective intervention for coma after out-of-hospital cardiac arrest (OHCA). However, controversies exist concerning the proper implementation and overall efficacy of post-CA TTM, particularly related to optimal timing and depth of TTM and cooling methods. A review of the literature finds that optimizing and individualizing TTM remains an open question requiring further clinical investigation. This paper will summarize the preclinical and clinical trial data to-date, current recommendations, and future directions of this therapy, including new cooling methods under investigation. For now, early induction, maintenance for at least 24 hours, and slow rewarming utilizing endovascular methods may be preferred. Moreover, timely and accurate neuro-prognostication is valuable for guiding ethical and cost-effective management of post-CA coma. Current evidence for early neuro-prognostication after TTM suggests that a combination of initial prediction models, biomarkers, neuroimaging, and electrophysiological methods is the optimal strategy in predicting neurological functional outcomes.

Simplified frontal EEG in adults under veno-arterial extracorporeal membrane oxygenation

Background

EEG-based prognostication studies in intensive care units often rely on a standard 21-electrode montage (_stdEEG) requiring substantial human, technical, and financial resources. We here evaluate whether a simplified 4-frontal electrode montage (_4-frontEEG) can detect EEG patterns associated with poor outcomes in adult patients under veno-arterial extracorporeal membrane oxygenation (VA-ECMO).

Methods

We conducted a reanalysis of EEG data from a prospective cohort on 118 adult patients under VA-ECMO, in whom EEG was performed on admission to intensive care. EEG patterns of interest included background rhythm, discontinuity, reactivity, and the Synek’s score. They were all reassessed by an intensivist on a _4-frontEEG montage, whose analysis was then compared to an expert’s interpretation made on _stdEEG recordings. The main outcome measure was the degree of correlation between _4-frontEEG and _stdEEG montages to identify EEG patterns of interest. The performance of the Synek scores calculated on _4-frontEEG and _stdEEG montage to predict outcomes (i.e., 28-day mortality and 90-day Rankin score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\ge {4}}$$\end{document}≥4) was investigated in a secondary exploratory analysis.

Results

The detection of EEG patterns using _4-frontEEG was statistically similar to that of _stdEEG for background rhythm (Spearman rank test, ρ = 0.66, p < 0.001), discontinuity (Cohen’s kappa, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\kappa$$\end{document}κ = 0.955), reactivity (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\kappa$$\end{document}κ = 0.739) and the Synek’s score (ρ = 0.794, p < 0.001). Using the Synek classification, we found similar performances between _4-frontEEG and _stdEEG montages in predicting 28-day mortality (AUC _4-frontEEG 0.71, AUC _stdEEG 0.68) and for 90-day poor neurologic outcome (AUC _4-frontEEG 0.71, AUC _stdEEG 0.66). An exploratory analysis confirmed that the Synek scores determined by 4 or 21 electrodes were independently associated with 28-day mortality and poor 90-day functional outcome.

Conclusion

In adult patients under VA-ECMO, a simplified 4-frontal electrode EEG montage interpreted by an intensivist, detected common EEG patterns associated with poor outcomes, with a performance similar to that of a standard EEG montage interpreted by expert neurophysiologists. This simplified montage could be implemented as part of a multimodal evaluation for bedside prognostication.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13613-021-00854-0.

Assessment of a standardized EEG reactivity protocol after cardiac arrest

OBJECTIVE

Reactivity assessment during EEG might provide important prognostic information in post-anoxic coma. It is still unclear how best to perform reactivity testing and how it might be affected by hypothermia. Our primary aim was to determine and compare the effectiveness, inter-rater reliability and prognostic value of different types of stimulus for EEG reactivity testing, using a standardized stimulation protocol and standardized definitions. Our secondary aims were to assess the effect of hypothermia on these measures, and to determine the prognostic value of a simplified sequence with the three most efficient stimuli.

METHODS

Prospective single-center cohort of post-anoxic comatose patients admitted to the intensive care unit of an academic medical center between January 1, 2016 and December 31, 2018 and receiving continuous EEG monitoring (CEEG). Reactivity was assessed using standardized definitions and standardized sequence of stimuli: auditory (mild noise and loud noise), tactile (shaking), nociceptive (nostril tickling, trapezius muscle squeezing, endotracheal tube suctioning), and visual (passive eye opening). Gwet's AC1 and percent agreement (PA) were used to measure inter-rater agreement (IRA). Ability to predict favorable neurological outcome (defined as a Cerebral Performance Category of 1 to 2: no disability to moderate disability) was measured with sensitivity (Se), specificity (Sp), accuracy, and odds ratio [OR]. These were calculated for each stimulus type and at the level of the entire sequence comprising all the stimuli.

RESULTS

One-hundred and fifteen patients were included and 242 EEG epochs were analyzed. Loud noise, shaking and trapezius muscle squeezing most frequently elicited EEG reactivity (42%, 38% and 38%, respectively) but were all inferior to the entire sequence, which elicited reactivity in 58% cases. The IRA for reactivity to individual stimuli varied from moderate to good (AC1:58-69%; PA:56-68%) and was the highest for loud noise (AC1:69%; PA:68%), trapezius muscle squeezing (AC1:67%; PA:65%) and passive eye opening (AC1:68%; PA:64%). Mild (odds ratio [OR]:11.0; Se:70% and Sp:86%) and loud noises (OR:27.0; Se:73% and Sp:75%), and trapezius muscle squeezing (OR:15.3; Se:76% and Sp:83%) during hypothermia had the best predictive value for favorable neurological outcome, although each was inferior to the whole sequence (OR:60.2; Se:91% and Sp:73%). A simplified sequence of loud noise, shaking and trapezius muscle squeezing had the same performance for predicting neurological outcome as the entire sequence. Hypothermia did not significantly affect the effectiveness of stimulation, but IRA was slightly better during hypothermia, for all stimuli. Similarly, the predictive value was higher during hypothermia than during normothermia.

CONCLUSIONS

Despite a standardized stimulation protocol and standardized definitions, the IRA of EEG reactivity testing in post-anoxic comatose patients was only good at best (AC1 < 70%), and its predictive value for neurological outcome remained imperfect, in particular with Sp values < 90%. While no single stimulus appeared superior to others, a full sequence using all stimuli or a simplified sequence comprising loud noise, shaking and trapezius muscle squeezing had the best combination of IRA and predictive value.

SIGNIFICANCE

This study stresses the necessity to use multiple stimulus types to improve the predictive value of reactivity testing in post-anoxic coma and confirms that it is not affected by hypothermia.