Hi, everyone, and welcome to today's session. This session is entitled Future Integrated Neurocritical Care for Brain Trauma. We have three fantastic speakers today. Our first title will be Prognostication in Traumatic Brain Injury, Opportunities for Patient and Family-Centered Care, with the discussion centering around shared decision making and neuroprognostication. Our next title and speaker will be speaking on Integrated Brain Monitoring in Coma, talking about advances in neuromonitoring for comatose patients in the intensive care unit. And our last but not least speaker will be using the title To Compromise or To Protect the Brain at All Costs, coming from a surgical perspective in the intensive care unit. So with that, I will stop there and let our speakers begin. And thanks again for joining us in this topic. Thank you. Hello. My name is Susanna Muschagel. I'm an associate professor of neurocritical care at the University of Massachusetts, and I've been asked to talk to you about prognostication and TBI, opportunities for patient and family-centered care. I want to thank the organizers for inviting me and for allowing me to speak about this topic that's very near and dear to my heart. My disclosures are listed here. The first thing I'd like to always start with is that prognosis matters. This is a very important paper written by Claiborne Johnston in Muscle and Nerve, and it really summarizes why that is the case. It is important in determining management. It also is the complex synthesis of clinical data and physician experience. And in a longer but really important explanation, it defines it in a more formal way with reproducible clinical variables that may educate practitioners with less experience and correct misconceptions and others. There's no standard way, however, to approach prognostication. So many of us think that maybe a prognostic scale may be important, comorbid conditions, the provider hunch and experience. We do know that early DNI, DNR matters in the neuro ICU, but also that ICU complications and the patient's trajectory in the ICU matters. And maybe there's a bunch of other things that also plays a role. One small study in interstitial hemorrhage gives us an example how many factors clinicians consider when they're prognosticating. This is a secondary qualitative analysis of a multicenter study in five centers of 121 patients, and clinicians were asked what factors they considered when prognosticating outcomes. And as you can see here, there's many, many different responses that clinicians provided. Some used a formal scale, but not many, many just use a component of the scale and many other factors. So one is kind of led to wonder whether that leads to variability. For many, many years, neuroprognostication has been considered an art. I've listed here papers from 2010, as well as all the way to 2020 that talk about that. And it really hasn't changed in more than a decade and probably longer. But an art also means that there is a lot of flexibility, freedom in how to do it. And again, the issue of variability comes up. So I'm asking, should it really be an art? We have some examples from traumatic brain injury. Here's a paper from Alexis Toujon, published in 2013, where over 600 physicians taking care of TBI patients that are neurosurgeons, intensivists, neurologists, were asked whether they're comfortable providing a definite outcome prognostication. This is not a real-life study. This is a clinical vignette-based study, but it became clear the wide variability. So here, asking the question, how comfortable are you in recommending withdrawal of life-sustaining therapies in the provided vignette? And you can see the difference in what certain physicians answered. So while the majority felt uncomfortable providing a definite prognostication, you can see that there is still a proportion that were very comfortable, and it really describes overall the existing variability here. And a similar variability of physicians' prognostication exists in ICH, and I'm just showing you here an important paper from Darren Sahouranek that documents the same issue. So one is left to ask, how should we really go about prognostication? And so I'd like to break down the process into two things. The first is the actual formulation of the prognosis by the clinician and physician. And we have to admit that this is not well understood how physicians do this in TBI, or in fact, in other diseases. There is the argument that we should do this scientifically. Some suggest that we should use a validated prediction model, but people also criticize that approach. We do know from research in elderly, critically ill patients, is that the norms, meaning where a physician is, at which institution, what role models that person has had, matters. There might be biases. How about the heuristics? So the very fast, quick decision making when we're in a rush or when we're stressed. And then the second part is the communication of this prognosis. There's no empiric research on this in TBI, or in fact, in any other neurocritical care diseases. And I put here yet, because at the end, I will show you some of the preliminary research that my group's been doing. And we only have some guidance from general ICU and communication experts. And I want to emphasize that this might actually be an opportunity for shared decision making between the surrogate decision maker and the family, because that's really what happens when we talk about the prognosis. Families want to know the prognosis for the good reason of making decisions for their loved one. One word about the prediction scales that some folks suggest might be a scientific way of approaching this. And we use this all the time in medicine, but we need to be aware of their limitations. So prediction scales are disease specific. They should be simple to apply. They were developed in populations, but we need to know what the population was derived from, because that's really only the population they should be applied to. They're based on a statistical model that's parsimonious, meaning as short as possible. Importantly, it needs to be externally validated. And we have to think about not just discrimination, the so-called area under the curve, but also the calibration. How well do they actually predict in real life? And so they can be helpful, no doubt about it. They can help us triage high versus low risk patients, decide maybe on the intensity of monitoring, facilitate some efficient communication between different providers. But we do need to know that they have important limitations, and we must be aware of them. So many models have never been externally validated, and so we really have to question their validity. Self-fulfilling prophecies in the derivation cohort then leads to potentially also self-fulfilling prophecy within the model. Many times confidence intervals are not shown to express the always present uncertainty. And again, many argue that because they were developed in populations, they really shouldn't be applied to individual patients. I've heard other argues, however, that why are we even developing them if we're not applying them to individual patients? Those are all valid opinions. We do know, and this is a qualitative study that my group has conducted in seven centers, in 20 physicians taking care of TBI patients. You might think, well, 20 is not a large study, but for a qualitative study, we did achieve theme saturation, meaning asking further physicians didn't bring us any further additional information. We focused on the impact model on TBI because we know it's the most widely validated prediction model. And so we asked them, what do you think about the impact model, and do you use it? And here were the four themes that kind of derived, and these were of nurse surgeons, neurointensivists, a palliative care physician, and trauma physicians. So many have said impact is for research, but not for individuals, so the same thing emerging again. Some had mistrust in the model itself because they didn't trust the derivation data. Impact was developed on a kind of conglomerate of many studies put together, and so they had to focus on the common data elements that were in those studies. Some do say it is helpful to have such a model. It's helpful in reducing physician variability in a very heterogeneous disease, a very important theme. And then some were worried that statistical models and numbers may mislead patient families. And so what was interesting, though, of the 20 physicians, they all take care of severe TBI patients, and these were all tertiary care centers across the United States. Only 60% of physicians, 12 of the 20, had even heard of the impact model. And so we were very surprised by this finding, given that it is the most widely validated model. So we must mention the notion of self-fulfilling prophecies in neurocritical care. I'm listing here some important papers, the landmark paper by Kira Becker in intracerebral hemorrhage, and then we confirmed the same findings in traumatic brain injury, and one of my favorite papers really nailing the issue is this paper by Claude Hemphill and Doug White about clinical nihilism and neuroemergencies, where the use of such prediction models can be very misleading because it may lead physicians to do less very early on. Another study that was very important, again by Alexis Turgeon, was this one where 120 patients at six centers were randomly selected, and then it was determined how many died from withdrawal of care. And you can see here the wide variability among these centers, all level one or two trauma centers in Canada. And so some centers had more than twice the odds of dying from withdrawal of care versus others less than half. So this large variability really made many people wonder what the underlying factors are. And again, I may want to say that this happened or this was found even after adjusting for important predictors of mortality in TBI. So one question that came up is, could it be that physicians are the culprit? Could it be that the variability among physician prognostication may be an issue? How about the variable language, and what language do physicians use? So I'm going to show you three preliminary studies or preliminary data collections that are not yet published. They've been shown in abstracts and are currently being submitted for publication where we put together the real-life audio-recorded family meetings at seven centers, and these were not TBI-specific. These were any neurocritically ill patients or any disease, although the majority of them were stroke and TBI, and then did a qualitative analysis of de-identified transcripts. And here's one paper where we looked at the actual statements and the actual types of languages that were used, and we differentiated between probabilistic statements using either percentages or things like most likely, less likely, very likely versus language where that wasn't used. And the other finding, first of all, you can see that there was a differential between what type of statements were used, but also depending on what outcomes were discussed, cognitive versus physical versus survival, there was a statistically significant difference depending on the type of outcome. So the bottom line is maybe it's the type of language that physicians use that make a difference. Also, when we talk about uncertainty or uncertainty statements, so there's a huge discrepancy in the number of uncertainty statements that we also found. So communicating prognosis and goals of cares, we don't really have guidelines, and this really leaves room for variability. We do have an expert opinion, and I urge everyone to read this paper, which is an expert opinion on how to do shared decision-making and how to communicate with families in ICUs. And I'm leaving you here with the definition of shared decision-making in ICUs. It's a collaborative process that allows patients, surrogates, and clinicians to make health decisions together, taking into account the best scientific evidence available and the patient's goals and preferences. So essentially, the physician provides information, it is then a collaborative process using the best scientific evidence, asking about patient values and preferences, honoring the expert opinion, and then the surrogates making a fully informed decision about all care options, and then thereby making the best individualized treatment decision. That's really what shared decision-making is. And then some of the data that we're looking at is how many surrogates actually want to share this and want to do shared decision-making. The vast majority actually do 85%, and very few want to just rely on the physician making the decision for them. We've listed here some of the expert recommended key communication skills to involve surrogates in the treatment decisions, and this is from this paper that I just mentioned. For the interest of time, I will skip over this, but I urge you to review this, and this is what is recommended by the experts and the only kind of guideline we have right now. When we look again at the audio-recorded family meetings from the seven centers, we looked at what type of approach do physicians take. We used a previous framework published by Doug White and differentiated into authoritative informational, advised, or responsive, authoritative being just telling the family member what to do, not eliciting values and preferences, informational, just giving out information without any kind of elicitation of further plans or decisions or values and preferences, advised or responsive, giving a prognosis and eliciting values and preferences or the vice versa. What we could find is that 44%, so almost half of the audio recordings, never elicited values or preferences for the patient in that family meeting. What we're trying to show here is that while experts recommend all that, we still have a long way to go to really do a family-centered type of care in the neuro-ICU. Then finally, in real life, how well do clinicians do with shared decision-making? We applied a validated 10-element shared decision-making instrument listed here, also previously published by Doug White, and applied that to the audio-recorded family meetings. What we found is that so far, this is preliminary, only 31 meetings analyzed, and median shared decision-making score is five. If we look at all of these and if everything were done, it should be a 10. Only one out of the 31 family meetings, 3%, used all 10 shared decision-making elements. When we look at which elements were most frequently used, it becomes evident that discussing pros and cons of choices was most frequently used versus assessing a need from others was least frequently applied. We still do have quite a way to go to really conduct true shared decision-making in the neuro-ICU. At the bottom line, in summary, TBI and other neurocritic care diseases, there's substantial variability in the prognostication and goals of care discussions. We have very little real-life research on it, but we really need that to move forward. Experts in clinician family communication in the ICU call for more shared decision-making. Clinicians receive very little to no training in prognostication and communication of it. As a result, we believe, at least preliminarily, we've observed very highly variable approaches and language used with very inconsistent use of shared decision-making elements. However, we also want to mention that until the findings have been confirmed in larger cohorts and we have developed validated interventions to mitigate the problem, we should probably go with what experts recommend. I urge you to review this policy statement by the SCCM and ATS. With that, I'd like to thank you for listening to my talk today. My contact information is listed here. Thank you very much. My name is Jan Claassen. I'm an Associate Professor of Neurology at Columbia University in New York. I was asked by the Society to talk about integrated brain monitoring in coma. I want to thank you for this invitation and the honor to speak to you today. Here you can see my disclosures, my funding, and in addition to that, I'm a minority shareholder at Eisner Systems, but that doesn't apply directly to the talk here today. I will talk first about the goal of monitoring, which you have to ask yourself before you start thinking about what monitoring to select, and then about coma assessment and prediction. The goal of monitoring can be, for example, to assess the cause of coma in these patients. Why is somebody comatose? You can see monitoring includes anything from brain imaging that you can see here, very traditional brain imaging like CAT scans, where you can see a huge amount of blood in this patient's ventricles and some parenchymal bleed that can be the cause of the coma. It can include EEG. For example, here you see a patient with an electrographic seizure. That's just a patient lying in front of you in a coma. You can use this to come up with an idea about what's the cause of coma. If you think about what's causing coma, the classic distinctions are that you can have a global, diffuse injury of the brain. You can have also very focal areas of the brain affected. So here you can have, this is a classic example for a patient with anoxic injury, where diffusely the cortex gets affected. Here is a classic patient with a brain tumor or a bleed in the midbrain, and you can have the patient comatose because of that. Or here you can have a patient with a locked-in syndrome, for example, with a deeper injury in the pontine area. Secondarily, you can think about detection of worsening of the patient's status. Potentially, you can have, for example, the classic example for that is a patient with subarachnoid hemorrhage that initially is not in a coma, but then develops delayed cerebral ischemia, and you can think about what monitoring you want to use to detect this worsening of the underlying brain physiology that then would lead to a comatose patient. Or in a comatose patient, you miss the onset of ischemia because you have a limited examination approach there. So here you can see a patient that was monitored with EEG, and you can see on day seven the patient diffused slowing, so these large delta waves here on the right side of the brain. That's a patient that developed delayed cerebral ischemia from vasospasm after subarachnoid hemorrhage. The Boston Group here put a very nice approach out where they take a number of EEG features and combine them and give you a daily prediction of how likely it is that your patient will be in developing delayed cerebral ischemia. So these are advanced approaches, how to utilize and integrate all of the EEG monitoring information to give you a detection of secondary worsening. And next here, you can also integrate that with invasive brain monitoring techniques. So here is an example of a patient that developed a big stroke here, and you can see that the brain, the intracranial EEG detected the worsening of this patient's status, and in other words, the stroke in development. These invasive brain monitoring techniques have been shown to be useful potentially in ischemia detection, brain edema detection, electrographic seizure detection, and periodic discharges amongst other things. But the major part of the talk, I want to talk about prediction of outcome, and this includes prediction of recovery of consciousness. How often are patients in coma? There's not actually great data on that, but I want to highlight the Curing Coma Campaign, which is an international initiative initiated by the Neurocritical Care Society that brings together experts from all over the world. One of the goals of this initiative is also to provide data and support to give us better assessments of comatose patients and allow us better predictions of outcomes. So as an initial approach here, one of the initiatives is going to try with a survey to assess what the prevalence of coma in different brain etiologies around the globe is. Very practically, what is the trajectory of an unconscious patient in the ICU? Unresponsive patients are very, very common, but it's very challenging actually to even assess the level of consciousness in these patients. Just think about delirium, where patients' consciousness fluctuates widely. Think about recovery from the initial insult, and then secondary worsening. So there's a lot of things, including medical complications that develop, that have to be taken into account. However, families demand guidance from us. This is actually challenging because the vast majority of patients with brain injury that are unconscious die from withdrawal of life-sustaining therapy. This was nicely shown in Alex Turgeon's paper in Canada, where he looked at the practice of how patients with acute trauma that are unconscious die. The vast majority died from withdrawal of life-sustaining therapy, but what was interesting was that the practice of how many of the same kind of patients were offered withdrawal of life support or were recommended to have withdrawal of life support varied vastly between centers. And so we're not very good at predicting who's going to recover and who's not going to recover. What we know is that consciousness, the assessment of consciousness, weighs heavily in the decision-making for families. Not everyone recovers, and there's very delayed recovery now being described in patients with TBI, cardiac arrest, subarachnoid hemorrhage, and other brain injury etiologies. What is important is that it is not just survival or recovery of consciousness that is achieved, but that some of these patients actually, not infrequently, also have high-level outcomes. So of those that were discharged unconscious with trauma, if you see here, 50% of the ones that recover consciousness have at least daytime independence at home, 22% return to school or work. So that is a high-level outcome. What does the recovery potential depend on? Of course, things like age, comorbidities, the environment, like the social environment matters, how much support you have, the injury factors, like mechanism of injury, the location of the injury and extent of the injury are important, the time to treatment and the time from injury are important, but also what's crucially important is what is the best neurological exam that a patient has in the acute care setting. So I'll talk a little bit about that, how that can be assessed and how that can be integrated in your prognostication algorithm. If you look at different behavioral states in unconscious patients or in patients that are admitted after acute brain injury, you can look at the two axes of motor function and then cognitive function. Patient in coma has very poor motor function and very poor cognitive function. In patients with vegetative state, they have some, they have spontaneous eye opening, but they are not interactive with the examiner. So while it's slightly better than the comatose patients, there's no interaction with the examiner. The minimally conscious state patient has some intermittent interaction with the examiner or at least some in the minimally conscious minus patients, some tracking, eye movements, orientation responses. And then the recovered patients have full motor function, full cognitive functions. In the locked-in patients, they have full cognitive function or some cognitive function that is better than their motor function. How do we assess behavior in the acute care setting? The most widely used scale is the Glasgow Coma Scale. Alcovetics proposed the four score here, which is a modification of the Glasgow Coma Scale, goes in a bit more detail through the brainstem functions. This is a good prediction tool overall for comatose patients, but it doesn't in detail assess behavioral function. What is the gold standard for behavioral assessment in patients with disorders of consciousness is the Coma Recovery Scale Revised, which was published by Gio Giacino, and it goes in detail through different levels of behavioral assessments that are features of consciousness. This scale is great for research. It's the gold standard for that. In clinical practice, at times, it takes too long to assess. And so there's efforts underway to come up with an abbreviated version of that. How can we go beyond behavioral assessments alone? You can look at some resting measures. I'll introduce some of those to you. You can look at some perturbation tasks. I'll talk a bit about these functional EEG and functional MRI studies, and then you can look at how the interventions like treatments change the physiologic measurements and what that means in terms of prediction. Let's talk about resting EEGs. So this is an EEG that's recorded while the patient is not perturbed. It's just lying there. And when you do a quantitative EEG analysis, so you look at the power in different frequency wave spectra of the EEG, you can see that comatose patients, much of this work was initially done in anesthesia settings. If you have somebody deeply anesthetized, the power in the alpha frequency range relocates in the awake state from the back of the brain to the front of the brain in the anesthetized state. Or you can look at how functionally connected parts of the brain are, and this is depicted in the alpha coherence. You can see how that relocates from the back of the brain to the front of the brain, and then when the patient goes under and then re-emerges from anesthesia, it relocates to the back of the brain. Even different anesthetics have distinct signatures in these quantitative EEG measures. I want to show you how in patients with acute brain injury, these quantitative measures may become useful. These are patients with subarachnoid hemorrhage, which have been acutely studied. These are patients with traumatic brain injury studied in the more subacute chronic setting. You can see that comatose patients in this alpha power measure and in this complexity measure here very easily are distinguishable from command following patients. The patterns that you see are very similar whether you study subarachnoid or traumatic brain injury patients or whether you study acute or chronic disorders of consciousness patients. What's interesting is if you take all of these EEG measures, put them into an automatic algorithm that based on the EEG classifies patients into conscious or not, you get some misclassification based on the EEG when you compare it to the behavioral assessment at that time. What's interesting is that potentially these patients are not misclassified, but the EEG detects something already that you cannot assess at the bedside. Because when these investigators followed the patients out weeks and months later, they found that the patients that were so-called misclassified based on the EEG actually had a much higher chance to recover behaviorally as well than those that were correctly classified. So the EEG may detect something already that we can otherwise not see. There's ongoing great work from Paris from this group with Jacobo Sitt and Leonel Nakashen, Benjamin Roehout, who developed automatic ways of classifying patients just based on the EEG signature that are generalizable between centers. So this might be a nice associative measure for behavioral state assessment. Next, the rest of the talk I will talk about perturbed EEG. This work all started in a paper by Adrian Ohn, now more than a decade ago, where he had put a patient in a vegetative state, so a patient that showed no interaction with the examiner, into a functional MRI scanner that visualizes areas in the brain that are active by looking at the blood flow. And he found that a patient that was in a vegetative state when asked to imagine playing tennis showed very similar activation patterns to healthy volunteers. And also when asked to imagine walking through her apartment, she showed very similar activation patterns compared to healthy volunteers. Later on, it was shown that this can also be done with EEGs. So Nico Schiff's group showed here at Cornell, showed that you can also use EEG analysis to classify these patients and detect this phenomenon, which is called cognitive motor dissociation, so a state where your cognition is better than your motor response. We then did a larger study where we presented a motor command to the patients over and over again. We asked them, keep opening and closing your right hand, stop opening and closing your right hand, repeated this over and over again, took the EEG, processed it, cleaned it, put it into an analysis algorithm to look at power in different, at each electrode in different frequency spectra, put this into what we call a machine learning algorithm. This was a support vector machine learning algorithm to determine whether the response that was seen on the EEG systematically was different when the patient was asked to keep opening and closing their right hand or their left hand, and stop opening and closing their right hand or their left hand. And if the response was systematically different, the patient was classified as being in a state of cognitive motor dissociation. So here you can see in black the patients that did not have motor dissociation. You can see in orange the healthy volunteers. So this is the decoding prediction. So when asked to move, it decoded move. When asked to stop moving, it decoded stop moving. And the patients in blue that had cognitive motor dissociation followed this same pattern. This was very striking because those patients behaviorally looked exactly the same as the patients that did not have cognitive motor dissociation. What was most interesting was when we followed, this was seen in 15% of our cohort. When we followed these patients out for a year later, we found that the patients that had cognitive motor dissociation, almost half of them had functional recovery defined as the GASCO outcome scale extended greater than three. So the ability to be left alone for at least eight hours in the day compared to only 14% of the patients that did not have cognitive motor dissociation. So are there limitations with all of these assessments? Yes, the REST measures that I introduced first, the EEG REST measures, you can only imply consciousness because it doesn't directly test whether somebody is conscious. You just say this pattern seen in other patients is associated with consciousness. And the generalizability of the models is a question. So if you'd record your data in a different setting, is this generalizable? There's some interesting work going on again from the Paris group and they found when they developed their prediction algorithm and applied it to data from another group, this was developed, the algorithm was developed in Paris, the data that was used as a test was taken from Liège, they found that it still predicted it very accurately. Are there limitations with the active tasks? So you have to be very careful that you're not just detecting artifacts and this may be computationally challenging. So there was a study published a couple of years in Lancet where patients were claimed to be in a cognitive motor dissociation state that behaviorally looked like vegetative state and when the data was reanalyzed, it looked like likely they only detected artifact. So this is a matter of concern, you have to be very careful when you design your algorithm and that's why we picked in our study this block design. All of the methods are nicely summarized in this great paper by Komoduchi which goes over all of the methodologies used in disorders of consciousness research and one of the things that I encourage everybody to do that we did with our paper with patients with cognitive motor dissociation, we published our code publicly so anybody that wants to replicate our data can actually go to this website with a new paper, it's actually the code is published there. So if you think about cognitive motor dissociation, those patients from minimally conscious state minus vegetative state or comatose state can be detected to actually be in a cognitive motor dissociation state and they have a higher chance to recover later on. Importantly, we want to go from just detection and prediction to treatment and so this is one of the big goals of the Curing Coma campaign. In a recent paper, this was summarized by Brian Adlow and myself and others and Nico Schiff and David Greer, where we looked at what we would need to do, what's the current state of the data, where is basically the circuitry broken, and if we identify that could we then go to a personalized treatment approach for patients with disorders of consciousness. So if we find the circuitry that's abnormal, that's faulty, can we then actually do a targeted treatment for individualized patients. Here in this paper by Thibault, I highly recommend this, she nicely went over all of the current treatment approaches for patients with disorders of consciousness that have been proposed at this point in time. If you do this cognitive, this disorders of consciousness research, you have to also ask yourself about ethical questions, how to best prognosticate. This has huge implications for patients, for families, but also for society. I want to add that currently we predict outcomes based on very, very poor data. So I think the ethical problem is there at the moment. The equity of the assessments, if you develop highly complicated assessment tools that predict outcomes, you have to make sure that they're going to be available at all care settings and across all hospitals, at least ultimately. And ideally, you want to develop something that can be applied in any place in the world and not just in developed countries. The burden to families, patients, and society and healthcare systems have to be considered, both of early death that could be actually of a patient that could be reintegrated into society as well as patients that would survive without meaningful recovery. And what is meaningful survival? This has to be asked, is not answered adequately. This should be determined by the patients themselves or by families that actually live with patients that survive. In conclusion, standing at the bedside will not allow you to figure out what's going on in the brain. I think this is clear. You have to go beyond that. Behavioral assessment is an important part of the assessment of these patients, but it will not be able to stand alone. Why do you want to monitor? You have to ask yourself, and based on that, you have to also choose your monitoring techniques. Monitoring of coma holds great promise for the future, and it may allow sort of insights into the trajectory of recovery, into the trajectory of what is the prognostication of these patients. And it may allow us to diagnose the cause of the coma. I want to again highlight the Curing Coma Campaign and encourage everybody to get involved with the Curing Coma Campaign. It is a great initiative. It is an international initiative that is not, that was started by the Neurocritical Care Society and is clearly not limited to the Neurocritical Care Society. It goes way beyond. It integrates people from all aspects of the healthcare system. There's not just physicians in there involved. There's nurses, there's physician's assistants, NPs, pharmacists. All of these are involved in this initiative, and also healthcare providers that are involved in the post-acute care setting are a crucial integrated part of this initiative. So I want to encourage you, anybody that sees any patient with a disorder of consciousness to reach out and get involved with the initiative. Thank you very much for your attention. And this is just the last picture of the people that I work with and my dearest and closest collaborators around the world and at other institutions. Thank you very much for your attention. Hi, good day, morning or afternoon, depending on where you are watching this from. It is my real pleasure to be here today. I really want to thank the Society of Critical Care Medicine for the invitation to talk to you about something that, a topic that I actually really enjoy thinking about and talking about. I spend a lot of time talking to my trainees on rounds at the bedside every day seemingly about this issue of managing the injured brain and all the things that we do to manage the injured brain and how those have very significant and widespread systemic effects and how we balance the risks and benefits of all those therapies. So for the next 15 minutes or so, I'm really excited to have this conversation with you all. Just to start with a little bit of epidemiology, we all know that traumatic brain injury is a huge public health problem. It is the leading cause of death following injury, followed by hemorrhage and sepsis and multiple organ failure for all comers, as well as injury being the leading cause of death for all persons ages 1 to 45 in the United States. According to CDC, traumatic brain injury accounts for about 2.5 ED visits, 290,000 hospital admissions, 56,000 deaths, 80,000 permanently disabled individuals. About a third of all injury-related deaths in the United States involve a traumatic brain injury. The costs are astronomical, 50 million individuals with sustained traumatic brain injury every year in the world with an annual cost of $400 billion. Just to put this in a little bit of context, the annual incidence of acquired brain injury or traumatic brain injury is 30 times more common than breast cancer, 400 times more common than HIV-AIDS. Traumatic brain injury occurs at a rate greater than all cases of MS, HIV-AIDS and breast cancer per year combined, just to give you a little bit of perspective. In addition, for those people who survived their brain injury and who are not severely disabled from their brain injury, patients who suffer from traumatic brain injury are at an increased risk of depression and PTSD, an increased risk of suicide, increased risk of substance abuse. They are at a three times greater risk of experiencing a second traumatic brain injury and face an eight times greater risk of sustaining subsequent other injuries. And what's a really fascinating and terrifying statistic is that estimates of 60 to 80 percent of people who are currently in prison actually have a history of TBI, establishing some relationship between criminality and previous history of traumatic brain injury. And how do we manage patients when they come in and they come into the hospital with their brain injury? What do we do to take care of these people? Well, a really important concept to understand is this one of primary versus secondary insult. The primary injury or primary insult is what occurs at the time of impact, right? Car hits the tree, tree hits the windshield, baseball bat hits the tree. That's your impact loading injury, which is collision of the head with a solid object at a tangible speed. You then also have these impulsive loading injuries. These are sudden motion without significant physical contact, and these are typically going to be your acceleration, deceleration type injuries, diffuse axonal injury, and patients who, for example, in babies and pediatrics, the corollary, the analogy is those patients who have shaken baby syndrome, for example. Currently in 2021, those injuries are unmodifiable. What's done is done. If a neuron is broken, the neuron is broken. We don't know how to fix that currently in 2021. But what happens subsequent to that initial primary injury or primary insult is a whole sequence of secondary insults and secondary injury, and it's really all of our therapies are really targeted against mitigating those and attenuating the effects of the ischemia, swelling and edema, inflammation, and cellular dysfunction. What I'm going to talk about when I talk about some therapies, just think about it. This is all about minimizing the secondary injury. That is what we are trying to do when we take care of a patient with traumatic brain injury. The two secondary injuries that are very clearly associated with poor outcome, these are old studies, but these are very compelling data that tells us hypoxia and hypotension are the two worst things that you can do for somebody with a brain injury. You can see from these slides here, again, this is old data, but a single episode of hypoxia, a single episode of hypotension, and the combination of the two result in not only increased mortality, but also poorer functional outcomes following traumatic brain injury. So preventing hypoxia and preventing hypotension are the two most important secondary insults that we can prevent. However, when we talk about hypoxia is bad, hyperoxia, action levels that are too high, is also not so good for your injured brain, and these are just two studies, one from my former institution on the left-hand side and one from a pre-hospital study on the right-hand side that show you that hyperoxia is not so good for you either, which means normoxia is probably the way to go. Similarly, we said hypotension is bad, but so is hypertension, and this is just a compilation of studies that demonstrate the association between poor outcome and patients who are hypertensive following severe traumatic brain injury. So again, normotension being kind of the name of the game, a concept we'll come back to a little bit later. So after we prevent, minimize hypotension, minimize the risks of hypoxia, what do we then do? Well, most of the time we tend to take patients with severe traumatic brain injury and we put an ICP monitor in. Well, why? What is the benefit of ICP monitoring or intracranial pressure monitoring? Well, it's definitely the most rapid and consistent way of determining the development of cerebral edema. It also can tell us about advanced warning of impending worsening from hematoma or contusion expansion, cerebral venous outflow compromise, loss of autoregulation, hyperemia, abnormalities of CSF absorption, as well as tell us about, lead us in, there may be post-operative complications. And currently, despite a randomized trial that failed to demonstrate benefit, the details of which are far too complicated for me to go into in this talk, but the Brain Trauma Foundation guideline, the last edition in 2016, still does recommend ICP monitoring for appropriate patients to reduce in-hospital and two-week mortality post-injury. And how do we do that? Well, this goes back to the Monroe-Kelley Doctrine, right? How do we manage intracranial pressure? How do we manage intracranial hypertension? There are three things that live in your rigid, thick skull, brain tissue, blood, and your cerebral spinal fluid. If the volume of any one of those three increases, the volume of one of the other three has to decrease or pressure will elevate, and if that pressure continues to elevate, you develop intracranial hypertension and ultimately can develop herniation. So all of our therapies are about reducing the secondary insults by decreasing one of the volume of one of those three compartments. It's actually pretty straightforward. And what's available to us? Obviously, there's CSF drainage. That's one. That's pretty obvious. How do we decrease the brain compartment? Well, sugar or salt, right? Doritos or cupcakes, mannitol or hypertonic saline. It doesn't really matter which one you prefer, one salt, one sugar. People try to compare them. People try to figure out which one's better. I don't think it really matters, right? Both of them will help to reduce cerebral edema by decreasing brain tissue water content. They both work very well. There are benefits to one in certain circumstances and another in other circumstances that is absolutely true. But generally, they both work and they should be used. They can be used in a complementary fashion. What about the brain compartment? The brain compartment? I'm sorry. The blood compartment? Well, we have a lot of things that we do that potentially can reduce the blood compartment. And this is really all about reducing cerebral blood flow. And these are a laundry list of things that we can do that potentially will reduce cerebral blood flow. Hyperventilation, sedation, head positioning, neuromuscular blockade, et cetera. And let me give you a couple of examples. One being what I would call appropriate cerebral blood flow reduction. So this is you take a patient, you adequately sedate them, that decreases cerebral metabolic rate, which decreases appropriately cerebral blood flow, which will then decrease intracranial pressure. That is an appropriate cerebral blood flow reduction strategy. As opposed to, potentially, lowering their PCO2. So hyperventilation, which induces a reflexive vasoconstriction, not because of reduction in cerebral metabolic rate, but induces a reflexive vasoconstriction, which reduces cerebral blood flow and therefore decreases intracranial pressure. It's very effective, as we all know, but it's not really an appropriate reduction in cerebral blood flow to meet cerebral metabolic demand. And so lowering their PCO2 will lower their intracranial pressure, but if you lower it too low or for too long, it can be associated with cerebral ischemia, and that's actually been well documented. That being said, just like hyperoxia's bad, hypertension's bad, hypercarbia is not so good for your brain either. Not only will it increase your intracranial pressure, but it can also increase hyperemia. Again, going back to this concept of normal parameters. Well, this is an algorithm that the Seattle International Severe Traumatic Brain Injury Consensus Conference came up with that really is kind of an extension of the Brain Trauma Foundation where we took all the different modalities that were available and tiered them and ranked them by expert opinion to kind of say how can we go about an algorithmic fashion managing a patient's intracranial pressure once we decide to put a monitor in them. And what you'll notice about this list and what's on here is that basically everything is systemic. All of the therapies that we are applying are applying systemically across the body. Now I want to just point out very quickly, and I apologize I jumped ahead a little bit, but ICP monitoring is not the only way to monitor a patient with severe traumatic brain injury. Clearly, there are tons of other types of monitors that we can use, cerebral microdialysis, transcranial doctor imaging, advanced neuroimaging, LICOX devices, but basically the concepts are all the same. The therapies that we apply to improve the PBRR2 or improve cerebral blood flow or improve the lactate to pyruvate ratios are all about the same things that we just talked about with respect to how we lower intracranial pressure. And attenuation of secondary insults. And all of these treatments, everything I've just talked about, everything I list on every single one of these slides, with the possible exception of CSF drainage, is a systemic therapy, right? Hyperventilation is not something you just apply to the brain, propofol is not something you just apply to the brain, barbiturates are certainly not something you just apply to the brain. These are all systemic therapies that are going to have widespread systemic effects, which comes into this issue of risk and benefit. Let's talk about temperature management as a kind of nice example. We know that brain temperature is typically about a degree to a degree and a half higher than body temperature. We know that fever increases cerebral metabolic rate and CO2 production, and that increase in cerebral metabolic rate will increase cerebral blood flow, which can elevate intracranial pressure. So we know that fever or hyperthermia is bad, right? So we know that if we, and this is just one study that demonstrates this, if we prevent hyperthermia, prevent fever, fever-managed patients, they do better, their ICPs are lower. So what happens if we induce hypothermia, right, therapeutic hypothermia? Well, there are two types of ways you can apply it for brain injury. One is when it's used as part of a tiered algorithmic approach to treating elevated intracranial pressure, and one is where it's used as a neuroprotectant, irrespective of what the ICP is doing. Well, the pros of doing this approach, reducing cerebral metabolic rate, like we talked about, is blocking deleterious chemical cascades, decreasing excessive post-traumatic release of excitatory neurotransmitters, attenuating the opening of the blood-brain barrier, and may prevent reperfusion injury and cellular damage. However, you see on the con list here, there is a number of systemic complications that are very well-established to be associated with therapeutic hyperthermia. In addition, there are rear-warming-related complications, and there are rear-warming-related cerebrovascular abnormalities, and what this has basically resulted in is a number of randomized trials. There are three, all of whom, all of these large, multicenter, randomized trials have either failed to demonstrate benefit, demonstrated worse neurologic outcomes, or demonstrated a marked significant increase in systemic complications in patients who were treated with therapeutic hypothermia as a part of their management of severe traumatic brain injury. Again, systemic therapies. These are not isolated to the brain by any means, and you can see here, these are all of the systemic effects that you can have of inducing therapeutic hypothermia to somebody. All of these therapies, as I said, have the potential to induce systemic harm, and so it's really a matter of how much harm are we willing to accept in order to manage the intracranial pressure, or the PBRO2, or the seizure, or the cerebral ischemia. How much harm are we willing to tolerate with respect to systemic effects? In addition, you will notice that there are very few things that have been demonstrated in high-quality studies to actually have benefit. All of these things that we do, outcome studies are pretty rare, and there are very few level 1A recommendations when it comes to brain injury other than avoiding steroids, and that's about it. All the therapies that we've tried, there's very few studies that have really demonstrated high-quality improved outcomes, so we're just potentially in some of these instances inducing harm. I say that a little bit, I'm being a little bit facetious when I say that, but we induce harm, right? I'll just give you an example. Increased intracranial pressure, so I went up in the propofol, which dropped the blood pressure, so I increased the amount of fluid I was giving, which meant I needed to increase my ventilator settings, which increased my intracranial pressure because I increased my thoracic pressure and reduced my cerebral venous drainage. You can see how we, at the bedside, can induce systemic effects. I'm not saying all therapies are bad at all, but I'm saying that we need to be thoughtful about how we apply them. This is a kid I took care of years ago, this is back in the high-volume crystalloid days, I fully admit that, who had an isolated brain injury, and this is what happened as a result of his therapy. We also have to keep in mind that when we talk about systemic effects, that there is data that suggests that many of our patients don't have isolated brain injuries. About 55% of our patients who are admitted to our ICU will have polytrauma, and so all of these systemic therapies that we are applying will affect their injured organs as well. And then we also have to keep in mind that the body has taken a hit with a brain injury. Traumatic brain injury is a multisystem disease. It is not isolated to the brain in and of itself, even irrespective of the therapies you apply. 89% of patients who have traumatic brain injury will develop dysfunction of at least one non-neurologic organ system. We know the catecholamine surge storm is a well-established phenomenon after TBI. We know that there's a very close relationship between the cytokine network and what happens in the brain in systemic inflammatory response. On the right-hand side there is some data from an old paper that we published that just looks at serum levels of a wide variety of cytokines in patients with isolated severe traumatic brain injury. We know that patient's respiratory failure is extremely common following brain injury, neurogenic pulmonary edema, aspiration, VAPS, systemic inflammatory response with ARDS, ventilator-induced lung injury. We know that cardiac dysfunction is very common. And we know that the brain-heart interactions in critically ill patients is increasingly recognized as an important contributor to human dynamic electrophysiologic and myocardial dysfunction. On the right-hand side there is some data that we published that looks at just isolated elevated in young, healthy people with severe traumatic brain injury. We know that TBI-induced liver injury and hepatocyte dysfunction is associated with metabolic perturbations, decrease in hepatic transcription factors, impaired host immune function, impaired protein synthesis, and impaired clearance of clotting factors. And we know that coagulopathy obviously occurs very frequently after traumatic brain injury, depending on how you define it, somewhere between 10% and 90%, again, depending on definition. Coagulopathy is associated with an injury-mediated local release of tissue factor. And we know that outcomes are worse in patients who have coagulopathy in association with their brain injury. So, all of those systemic therapies we applied, many of these patients have polytrauma, and then they also get sick systemically just by the nature of their brain injury in and of itself. So, when we ask ourselves this question about should we compromise or protect the brain at all costs, it depends. It depends on an individual risk-benefit analysis for an individual patient. There's no right answer here. This is about doing the best you can, because it really is about being middle of the road for most of our therapies, normocarbia, normovulemia, normoxia, normothermia, normotension, because at the end of the day, it's about protecting the brain while managing the rest of the patient, but protecting the rest of the patient while managing the brain. So I very much thank you for your attention. I very much look forward to the question and answer session, where hopefully we can discuss this further. Thank you very much. ♪♪♪

neurocritical care

brain trauma

traumatic brain injury

prognosis

integrated brain monitoring

comatose patients

intracranial pressure

systemic effects

individualized care