EEG in the ICU: Key to Detecting Covert Consciousness?

Nancy A. Melville

November 05, 2018

ATLANTA — Efforts to better detect covert consciousness and to increase the chances for recovery among patients who appear to be in a vegetative state in the intensive care unit (ICU) are making headway as researchers refine ways to interpret and improve the prognostic capabilities of electroencephalography (EEG) and fMRI.

Monitoring with continuous EEG is a standard protocol for patients who have no apparent consciousness in the ICU, but a key challenge to using these measures is that ICUs are noisy, lead author Abhinaba Chatterjee, medical student at Weill Cornell Medicine, New York City, told Medscape Medical News.

He presented the findings here at ANA 2018: 143rd Annual Meeting of the American Neurological Association.

"Noise can come from 60-Hz AC electric sockets, alarms, ventilators, IV pumps, vibrations of other sensors, muscle artefacts, and more," said Chatterjee, who has been studying the properties of EEG recordings obtained from the ICU at the Feil Family Brain and Mind Research Institute at Weill Cornell for 2 years.

"One well-known example of how ICU noise can affect electrical signal quality is in the ECG [electrocardiogram], where noise often leads to false alarms regarding irregular heartbeats and other pathological heart conditions. The same kind of problem can manifest itself in the analysis of EEG data," he said.

Novel Signal Algorithm

In a previous study, Chatterjee and colleagues showed that EEG signals indicative of the integrity of the thalamus and cortex regions of the brain could be detected via key EEG measures known as spectral power. They found that these findings were significantly associated with the recovery of consciousness after cardiac arrest in a group of 54 patients.

To build on these results, the investigators worked to characterize a different EEG-based measure, network connectivity, in a similar cohort of ICU patients.

Although network connectivity is highly sensitive to the noise that is common in the ICU, the researchers developed a novel signal quality assessment algorithm to overcome the problem and distinguish signals of relevant activity from the noise. The algorithm allowed them to identify significant connectivity between signals from different EEG electrodes in theta and alpha frequency bands.

To test this approach, the investigators enrolled eight healthy volunteers who underwent two EEG assessments 6 months apart. Their EEG results were then compared to those of 20 post–cardiac arrest patients. These later EEGs were recorded in the ICU less than 24 hours after completion of a therapeutic hypothermia protocol.

The post–cardiac arrest EEGs were recorded during periods when patients were considered to be in "maximally aroused states," such as immediately following a neurologic examination after sedation had been lifted or when the patient's eyes were open.

"Striking" Differences

Results showed that the connection density and node degree were lower in all 20 patients who had experienced a cardiac arrest than in the healthy volunteers. In these 20 patients, the node degree in the theta band was also generally higher than in the alpha band.

Three members of the cardiac arrest group recovered consciousness by the time of hospital discharge.

"The differences in these measures between patients who recovered and did not recover were striking, as all three patients who ultimately recovered consciousness demonstrated increased node degree over the midline centroparietal electrodes compared to patients who died at the end of their hospitalization period," said Chatterjee.

"The patients who did not survive generally had very sparse networks with few connections," he said.

Importantly, the results were obtained despite the fact that the EEG recordings were made in the ICU, a setting in which this type of approach has not been thoroughly explored.

Whereas fMRI can be useful in detecting measures of consciousness, EEG is more accessible in the ICU, noted study senior author Peter Forgacs, MD, attending neurologist at Weill Cornell and neuroscience instructor at the Feil Family Brain and Mind Research Institute.

"It is often challenging to obtain an MRI in patients who are in the ICUs. It requires transport to the imaging areas, and therefore the patients need to be medically stable enough to safely tolerate the procedure," Fogacs told Medscape Medical News.

"In contrast, EEG can be obtained very easily and quickly in the ICU and can be repeated as many times as needed or can be left on the patient for prolonged periods of time, days to weeks, to continuously monitor the brain activity of these patients," he said.

"This allows recording and analysis of EEG during various levels of awareness or activity as well as during the most optimal functional state, for instance, when sedative medications are lifted," he said.

The authors note that larger trials are needed before the model can be implemented in clinical settings and that their research into the approach is continuing.

Feasible Recovery Predictor?

Asked by Medscape Medical News to comment on the study, R. Edward Faught Jr, MD, director of the Emory Epilepsy Center at the Emory University School of Medicine, Atlanta, Georgia, said the findings offer valuable insights on how EEG could feasibly have more usefulness as a predictor of recovery.

"One of the difficult things in patients who are comatose, particularly after cardiac arrest, is knowing what their outlook is, because it's always difficult to talk to families and give them as accurate as possible an outlook," Faught said.

"EEG is an old technology that has been around for a long time — since 1929; but it hasn't been well analyzed like this with modern mathematical methods. So I think this kind of thing can be useful for prognosis," he added. "Anything we can come up with to give us more assurance of probability of a recovery is useful."

The pressing need for more advanced methods to better assess patients' level of consciousness in the ICU was the theme of the ANA session on covert consciousness. In another talk, Brian L. Edlow, MD, assistant professor of neurology at Harvard Medical School, Boston, Massachusetts, noted that as many as 40% of patients are misdiagnosed as being in a vegetative state.

"Multiple studies show that if we, as experts, go in and examine patients using a Glasgow Coma Scale–style evaluation and reach a consensus that the patient is vegetative, in as many as 40% of cases, if we go back in and do a more comprehensive behavioral assessment with the Coma Recovery Scale Revised, we will find evidence of consciousness," he said.

Edlow noted that those findings largely involve trauma centers, whereas studies specific to ICUs have suggested slightly lower misdiagnosis rates of about 27%; but that "is still an alarming rate of misdiagnosis."

Such misdiagnoses can have life-altering implications in terms of decision making, such as influencing whether to withhold life-sustaining therapies in patients deemed to be in a vegetative state.

In his own previous research on the issue, Edlow and colleagues found that task-based fMRI could indeed detect consciousness in patients with traumatic brain injury in the ICU even before they were able to demonstrate any self-expression.

In the study, published in 2017, 4 of 16 patients with acute, severe brain injury, including three who had been diagnosed as being in a vegetative state and one as being in a minimally conscious state without language, were found to have cognitive motor dissociation or covert consciousness on task-based fMRI.

The tasks included motor prompts in which the patient was asked to imagine opening and closing a hand, listen to John F. Kennedy's inaugural address, and listen to classical music. After 6 months, 3 of the 4 patients survived and recovered full consciousness.

Although task-based EEG in the study did not yield similar results, Edlow said that that was a "statistical anomaly" and that EEG likely is as sensitive as fMRI.

"It is feasible to use task-based fMRI and EEG in the ICU to identify covert consciousness that evades detection on bedside behavioral examination," he said. "Confirmation of the diagnostic and prognostic relevance of covert consciousness awaits corroborating evidence from larger studies."

Dr Chatterjee and Dr Edlow have disclosed no relevant financial relationships

ANA 2018: 143rd Annual Meeting of the American Neurological Association. Abstract M261, presented October 21, 2018.

For more Medscape Neurology news, join us on Facebook and Twitter.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.