The Narcotrend Depth of Anaesthesia Monitor as an Indicator of Seizures and Burst Suppression in the Intensive Care Unit

Sneath, Helen L. (2025) The Narcotrend Depth of Anaesthesia Monitor as an Indicator of Seizures and Burst Suppression in the Intensive Care Unit. Doctoral thesis (DClinSci), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (14MB) | Preview

Abstract

Aim The aim of this prospective study was to investigate the utility of the Narcotrend Depth of Anaesthesia (DoA) device in monitoring for seizures and burst suppression on the adult and paediatric intensive care unit at Nottingham University Hospitals (NUH) Trust. Method Standard 2-channel cEEG / CFAM and 2-channel frontal Narcotrend monitoring was recorded concurrently in all patients referred for routine cEEG / CFAM monitoring. Twice-daily, time-matched review periods were assessed for the presence of seizures and burst suppression using raw EEG, amplitude EEG (aEEG) and density spectral array (DSA). Review was conducted independently by 3 experienced clinical physiologists / scientists in clinical neurophysiology and a group of 7 non-expert advanced critical care practitioners (ACCPs) who reviewed recordings independently but acted as a pool of reviewers. ACCPs reviewed both CFAM and Narcotrend recordings to differentiate skill of interpretation and utility of the Narcotrend device. Cohen’s kappa, McNemar’s test, Fleiss’ kappa and Receiver Operating Characteristic (ROC) AUC values were calculated across the analyses of seizures and burst suppression where appropriate. Results Seizures – experienced reviewers There was moderate Κ = 0.546 (95% CI 0.713 to 0.38), p < 0.001, moderate Κ = 0.436 (95% CI 0.61 to 0.262), p < 0.001 and fair 0.307 (95% CI 0.481 to 0.133), p < 0.001 agreement of seizures detected on Narcotrend compared with the original CFAM recording. McNemar’s test determined statistically significant differences in the two devices p < 0.001. There was fair K = 0.397 (95% CI 0.321 to 0.473), p < 0.001 interrater agreement of seizures detected on Narcotrend. Individual Kappa for the ‘no seizure, ‘seizure’ and ‘unclear categories’ was 0.394, 0.438 and -0.005 respectively. Area under the ROC curve (AUC) 0.791, 0.704, 0.642. Seizures - non-expert reviewers There was poor agreement Κ = 0.151 (95% CI, -0.267 to 0.568), p = 0.480 of seizures on CFAM when compared to the original CFAM recording and poor agreement of seizures detected on Narcotrend when compared to the original CFAM recording, Κ = -0.222 (95% CI, -0.640 to 0.196), p = 0.297. Area under the ROC curve (AUC) seizures on CFAM 0.881 and the Narcotrend 0.333. Burst suppression – experienced reviewers There was good Κ = 0.762 (95% CI, 0.854 to 0.670), p < 0.001, very good Κ = 0.825 (95% CI, 0.905 to 0.745), p < 0.001 and good Κ = 0.771 (95% CI, 0.863 to 0.679), p < 0.001 agreement of burst suppression on Narcotrend compared with the original CFAM recording. McNemar’s test determined statistically significant differences in the two devices by two out of the three expert reviewers p = 0.007, p < 0.001. McNemar’s test determined no statistically significant difference in the two devices p = 0.052, by the remaining reviewer. 4 There was very good K = 0.818 (95% CI, 0.737 to 0.899), p < 0.001 interrater agreement of burst suppression detected on the Narcotrend. Individual Kappa for burst suppression and no burst suppression was 0.818. Area under the ROC curve (AUC) 0.873, 0.901 and 0.875. Burst suppression – non-expert reviewers There was good agreement Κ = 0.616 (95% CI, 0.167 to 1.066), p = 0.007 of burst suppression on CFAM when compared to the original CFAM recording and good agreement of burst suppression on Narcotrend when compared to original CFAM recording, Κ = 0.649 (95% CI, 0.200 to 1.099), p = 0.005. Area under the ROC curve (AUC) burst suppression on CFAM 0.829 and the Narcotrend 0.893. Conclusion This study does not support the use of Narcotrend in the detection of seizures on the intensive care unit when used with standard frontal electrode placement. It’s utility in the monitoring of burst suppression shows promise when using raw EEG, aEEG and DSA, including its use by the non-expert. However, use of the Narcotrend in monitoring patients with status epilepticus requires further study.