Hi, I'm Diane McLaughlin, I'm a nurse practitioner with Baptiste Lyerly Neurosurgery and Neurocritical Care and Mayo Clinic Neurocritical Care. I'm here to talk to you about new initiatives in the diagnosis, management, and prognostication of coma, which we'll collectively call neuroprognostication. I have no disclosures. We're going to talk a little bit about neuroprognostication by pathology, we're going to talk about the value of ancillary testing, then we're going to talk a little bit about how to effectively communicate with families and decision makers. So neuroprognostication is essentially how you look in your crystal ball to predict the future. And it's important that the provider has some ability to prognosticate because this often becomes the basis for end-of-life decision-making and poor predictions determine withdrawal of life-sustaining treatment. Historically, these predictions have been highly variable and often inaccurate. So there's been a lot of effort, particularly in the last two years, to standardize the approach to what we now will refer to as neuroprognostication. So the patients that we're going to be prognosticating are typically in a comatose state, which is a state of unresponsiveness typically prolonged. Definitions can vary anywhere from an eyes-closed state with reflexic, non-purposeful, or unresponsive state to stimuli, or looking at absence of arousal, which anatomically resides in the brainstem, and awareness and that content component, which anatomically resides within the cerebral cortices. We are looking at a period of time greater than one hour in this two-component model. It can be caused by a variety of insults such as direct damage to the brain tissue, so structural damage from stroke, hemorrhage, or metabolic factors such as hypo- or hyperglycemia, electrolyte disorders, drug overdose, cardiac arrest. Treatment for most comatose patients typically consists of supportive care while correcting the underlying cause of pathology. The prognosis for these patients is then unclear, particularly upon presentation, which is why there's been so much effort to try to come up with strategies to be able to accurately do this. Literature on discrepancies in physicians' prognostication is heterogeneous and sparse. Literature that is available shows that physicians are mostly pessimistic regarding patient outcomes. Your medical specialty strongly influences prognosis, with specialties exposed to critically ill patients being more pessimistic, whereas specialties that follow patients longitudinally might be more optimistic. Early career physicians were more pessimistic than more experienced physicians. Literature also shows that a physician's personality and psychology influences prognostic accuracy and can be improved with evidence-based debiasing strategies. So the optimal timing for prognostication is not completely clear. As you'll see and will be repetitive throughout this presentation is waiting at least 72 hours to allow for delayed awakening. And really you're going to do it within the first 28 days, which is every patient we're going to see. When I talk about heuristics, what does this mean? These are mental shortcuts that we take over time based upon guesswork, trial and error, process of elimination, and past experience. I've seen this before. This is what these patients typically look like. It's a type of reasoning that simplifies the process. And even though this is good enough for a lot of things, it's probably not good enough when we're talking about life and death, which is why we need a more structured analytical approach. And then the other term you're going to hear throughout this presentation is the self-fulfilling prophecy. So this is like a positive feedback loop, a prediction that comes true at least in part as a result of a person's belief or expectation that said prediction would come true. This is most commonly applied to intracerebral hemorrhage, where people say, oh, they typically don't do well, high rate of morbidity and mortality, reasonable to withdraw life support. You withdraw life support and the patient dies and you say, well, see, these patients never do well. And this becomes particularly problematic when the number one cause of death in most of these pathologies we're going to talk about is withdrawal of life support. So if it was just looking at the brain, maybe it would be a little simpler, but it's not because most of these patients are sick and have other things going on. They receive medications which can affect or suppress their neurological examination, seizure medication, sedation to prevent tolerance or for patients that have an acute lung injury, patients that are hypothermic, hyperthermic, that are hemodynamically unstable. Until the day I die, I'll tell you that the brain is a barometer of critical illness. So if you're critically ill, there's a very good chance that you have some type of encephalopathy and your brain's not working to full capacity. So some of these confounders can be removed, some cannot. And so we try to augment what we see in terms of the exam with the addition of various tests. And again, some are good, some maybe not so much. If you have a patient that's in coma and we put EEG on, it turns out the patient's in non-convulsive status epilepticus, well, that's helpful. It's something that we could potentially treat and the patient could potentially get better. If we put a patient on EEG and it shows generalized slowing consistent with a toxic metabolic disturbance of unknown significance, or whatever phrase whoever's reading your EEGs puts on 90% of the EEGs we order, maybe it's not so helpful. Neuroimaging in a lot of ways, it's incredibly useful. It tells you if you have a structural damage or not, but you have to know your test and what you're looking for. If a CT shows anoxic injury, you don't need an MRI to show the same. That being said, if you have normal imaging, and we'll talk about that more a little later, and this is including MRI, then there's probably an alternative cause of what's making this patient comatose. When you hear the phrase neuroprognostication, most of the time people are talking about in cardiac arrest patients. We've been lucky enough to have not one, not two, but three guidelines come out in the last few years. So we'll jump in and give you some key points. So if you don't know why this is an issue, again, the brain's particularly susceptible to injury after cardiac arrest, barometer of critical illness, and delayed awakening can cause some to doubt the chance for a meaningful recovery. Up to 80% of patients who survive out of hospital cardiac arrest are comatose on admission. 50% of those are still comatose 72 hours later. So with all the new knowledge that's come out over the past few years, and what we used to believe was true, we found some things that we used to believe really aren't true anymore. We used to think that hypothermia preserved these patients' brains for better outcomes. Now we know that really it's avoidance of fever is just as good with less hemodynamic compromise. We used to think that time until return of spontaneous circulation was a predictor of poor outcome. It's not really true. Maybe depends more on the quality of ACLS that the patient's receiving. So you can see what we know is continuing to evolve. What has become more explicitly obvious is the need for patience and to allow for delayed awakening with most guidelines recommending at least 72 hours. So in the meantime, as we're going to do with most of these pathologies, you wait. Limit sedation as tolerated, do serial neurologic assessment, avoid severe metabolic disturbances, avoid fever, and start moving towards ancillary testing. And watch for a clear trajectory to emerge. So the European Society of Intensive Care Medicine released guidelines in 2021 that were being used up until this year when we had some new recommendations come out. So some of what they're recommending, we have new literature that has showed not to do these things. So they still recommend targeted temperature management, 32 to 36 degrees times 24 hours in unresponsive patients, post-cardiac arrest. They actually want you to start ancillary testing within that 72-hour period of time in all patients that don't withdraw from noxious stimuli. Additional recommended testing include a CT followed by MRI 48 to 72 hours time and EEG. Patients are considered to have a poor prognosis if after 72 hours, they still don't withdraw from noxious stimuli and had at least two of the following, absent pupillary or corneal reflex, absent N20 SSEP waves, which is somatosensory evoked potentials, burst suppression or suppressed background on EEG, early myoclonic status, diffuse extensive anoxic changes seen on CT or MRI, or a neuron-specific NLAs greater than 60 at 48 to 72 hours. So that's your NSE. But we've had two new guidelines emerge in 2023, including the Canadian guidelines that say the first 48 hours solely supportive care, don't talk about neuroprognostication or anything until after the 72-hour mark from hospital admission. They also recognize findings that are more likely, particularly in specific patients, to suggest poor anticipated prognosis. These are somewhat similar to the European guidelines. So a neuro exam that doesn't get better with time, absent pupils, presence of status myoclonus, an isoelectric suppressed or burst suppressed EEG, or periods of generalized periodic discharges with suppression, a head CT with reduced gray-white ratio or MRI with increased large-volume DWI changes, absent N20 responses on SSEP. And so these all factor in to equaling a likely worse prognosis. And they recommend this be discussed with family as long as it's disclosed that the examination might be confounded for whatever reasons, and that there's not strong data available to support early prediction of poor outcome short of brain death. So they put that little catchphrase in to say even though these seem to suggest a poor prognosis, there's actually nothing out there that says for sure that these patients can't get better. Now NCS guidelines simplify this even further, and they gave us some indicators of good outcomes, which unsurprisingly is somewhat opposite of the Canadian guidelines. So an improving neuro exam, absent or small DWI changes on MRI two to seven days following a rest, EEG with reactivity within 24 hours, or N20 amplitudes greater than four. And their poor prognostic indicators at more than 72 hours were similar. And they also recommend that with poor examination and absence of reliable indicators of poor outcome, the recommendation is further observation to allow for delayed awakening. And we're actually seeing post-cardiac arrest patients housed in neuro ICUs now where there's a little more patience to wait and see if these patients recover. And this is a summary of some of those findings. So since 2005, the chances of having an acute ischemic stroke have increased 50% worldwide. This is World Health Organization data. Another report by the World Stroke Organization states upwards of 12 million new strokes every year. And all this combined, the data suggests that approximately one in four people globally can be expected to have a stroke during their lifetime. So if you're listening to me, you know somebody that's had a stroke, you've taken care of a patient with a stroke. So the first question that most clinicians are asked when a patient is admitted with stroke is, how is this patient going to do? And there's a lot of factors that influence stroke prognosis, age, stroke severity, stroke mechanism, the location of the infarct, comorbid conditions, clinical findings, and other related complications. In addition, interventions such as thrombolysis, endovascular therapy, stroke unit care, rehab also play an important role in the outcomes. So knowledge of these factors that affect prognosis is necessary for the clinician to make a reasonable prediction for individual patients and to provide a rational approach to management and really help patient and family understand what the course of the disease looks like. Now stroke mortality has been greatly affected by advances in endovascular therapy, however significant disability remains. So goals of care prior to initiating treatment for patients with large facial occlusions was previously based on achieving functional independence or having a modified Rankin scale score, which we'll talk about 0 to 2, indicating that the patient could ambulate independently. So you can see the modified Rankin scale here, 0 is you're fine, 6 is you're dead, and kind of the terrible joke is that 5 is that you wish you were dead. And so anything from 1 to 2 was initially also considered a good outcome with 3, 4, and 5 considered a poor outcome. But they actually looked at this a little bit and maybe it's not as clear cut as it seems because when they looked at ischemic stroke patients who achieved a modified Rankin of 2 and 3, functional outcomes seem to have similar health-related quality of life scores. So maybe 0 to 3 should be considered a good outcome rather than 0 to 2 in moderate to severe ischemic stroke. So it's historically difficult to predict treatment effect when patients are being triaged to receive therapy or not. In the stroke treatment assessments prior to thrombectomy and neurointervention or the Satin study, they had the interventionalists try to determine what these patients were going to look like after intervention. And only 44% of the time were they able to do this correctly, which means over 50% of the time we weren't able to predict functional outcome for these patients. And that's affected who we're taking for endovascular treatment now. So studies are showing that more aggressive therapy, particularly in endovascular treatments, leading to better outcomes. So the randomized controlled trial to optimize patient selection for endovascular treatment in acute ischemic stroke or Select-2 demonstrated that for one additional patient to achieve modified Rankin 0 to 2 and 0 to 3, a total of 7 in 5 patients needed to be treated respectively. Also greater than 50% reduction in patients with modified Rankin 5. Similar, the endovascular therapy in acute and interior circulation, large vessel occlusion, patients with a large infarct core or angel aspect demonstrated modified Rankin improvement in 30% enrolled mechanical thrombectomy patients had a 90-day functional outcome or a modified Rankin of 0 to 2 as opposed to 11% improvement in MRS in the control group who didn't receive thrombectomy. So again, in this trial, patients with modified Rankin 5 at discharge was reduced by half. And it's important because in these trials, they just took these patients regardless of the aspects score, as opposed to trying to predict how they were gonna do even though they already had infarct on CT. So with acute ischemic stroke being so common, you think we'd have good scales to predict outcomes. So there are actually a lot of prediction scores out there, but it's interesting because we don't really use any of them which tells you that maybe they're not that great. And some of this is the recent advances that we've had with thrombolysis and endovascular treatment tells us that there's still room for improvement. And these scores aren't based upon that new data that's available. So they're not recommended to be utilized as of this point. So ICH for the purpose of this discussion will include intraparenchymal hemorrhage, intraventricular hemorrhage, non-traumatic subarachnoid hemorrhage, typically aneurysmal, and non-traumatic subdural hematoma is associated with high mortality. Accounts for approximately 10% of strokes for ICH and having an incidence of about 25 per a hundred thousand person years. Although the worldwide incidence of ICH and its associated morbidity and mortality have remained stable or decreased since the 1970s, there really hasn't been a lot new that's come out in terms of management or definitive therapy for these patients. Even though that might be changing if anybody's aware of some recent trials that haven't been published yet. And actually there's some signal that in the United States incidence of ICH is actually increasing. So in the absolute sense, it's ICH still remains associated with high mortality and poor functional outcome. But you can't help but ask yourself why? So the most common reason that patients with ICH die is withdrawal of life support. And one study reported that 26% of its ICH patients underwent withdrawal of life support and patients who experienced ICH are thought to be particularly vulnerable to this withdrawal bias with early withdrawal occurring more commonly in this population than in patients with ischemic stroke. So that's self-fulfilling prophecy. The most used scale is the ICH score, which is associated with 30 day mortality in a one year modified Rankin. The score is relatively simple to complete and includes variables of age, examination, the volume of hemorrhage, presence of intraventricular hemorrhage and then the location. But the truth is this was never meant to be a prognostic tool, but rather a measure of injury severity. The creators of the scale studied this further by prospectively observing patients without do not resuscitate orders, despite having elevated ICH scores and receiving full support for five days. They found that these patients had a significantly lower mortality than predicted by ICH score without increased disability with the biggest changes occurring in patients with high ICH scores. In fact, the most recent AHA, ASA guidelines for patients with spontaneous ICH recommend postponing changes in code status or even discussion of withdrawal of life support until at least a second full hospital day. And those were the 2022 guidelines. And this month, as I'm recording this, so November, 2023, we had the Neurocritical Care Society, Neuroprognostication and ICH guidelines come out. And they say, although many clinical variables and grading scales are associated with ICH poor outcome, there's not any one variable or sole grading scale suggested as being reliable by itself to use in counseling patients with ICH or their families regarding functional outcome at three month and beyond or a 30 day mortality. So the truth is there's not one thing that's gonna say that this patient's gonna do well. The one for sure thing to have patients not do well is to remove all of their support. Non-traumatic or aneurysmal subarachnoid hemorrhage also recently received updated guidelines by AHA and ASA. And they advise that treatment of high-grade aneurysmal subarachnoid at even advanced stage is reasonable following discussion of prognosis. They also discuss that some patients with high-grade subarachnoid hemorrhage might not benefit from treatment with factors associated with poor prognosis, including absent brain stem reflexes, lack of response to noxious stimuli, large completed infarct on CT or presence of global cerebral edema consistent with anoxic brain injury. But even in the presence of these factors, the guidelines recommend full treatment and postponement of DNR orders for at least 72 hours. So traumatic brain injury is one of the areas that's actually best studied because of large databases. So TBI is devastating, obviously, in terms of personal, societal, and wider economic impact. And prognosis is difficult, but important because it guides appropriate treatment, tries to limit the proportion of patients left in a persistent vegetative state, and it aids the family in coming to terms with what their loved one's condition might be in the future. Clinical physiologic variables and radiological predictors and biological markers do exist, and all are more useful for prediction at the population level than for guiding decisions concerning individual patients. Historic data, such as the MRC crash study in 2008, which was approximately 10,000 TBI patients with GCS less than 15 had these outcomes. One in five were dead at two weeks, one in four dead at six months, one in three dead or severely disabled at six months. But we have some new data that's come out that actually suggests that a lot of these patients do well. So TBI is a global term and encompasses a lot of different pathologies, including cerebral contusions, traumatic subarachnoids, subdural or epidural hematomas, and traumatic axonal injuries. Combined, TBI is responsible for 30% of all injury-related deaths. Withdrawal of life support has been deemed the most important predictor of TBI outcome. Like in cardiac arrest, it's recommended that neuroprognostication not occur until after 72 hours. But unlike cardiac arrest, forthcoming NCS guidelines prefer a period of two weeks allowing polytrauma to be appropriately managed. So the CRASH model, or corticosteroid randomization after a significant head injury, it was developed in a large cohort of TBI patients to predict short-term, so less than 14-day mortality in death and severe disability at six months. Variables were age, GCS, pupils, presence of other extracranial injuries. They expanded this to the CRASH-CT model and included CT findings. The International Mission for Prognosis and Analysis of Clinical Trials in TBI or IMPACT model also developed with a large cohort pulled from 11 studies. Variables included age, GCS, motor subscore specifically in pupil response. And they had an extended model as well, which included what I just said, plus the presence of hypotension, hypoxia, the Marshall CT classification and presence of traumatic subarachnoid or epidural hematoma. So these two scores have been able to accurately predict six-month mortality and unfavorable outcomes. IMPACT performed slightly better than CRASH when predicting unfavorable outcomes in six-month mortality. A more recent systematic review and meta-analysis showed more similar performance between the two models with six-month mortality. But even more recent clinical trials have demonstrated that recovery continues beyond six months, suggesting that at this point might be too early to prognosticate functional recovery. The RESQ-ICP trial randomized 408 patients with TBI and refractory ICP to decompressive craniectomy or medical therapy, including barbiturate coma. This 2016 publication found that decompression in patients with TBI and refractory intracranial hypertension decreased mortality, but entailed higher rates of severe disability at six months. So then more recently, the study went secondary analysis to evaluate functional outcomes at 24 months. As observed with the first analysis, the surgical group was lower mortality than the medical group with higher rates of vegetative state and severe disability, but similar moderate disability and good recovery. But with this additional year of observation, the study found that the surgical group was more likely to improve over time than the medical group with 30% of the surgical group in proving their Glasgow outcome scale extended score by one compared with 14.5% of the medical group. So what that means is at 24 months, the rate of moderate disability in the surgical group was higher than in the medical group, whereas rates of good recovery stayed the same, demonstrating this gradual improvement in the ghost score. So the TREC TBI study, which came out in 2021, prospectively analyzed functional outcome in 484 patients with moderate to severe TBI from 18 level one trauma centers in the US between 2014 and 2018. They looked at ghost score and disability rating scale at two weeks, three months, six months, and 12 months post-injury. Ghost scores of four to eight with eight indicating full recovery were considered favorable, and scores one to three, one is death, considered unfavorable. They demonstrated incremental improvement in ghosts at each time point for most patients. In the group with severe TBI, 12% of patients achieved favorable recovery or outcome at two weeks, rising to 45% at three months and 52% by 12 months. Full recovery was achieved by 12.5% of patients. So in the moderate TBI cohort, 41% had a favorable outcome at two weeks, increasing to 69% at three months and 75% at 12 months. No participant in either the moderate or severe demonstrated a decline from favorable to unfavorable outcome from two weeks to 12 months post-injury. So nobody got worse. Everybody either stayed the same or got better, but a lot of them got better. The degree of disability also continued to improve as the study went on, with 19% of those patients with severe TBI having no disability by one year, and 32% of patients in the moderate TBI having no disability. Maybe most compelling or exciting is that of the patients in a vegetative state at two weeks after injury, that 78% of these patients regained consciousness and 25% regained orientation by 12 months. So what did guidelines say? Shockingly, delay neuroprognostication until after 72 hours. NCIS suggests waiting at least two weeks and really this just hits home the need for patients and the importance of long-term care and rehab to optimize recovery. So if we're taking the key points from all of these is you need to allow time for recovery. If there's radiographic evidence of structural damage, most of the time this is not reversible, that confounding factors can severely influence a patient's examination. You need to remove whatever you can and we'll talk a little bit more about how to talk to families about this. So brain death should be easy, but unfortunately it is not. There's been a lot of hospital to hospital variability in the declaration of brain death in the U.S. Study shows discrepancy and adherence to AAN guidelines with 76% of patients trained but only 25% performing exams consistent with AAN, including apnea testing that if the patient breathes and they continue on for more confirmatory tests for brain death, even though the patient is breathing and they are not brain dead. And also cases where brain function has been observed after the diagnosis. And honestly, it is getting more challenging. We don't have good guidelines for how to perform brain death testing with ECMO. There've been some proposed methods such as to induce PaCO2 challenges or changes but that has risks. And then new technologies might actually restore brain function post-mortem. There were a couple of reanimation studies that came out which challenges the whole concept of irreversibility in brain death and that NRP now exists which may conflict with UDDA's definition of death and the ethical concerns over reversible cessation of circulation. So there are proposals to redefine brain death in light of new evidence with possible inclusion of hypothalamic function and assessments. But the truth is that there are unresolved controversies with debate over validity of death by neurologic criteria. And there's a need for alignment between clinical practice and advancing technology. Right now, things haven't changed but it'll be interesting to see where this goes. And at the end of this, probably the most important thing from everything that I've said is how do you communicate with the family? So it's really, really important early on to talk to the family and say that we're gonna be continuing to gather data to try to paint a picture of what the future will look like but knowing that it's gonna be more of an impressionist painting that we're putting together. It's also really important to know what's important to the patient, whether or not living in a certain disabled state would be acceptable to them or not. And then to acknowledge that there's not one clear test that's gonna tell us exactly how they're gonna do, that some patients surprise us for the better and other patients we do everything right and then they continue to get worse. It's also important to acknowledge that if families want a second opinion, that that's reasonable and to encourage provider awareness of palliatives role, which does not necessarily mean a withdrawal but how to make the patient and the family comfortable. One good communication technique is first having the family or the surrogate explain to you the patient's condition and if they've ever seen other patients with a similar condition and what their prognosis and outcome was. This gives you kind of a landscape for the family's views. Also again, frequent check-ins with family and giving them more information as it becomes available and reiterating the diagnostic uncertainty for prognosis but putting it in context to what would be acceptable to both the patient and the family. Also discussion when you get towards the end of the ICU stay regarding patient disposition, whether they'd be a candidate for inpatient rehab versus nursing home versus LTAC is really, really important and might influence what the family wants to do. So that being said, we're at time. So any questions you have, please X me or tweet me or email me and I'm happy to answer them or if you're in the live class and we'll have some opportunity to talk now. Thank you.

Diane McLaughlin

nurse practitioner

neuroprognostication

coma patients

end-of-life decisions

neuroimaging

brain death diagnosis