Accuracy of Limited-Montage Electroencephalography in Monitoring Postanoxic Comatose Patients

BACKGROUND

Continuous EEG (cEEG) monitoring may help to identify the small percentage of adults with hypoxic-ischemic encephalopathy (HIE) who will regain consciousness if allowed sufficient time. However, the limited yield in this population has led some to question the cost-effectiveness cEEG monitoring in this population. We hypothesized that limited-montage cEEG could provide essentially the same neurophysiologic information at lower cost. In this proof of concept study, we aim to demonstrate the potentials of limited channel EEG in prognostication in postanoxic patients.

METHODS

We retrospectively reviewed cEEG data from cases monitored at our institution with conventional 21-channel EEG over a 6-month period. Twenty-eight cases were identified in which patients with HIE underwent cEEG for at least 24 hours. Gold-standard findings were determined by conventional visual analysis of the full cEEG, and 2 independent electroencephalographers scored the same data using only limited-montage (4-channel) views. The sensitivity and specificity of limited-montage cEEG review were compared with conventional analysis. We also compared the relative costs of conventional and limited-montage EEG.

RESULTS

Using 4-channel limited montage cEEG, reviewers were able to classify accurately background continuity (in 88%), background amplitude (in 81%), maximum background frequency (in 70%), periodic epileptiform discharges, including a seizure (in 92%) and sporadic discharges (in 91%). All epileptiform features were detected with greater than 90% sensitivity and specificity. Eye movement artifact seen over bifrontal electrodes gave false positive detections of periodic epileptiform discharges in 31% of cases.

CONCLUSIONS

Limited-channel continuous EEG monitoring can provide meaningful electrophysiological data that can be used for prognostication in postanoxic comatose patients. Limited channel EEG can be a cost-effective alternative to conventional EEG monitoring in post-anoxic comatose patients.

Default Mode Network Perfusion in Aneurysmal Subarachnoid Hemorrhage

BACKGROUND

The etiology of altered consciousness in patients with high-grade aneurysmal subarachnoid hemorrhage (SAH) is not thoroughly understood. We hypothesized that decreased cerebral blood flow (CBF) in brain regions critical to consciousness may contribute.

METHODS

We retrospectively evaluated arterial-spin labeled (ASL) perfusion magnetic resonance imaging (MRI) measurements of CBF in 12 patients with aneurysmal SAH admitted to our neurocritical care unit. CBF values were analyzed within gray matter nodes of the default mode network (DMN), whose functional integrity has been shown to be necessary for consciousness. DMN nodes studied were the bilateral medial prefrontal cortices, thalami, and posterior cingulate cortices. Correlations between nodal CBF and admission Glasgow Coma Scale (GCS) score, admission Hunt and Hess (HH) class, and GCS score at the time of MRI (MRI GCS) were tested.

RESULTS

Spearman's correlation coefficients were not significant when comparing admission GCS, admission HH, and MRI GCS versus nodal CBF (p > 0.05). However, inter-rater reliability for nodal CBF was high (r = 0.71, p = 0.01).

CONCLUSIONS

In this retrospective pilot study, we did not identify significant correlations between CBF and admission GCS, admission HH class, or MRI GCS for any DMN node. Potential explanations for these findings include small sample size, ASL data acquisition at variable times after SAH onset, and CBF analysis in DMN nodes that may not reflect the functional integrity of the entire network. High inter-rater reliability suggests ASL measurements of CBF within DMN nodes are reproducible. Larger prospective studies are needed to elucidate whether decreased cerebral perfusion contributes to altered consciousness in SAH.

Large-Scale Automated Sleep Staging

Study Objectives

Automated sleep staging has been previously limited by a combination of clinical and physiological heterogeneity. Both factors are in principle addressable with large data sets that enable robust calibration. However, the impact of sample size remains uncertain. The objectives are to investigate the extent to which machine learning methods can approximate the performance of human scorers when supplied with sufficient training cases and to investigate how staging performance depends on the number of training patients, contextual information, model complexity, and imbalance between sleep stage proportions.

Methods

A total of 102 features were extracted from six electroencephalography (EEG) channels in routine polysomnography. Two thousand nights were partitioned into equal (n = 1000) training and testing sets for validation. We used epoch-by-epoch Cohen's kappa statistics to measure the agreement between classifier output and human scorer according to American Academy of Sleep Medicine scoring criteria.

Results

Epoch-by-epoch Cohen's kappa improved with increasing training EEG recordings until saturation occurred (n = ~300). The kappa value was further improved by accounting for contextual (temporal) information, increasing model complexity, and adjusting the model training procedure to account for the imbalance of stage proportions. The final kappa on the testing set was 0.68. Testing on more EEG recordings leads to kappa estimates with lower variance.

Conclusion

Training with a large data set enables automated sleep staging that compares favorably with human scorers. Because testing was performed on a large and heterogeneous data set, the performance estimate has low variance and is likely to generalize broadly.

Interictal epileptiform discharge characteristics underlying expert interrater agreement

OBJECTIVE

The presence of interictal epileptiform discharges (IED) in the electroencephalogram (EEG) is a key finding in the medical workup of a patient with suspected epilepsy. However, inter-rater agreement (IRA) regarding the presence of IED is imperfect, leading to incorrect and delayed diagnoses. An improved understanding of which IED attributes mediate expert IRA might help in developing automatic methods for IED detection able to emulate the abilities of experts. Therefore, using a set of IED scored by a large number of experts, we set out to determine which attributes of IED predict expert agreement regarding the presence of IED.

METHODS

IED were annotated on a 5-point scale by 18 clinical neurophysiologists within 200 30-s EEG segments from recordings of 200 patients. 5538 signal analysis features were extracted from the waveforms, including wavelet coefficients, morphological features, signal energy, nonlinear energy operator response, electrode location, and spectrogram features. Feature selection was performed by applying elastic net regression and support vector regression (SVR) was applied to predict expert opinion, with and without the feature selection procedure and with and without several types of signal normalization.

RESULTS

Multiple types of features were useful for predicting expert annotations, but particular types of wavelet features performed best. Local EEG normalization also enhanced best model performance. As the size of the group of EEGers used to train the models was increased, the performance of the models leveled off at a group size of around 11.

CONCLUSIONS

The features that best predict inter-rater agreement among experts regarding the presence of IED are wavelet features, using locally standardized EEG. Our models for predicting expert opinion based on EEGer's scores perform best with a large group of EEGers (more than 10).

SIGNIFICANCE

By examining a large group of EEG signal analysis features we found that wavelet features with certain wavelet basis functions performed best to identify IEDs. Local normalization also improves predictability, suggesting the importance of IED morphology over amplitude-based features. Although most IED detection studies in the past have used opinion from three or fewer experts, our study suggests a "wisdom of the crowd" effect, such that pooling over a larger number of expert opinions produces a better correlation between expert opinion and objectively quantifiable features of the EEG.

Time-dependent risk of seizures in critically ill patients on continuous electroencephalogram

OBJECTIVE

Find the optimal continuous electroencephalographic (CEEG) monitoring duration for seizure detection in critically ill patients.

METHODS

We analyzed prospective data from 665 consecutive CEEGs, including clinical factors and time-to-event emergence of electroencephalographic (EEG) findings over 72 hours. Clinical factors were selected using logistic regression. EEG risk factors were selected a priori. Clinical factors were used for baseline (pre-EEG) risk. EEG findings were used for the creation of a multistate survival model with 3 states (entry, EEG risk, and seizure). EEG risk state is defined by emergence of epileptiform patterns.

RESULTS

The clinical variables of greatest predictive value were coma (31% had seizures; odds ratio [OR] = 1.8, p < 0.01) and history of seizures, either remotely or related to acute illness (34% had seizures; OR = 3.0, p < 0.001). If there were no epileptiform findings on EEG, the risk of seizures within 72 hours was between 9% (no clinical risk factors) and 36% (coma and history of seizures). If epileptiform findings developed, the seizure incidence was between 18% (no clinical risk factors) and 64% (coma and history of seizures). In the absence of epileptiform EEG abnormalities, the duration of monitoring needed for seizure risk of <5% was between 0.4 hours (for patients who are not comatose and had no prior seizure) and 16.4 hours (comatose and prior seizure).

INTERPRETATION

The initial risk of seizures on CEEG is dependent on history of prior seizures and presence of coma. The risk of developing seizures on CEEG decays to <5% by 24 hours if no epileptiform EEG abnormalities emerge, independent of initial clinical risk factors. Ann Neurol 2017;82:177-185.

Automated epileptiform spike detection via affinity propagation-based template matching

Interictal epileptiform spikes are the key diagnostic biomarkers for epilepsy. The clinical gold standard of spike detection is visual inspection performed by neurologists. This is a tedious, time-consuming, and expert-centered process. The development of automated spike detection systems is necessary in order to provide a faster and more reliable diagnosis of epilepsy. In this paper, we propose an efficient template matching spike detector based on a combination of spike and background waveform templates. We generate a template library by clustering a collection of spikes and background waveforms extracted from a database of 50 patients with epilepsy. We benchmark the performance of five clustering techniques based on the receiver operating characteristic (ROC) curves. In addition, background templates are integrated with existing spike templates to improve the overall performance. The affinity propagation-based template matching system with a combination of spike and background templates is shown to outperform the other four conventional methods with the highest area-under-curve (AUC) of 0.953.

Heart rate variability as a biomarker for sedation depth estimation in ICU patients

An automated patient-specific system to classify the level of sedation in ICU patients using heart rate variability signal is presented in this paper. ECG from 70 mechanically ventilated adult patients with administered sedatives in an ICU setting were used to develop a support vector machine based system for sedation depth monitoring using several heart rate variability measures. A leave-one-subject-out cross validation was used for classifier training and performance evaluations. The proposed patient-specific system provided a sensitivity, specificity and an AUC of 64%, 84.8% and 0.72, respectively. It is hoped that with the help of additional physiological signals the proposed patient-specific sedation level prediction system could lead to a fully automated multimodal system to assist clinical staff in ICUs to interpret the sedation level of the patient.

Recording, analysis, and interpretation of spreading depolarizations in neurointensive care: Review and recommendations of the COSBID research group

Spreading depolarizations (SD) are waves of abrupt, near-complete breakdown of neuronal transmembrane ion gradients, are the largest possible pathophysiologic disruption of viable cerebral gray matter, and are a crucial mechanism of lesion development. Spreading depolarizations are increasingly recorded during multimodal neuromonitoring in neurocritical care as a causal biomarker providing a diagnostic summary measure of metabolic failure and excitotoxic injury. Focal ischemia causes spreading depolarization within minutes. Further spreading depolarizations arise for hours to days due to energy supply-demand mismatch in viable tissue. Spreading depolarizations exacerbate neuronal injury through prolonged ionic breakdown and spreading depolarization-related hypoperfusion (spreading ischemia). Local duration of the depolarization indicates local tissue energy status and risk of injury. Regional electrocorticographic monitoring affords even remote detection of injury because spreading depolarizations propagate widely from ischemic or metabolically stressed zones; characteristic patterns, including temporal clusters of spreading depolarizations and persistent depression of spontaneous cortical activity, can be recognized and quantified. Here, we describe the experimental basis for interpreting these patterns and illustrate their translation to human disease. We further provide consensus recommendations for electrocorticographic methods to record, classify, and score spreading depolarizations and associated spreading depressions. These methods offer distinct advantages over other neuromonitoring modalities and allow for future refinement through less invasive and more automated approaches.

The continuum of spreading depolarizations in acute cortical lesion development: Examining Leão's legacy

A modern understanding of how cerebral cortical lesions develop after acute brain injury is based on Aristides Leão's historic discoveries of spreading depression and asphyxial/anoxic depolarization. Treated as separate entities for decades, we now appreciate that these events define a continuum of spreading mass depolarizations, a concept that is central to understanding their pathologic effects. Within minutes of acute severe ischemia, the onset of persistent depolarization triggers the breakdown of ion homeostasis and development of cytotoxic edema. These persistent changes are diagnosed as diffusion restriction in magnetic resonance imaging and define the ischemic core. In delayed lesion growth, transient spreading depolarizations arise spontaneously in the ischemic penumbra and induce further persistent depolarization and excitotoxic damage, progressively expanding the ischemic core. The causal role of these waves in lesion development has been proven by real-time monitoring of electrophysiology, blood flow, and cytotoxic edema. The spreading depolarization continuum further applies to other models of acute cortical lesions, suggesting that it is a universal principle of cortical lesion development. These pathophysiologic concepts establish a working hypothesis for translation to human disease, where complex patterns of depolarizations are observed in acute brain injury and appear to mediate and signal ongoing secondary damage.

cEEG electrode-related pressure ulcers in acutely hospitalized patients

BACKGROUND

Pressure ulcers resulting from continuous EEG (cEEG) monitoring in hospitalized patients have gained attention as a preventable medical complication. We measured their incidence and risk factors.

METHODS

We performed an observational investigation of cEEG-electrode-related pressure ulcers (EERPU) among acutely ill patients over a 22-month period. Variables analyzed included age, sex, monitoring duration, hospital location, application methods, vasopressor usage, nutritional status, skin allergies, fever, and presence/severity of EERPU. We examined risk for pressure ulcers vs monitoring duration using Kaplan-Meyer survival analysis, and performed multivariate risk assessment using Cox proportional hazard model.

RESULTS

Among 1,519 patients, EERPU occurred in 118 (7.8%). Most (n = 109, 92.3%) consisted of hyperemia only without skin breakdown. A major predictor was monitoring duration, with 3-, 5-, and 10-day risks of 16%, 32%, and 60%, respectively. Risk factors included older age (mean age 60.65 vs 50.3, p < 0.01), care in an intensive care unit (9.37% vs 5.32%, p < 0.01), lack of a head wrap (8.31% vs 27.3%, p = 0.02), use of vasopressors (16.7% vs 9.64%, p < 0.01), enteral feeding (11.7% vs 5.45%, p = 0.04), and fever (18.4% vs 9.3%, p < 0.01). Elderly patients (71-80 years) were at higher risk (hazard ratio 6.84 [1.95-24], p < 0.01), even after accounting for monitoring time and other pertinent variables in multivariate analysis.

CONCLUSIONS

EERPU are uncommon and generally mild. Elderly patients and those with more severe illness have higher risk of developing EERPU, and the risk increases as a function of monitoring duration.

ADARRI: a novel method to detect spurious R-peaks in the electrocardiogram for heart rate variability analysis in the intensive care unit

We developed a simple and fully automated method for detecting artifacts in the R-R interval (RRI) time series of the ECG that is tailored to the intensive care unit (ICU) setting. From ECG recordings of 50 adult ICU-subjects we selected 60 epochs with valid R-peak detections and 60 epochs containing artifacts leading to missed or false positive R-peak detections. Next, we calculated the absolute value of the difference between two adjacent RRIs (adRRI), and obtained the empirical probability distributions of adRRI values for valid R-peaks and artifacts. From these, we calculated an optimal threshold for separating adRRI values arising from artifact versus non-artefactual data. We compared the performance of our method with the methods of Berntson and Clifford on the same data. We identified 257,458 R-peak detections, of which 235,644 (91.5%) were true detections and 21,814 (8.5%) arose from artifacts. Our method showed superior performance for detecting artifacts with sensitivity 100%, specificity 99%, precision 99%, positive likelihood ratio of 100 and negative likelihood ratio <0.001 compared to Berntson's and Clifford's method with a sensitivity, specificity, precision and positive and negative likelihood ratio of 99%, 78%, 82%, 4.5, 0.013 for Berntson's method and 55%, 98%, 96%, 27.5, 0.460 for Clifford's method, respectively. A novel algorithm using a patient-independent threshold derived from the distribution of adRRI values in ICU ECG data identifies artifacts accurately, and outperforms two other methods in common use. Furthermore, the threshold was calculated based on real data from critically ill patients and the algorithm is easy to implement.

Big data in sleep medicine: prospects and pitfalls in phenotyping

Clinical polysomnography (PSG) databases are a rich resource in the era of "big data" analytics. We explore the uses and potential pitfalls of clinical data mining of PSG using statistical principles and analysis of clinical data from our sleep center. We performed retrospective analysis of self-reported and objective PSG data from adults who underwent overnight PSG (diagnostic tests, n=1835). Self-reported symptoms overlapped markedly between the two most common categories, insomnia and sleep apnea, with the majority reporting symptoms of both disorders. Standard clinical metrics routinely reported on objective data were analyzed for basic properties (missing values, distributions), pairwise correlations, and descriptive phenotyping. Of 41 continuous variables, including clinical and PSG derived, none passed testing for normality. Objective findings of sleep apnea and periodic limb movements were common, with 51% having an apnea-hypopnea index (AHI) >5 per hour and 25% having a leg movement index >15 per hour. Different visualization methods are shown for common variables to explore population distributions. Phenotyping methods based on clinical databases are discussed for sleep architecture, sleep apnea, and insomnia. Inferential pitfalls are discussed using the current dataset and case examples from the literature. The increasing availability of clinical databases for large-scale analytics holds important promise in sleep medicine, especially as it becomes increasingly important to demonstrate the utility of clinical testing methods in management of sleep disorders. Awareness of the strengths, as well as caution regarding the limitations, will maximize the productive use of big data analytics in sleep medicine.

Epileptiform abnormalities predict delayed cerebral ischemia in subarachnoid hemorrhage

OBJECTIVE

To identify whether abnormal neural activity, in the form of epileptiform discharges and rhythmic or periodic activity, which we term here ictal-interictal continuum abnormalities (IICAs), are associated with delayed cerebral ischemia (DCI).

METHODS

Retrospective analysis of continuous electroencephalography (cEEG) reports and medical records from 124 patients with moderate to severe grade subarachnoid hemorrhage (SAH). We identified daily occurrence of seizures and IICAs. Using survival analysis methods, we estimated the cumulative probability of IICA onset time for patients with and without delayed cerebral ischemia (DCI).

RESULTS

Our data suggest the presence of IICAs indeed increases the risk of developing DCI, especially when they begin several days after the onset of SAH. We found that all IICA types except generalized rhythmic delta activity occur more commonly in patients who develop DCI. In particular, IICAs that begin later in hospitalization correlate with increased risk of DCI.

CONCLUSIONS

IICAs represent a new marker for identifying early patients at increased risk for DCI. Moreover, IICAs might contribute mechanistically to DCI and therefore represent a new potential target for intervention to prevent secondary cerebral injury following SAH.

SIGNIFICANCE

These findings imply that IICAs may be a novel marker for predicting those at higher risk for DCI development.

Response Rates to Anticonvulsant Trials in Patients with Triphasic-Wave EEG Patterns of Uncertain Significance

BACKGROUND

Generalized triphasic waves (TPWs) occur in both metabolic encephalopathies and non-convulsive status epilepticus (NCSE). Empiric trials of benzodiazepines (BZDs) or non-sedating AED (NSAEDs) are commonly used to differentiate the two, but the utility of such trials is debated. The goal of this study was to assess response rates of such trials and investigate whether metabolic profile differences affect the likelihood of a response.

METHODS

Three institutions within the Critical Care EEG Monitoring Research Consortium retrospectively identified patients with unexplained encephalopathy and TPWs who had undergone a trial of BZD and/or NSAEDs to differentiate between ictal and non-ictal patterns. We assessed responder rates and compared metabolic profiles of responders and non-responders. Response was defined as resolution of the EEG pattern and either unequivocal improvement in encephalopathy or appearance of previously absent normal EEG patterns, and further categorized as immediate (within <2 h of trial initiation) or delayed (>2 h from trial initiation).

RESULTS

We identified 64 patients with TPWs who had an empiric trial of BZD and/or NSAED. Most patients (71.9%) were admitted with metabolic derangements and/or infection. Positive clinical responses occurred in 10/53 (18.9%) treated with BZDs. Responses to NSAEDs occurred in 19/45 (42.2%), being immediate in 6.7%, delayed but definite in 20.0%, and delayed but equivocal in 15.6%. Overall, 22/64 (34.4%) showed a definite response to either BZDs or NSAEDs, and 7/64 (10.9%) showed a possible response. Metabolic differences of responders versus non-responders were statistically insignificant, except that the 48-h low value of albumin in the BZD responder group was lower than in the non-responder group.

CONCLUSIONS

Similar metabolic profiles in patients with encephalopathy and TPWs between responders and non-responders to anticonvulsants suggest that predicting responders a priori is difficult. The high responder rate suggests that empiric trials of anticonvulsants indeed provide useful clinical information. The more than twofold higher response rate to NSAEDs suggests that this strategy may be preferable to BZDs. Further prospective investigation is warranted.

Selection of first-line therapy in multiple sclerosis using risk-benefit decision analysis

OBJECTIVE

To integrate long-term measures of disease-modifying drug efficacy and risk to guide selection of first-line treatment of multiple sclerosis.

METHODS

We created a Markov decision model to evaluate disability worsening and progressive multifocal leukoencephalopathy (PML) risk in patients receiving natalizumab (NTZ), fingolimod (FGL), or glatiramer acetate (GA) over 30 years. Leveraging publicly available data, we integrated treatment utility, disability worsening, and risk of PML into quality-adjusted life-years (QALYs). We performed sensitivity analyses varying PML risk, mortality and morbidity, and relative risk of disease worsening across clinically relevant ranges.

RESULTS

Over the entire reported range of NTZ-associated PML risk, NTZ as first-line therapy is predicted to provide a greater net benefit (15.06 QALYs) than FGL (13.99 QALYs) or GA (12.71 QALYs) treatment over 30 years, after accounting for loss of QALYs due to PML or death (resulting from all causes). NTZ treatment is associated with delayed worsening to an Expanded Disability Status Scale score ≥6.0 vs FGL or GA (22.7, 17.0, and 12.4 years, respectively). Compared to untreated patients, NTZ-treated patients have a greater relative risk of death in the early years of treatment that varies according to PML risk profile.

CONCLUSIONS

NTZ as a first-line treatment is associated with the highest net benefit across full ranges of PML risk, mortality, and morbidity compared to FGL or GA. Integrated modeling of long-term treatment risks and benefits informs stratified clinical decision-making and can support patient counseling on selection of first-line treatment options.

Decision Modeling in Sleep Apnea: The Critical Roles of Pretest Probability, Cost of Untreated Obstructive Sleep Apnea, and Time Horizon

STUDY OBJECTIVES

Obstructive sleep apnea (OSA) is associated with increased morbidity and mortality, and treatment with positive airway pressure (PAP) is cost-effective. However, the optimal diagnostic strategy remains a subject of debate. Prior modeling studies have not consistently supported the widely held assumption that home sleep testing (HST) is cost-effective.

METHODS

We modeled four strategies: (1) treat no one; (2) treat everyone empirically; (3) treat those testing positive during in-laboratory polysomnography (PSG) via in-laboratory titration; and (4) treat those testing positive during HST with auto-PAP. The population was assumed to lack independent reasons for in-laboratory PSG (such as insomnia, periodic limb movements in sleep, complex apnea). We considered the third-party payer perspective, via both standard (quality-adjusted) and pure cost methods.

RESULTS

The preferred strategy depended on three key factors: pretest probability of OSA, cost of untreated OSA, and time horizon. At low prevalence and low cost of untreated OSA, the treat no one strategy was favored, whereas empiric treatment was favored for high prevalence and high cost of untreated OSA. In-laboratory backup for failures in the at-home strategy increased the preference for the at-home strategy. Without laboratory backup in the at-home arm, the in-laboratory strategy was increasingly preferred at longer time horizons.

CONCLUSION

Using a model framework that captures a broad range of clinical possibilities, the optimal diagnostic approach to uncomplicated OSA depends on pretest probability, cost of untreated OSA, and time horizon. Estimating each of these critical factors remains a challenge warranting further investigation.

Clinical Prediction Models for Sleep Apnea: The Importance of Medical History over Symptoms

STUDY OBJECTIVE

Obstructive sleep apnea (OSA) is a treatable contributor to morbidity and mortality. However, most patients with OSA remain undiagnosed. We used a new machine learning method known as SLIM (Supersparse Linear Integer Models) to test the hypothesis that a diagnostic screening tool based on routinely available medical information would be superior to one based solely on patient-reported sleep-related symptoms.

METHODS

We analyzed polysomnography (PSG) and self-reported clinical information from 1,922 patients tested in our clinical sleep laboratory. We used SLIM and 7 state-of-the-art classification methods to produce predictive models for OSA screening using features from: (i) self-reported symptoms; (ii) self-reported medical information that could, in principle, be extracted from electronic health records (demographics, comorbidities), or (iii) both.

RESULTS

For diagnosing OSA, we found that model performance using only medical history features was superior to model performance using symptoms alone, and similar to model performance using all features. Performance was similar to that reported for other widely used tools: sensitivity 64.2% and specificity 77%. SLIM accuracy was similar to state-of-the-art classification models applied to this dataset, but with the benefit of full transparency, allowing for hands-on prediction using yes/no answers to a small number of clinical queries.

CONCLUSION

To predict OSA, variables such as age, sex, BMI, and medical history are superior to the symptom variables we examined for predicting OSA. SLIM produces an actionable clinical tool that can be applied to data that is routinely available in modern electronic health records, which may facilitate automated, rather than manual, OSA screening.

COMMENTARY

A commentary on this article appears in this issue on page 159.

Alternative remedies for insomnia: a proposed method for personalized therapeutic trials

Insomnia is a common symptom, with chronic insomnia being diagnosed in 5-10% of adults. Although many insomnia patients use prescription therapy for insomnia, the health benefits remain uncertain and adverse risks remain a concern. While similar effectiveness and risk concerns exist for herbal remedies, many individuals turn to such alternatives to prescriptions for insomnia. Like prescription hypnotics, herbal remedies that have undergone clinical testing often show subjective sleep improvements that exceed objective measures, which may relate to interindividual heterogeneity and/or placebo effects. Response heterogeneity can undermine traditional randomized trial approaches, which in some fields has prompted a shift toward stratified trials based on genotype or phenotype, or the so-called n-of-1 method of testing placebo versus active drug in within-person alternating blocks. We reviewed six independent compendiums of herbal agents to assemble a group of over 70 reported to benefit sleep. To bridge the gap between the unfeasible expectation of formal evidence in this space and the reality of common self-medication by those with insomnia, we propose a method for guided self-testing that overcomes certain operational barriers related to inter- and intraindividual sources of phenotypic variability. Patient-chosen outcomes drive a general statistical model that allows personalized self-assessment that can augment the open-label nature of routine practice. The potential advantages of this method include flexibility to implement for other (nonherbal) insomnia interventions.

Causal inference as an emerging statistical approach in neurology: an example for epilepsy in the elderly

The elderly population faces an increasing number of cases of chronic neurological conditions, such as epilepsy and Alzheimer's disease. Because the elderly with epilepsy are commonly excluded from randomized controlled clinical trials, there are few rigorous studies to guide clinical practice. When the elderly are eligible for trials, they either rarely participate or frequently have poor adherence to therapy, thus limiting both generalizability and validity. In contrast, large observational data sets are increasingly available, but are susceptible to bias when using common analytic approaches. Recent developments in causal inference-analytic approaches also introduce the possibility of emulating randomized controlled trials to yield valid estimates. We provide a practical example of the application of the principles of causal inference to a large observational data set of patients with epilepsy. This review also provides a framework for comparative-effectiveness research in chronic neurological conditions.

Predicting neurosurgical outcomes in focal epilepsy patients using computational modelling

See Eissa and Schevon (doi:10.1093/aww332) for a scientific commentary on this article.

Surgery can be a last resort for patients with intractable, medically refractory epilepsy. For many of these patients, however, there is substantial risk that the surgery will be ineffective. The prediction of who is likely to benefit from a surgical approach is crucial for being able to inform patients better, conduct principled prospective clinical trials, and ultimately tailor therapeutic approaches to these patients more effectively. Dynamical computational models, informed with patient data, can be used to make predictions and give mechanistic insight. In this study, we develop patient-specific dynamical network models of epileptogenic cortex. We infer the network connectivity matrix from non-seizure electrographic recordings of patients and use these connectivity matrices as the network structure in our model. The model simulates the dynamics of a bi-stable switch at every node in this network, meaning that every node starts in a background state, but has the ability to transit to a co-existing seizure state. Whether a transition happens in a node is partly determined by the stochastic nature of the input to the node, but also by the input the node receives from other connected nodes in the network. By conducting simulations with such a model, we can detect the average transition time for nodes in a given network, and therefore define nodes with a short transition time as highly epileptogenic. In a retrospective study, we found that in some patients the regions with high epileptogenicity in the model overlap with those identified clinically as the seizure onset zone. Moreover, it was found that the resection of these regions in the model reduces the overall likelihood of a seizure. Following removal of these regions in the model, we predicted surgical outcomes and compared these to actual patient outcomes. Our predictions were found to be 81.3% accurate on a dataset of 16 patients with intractable epilepsy. Intriguingly, in patients with unsuccessful outcomes, the proposed computational approach is able to suggest alternative resection sites. The model presented here gives mechanistic insight as to why surgery may be unsuccessful in some patients. This may aid clinicians in presurgical evaluation by providing a tool to explore various surgical options, offering complementary information to existing clinical techniques.

Rapid annotation of interictal epileptiform discharges via template matching under Dynamic Time Warping

BACKGROUND

EEG interpretation relies on experts who are in short supply. There is a great need for automated pattern recognition systems to assist with interpretation. However, attempts to develop such systems have been limited by insufficient expert-annotated data. To address these issues, we developed a system named NeuroBrowser for EEG review and rapid waveform annotation.

NEW METHODS

At the core of NeuroBrowser lies on ultrafast template matching under Dynamic Time Warping, which substantially accelerates the task of annotation.

RESULTS

Our results demonstrate that NeuroBrowser can reduce the time required for annotation of interictal epileptiform discharges by EEG experts by 20-90%, with an average of approximately 70%.

COMPARISON WITH EXISTING METHOD(S)

In comparison with conventional manual EEG annotation, NeuroBrowser is able to save EEG experts approximately 70% on average of the time spent in annotating interictal epileptiform discharges. We have already extracted 19,000+ interictal epileptiform discharges from 100 patient EEG recordings. To our knowledge this represents the largest annotated database of interictal epileptiform discharges in existence.

CONCLUSION

NeuroBrowser is an integrated system for rapid waveform annotation. While the algorithm is currently tailored to annotation of interictal epileptiform discharges in scalp EEG recordings, the concepts can be easily generalized to other waveforms and signal types.

Clustering analysis to identify distinct spectral components of encephalogram burst suppression in critically ill patients

Millions of patients are admitted each year to intensive care units (ICUs) in the United States. A significant fraction of ICU survivors develop life-long cognitive impairment, incurring tremendous financial and societal costs. Delirium, a state of impaired awareness, attention and cognition that frequently develops during ICU care, is a major risk factor for post-ICU cognitive impairment. Recent studies suggest that patients experiencing electroencephalogram (EEG) burst suppression have higher rates of mortality and are more likely to develop delirium than patients who do not experience burst suppression. Burst suppression is typically associated with coma and deep levels of anesthesia or hypothermia, and is defined clinically as an alternating pattern of high-amplitude "burst" periods interrupted by sustained low-amplitude "suppression" periods. Here we describe a clustering method to analyze EEG spectra during burst and suppression periods. We used this method to identify a set of distinct spectral patterns in the EEG during burst and suppression periods in critically ill patients. These patterns correlate with level of patient sedation, quantified in terms of sedative infusion rates and clinical sedation scores. This analysis suggests that EEG burst suppression in critically ill patients may not be a single state, but instead may reflect a plurality of states whose specific dynamics relate to a patient's underlying brain function.

Therapeutic coma for status epilepticus: Differing practices in a prospective multicenter study

OBJECTIVE

Our aim was to analyze and compare the use of therapeutic coma (TC) for refractory status epilepticus (SE) across different centers and its effect on outcome.

METHODS

Clinical data for all consecutive adults (>16 years) with SE of all etiologies (except postanoxic) admitted to 4 tertiary care centers belonging to Harvard Affiliated Hospitals (HAH) and the Centre Hospitalier Universitaire Vaudois (CHUV) were prospectively collected and analyzed for TC details, mortality, and duration of hospitalization.

RESULTS

Two hundred thirty-six SE episodes in the CHUV and 126 in the HAH were identified. Both groups were homogeneous in demographics, comorbidities, SE characteristics, and Status Epilepticus Severity Score (STESS); TC was used in 25.4% of cases in HAH vs 9.75% in CHUV. After adjustment, TC use was associated with younger age, lower Charlson Comorbidity Index, increasing SE severity, refractory SE, and center (odds ratio 11.3 for HAH vs CHUV, 95% confidence interval 2.47-51.7). Mortality was associated with increasing Charlson Comorbidity Index and STESS, etiology, and refractory SE. Length of stay correlated with STESS, etiology, refractory SE, and use of TC (incidence rate ratio 1.6, 95% confidence interval 1.22-2.11).

CONCLUSIONS

Use of TC for SE treatment seems markedly different between centers from the United States and Europe, and did not affect mortality considering the whole cohort. However, TC may increase length of hospital stay and related costs.

CLASSIFICATION OF EVIDENCE

This study provides Class III evidence that for patients with SE, TC does not significantly affect mortality. The study lacked the precision to exclude an important effect of TC on mortality.

Spatial variation in automated burst suppression detection in pharmacologically induced coma

Burst suppression is actively studied as a control signal to guide anesthetic dosing in patients undergoing medically induced coma. The ability to automatically identify periods of EEG suppression and compactly summarize the depth of coma using the burst suppression probability (BSP) is crucial to effective and safe monitoring and control of medical coma. Current literature however does not explicitly account for the potential variation in burst suppression parameters across different scalp locations. In this study we analyzed standard 19-channel EEG recordings from 8 patients with refractory status epilepticus who underwent pharmacologically induced burst suppression as medical treatment for refractory seizures. We found that although burst suppression is generally considered a global phenomenon, BSP obtained using a previously validated algorithm varies systematically across different channels. A global representation of information from individual channels is proposed that takes into account the burst suppression characteristics recorded at multiple electrodes. BSP computed from this representative burst suppression pattern may be more resilient to noise and a better representation of the brain state of patients. Multichannel data integration may enhance the reliability of estimates of the depth of medical coma.

Automated information extraction from free-text EEG reports

In this study we have developed a supervised learning to automatically detect with high accuracy EEG reports that describe seizures and epileptiform discharges. We manually labeled 3,277 documents as describing one or more seizures vs no seizures, and as describing epileptiform discharges vs no epileptiform discharges. We then used Naïve Bayes to develop a system able to automatically classify EEG reports into these categories. Our system consisted of normalization techniques, extraction of key sentences, and automated feature selection using cross validation. As candidate features we used key words and special word patterns called elastic word sequences (EWS). Final feature selection was accomplished via sequential backward selection. We used cross validation to predict out of sample performance. Our automated feature selection procedure resulted in a classifier with 38 features for seizure detection, and 23 features for epileptiform discharge detection. The average [95% CI] area under the receiver operating curve was 99.05 [98.79, 99.32]% for detecting reports with seizures, and 96.15 [92.31, 100.00]% for detecting reports with epileptiform discharges. The methodology described herein greatly reduces the manual labor involved in identifying large cohorts of patients for retrospective neurophysiological studies of patients with epilepsy.

Continuous EEG monitoring enhances multimodal outcome prediction in hypoxic-ischemic brain injury

OBJECTIVE

Hypoxic brain injury is the largest contributor to disability and mortality after cardiac arrest. We aim to identify electroencephalogram (EEG) characteristics that can predict outcome on cardiac arrest patients treated with targeted temperature management (TTM).

METHODS

We retrospectively examined clinical, EEG, functional outcome at discharge, and in-hospital mortality for 373 adult subjects with return of spontaneous circulation after cardiac arrest. Poor outcome was defined as a Cerebral Performance Category score of 3-5. Pure suppression-burst (SB) was defined as SB not associated with status epilepticus (SE), seizures, or generalized periodic discharges.

RESULTS

In-hospital mortality was 68.6% (N=256). Presence of both unreactive EEG background and SE was associated with a positive predictive value (PPV) of 100% (95% confidence interval: 0.96-1) and a false-positive rate (FPR) of 0% (95% CI: 0-0.11) for poor functional outcome. A prediction model including demographics data, admission exam, presence of status epilepticus, pure SB, and lack of EEG reactivity had an area under the curve of 0.92 (95% CI: 0.87-0.95) for poor functional outcome prediction, and 0.96 (95% CI: 0.94-0.98) for in-hospital mortality. Presence of pure SB (N=87) was confounded by anesthetics use in 83.9% of the cases, and was not an independent predictor of poor functional outcome, having a FPR of 23% (95% CI: 0.19-0.28).

CONCLUSIONS

An unreactive EEG background and SE predicted poor functional outcome and in-hospital mortality in cardiac arrest patients undergoing TTM. Prognostic value of pure SB is confounded by use of sedative agents, and its use on prognostication decisions should be made with caution.

Sensitivity of quantitative EEG for seizure identification in the intensive care unit

OBJECTIVE

To evaluate the sensitivity of quantitative EEG (QEEG) for electrographic seizure identification in the intensive care unit (ICU).

METHODS

Six-hour EEG epochs chosen from 15 patients underwent transformation into QEEG displays. Each epoch was reviewed in 3 formats: raw EEG, QEEG + raw, and QEEG-only. Epochs were also analyzed by a proprietary seizure detection algorithm. Nine neurophysiologists reviewed raw EEGs to identify seizures to serve as the gold standard. Nine other neurophysiologists with experience in QEEG evaluated the epochs in QEEG formats, with and without concomitant raw EEG. Sensitivity and false-positive rates (FPRs) for seizure identification were calculated and median review time assessed.

RESULTS

Mean sensitivity for seizure identification ranged from 51% to 67% for QEEG-only and 63%-68% for QEEG + raw. FPRs averaged 1/h for QEEG-only and 0.5/h for QEEG + raw. Mean sensitivity of seizure probability software was 26.2%-26.7%, with FPR of 0.07/h. Epochs with the highest sensitivities contained frequent, intermittent seizures. Lower sensitivities were seen with slow-frequency, low-amplitude seizures and epochs with rhythmic or periodic patterns. Median review times were shorter for QEEG (6 minutes) and QEEG + raw analysis (14.5 minutes) vs raw EEG (19 minutes; p = 0.00003).

CONCLUSIONS

A panel of QEEG trends can be used by experts to shorten EEG review time for seizure identification with reasonable sensitivity and low FPRs. The prevalence of false detections confirms that raw EEG review must be used in conjunction with QEEG. Studies are needed to identify optimal QEEG trend configurations and the utility of QEEG as a screening tool for non-EEG personnel.

CLASSIFICATION OF EVIDENCE REVIEW

This study provides Class II evidence that QEEG + raw interpreted by experts identifies seizures in patients in the ICU with a sensitivity of 63%-68% and FPR of 0.5 seizures per hour.

Decision analysis of intracranial monitoring in non-lesional epilepsy

PURPOSE

Up to one third of epilepsy patients develop pharmacoresistant seizures and many benefit from resective surgery. However, patients with non-lesional focal epilepsy often require intracranial monitoring to localize the seizure focus. Intracranial monitoring carries operative morbidity risk and does not always succeed in localizing the seizures, making the benefit of this approach less certain. We performed a decision analysis comparing three strategies for patients with non-lesional focal epilepsy: (1) intracranial monitoring, (2) vagal nerve stimulator (VNS) implantation and (3) medical management to determine which strategy maximizes the expected quality-adjusted life years (QALYs) for our base cases.

METHOD

We constructed two base cases using parameters reported in the medical literature: (1) a young, otherwise healthy patient and (2) an elderly, otherwise healthy patient. We constructed a decision tree comprising strategies for the treatment of non-lesional epilepsy and two clinical outcomes: seizure freedom and no seizure freedom. Sensitivity analyses of probabilities at each branch were guided by data from the medical literature to define decision thresholds across plausible parameter ranges.

RESULTS

Intracranial monitoring maximizes the expected QALYs for both base cases. The sensitivity analyses provide estimates of the values of key variables, such as the surgical risk or the chance of localizing the focus, at which intracranial monitoring is no longer favored.

CONCLUSION

Intracranial monitoring is favored over VNS and medical management in young and elderly patients over a wide, clinically-relevant range of pertinent model variables such as the chance of localizing the seizure focus and the surgical morbidity rate.

EPILEPTIFORM SPIKE DETECTION VIA CONVOLUTIONAL NEURAL NETWORKS

The EEG of epileptic patients often contains sharp waveforms called "spikes", occurring between seizures. Detecting such spikes is crucial for diagnosing epilepsy. In this paper, we develop a convolutional neural network (CNN) for detecting spikes in EEG of epileptic patients in an automated fashion. The CNN has a convolutional architecture with filters of various sizes applied to the input layer, leaky ReLUs as activation functions, and a sigmoid output layer. Balanced mini-batches were applied to handle the imbalance in the data set. Leave-one-patient-out cross-validation was carried out to test the CNN and benchmark models on EEG data of five epilepsy patients. We achieved 0.947 AUC for the CNN, while the best performing benchmark model, Support Vector Machines with Gaussian kernel, achieved an AUC of 0.912.

CLUSTERING OF INTERICTAL SPIKES BY DYNAMIC TIME WARPING AND AFFINITY PROPAGATION

Epilepsy is often associated with the presence of spikes in electroencephalograms (EEGs). The spike waveforms vary vastly among epilepsy patients, and also for the same patient across time. In order to develop semi-automated and automated methods for detecting spikes, it is crucial to obtain a better understanding of the various spike shapes. In this paper, we develop several approaches to extract exemplars of spikes. We generate spike exemplars by applying clustering algorithms to a database of spikes from 12 patients. As similarity measures for clustering, we consider the Euclidean distance and Dynamic Time Warping (DTW). We assess two clustering algorithms, namely, K-means clustering and affinity propagation. The clustering methods are compared based on the mean squared error, and the similarity measures are assessed based on the number of generated spike clusters. Affinity propagation with DTW is shown to be the best combination for clustering epileptic spikes, since it generates fewer spike templates and does not require to pre-specify the number of spike templates.

FAST AND EFFICIENT REJECTION OF BACKGROUND WAVEFORMS IN INTERICTAL EEG

Automated annotation of electroencephalograms (EEG) of epileptic patients is important in diagnosis and management of epilepsy. Epilepsy is often associated with the presence of epileptiform transients (ET) in the EEG. To develop an efficient ET detector, a vast amount of data is required to train and evaluate the performance of the detector. Interictal EEG data contains mostly background waveforms, since ETs only occur occasionally in most patients. In order to detect ETs in an automated fashion, it is meaningful to first try to eliminate most background waveforms by means of simple, fast classifiers. The remaining waveforms can in a following step be processed by more sophisticated and computationally demanding classification algorithms, such as deep learning systems. In this study, we design a cascade of simple thresholding steps to reject most background waveforms in interictal EEG, while maintaining most ETs. Several simple and quick-to-compute EEG features are chosen. By thresholding these features in consecutive steps, background waveforms are rejected sequentially. In our numerical experiments, a cascade of 10 steps is able to reject 98.65% of all background segments in the dataset, while preserving 90.6% of the ETs.

The challenge of undiagnosed sleep apnea in low-risk populations: a decision analysis

OBJECTIVES

Obstructive sleep apnea (OSA) may contribute to impaired performance among otherwise healthy active duty military personnel. We used decision analysis to evaluate three approaches to identifying and treating OSA in low-risk populations, which may differ from current standard practice for high-risk populations.

METHODS

We developed a decision tree to compare two simple strategies for diagnosis and management of sleep apnea in a low-risk population. In one strategy, a simple screening inventory was followed by conventional laboratory polysomnography (split-night), whereas the alternative strategy involved performing home testing in all individuals. This allowed us to weigh the costs associated with large-scale diagnostic approaches against the costs of untreated OSA in a small fraction of the population.

RESULTS

We found that the home testing approach was less expensive than the screen-then-test approach across a broad range of other important parameters, including the annual performance cost associated with untreated OSA, the prevalence of OSA, and the duration of active duty.

CONCLUSIONS

Assuming even modest annual performance costs associated with untreated OSA, a population strategy involving large-scale home testing is less expensive than a screening inventory approach. These results may inform either targeted or large-scale investigation of undiagnosed OSA in low-risk populations such as active duty military.

Monitoring burst suppression in critically ill patients: Multi-centric evaluation of a novel method

OBJECTIVE

To develop a computational method to detect and quantify burst suppression patterns (BSP) in the EEGs of critical care patients. A multi-center validation study was performed to assess the detection performance of the method.

METHODS

The fully automatic method scans the EEG for discontinuous patterns and shows detected BSP and quantitative information on a trending display in real-time. The method is designed to work without setting any patient specific parameters and to be insensitive to EEG artifacts and periodic patterns. For validation a total of 3982 h of EEG from 88 patients were analyzed from three centers. Each EEG was annotated by two reviewers to assess the detection performance and the inter-rater agreement.

RESULTS

Average inter-rater agreement between pairs of reviewers was κ=0.69. On average 22% of the review segments included BSP. An average sensitivity of 90% and a specificity of 84% were measured on the consensus annotations of two reviewers. More than 95% of the periodic patterns in the EEGs were correctly suppressed.

CONCLUSION

A fully automatic method to detect burst suppression patterns was assessed in a multi-center study. The method showed high sensitivity and specificity.

SIGNIFICANCE

Clinically applicable burst suppression detection method validated in a large multi-center study.

Age-dependency of sevoflurane-induced electroencephalogram dynamics in children

BACKGROUND

General anaesthesia induces highly structured oscillations in the electroencephalogram (EEG) in adults, but the anaesthesia-induced EEG in paediatric patients is less understood. Neural circuits undergo structural and functional transformations during development that might be reflected in anaesthesia-induced EEG oscillations. We therefore investigated age-related changes in the EEG during sevoflurane general anaesthesia in paediatric patients.

METHODS

We analysed the EEG recorded during routine care of patients between 0 and 28 yr of age (n=54), using power spectral and coherence methods. The power spectrum quantifies the energy in the EEG at each frequency, while the coherence measures the frequency-dependent correlation or synchronization between EEG signals at different scalp locations. We characterized the EEG as a function of age and within 5 age groups: <1 yr old (n=4), 1-6 yr old (n=12), >6-14 yr old (n=14), >14-21 yr old (n=11), >21-28 yr old (n=13).

RESULTS

EEG power significantly increased from infancy through ∼6 yr, subsequently declining to a plateau at approximately 21 yr. Alpha (8-13 Hz) coherence, a prominent EEG feature associated with sevoflurane-induced unconsciousness in adults, is absent in patients <1 yr.

CONCLUSIONS

Sevoflurane-induced EEG dynamics in children vary significantly as a function of age. These age-related dynamics likely reflect ongoing development within brain circuits that are modulated by sevoflurane. These readily observed paediatric-specific EEG signatures could be used to improve brain state monitoring in children receiving general anaesthesia.

Robust control of burst suppression for medical coma

OBJECTIVE

Medical coma is an anesthetic-induced state of brain inactivation, manifest in the electroencephalogram by burst suppression. Feedback control can be used to regulate burst suppression, however, previous designs have not been robust. Robust control design is critical under real-world operating conditions, subject to substantial pharmacokinetic and pharmacodynamic parameter uncertainty and unpredictable external disturbances. We sought to develop a robust closed-loop anesthesia delivery (CLAD) system to control medical coma.

APPROACH

We developed a robust CLAD system to control the burst suppression probability (BSP). We developed a novel BSP tracking algorithm based on realistic models of propofol pharmacokinetics and pharmacodynamics. We also developed a practical method for estimating patient-specific pharmacodynamics parameters. Finally, we synthesized a robust proportional integral controller. Using a factorial design spanning patient age, mass, height, and gender, we tested whether the system performed within clinically acceptable limits. Throughout all experiments we subjected the system to disturbances, simulating treatment of refractory status epilepticus in a real-world intensive care unit environment.

MAIN RESULTS

In 5400 simulations, CLAD behavior remained within specifications. Transient behavior after a step in target BSP from 0.2 to 0.8 exhibited a rise time (the median (min, max)) of 1.4 [1.1, 1.9] min; settling time, 7.8 [4.2, 9.0] min; and percent overshoot of 9.6 [2.3, 10.8]%. Under steady state conditions the CLAD system exhibited a median error of 0.1 [-0.5, 0.9]%; inaccuracy of 1.8 [0.9, 3.4]%; oscillation index of 1.8 [0.9, 3.4]%; and maximum instantaneous propofol dose of 4.3 [2.1, 10.5] mg kg(-1). The maximum hourly propofol dose was 4.3 [2.1, 10.3] mg kg(-1) h(-1). Performance fell within clinically acceptable limits for all measures.

SIGNIFICANCE

A CLAD system designed using robust control theory achieves clinically acceptable performance in the presence of realistic unmodeled disturbances and in spite of realistic model uncertainty, while maintaining infusion rates within acceptable safety limits.

The number of seizures needed in the EMU

OBJECTIVE

The purpose of this study was to develop a quantitative framework to estimate the likelihood of multifocal epilepsy based on the number of unifocal seizures observed in the epilepsy monitoring unit (EMU).

METHODS

Patient records from the EMU at Massachusetts General Hospital (MGH) from 2012 to 2014 were assessed for the presence of multifocal seizures as well the presence of multifocal interictal discharges and multifocal structural imaging abnormalities during the course of the EMU admission. Risk factors for multifocal seizures were assessed using sensitivity and specificity analysis. A Kaplan-Meier survival analysis was used to estimate the risk of multifocal epilepsy for a given number of consecutive seizures. To overcome the limits of the Kaplan-Meier analysis, a parametric survival function was fit to the EMU subjects with multifocal seizures and this was used to develop a Bayesian model to estimate the risk of multifocal seizures during an EMU admission.

RESULTS

Multifocal interictal discharges were a significant predictor of multifocal seizures within an EMU admission with a p < 0.01, albeit with only modest sensitivity 0.74 and specificity 0.69. Multifocal potentially epileptogenic lesions on MRI were not a significant predictor p = 0.44. Kaplan-Meier analysis was limited by wide confidence intervals secondary to significant patient dropout and concern for informative censoring. The Bayesian framework provided estimates for the number of unifocal seizures needed to predict absence of multifocal seizures. To achieve 90% confidence for the absence of multifocal seizure, three seizures are needed when the pretest probability for multifocal epilepsy is 20%, seven seizures for a pretest probability of 50%, and nine seizures for a pretest probability of 80%.

SIGNIFICANCE

These results provide a framework to assist clinicians in determining the utility of trying to capture a specific number of seizures in EMU evaluations of candidates for epilepsy surgery.

High risk for seizures following subarachnoid hemorrhage regardless of referral bias

BACKGROUND

To investigate the frequency, predictors, and clinical impact of electrographic seizures in patients with high clinical or radiologic grade non-traumatic subarachnoid hemorrhage (SAH), independent of referral bias.

METHODS

We compared rates of electrographic seizures and associated clinical variables and outcomes in patients with high clinical or radiologic grade non-traumatic SAH. Rates of electrographic seizure detection before and after institution of a guideline which made continuous EEG monitoring routine in this population were compared.

RESULTS

Electrographic seizures occurred in 17.6 % of patients monitored expressly because of clinically suspected subclinical seizures. In unselected patients, seizures still occurred in 9.6 % of all cases, and in 8.6 % of cases in which there was no a priori suspicion of seizures. The first seizure detected occurred 5.4 (IQR 2.9-7.3) days after onset of subarachnoid hemorrhage with three of eight patients (37.5 %) having the first recorded seizure more than 48 h following EEG initiation, and 2/8 (25 %) at more than 72 h following EEG initiation. High clinical grade was associated with poor outcome at time of hospital discharge; electrographic seizures were not associated with poor outcome.

CONCLUSIONS

Electrographic seizures occur at a relatively high rate in patients with non-traumatic SAH even after accounting for referral bias. The prolonged time to the first detected seizure in this cohort may reflect dynamic clinical features unique to the SAH population.

Estimating Total Cerebral Microinfarct Burden From Diffusion-Weighted Imaging

BACKGROUND AND PURPOSE

Cerebral microinfarcts (CMI) are important contributors to vascular cognitive impairment. Magnetic resonance imaging diffusion-weighted imaging (DWI) hyperintensities have been suggested to represent acute CMI. We aim to describe a mathematical method for estimating total number of CMI based on the presence of incidental DWI lesions.

METHODS

We reviewed magnetic resonance imaging scans of subjects with cognitive decline, cognitively normal subjects and previously reported subjects with past intracerebral hemorrhage (ICH). Based on temporal and spatial characteristics of DWI lesions, we estimated the annual rate of CMI needed to explain the observed rate of DWI lesion detection in each group. To confirm our estimates, we performed extensive sampling for CMI in the brain of a deceased subject with past lobar ICH who found to have a DWI lesion during life.

RESULTS

Clinically silent DWI lesions were present in 13 of 343 (3.8%) cognitively impaired and 10 of 199 (5%) cognitively intact normal non-ICH patients, both lower than the incidence in the past ICH patients (23 of 178; 12.9%; P<0.0006). The predicted annual incidence of CMI ranges from 16 to 1566 for non-ICH and 50 to 5041 for ICH individuals. Histological sampling revealed a total of 60 lesions in 32 sections. Based on previously reported methods, this density of CMI yields an estimated total brain burden maximum likelihood estimate of 9321 CMIs (95% confidence interval, 7255-11 990).

CONCLUSIONS

Detecting even a single DWI lesion suggests an annual incidence of hundreds of new CMI. The cumulative effects of these lesions may directly contribute to small-vessel-related vascular cognitive impairment.

Practice variability and efficacy of clonazepam, lorazepam, and midazolam in status epilepticus: A multicenter comparison

OBJECTIVE

Benzodiazepines (BZD) are recommended as first-line treatment for status epilepticus (SE), with lorazepam (LZP) and midazolam (MDZ) being the most widely used drugs and part of current treatment guidelines. Clonazepam (CLZ) is also utilized in many countries; however, there is no systematic comparison of these agents for treatment of SE to date.

METHODS

We identified all patients treated with CLZ, LZP, or MDZ as a first-line agent from a prospectively collected observational cohort of adult patients treated for SE in four tertiary care centers. Relative efficacies of CLZ, LZP, and MDZ were compared by assessing the risk of developing refractory SE and the number of antiseizure drugs (ASDs) required to control SE.

RESULTS

Among 177 patients, 72 patients (40.62%) received CLZ, 82 patients (46.33%) LZP, and 23 (12.99%) MDZ; groups were similar in demographics and SE characteristics. Loading dose was considered insufficient in the majority of cases for LZP, with a similar rate (84%, 95%, and 87.5%) in the centers involved, and CLZ was used as recommended in 52% of patients. After adjustment for relevant variables, LZP was associated with an increased risk of refractoriness as compared to CLZ (odds ratio [OR] 6.4, 95% confidence interval [CI] 2.66-15.5) and with an increased number of ASDs needed for SE control (OR 4.35, 95% CI 1.8-10.49).

SIGNIFICANCE

CLZ seems to be an effective alternative to LZP and MDZ. LZP is frequently underdosed in this setting. These findings are highly relevant, since they may impact daily practice.

Quantification of EEG reactivity in comatose patients

OBJECTIVE

EEG reactivity is an important predictor of outcome in comatose patients. However, visual analysis of reactivity is prone to subjectivity and may benefit from quantitative approaches.

METHODS

In EEG segments recorded during reactivity testing in 59 comatose patients, 13 quantitative EEG parameters were used to compare the spectral characteristics of 1-minute segments before and after the onset of stimulation (spectral temporal symmetry). Reactivity was quantified with probability values estimated using combinations of these parameters. The accuracy of probability values as a reactivity classifier was evaluated against the consensus assessment of three expert clinical electroencephalographers using visual analysis.

RESULTS

The binary classifier assessing spectral temporal symmetry in four frequency bands (delta, theta, alpha and beta) showed best accuracy (Median AUC: 0.95) and was accompanied by substantial agreement with the individual opinion of experts (Gwet's AC1: 65-70%), at least as good as inter-expert agreement (AC1: 55%). Probability values also reflected the degree of reactivity, as measured by the inter-experts' agreement regarding reactivity for each individual case.

CONCLUSION

Automated quantitative EEG approaches based on probabilistic description of spectral temporal symmetry reliably quantify EEG reactivity.

SIGNIFICANCE

Quantitative EEG may be useful for evaluating reactivity in comatose patients, offering increased objectivity.

Characteristics and role in outcome prediction of continuous EEG after status epilepticus: A prospective observational cohort

OBJECTIVE

Continuous electroencephalography (cEEG) is important for treatment guidance in status epilepticus (SE) management, but its role in clinical outcome prediction is unclear. Our aim is to determine which cEEG features give independent outcome information after correction for clinical predictor.

METHODS

cEEG data of 120 consecutive adult patients with SE were prospectively collected in three academic medical centers using the 2012 American Clinical Neurophysiology Society's Standardized Critical Care EEG Terminology. Association between cEEG features and two clinical outcome measures (mortality and complete recovery) was assessed.

RESULTS

In the first 24 h of EEG recording, 49 patients (40.8%) showed no periodic or rhythmic pattern, 45 (37.5%) had periodic discharges, 20 (16.7%) had rhythmic delta activity, and 6 (5%) had spike-and-wave discharges. Seizures were recorded in 68.3% of patients. After adjusting for known clinical predictive factors for mortality including the STatus Epilepticus Severity Score (STESS) and the presence of a potentially fatal etiology, the only EEG features (among rhythmic and periodic patterns, seizures, and background activity) that remained significantly associated with outcome were the absence of a posterior dominant rhythm (odds ratio [OR] 9.8; p = 0.033) for mortality and changes in stage II sleep pattern characteristics (OR 2.59 for each step up among these categories: absent, present and abnormal, present and normal; p = 0.002) for complete recovery.

SIGNIFICANCE

After adjustment for relevant clinical findings, including SE severity and etiology, cEEG background information (posterior dominant rhythm and sleep patterns) is more predictive for clinical outcome after SE than are rhythmic and periodic patterns or seizures.

Stimulus information stored in lasting active and hidden network states is destroyed by network bursts

In both humans and animals brief synchronizing bursts of epileptiform activity known as interictal epileptiform discharges (IEDs) can, even in the absence of overt seizures, cause transient cognitive impairments (TCI) that include problems with perception or short-term memory. While no evidence from single units is available, it has been assumed that IEDs destroy information represented in neuronal networks. Cultured neuronal networks are a model for generic cortical microcircuits, and their spontaneous activity is characterized by the presence of synchronized network bursts (SNBs), which share a number of properties with IEDs, including the high degree of synchronization and their spontaneous occurrence in the absence of an external stimulus. As a model approach to understanding the processes underlying IEDs, optogenetic stimulation and multielectrode array (MEA) recordings of cultured neuronal networks were used to study whether stimulus information represented in these networks survives SNBs. When such networks are optically stimulated they encode and maintain stimulus information for as long as one second. Experiments involved recording the network response to a single stimulus and trials where two different stimuli were presented sequentially, akin to a paired pulse trial. We broke the sequential stimulus trials into encoding, delay and readout phases and found that regardless of which phase the SNB occurs, stimulus-specific information was impaired. SNBs were observed to increase the mean network firing rate, but this did not translate monotonically into increases in network entropy. It was found that the more excitable a network, the more stereotyped its response was during a network burst. These measurements speak to whether SNBs are capable of transmitting information in addition to blocking it. These results are consistent with previous reports and provide baseline predictions concerning the neural mechanisms by which IEDs might cause TCI.

Sensitivity of compressed spectral arrays for detecting seizures in acutely ill adults

BACKGROUND

Continuous EEG recordings (cEEGs) are increasingly used in evaluation of acutely ill adults. Pre-screening using compressed data formats, such as compressed spectral array (CSA), may accelerate EEG review. We tested whether screening with CSA can enable detection of seizures and other relevant patterns.

METHODS

Two individuals reviewed the CSA displays of 113 cEEGs. While blinded to the raw EEG data, they marked each visually homogeneous CSA segment. An independent experienced electroencephalographer reviewed the raw EEG within 60 s on either side of each mark and recorded any seizures (and isolated epileptiform discharges, periodic epileptiform discharges (PEDs), rhythmic delta activity (RDA), and focal or generalized slowing). Seizures were considered to have been detected if the CSA mark was within 60 s of the seizure. The electroencephalographer then determined the total number of seizures (and other critical findings) for each record by exhaustive, page-by-page review of the entire raw EEG.

RESULTS

Within each of the 39 cEEG recordings containing seizures, one CSA reviewer identified at least one seizure, while the second CSA reviewer identified 38/39 patients with seizures. The overall detection rate was 89.0 % of 1,190 total seizures. When present, an average of 87.9 % of seizures were detected per individual patient. Detection rates for other critical findings were as follows: epileptiform discharges, 94.0 %; PEDs, 100 %; RDA, 97.9 %; focal slowing, 100 %; and generalized slowing, 100 %.

CONCLUSIONS

CSA-guided review can support sensitive screening of critical pathological information in cEEG recordings. However, some patients with seizures may not be identified.

Treating seizures in Creutzfeldt-Jakob disease

Seizures are known to occur in Creutzfeldt–Jakob disease (CJD). In the setting of a rapidly progressive condition with no effective therapy, determining appropriate treatment for seizures can be difficult if clinical morbidity is not obvious yet the electroencephalogram (EEG) demonstrates a worrisome pattern such as status epilepticus. Herein, we present the case of a 39-year-old man with CJD and electrographic seizures, discuss how this case challenges conventional definitions of seizures, and discuss a rational approach toward treatment. Coincidentally, our case is the first report of CJD in a patient with Stickler syndrome.

Physiological consequences of abnormal connectivity in a developmental epilepsy

OBJECTIVE

Many forms of epilepsy are associated with aberrant neuronal connections, but the relationship between such pathological connectivity and the underlying physiological predisposition to seizures is unclear. We sought to characterize the cortical excitability profile of a developmental form of epilepsy known to have structural and functional connectivity abnormalities.

METHODS

We employed transcranial magnetic stimulation (TMS) with simultaneous electroencephalographic (EEG) recording in 8 patients with epilepsy from periventricular nodular heterotopia and matched healthy controls. We used connectivity imaging findings to guide TMS targeting and compared the evoked responses to single-pulse stimulation from different cortical regions.

RESULTS

Heterotopia patients with active epilepsy demonstrated a relatively augmented late cortical response that was greater than that of matched controls. This abnormality was specific to cortical regions with connectivity to subcortical heterotopic gray matter. Topographic mapping of the late response differences showed distributed cortical networks that were not limited to the stimulation site, and source analysis in 1 subject revealed that the generator of abnormal TMS-evoked activity overlapped with the spike and seizure onset zone.

INTERPRETATION

Our findings indicate that patients with epilepsy from gray matter heterotopia have altered cortical physiology consistent with hyperexcitability, and that this abnormality is specifically linked to the presence of aberrant connectivity. These results support the idea that TMS-EEG could be a useful biomarker in epilepsy in gray matter heterotopia, expand our understanding of circuit mechanisms of epileptogenesis, and have potential implications for therapeutic neuromodulation in similar epileptic conditions associated with deep lesions.

Propofol and sevoflurane induce distinct burst suppression patterns in rats

Burst suppression is an EEG pattern characterized by alternating periods of high-amplitude activity (bursts) and relatively low amplitude activity (suppressions). Burst suppression can arise from several different pathological conditions, as well as from general anesthesia. Here we review current algorithms that are used to quantify burst suppression, its various etiologies, and possible underlying mechanisms. We then review clinical applications of anesthetic-induced burst suppression. Finally, we report the results of our new study showing clear electrophysiological differences in burst suppression patterns induced by two common general anesthetics, sevoflurane and propofol. Our data suggest that the circuit mechanisms that generate burst suppression activity may differ among general anesthetics.