The use of continuous electroencephalography (cEEG) in the care of patients who are comatose after cardiac arrest has become common at tertiary care centers, but there are few data to clarify whether this costly and resource-intensive approach is superior to routine (ie, 20- to 40-minute duration) EEG assessments.1,2 There are 2 potential indications for EEG monitoring in this population of patients, as follows: to identify epileptic activity that is potentially amenable to treatment and to obtain prognostic information. Apart from clinical observation of convulsive movements, it is often impossible to predict whether potentially treatable findings will be observed, so EEG is often pursued based on both indications.
Using an observed cohort of 759 patients who were comatose after cardiac arrests and underwent cEEG at 2 academic medical centers, Elmer and colleagues3 have provided helpful insights into the utility of longer-duration EEG monitoring. Using random sampling of the cEEG record, they simulated multiple scenarios that were compared with full cEEG recording, as follows: no EEG, routine (ie, 40-minute) EEG obtained randomly within 24 hours of admission or during first available daytime hours, or use of routine EEG with conversion to cEEG if the routine EEG showed high-risk findings.3 All EEG testing scenarios were evaluated separately for their sensitivity in detecting findings relevant to prognosis and findings considered to be potentially treatable pathologies. The finding that cEEG was more sensitive for detection of pathology was an inevitable conclusion by virtue of the study design, but the novel and interesting information was the comparison of how these different methods could improve prognostic accuracy for a favorable outcome, defined as nonvegetative status at discharge.
A multimodality model was constructed to estimate favorable discharge status, incorporating demographic information, neuroimaging findings, and other clinical data but no EEG information. That basic model estimated that 7% of patients would have very low (ie, <1%) likelihood of favorable recovery and achieved an observed-to-estimated concordance statistic of 0.87. Compared with the model with no EEG information, models incorporating findings from the 4 variations of routine EEG sampling identified 20% to 22% of patients as having very low likelihood of favorable recovery, and all achieved concordance statistics of 0.91. Maximal data obtained by cEEG yielded a slightly better concordance statistic (0.92) and classified a greater proportion of patients (26%) as very unlikely to experience a favorable recovery. Using the narrower definition of potentially treatable events, the concordance statistic was 0.90 for all variations of routine EEG as well as for cEEG, and the proportion of patients classified as very unlikely to experience favorable recovery did not differ among approaches.
Compared with the criterion-standard time of first EEG event captured by cEEG, the delay to event detection by the routine EEG methods was approximately 11 to 12 hours. The reason for the relatively good performance of routine EEG compared with cEEG was that most prognostic epileptiform findings were persistent rather than transient events. Withdrawal of life support due to poor prognosis rarely occurs within the first 12 hours of admission, especially because other prognostic variables (eg, cranial nerve and motor responses) become more informative during the course of 72 hours, so a 12-hour delay in obtaining EEG data for prognostic purposes is unlikely to be important. The key benefit of longer-term monitoring demonstrated here is the capacity to accurately characterize a greater number of patients as unlikely to experience favorable outcomes, a factor that may facilitate establishing care decisions.
Phenotyping of anoxic brain injury by EEG remains in the early stages. At this time, it is not conclusively established whether treating epileptiform pathology could improve functional outcomes, but if beneficial treatments are later proven, cEEG would likely be required to guide those interventions. Elmer and colleagues3 used a well-established and validated approach to identify pathologic background activity and discrete epileptiform events using the American Clinical Neurophysiology Society guideline framework and prior related research, but qualitative methods based on visual inspection of the EEG record may not capture all the important information contained in the EEG record.5 More advanced methods to quantify EEG signal characteristics along with machine learning analyses, some of which have been reported by this group,6 require longer sampling intervals. Measurements that assess longitudinal change in EEG will require either intermittent EEG resampling or cEEG.
Should all patients who are initially comatose after cardiac arrest resuscitation undergo EEG, and should continuous monitoring be used? These data provide convincing evidence that EEG can improve our ability to identify patients with a very low probability of favorable recovery, which is valuable clinical information to inform goals of care. Nearly 80% of neurointensivists and epileptologists at academic medical centers in the United States report using cEEG in some patients after cardiac arrest, generally for 24 to 48 hours, although it is unknown how consistently EEG is performed.1 Use of cEEG is growing in the United States and Europe, but cEEG resources are heavily concentrated at large referral centers.1,2 A 2019 study of cEEG use in the United States from 2004 to 20134 found that 94% of hospitals never used cEEG in critically ill patients. Even in high-income countries, existing medical systems lack the capacity to provide cEEG as standard care for post–cardiac arrest management. These data reassure us that spot recordings are a reasonably good substitute. As quantitative EEG analysis becomes more sophisticated, the marginal benefit of continuous recording is likely to increase. It will be imperative to increase the capacity for cEEG throughout the care system, through some combination of remote monitoring, quantitative analyses, and simpler device applications, for advances in multimodality prognostication to translate into a public health benefit.
Published: April 28, 2020. doi:10.1001/jamanetworkopen.2020.3743
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Maas MB. JAMA Network Open.
Corresponding Author: Matthew B. Maas, MD, MS, Department of Neurology, Northwestern University, 625 N Michigan Ave, Ste 1150, Chicago, IL 60611 (email@example.com).
Conflict of Interest Disclosures: Dr Maas reported receiving support from the National Institutes of Health and the Davee Foundation outside the submitted work.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Maas MB. Evaluating the Prognostic Utility of Intermittent vs Continuous Electroencephalography in Comatose Survivors of Cardiac Arrest. JAMA Netw Open. 2020;3(4):e203743. doi:10.1001/jamanetworkopen.2020.3743
Customize your JAMA Network experience by selecting one or more topics from the list below.