Searching for Signs of Consciousness

Q&A with Jan Claassen, M.D.
Kayt Sukel
February 11, 2020

Jan Claassen, M.D.
Associate Professor of Neurology
Head of Neurocritical Care
Medical Director, Neurological Intensive Care Unit
New York Presbyterian Hospital
Columbia University Medical Center
Dana Grantee:

When patients suffer a severe brain injury, due to a stroke or trauma, they often lose consciousness for weeks, sometimes longer. When they are in this sort of unresponsive state, it can be difficult to tell if they have retained any degree of awareness—that is, the cognitive function that allows them to sense what is happening around them even if they cannot respond to simple commands. Fifty years ago, patients who suffered from impairments of consciousness were not likely to survive the acute injury they experienced. But as life support technology has advanced, clinicians and scientists now recognize that patients with these brain injuries can experience different states of consciousness. And patients who exhibit a certain level of “covert consciousness,” or the presence of a subjective experience in the absence of a behavioral response, are much more likely to recover from their injuries than those without it.

Jan Claassen, M.D., is an associate professor of neurology and the medical director of the neurological ICU at Columbia University Medical Center. His research has centered on trying to find ways to better detect covert consciousness, specifically a dissociation between cognitive and motor activity in the brain. He and his colleagues recently demonstrated, using electroencephalogram (EEG) recordings, the ability to detect patterns of brain activation suggesting consciousness in unresponsive patients; their results were published in the New England Journal of Medicine (NEJM). If the results can be successfully replicated, this technique could help physicians make more informed decisions about the course of treatment, offer more accurate prognoses for recovery, and provide valuable information to family members who must make heart-wrenching decisions about whether to continue life support in the future. Here, Claassen discusses why consciousness is a continuum, the challenges of coming up with a detection method that works in the ICU, and why physicians should always assume that unresponsive appearing patients can understand what they say and do.

What is “covert consciousness”?

If you look at patients who have experienced a brain injury and have impairments of consciousness, there are different states that have been described. The classic one is coma. It was actually mentioned in the Hippocratic Corpus more than a thousand years ago. Basically, it’s a very deep sleep you can’t wake up from. On the other side of the spectrum, we have full consciousness. You are awake, you are responsive to your environment, and you can express this with movement. But if you think of consciousness living on two axes, motor function and cognitive function, you soon understand there are many other states along those lines. You may have a vegetative state, where people can open their eyes, there’s some arousal there, but they don’t exhibit consciousness. On the other end of the spectrum, the locked-in patient is fully conscious but has limited to no ability to express this consciousness often in the context of an injury to the pons. There’s also a minimally conscious state where you may see some orientation responses from the patient, but the person can’t interact efficiently with others or the environment.

More recently, cognitive-motor dissociation has been described, referring to a state in which consciousness cannot be detected by the examiner by observation or a behavioral examination of the patient but can be by means of technology analyzing the brain’s response. About a decade ago, Adrian Owen, a British neuroscientist, studied a woman in a persistent vegetative state. She could open her eyes but did could not follow movement commands when examined. Was she conscious? From observation only, it would be impossible to say. Owen put her in a functional magnetic resonance imaging (fMRI) scanner and asked her to do things like imagine walking through her apartment or imagine playing tennis. He found that the same areas in her brain were activated as those seen when he asked a completely conscious patient to imagine things. She had that “covert consciousness.”

After reading that study, I became very interested in how we might be able to better determine this kind of conscious activity in patients in the ICU setting. When you have patients in a minimally conscious state, those who have this consciousness have a higher likelihood of recovering – and of having a better functional outcome a year later. Owen had described a new phenomenon, and I wanted to find a way to link it to better patient outcomes.

What are some of the biggest challenges of trying to detect covert consciousness in a working ICU unit?

There is no direct test for consciousness. The behavioral reflection of an individual to a stimulus in the environment is central to our conventional concept of consciousness. However, a number of assumptions have to be true in individuals who do not respond to a command in order to call them unconscious. They have to be able to hear, they have to be able to comprehend the command, they have to be able to convert the comprehension of the command into an activation of the relevant motor program, and they have to be able to activate the muscles to express the command action. You can talk to the patient and you can ask a patient to move their fingers or blink their eyes. But if they don’t do it, it doesn’t mean they didn’t understand what you asked. Patients with cognitive-motor dissociation are able to hear and understand the command but despite the ability to move their muscles they do not show behavioral expression of the command. There is that cognitive-motor dissociation there. To detect this, you need a way to see what is happening in the brain.

MRI scans have revolutionized our field, and they have great advantages other techniques because they allow you to see the spatial distribution of brain injury as well as activation within the brain. But they always require transport of the patient, which can be a risk to a patient’s health depending on the condition. They are also quite costly and logistically difficult to do. You must have physicians, nurses, and respiratory technologists accompany the patient to have a scan done. Furthermore, consciousness is not completely static. It can change over time and during recovery it often fluctuates. I wanted to develop a test that could be repeated many times, even many times within a single day, to follow those changes. Even in an ideal environment with a very sick patient, you are lucky to get an MRI once or twice during an ICU stay. So, I wanted to use a different technique. Some other studies from Nicholas D. Schiff at Weill Cornell Medical College suggested we might be able to use electroencephalogram (EEG) to measure conscious brain activity.

Why think that EEG would be sensitive enough?

Certainly, EEG has limitations in detecting activation. You may not get a good spatial resolution from deeper areas in the brain. But using motor activation paradigms, you can activate large enough areas on the cortex to differentiate between a command to move and a command to not move. That made me think it would be sensitive enough to detect cognitive-motor dissociation.

Your EEG detection system found that 15 percent of the patients tested in your study had some sort of covert consciousness. Did that surprise you?

I was shocked. I thought we may be able to detect one or two, but I was very surprised that we found such a high rate of people showing this activity suggesting a level of consciousness that was completely impossible to detect by a clinical examination of the patient.

When you stood at the bedside of these patients, you could not tell whether they had a cognitive-motor dissociation or not. We looked very carefully at all the data we had collected on these patients to see if something else might differentiate patients with and patients without the activation. We looked at sedation levels, injury time, demographics, and every clinical feature we could think of to see whether there was something else that could explain the difference between those that showed a cognitive-motor dissociation on the EEG and those who didn’t. We couldn’t find anything. We don’t know yet if perhaps this dissociation is more common in certain types of medical conditions, but we found it in several different types of brain injury.

How does having this kind of information inform care decisions?

We know that covert consciousness is linked to clinical outcomes. The patients who had cognitive-motor dissociation were much more likely to be able to follow clinical commands before hospital discharge. Looking at the patients in our study, about half of the cognitive-motor dissociation patients and only about one quarter of the patients without a cognitive-motor dissociation started following clinical commands later in time. The cognitive-motor dissociation patients also were able to do this this much earlier than the others. That is very interesting to know.

But, as an ICU physician, this is also helpful for conversations with families. The families don’t want to know what the patient will look like tomorrow – rather, what will their loved one look like in a year? What will the functional level be? So, we also looked at the long-term outcomes for these patients, reconnecting with them a year after injury, and found almost half of the cognitive-motor dissociation patients had recovered to a good functional outcome as measured by the Glasgow Outcome scale. It basically translates into the patient being recovered enough so they can be left alone for at least eight hours during the day. A much smaller percentage, about 14 percent, of patients without a cognitive-motor dissociation, showed the same kind of recovery outcome.

Perhaps, most important, I think the results also reconfirmed—and this is something that residents and trainees have been told for decades—that when you are in the room and examining the patient, you have to assume the patient is conscious, even if there are no signs of that. When you are in the room, talk to the patient, not about the patient. Don’t say things that could be distressing. This is a good reminder of why that’s so important.

Our data still needs to be replicated and confirmed. It’s too early to integrate this into recovery prediction algorithms. But it does show the potential that this sort of test could help predict outcomes in a different way than we are doing right now.

What do these results add to our understanding of consciousness and the brain in general?

It’s a big question. How consciousness is defined can depend so much on who you talk to—whether it’s a philosopher or a physician. There are some philosophers who have said consciousness is just an illusion. Not attempting to address the “hard” problem of consciousness research best defined by David Chalmers, our work is in the tradition of detecting physical signals as a reflection of the conscious experience, which some might say are the “the “easy” problem. In this most rudimentary concept of recovering consciousness following acute brain injury, two fundamental concepts are required for a conscious experience: a minimum level of arousal and some degree content processing. There must be an interaction with the environment and there has to be some activity to show that consciousness is there. Our results supporting the relevance of cognitive-motor dissociation add a different dimension to the discuss because, in the past, we have relied to a large degree only on what we can observe with our eyes. Science is showing us this is increasingly a less tenable approach to measure consciousness. There are data now that we may need to account for that are not easily observable, but we may find they matter a great deal in terms of prognostication and making choices for the patient in the future.

Can this kind of EEG detector work in your average neurological ICU?

It’s certainly possible. As we designed this study, we intentionally didn’t use high-density EEG arrays with 256 electrodes. They have advantages with spatial distribution, but it would be hard to integrate them into what we currently use in clinical practice. We decided to use the common EEG used in clinical practice worldwide. We also use single-use headphones, to avoid spreading infections between patients, and then the whole set-up connects to a regular computer desktop. It’s very simple.

The experimental design is also straightforward. The study’s block design had two basic commands. Keep opening and closing your hand and stop opening and closing your hand. We tried to keep the move and rest commands as similar as possible and only changed one word. With switched out “keep” and “stop.” This was intentionally done so we could make sure the activation we saw wasn’t just a subconscious response or some other artifact but a conscious understanding of the difference between keep and stop. Most importantly our paradigm used a block design alternating the move and rest commands many times to minimize the risk of systematic artifact contamination.

We also used an algorithm to help analyze the EEG activity to help detect the conscious activity. We published that code with the NEJM paper. We wanted to make it freely available to anyone who wants to use it. Anyone anywhere can take that code and run the same analysis and have the results within a day, even within a few hours. I think that is going to be doable pretty much anywhere.

How do you follow this study?

We are currently working on a study, funded by NIH called RECONFIG. We are studying patients who have a single condition, intracerebral hemorrhage. We chose this condition because it is primarily a focal injury and we think it will allow us to get a better look at the mechanisms underlying these cognitive-motor dissociations. Right now, why patients have the ability to strongly distinguish differences between motor commands but are not able to move in response to them is unclear.

None of these patients are paralyzed or unable to move. They move the arm when we stimulate it but none of them can follow these two basic commands to keep and stop opening and closing the hand. So, we are using regular MRI scans as well as diffusion tensor imaging to look at the white matter tracts to see how they are connected.

What do you hope people take away from this study?

The big thing is that, when you stand at the bedside of an unresponsive patient, what you see and what you think may not be the reality. There is a lot going on under that skull that we have no idea about. That is important. On a larger scale, nearly 70 percent of patients with acute brain injury who are unconscious die because we withdraw care. Some of these patients may have some sort of consciousness and, since we don’t know it, we are not giving them a chance to show whether they can recover or not. I think we need to get much, much better at prognosticating for these patients.

Currently, I’m involved in a large initiative between the Neurocritical Care Society and scientists across the globe called the “Curing Coma Initiative.” We are trying to develop large-scale studies to see whether we can support the recovery process for patients who have an impairment of consciousness. We need to understand what the mechanisms are that underlie these impairments. What do they mean for the long-term recovery of a patient? And knowing that, how can we, as physicians, support that recovery in an active way? That’s what has been missing from care. That’s where we need to get to.


Detection of Brain Activation in Unresponsive Patients with Acute Brain Injury, Jan Claassen, M.D., et al., N Engl J Med 2019; 380:2497-2505. DOI: 10.1056/NEJMoa1812757

Twitter hashtag #curingcoma