The Challenges of Neuroprognostication in Critical Care

Patients with severe acute brain injury often lack the capacity to make their own medical decisions, leaving surrogate decision makers responsible for life-or-death choices. Patient-centered approaches and scientific methodologies can guide clinicians' prognostications.

In this episode, Teshamae Monteith, MD, FAAN, speaks with Susanne Muehlschlegel, MD, MPH, FNCS, FCCM, FAAN, author of the article "Prognostication in Neurocritical Care," in the Continuum® June 2024 Neurocritical Care issue.

Dr. Monteith is the associate editor of Continuum® Audio and an associate professor of clinical neurology at the University of Miami Miller School of Medicine in Miami, Florida.

Dr. Muehlschlegel is a professor (PAR) in the departments of neurology, anesthesiology/critical care medicine and neurosurgery, division of neurosciences critical care at Johns Hopkins University School of Medicine in Baltimore, Maryland.

Additional Resources

Read the article: Prognostication in Neurocritical Care

Subscribe to Continuum: shop.lww.com/Continuum

Earn CME (available only to AAN members): continpub.com/AudioCME

Continuum® Aloud (verbatim audio-book style recordings of articles available only to Continuum® subscribers): continpub.com/Aloud

More about the American Academy of Neurology: aan.com

Social Media

facebook.com/continuumcme

@ContinuumAAN

Host: @headacheMD

Guest: @SMuehlschMD

Transcript

Full transcript available here

Dr Jones: This is Dr Lyell Jones, Editor-in-Chief of Continuum, the premier topic- based neurology clinical review and CME journal from the American Academy of Neurology. Thank you for joining us on Continuum Audio, a companion podcast to the Journal. Continuum Audio features conversations with the guest editors and authors of Continuum, who are the leading experts in their fields. Subscribers to the Continuum journal can read the full article or listen to verbatim recordings of the article by visiting the link in the show notes. Subscribers also have access to exclusive audio content not featured on the podcast. As an ad-free journal entirely supported by subscriptions, if you're not already a subscriber, we encourage you to become one. For more information on subscribing, please visit the link in the show notes. AAN members, stay tuned after the episode to hear how you can get CME for listening.

Dr Monteith: This is Dr Tesha Monteith, Associate Editor of Continuum Audio. Today, I'm interviewing doctor Susanne Muehlschlegel about her article on prognostication in neurocritical care, which is part of the June 2024 Continuum issue on neurocritical care. Well, Susanne, thank you so much for coming on the podcast, and thank you for writing that beautiful article.

Dr Muehlschlegel: Thank you so much for having me. Excited to be here.

Dr Monteith: Why don't we start with you just introducing yourself?

Dr Muehlschlegel: Yeah, sure. My name is Susanne Muehlschlegel. I'm a neurointensivist at Johns Hopkins in Baltimore, Maryland. I have been a neurointensivist for about eighteen years or so. I worked previously at the University of Massachusetts and recently arrived here at Hopkins.

Dr Monteith: Cool. So, what were you thinking about - What information did you want to convey - when you set out to write your article?

Dr Muehlschlegel: Yeah. So, the article about neuroprognostication is really near and dear to my heart and my research focus, and I'm very passionate about that part. And as neurologist and neurointensivist, prognostication, you know, might be considered the bread and butter of what we're asked to do by families and other services, but as the article states, is that we don't usually do a great job (or physicians sometimes believe they do). But when you actually do research and look at data, it's probably not as good as we think, and there's a lot of room for improvement. And, so, the reason for this article really was to shine the light at the fact that I think we need to really make neuroprognostication a science, just like we make prediction models a science - and, so, that is the main topic of my research, as well as the article.

Dr Monteith: So, we know about your interest in research in this area, but what got you into critical care to begin with?

Dr Muehlschlegel: Yeah. It's, pretty much, a story of always being drawn to what's exciting and what others may want to avoid. So, in medical school, people were afraid of neurology and learning all the anatomy, and I just loved that and loved interacting with these patients. And then, in neurology residency, I was drawn to not just treating the brain and the spinal cord, but also the entire patient (so the lung and the heart and the interaction of all the organs). And then, naturally, I'm a little bit of an impatient person, and so I like the environment of the ICU of rapid change and always having to be on my toes. And so that's what drew me into neurocritical care. It was a very new field when I was training, and so, I was probably, you know, one of the, maybe, first- or second-generation neurointensivists.

Dr Monteith: And it sounds like you're maybe okay with uncertainty and a lot of variability?

Dr Muehlschlegel: Well, you know, neuroprognostication - I think everyone has to acknowledge that we cannot take away uncertainty, right? So, folks who pretend that they know for sure what's going to happen - I think the only time we can say that is in a patient who's braindead. But everyone else, we really don't know for sure, and all we can do is do the best to our ability to give a rough outlook - but we need to acknowledge uncertainty, that's for sure.

Dr Monteith: So, can you just give us a few of the biggest causes of variability when it comes to withdrawing life-sustaining therapies in patients with severe acute brain injuries?

Dr Muehlschlegel: So, that's the focus of quite some research. And, of course, there are many epidemiological factors, patient severity of disease, and, you know, how fast someone might arrive to the hospital, ethnic, racial, social demographic factors (and there's research on that), but when you adjust and control for all of those factors, variability remains. And so, what I've observed in my practice and what I also describe in the article is that maybe it's the way physicians describe prognostication or communicate with families, meaning there is potentially the chance for physician bias - that may also drive prognostication. And I can tell you from my own experience, what really drove me into this area is anecdotal experience that probably we've all had of other physicians kind of nihilistically prognosticating, thinking, you know, "This is going to be bad no matter what", and not even wanting to try to provide aggressive care to patients. So, I think these what we call "self-fulfilling prophecies" we need to be very aware of. So, I think some of the variability may be driven by other factors other than family, patient, or health system factors.

Dr Monteith: And you outline that really nicely in the article, so thank you for that. Why don't you just give us an example of a challenging case that maybe you're still thinking about today, that maybe happened years ago, that helps us understand what you go through?

Dr Muehlschlegel: Yeah, I'll rephrase the case. I still have, you know, very vivid memories about this, but I tell my residents about this case. When I was a fellow, there was a young patient in his early forties, a father of several children, a young family man who had a big right MCA stroke and really was progressing to the point that it was clear that he needed a hemicraniectomy or he was going to die. Discussed this with my attending, who said I should consult neurosurgery. At the time, the neurosurgical service had a transition to practice service for these emergencies - and so, these were fairly young, chief residents or early-year attendings. And the person came in, went into the patient's room, and I didn't even know about it, and came out and then just said, "Family decided for CMO". I was very surprised and shocked and was trying to understand how this happened, and this provider, all he said was, "Well, it's all how you put it to the family. I told him that he probably shouldn't be a vegetable. They didn't want him to be a vegetable, and so this was the only option." And, so, I was very shocked, and the patient did progress to die within a few days. And, so, that was a dire example of how biased prognostication can drive families to maybe an unnecessary outcome.

Dr Monteith: And what's CMO?

Dr Muehlschlegel: I'm sorry. Comfort measures only - so, essentially, a withdrawal of life-sustaining therapies.

Dr Monteith: Yeah. That is a good example of that and how our bias can inform families and maybe not with the exact amount of data to support that, as you outlined so nicely in your article.

Dr Muehlschlegel: And I do want to emphasize, I don't want to generalize that all providers are like that, but it is an example that really still sticks in the back of my mind, and I think, you know, we need to shine a light at how we do this and how we do it right or wrong.

Dr Monteith: And wouldn't it be nice to just have more objective measures (right?) to guide us? So why don't we talk about existing tools that are used to help guide neuroprognostication?

Dr Muehlschlegel: Yeah, so I think, in general, we can break down prognostication to two pieces (and I outline that in the article as well). So, one is, kind of, a derivation of prognostication in the head of a physician or, you know, clinician – and what may go into that is how the patient presented, examination, radiology or other diagnostics, biomarkers, you name it. But, then the second part of it (that also is really important) is how we put it to the family, right? Because we can influence families in a way that we may not even be aware of, and I think we all have unconscious biases, and how we talk to families is really important and may drive what happens to the patient as well. So, I always say there's two pieces to that – so, first of all, how we come up with a prognosis, and then how we disclose that to the family.

Dr Monteith: So how can we better handle uncertainty?

Dr Muehlschlegel: So, we actually did some research on that and we asked stakeholders, "How do you want physicians to handle uncertainty?". People are aware that no physician can be certain (again, other than in the case of brain death), and so families are very aware of that. And there's quite some data out there to suggest that if physicians have very absolute statements - you know, want to close the door by saying something very absolute - is that the optimistic bias in families goes up. So, the mistrust in what the physician is saying, coming up with their own (you know, "This is a fighter, and he or she is going to do better than what you're saying") - and, so, I think, you know, there's no true answer to what's the absolute right way to do it, but some have suggested to maybe fully acknowledge that there is uncertainty. That's actually what families want you to do, based on some qualitative research we've done – is to say, "I do not have a crystal ball. There will be uncertainty", but then to potentially go into a best/ worst-case scenario. But again, there, all we can do is give a best gross estimate and guess. And so, the work is not really clear at this point. There's research ongoing as to what should be the best way of doing it, but currently, that's what is suggested.

Dr Monteith: And in your article, you spoke about some pretty innovative approaches, such as modeling, to help guide shared decision-making. And, so, you know, how reliable is that?

Dr Muehlschlegel: That's a good point, right? So, that is up to statisticians or those who are inventing these new models. So, you know, in the old days we used logistic regression, maybe linear regression. Now, there are fancy machine-learning modeling and other Bayesian models that people use, and they certainly have some advantages that I outlined in the article. Bayesian models, for example, may use serial data as it comes in throughout the patient's hospital course - and that's kind of how we do it in real life. But, I think what's really important before we apply models is that we know that there's always outliers, and we don't know if this one patient might be the outlier, and that we need to validate these models, and most importantly, look at calibration. So, I talk in the article about how, you know, all models always report the what's called "area under the receiver-operating curve (the AUC)", which is discrimination. But, what's actually more important for a model to be applied to a patient at the bedside is calibration, meaning how well does it actually predict a potential outcome. And, you know, there's a lot of research into that, that only maybe half of the papers that report on a new model actually report calibration - so, I think it's really important to pay attention to that (has the model been validated and calibrated before we actually use these models?). I think prediction models have definitely a important role. But, then again, as the article says, we also have to think about how we then apply that to the patient and how we do it in individual patients.

Dr Monteith: And then, of course, there's some variability between institutions.

Dr Muehlschlegel: That's for sure. You know, there's these systematic approaches or system-based cultures in certain institutions. And then, of course, you know, there's still this model of learning from a role model or a mentor or an attending - meaning you look at how this person does it and then you may adapt it to your own practice. I think we need to critically examine whether we need to continue with that kind of apprenticeship model of learning how to neuroprognosticate, or whether we need to have other educational ways of doing that. So, especially in the field of palliative care, there's a lot of education now around communication - and I think med students get that exposure, and residents may get that exposure, too - but I think we need to practice it and study it systematically, whether having a standardized approach to do this leads to more patient-congruent decisions.

Dr Monteith: And, you know, we do have a lot of trainees, residents, and fellows that listen in. So, what are some key messages that you want to make sure gets conveyed?

Dr Muehlschlegel: Key messages is that, I think, we need to move away from looking at a patient the first one or two weeks and then concluding that we will know what will happen to this patient in six months or a year or further down the line. I think there's not a lot of longitudinal studies out there now that show that patients actually probably do better than expected if they're allowed to live. And what I mean by that is many studies allow early withdrawal of life-sustaining therapies within the first three days or maybe two weeks - but if we actually allow these patients to live, people wake up more than we thought, people may do better than we thought. So, referring to the article, I discuss in detail some twelve-month data from the TRACK-TBI study or very interesting results from South Korea where withdrawal of life-sustaining therapies is forbidden by law. And, so, you can actually do a true natural-history study of what happens with these patients if you allow them to live. And, surprisingly, a lot of people that, you know, within the first two weeks were still comatose actually ended up waking up. And, I think it's really important to look at those studies and to continue to conduct those studies so that we know better what might happen. I always shudder a little bit when I hear, "We need an MRI in the first few days or first week for neuroprognostication". And then I always question, "Well, what is it really going to tell you about that patient who clearly isn't brain dead and still has certain, you know, exam findings?" and "Shouldn't we just give those patients time?". I think some of those were a bit too quick to provide poor prognostication if we really don't know.

Dr Monteith: And, so, I want to know how did you get into research? You know, it can be competitive to get funding, grant funding - so, tell us about that in terms of, you know, your day-to-day, what's it like? And then, also, what makes you most excited about research happening in this area?

Dr Muehlschlegel: Yeah, I mean, there's a lot of research happening in that area. I think there's a huge focus on biomarkers and models and all sorts of new diagnostic tools to predict outcome, big push over decades now to do large longitudinal epidemiological studies - and all of those are very, very important, you know. I just mentioned as an example, the TRACK-TBI study is one of many other examples. I'm also excited about doing research in the second part of neuroprognostication that I mentioned - the communication and disclosure part - and the potential of bias as we speak to families. So, I get very excited about that part. It's not easy to get funding, but I think what's important is to focus on the potential impact. And, of course, then you try to convince funders that this is important research that has to be done in addition to funding model development and large epidemiological studies. What my day-to-day looks like? Well, you know, we have several ongoing projects (I won't get into details on that), but to get involved would probably be the best time as a trainee - so, I have medical students working with us, residents and fellows (although their time can be limited). And then to continue to just be curious and ask questions.

Dr Monteith: And what do you find most exciting about the work that you do? Just, kind of, overall?

Dr Muehlschlegel: I mean, without a doubt, the potential impact, right? So, changing the field a little bit. I'm not claiming that my research is doing that - I hope it might. But, most importantly, it's the potential impact on families and patients. I think our goal is not to have less withdrawal of care (although, sometimes, I just think we need to give people more time), but I think it's important to focus and ask about what patients might want, and then really focus families onto that. I think that can be difficult, because patients don't always tell families what they would want or families want something different than what they know the patient might want - and so, we spend quite some time on that when we speak to families. And then, I also talk about the disability paradox. So, you know, at one point, the family might say, "Well, he would not want to live if he can't walk", but then, patients, as they learn to live with this new normal, may actually later say, "Well, it's not as bad as I expected it to be, and I'm actually very happy to be alive, even if I'm not able to walk". And so, that's something that others are doing research on, and that's also important to consider.

Dr Monteith: Yeah, that's cool. Thinking about outside of the ICU, right?

Dr Muehlschlegel: For sure. Yes.

Dr Monteith: Great. Thank you so much for being on our podcast. I know that our listeners are going to really enjoy reading your article and all the thought that you put into that.

Dr Muehlschlegel: Thank you so much for having me.

Dr Monteith: Again, today, we've been interviewing Dr Susanne Muehlschlegel whose article on prognostication in neurocritical care appears in the most recent issue of Continuum on neurocritical care. Be sure to check out Continuum Audio episodes from this and other issues. And thank you to our listeners for joining today.

Dr Monteith: This is doctor Teshamae Monteith, Associate Editor of Continuum Audio. If you've enjoyed this episode, please consider subscribing to the journal. There's a link in the episode notes. We'd also appreciate you following the podcast and rating or reviewing it. AAN members, go to the link in the episode notes and complete the evaluation to get CME for this episode. Thank you for listening to Continuum Audio.