Nurses

The past decade has been marked by the advent of continuous EEG (cEEG) monitoring, which is now recommended as the standard of care in numerous medical conditions seen in the neurologic intensive care unit (ICU).^1,2 In clinical practice, its main indications are seizure detection, treatment monitoring, and prognostication in various conditions such as seizures/status epilepticus, ischemic stroke, subarachnoid hemorrhage, intracerebral hemorrhage, traumatic brain injury, brain tumor, encephalitis/sepsis-associated encephalopathy, extracorporeal membrane oxygenation, targeted temperature management, and cardiac arrest.^2,3

Implementation of cEEG monitoring in the ICU is challenging. It is recommended that raw EEG traces be reviewed at least twice a day by trained electroencephalographers, which may be difficult to achieve given the scarcity of epileptologists and neurophysiologists.⁴ In addition, technical issues may arise, and the availability of EEG technologists during off hours may be limited. The guidelines therefore suggest that the initial EEG review may be performed by trained but nonexpert EEG readers such as neurology or clinical neurophysiology trainees, intensivists, or attending EEG staff.⁴ However, given the degree of urgency related to most of the conditions in which cEEG monitoring is used,^3,5 there is a need for prompt bedside EEG interpretation in the ICU. The increasing use of cEEG was first made possible by digitization of the recordings and then by the development of computerized signal processing techniques that use algorithms to quantify the raw EEG activity. Quantitative EEG (qEEG) provides amplitude-integrated EEG recordings, color spectral arrays, color density spectral arrays, density spectral arrays, rhythmicity spectrograms, asymmetry spectrograms, and even seizure/pattern indicators that facilitate EEG interpretation and seizure detection.⁴

In this issue of Neurology® Clinical Practice, Kaleem et al.⁶ report on a prospective single-center study that evaluated the performance of neuro-ICU nurses for seizure detection by real-time bedside qEEG interpretation. In addition, the authors evaluated time to seizure detection and factors associated with interpretation inaccuracy. The cEEG recordings were processed using Persyst 12. Each of the 65 participating ICU nurses attended a 5–10-minute individual training session based on illustrative qEEG panel printouts. The nurses then prospectively counted the number of seizures they detected during a 12-hour shift. Their results were compared with those obtained post hoc by board-certified neurophysiologists. Time to seizure detection was compared between the ICU nurses and the standard of care defined as seizure detection by neurophysiology fellows. Factors associated with nurse accuracy in detecting seizures were sought. Of the 109 patients receiving cEEG monitoring, only 8 (7%) experienced seizures. Seizure detection by nurses was 74% sensitive and 92% specific. Of interest, the nurses detected seizures at a mean time from the onset of 53 minutes compared with 185 minutes with the standard of care. Finally, inaccurate nurse interpretations were associated with longer cEEG monitoring. This pragmatic prospective study provides important information on the potential role for nurses in improving seizure detection in ICU patients. However, several points warrant discussion.

First, the small sample size and low number of seizures limit the robustness of the evaluation of nurse performance. Ordering cEEG monitoring using a seizure risk tools such as the 2HELPS2B score could have increased the diagnostic efficiency of cEEG and may also help dealing with the incredible flow of digital information in the ICU.⁷ Other potential weaknesses are the single-center recruitment and potential contamination bias related to communication among the nurses.

Second, when seeking to implement cEEG monitoring, reading of the recordings is only one of many concerns. The first step is to ensure that high-quality cEEG recording is available 24 hours a day, 7 days a week. However, EEG technologists may not be available around the clock. The setting up of cEEG by ICU teams is probably the most efficient method but requires rigorous training of all ICU staff. The use of simplified devices and montages can also help.^8,9 Thus, we must keep in mind that interpretation of cEEG recordings requires thorough knowledge of all the technical aspects involved, notably the management of EEG artifacts, as these are common in ICU patients.

Third, the approach to cEEG interpretation and seizure detection used in the study was considerably simplified because the nurses were not trained to interpret the raw EEG data. The training program lasted only 5–10 minutes, and seizure detection was based on a rhythmicity spectrogram produced by automated seizure-alarm software and on amplitude-integrated EEG displays. Although such use of artificial intelligence can undeniably contribute to facilitate the implementation of cEEG monitoring, automatic seizure detection tools are not infallible. They are intended to assist in real-time event screening, facilitate interpretation, and help in the review of long recordings by displaying summary EEG trends. Their contribution is notable for nonexpert EEG readers, but they cannot replace in-depth knowledge of raw EEG data interpretation.¹⁰ It is worth noting, however, that recent studies of large data sets of cEEG recordings suggested that the Persyst algorithm was not inferior to human experts for seizure detection.¹¹

Finally, despite the limitations inherent in the study design, this study provides valuable knowledge about the potential role of ICU nurses in the interpretation of cEEG monitoring recordings in clinical practice. The use of automated seizure detection tools was associated with good nurse performance. However, the nurses should also be involved in the other aspects of cEEG monitoring and should therefore receive extensive training to ensure rapid real-time bedside detection of seizures.

Study Funding

No targeted funding reported.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.