Prognosis of comatose patients with reduced EEG montage by combining quantitative EEG features in various domains.

Objective: As the frontoparietal network underlies recovery from coma, a limited frontoparietal montage was used, and the prognostic values of EEG features for comatose patients were assessed. Methods: Collected with a limited frontoparietal EEG montage, continuous EEG recordings of 81 comatose patients in ICU were used retrospectively. By the 60-day Glasgow outcome scale (GOS), the patients were dichotomized into favorable and unfavorable outcome groups. Temporal-, frequency-, and spatial-domain features were automatically extracted for comparison. Partial correlation analysis was applied to eliminate redundant factors, and multiple correspondence analysis was used to explore discrimination between groups. Prognostic characteristics were calculated to assess the performance of EEG feature-based predictors established by logistic regression. Analyses were performed on all-patients group, strokes subgroup, and traumatic brain injury (TBI) subgroup. Results: By analysis of all patients, raised burst suppression ratio (BSR), suppressed root mean square (RMS), raised power ratio of ß to α rhythm (ß/α), and suppressed phase-lag index between F3 and P4 (PLI [F3, P4]) were associated with unfavorable outcome, and yielded AUC of 0.790, 0.811, 0.722, and 0.844, respectively. For the strokes subgroup, the significant variables were BSR, RMS, θ/total, θ/δ, and PLI (F3, P4), while for the TBI subgroup, only PLI (F3, P4) was significant. BSR combined with PLI (F3, P4) gave the best predictor by cross-validation analysis in the all-patients group (AUC = 0.889, 95% CI: 0.819-0.960). Conclusion: Features extracted from limited frontoparietal montage EEG served as valuable coma prognostic tools, where PLI (F3, P4) was always significant. Combining PLI (F3, P4) with features in other domains may achieve better performance. Significance: A limited-montage EEG coupled with an automated algorithm is valuable for coma prognosis.

An automatic detection method for 40-Hz auditory steady state response and its application in prognosis of comatose patients.

OBJECTIVE: We proposed a 40-Hz auditory steady-state response (ASSR) automatic detection method, and studied the prognosis of comatose patients by combining the 40-Hz ASSR detection results of multiple paradigms of auditory stimulation. METHODS: The 40-Hz ASSR elicitation experiments were carried out on 32 comatose patients, with the detection results used as prognosis predictors. To achieve automatic detection, the detection was modeled as a binary hypothesis test for a sinusoidal waveform with unknown amplitude and phase, based on the generalized likelihood ratio test (GLRT). The patients were followed up for 6 months, and each patient's outcome was classified as either favorable outcome (severe disability, moderate disability or good recovery) or unfavorable outcome (vegetative state/unresponsive wakefulness syndrome or death) according to the Glasgow outcome scale (GOS). The performance of the prognosis predictors was assessed using the area under the receiver operating characteristic curve (AUC-ROC). RESULTS: The largest AUC in univariate analysis involving a single stimulation paradigm was 0.849, while the AUC obtained by combining multiple predictors was increased to 0.966. CONCLUSIONS: For comatose patients, the absence of 40-Hz ASSR in multiple stimulation paradigms may indicate an unfavorable prognosis. Furthermore, the combination of multiple auditory stimulation paradigms may increase the outcome prediction accuracy. SIGNIFICANCE: The combination of multi-paradigm 40-Hz ASSR automatic detection results may provide a feasible automatic outcome prediction method for comatose patients.

Recommendations for the use of electroencephalography and evoked potentials in comatose patients.

Predicting the outcome of a comatose or poorly responsive patient is a major issue for intensive care unit teams, in order to give the most accurate information to the family and to choose the best therapeutic option. However, determining the level of cortical activity in patients with disorders of consciousness is a real challenge. Reliable criteria are required to help clinicians in the decision-making process, especially in the acute phase of coma. In this paper, we propose recommendations for recording and interpreting electroencephalography and evoked potentials in comatose patients based on the literature and the clinical experience of a group of neurophysiologists trained in the management of comatose patients. We propose methodological guidelines and discuss prognostic value of each test as well as the limitations concerning recording and interpretation. Recommendations for the strategy and timing of neurophysiological assessments are also proposed according to various clinical situations.

Large inter-rater variability on EEG-reactivity is improved by a novel quantitative method.

OBJECTIVE: To assess inter-rater agreement on EEG-reactivity (EEG-R) in comatose patients and compare it with a quantitative method (QEEG-R). METHODS: Six 30-s stimulation epochs (noxious, visual and auditory) were performed during EEG on 19 neurosurgical and 11 cardiac arrest patients. Six experts analysed EEGs for reactivity using their habitual methods. QEEG-R was defined as present if ≥2/6 epochs were reactive (stimulation/rest power ratio exceeding noise level). Three-months patient outcome was assessed by the Cerebral Performance Category Score (CPC) dichotomized in good (1-2) or poor (3-5). RESULTS: Agreement among experts on overall EEG-R varied from 53% to 83% (κ: 0.05-0.64) and reached 100% (κ: 1) between two QEEG-R calculators. For the experts, absence of EEG-R yielded sensitivities for poor outcome between 40-85% and specificities between 20-90%, for QEEG-R sensitivity was 40% (CI: 23-68%) and specificity 100% (CI: 69-100%). CONCLUSIONS: There is a large inter-rater variation among experts on EEG-R assessment in comatose patients. QEEG-R is a promising objective prognostic parameter with low inter-rater variation and a high specificity for prediction of poor outcome. SIGNIFICANCE: Clinicians should be cautious when using the traditional, qualitative method, in particular in end-of-life decisions. Implementation of the quantitative method in clinical practice may improve reliability of reactivity assessments.

The Nature of EEG Reactivity to Light, Sound, and Pain Stimulation in Neurosurgical Comatose Patients Evaluated by a Quantitative Method.

EEG reactivity (EEG-R) is regarded as an important parameter in coma prognosis but knowledge is sparse on the nature of EEG changes due to different kinds of stimulation and their prognostic significance. EEG-R was quantified in a study of 39 comatose neurosurgical patients. Six 30-second standardized visual, auditory, and painful stimulations were applied. EEG-R in the delta, theta, alpha, and beta band was normalized in z-scores as the power of a stimulation epoch relative to average power of 6 resting epochs. Outcome measure was 3 months Glasgow Outcome Scale. Increase in EEG activity was related to poor outcome, was more common (13.4% of tests), and grew continuously during the 30-second stimulation epoch. Decrease in EEG activity was related to good outcome, was rarer (2.5%), and peaked around 15 seconds. Pain was the most provocative stimulation (20.4%) followed by sound (8.7%) and eye-opening (6.7%). Discrimination between good (n = 6) and poor (n = 33) outcome was best in the theta and alpha bands for pain stimulation in the first 10-20 seconds and for sound stimulation in the first 5 to 10 seconds, eye-opening did not discriminate. Increase in activity predicted poor outcome with a high specificity 100% (CI = 52%-100%) and a modest sensitivity of 39% (CI = 23%-58%). Decrease in activity predicted good outcome with a high specificity of 100% (CI = 87%-100%) and a modest sensitivity of 33% (CI = 6%-76%). This quantitative study reveals new knowledge about the nature of EEG-R, which contribute to the development of more reliable and objective clinical procedures for outcome prediction.