SCCMPod-573 CCE: Long COVID After Critical Illness

Dr. Enfield: Hello, and welcome to the Society of Critical Care Medicine's podcast. I'm your host, Kyle Enfield. Today, I'm speaking with Dr. Matthew Baldwin, MD, MS, about the article post-acute sequelae of COVID-19 persist over three years in acute lung injury, acute respiratory distress syndrome survivors, but are not associated with persistent thromboinflammation or endothelial dysfunction. It was published in the March 2026 issue of Critical Care Explorations. To access the article, visit ccejournal.org. Dr. Baldwin is a board certified pulmonary and critical care physician and clinical investigator with a master's in biostatistics. He's associate professor of medicine at Columbia University Irving Medical School in New York City, New York. Welcome, and before we start, do you have any disclosures pertinent to this talk that we should know about?

Dr. Baldwin: No, I have no disclosures.

Dr. Enfield: Matt, I was excited to read your article. I see a lot of long-term COVID survivors, both those with ARDS and were on a ventilator from sort of the first wave and those that have survived afterwards. When you think about these patients, especially those that were critically ill and mechanically ventilated, what did we know about their long-term outcomes before your study?

Dr. Baldwin: Yeah, so before this study, most of our long-term outcomes really were extending out only six months to one year. In one of the most prominent studies, a large group in Australia published a comparison of mechanical ventilation survivors in the era just prior to COVID, and then they changed to enrolling only COVID mechanical ventilation survivors in 2020 and 2021. Australia wasn't overwhelmed as we were, so critical care was consistent over that time period.

And what they found was that long COVID and post-intensive care syndrome in ARDS survivors was really more similar than not at six months in terms of incident disability, mental health problems, and cognitive impairment. But we now know a couple of years later that long COVID can last more than six or 12 months and that it can affect individuals who are not just ICU survivors, but even individuals who have had COVID and were never hospitalized.

Dr. Enfield: So your cohort for this study encompasses those that were in the first wave of the pandemic. For critical care physicians, that's a pretty marked time point in our history. But if you could help the listeners who might not be as familiar with that cohort of patients, what were they like?

How sick were they? And what did it mean that half of them required mechanical ventilation?

Dr. Baldwin: Yeah, in the first part of spring 2020 in New York City, we were just really overwhelmed. I remember I intubated the first patient who had acute respiratory distress syndrome from COVID in New York City. And that was day one.

And then on day two, we intubated two patients. And then on day three, we intubated 12. And then on day four, it was nearly 100.

The way it worked at our hospital is we changed wards to ICU rooms. We had three patients mechanically ventilated with COVID ARDS in each operating room that were originally used for elective surgery. So it was an overwhelming number of patients.

And I think we still provided a high level of care. Our patients really were on ventilators for a long time. The median duration of mechanical ventilation, that's just the 50th percentile, was 34 days in this cohort.

I was particularly interested in following these patients, but I wanted some heterogeneity as well. So we specifically enrolled about 50% mechanical ventilation survivors. That was intentional.

About 25%, I call non-invasive mechanical ventilation survivors, so that's high flow nasal cannula, non-invasive positive pressure ventilation, and about 25% of those with nasal cannula. And our goal really was to kind of break that standard barrier of following up people for six to 12 months and to go for three years and really say, what does happen after three years? In the pre-COVID era, we know that one of the most famous cohort studies, ARDS survivors published by Margaret Herridge and the Toronto Group, followed survivors out at one, two, and then five years.

And we really wanted to try to emulate them in many ways with this group of COVID survivors.

Dr. Enfield: So you chose to use the recovered definition to classify post-acute sequelae of COVID. There are a lot of different definitions, but that one has become sort of the gold standard. Why did you choose that definition and why is it important for us to understand the definitions used when we're looking at different cohorts?

Dr. Baldwin: Yeah, I think that's a great question. So when we create standardized definitions of heterogeneous and complex syndromes, we do it so that we can increase our knowledge and understanding of the syndrome by comparing and contrasting findings across different studies. We did this with ARDS in the Berlin criteria originally.

In order to kind of unify the literature and to make comparisons, we had to have a standard definition and the NIH created the recovered cohort, which consisted of about 10,000 individuals, only 1,000 of which who were hospitalized, who were followed out systematically with testing for SARS-CoV-2 infection, for antibodies, and for symptoms in a very systematic way. And then they analyzed these symptoms and came up with 12 symptoms that were most common and debilitating, and then weighted scores for each of those symptoms and figured out that based upon these 12 symptoms with different scores, if you have a score of 12 or more, you are many more times likely to have passed than someone with such a similar score who has never been infected with COVID proven with SARS-CoV-2 antibody testing.

And so as you stated, this has really become an anchor or a framework on which to study long COVID. We think our manuscript adds to the literature significantly because this was really the first time we were really drilling down and asking the question whether this phenotypic definition of long COVID applied to acute lung injury and ARDS survivors, remembering that it was derived in a 90% outpatient treated population of COVID survivors.

Dr. Enfield: So you hit on some of the things that I found fascinating, one being that the recovered definition really focused on those outpatient survivors. I guess one of the other questions that I had is when you were designing it, you chose a series of biomarkers that you were going to follow over time. What was the biological rationale for those biomarkers?

And how did you go about with your team choosing what you were going to follow? Because that has a cost associated with it as well as both monetarily and time-wise for the patient and the team to collect them.

Dr. Baldwin: Yeah. So in terms of the biomarkers, this was really hypothesis driven. And so I went back through the pre-COVID acute lung injury ARDS literature.

I went through the COVID literature that existed at the time around 2021-2022 when we were first choosing these biomarkers for this study that was funded by the US Department of War. And what I decided is let's go with markers of activated innate immunity in myeloid cells. And so that included things like IL-6, IL-8, tumor necrosis factor receptor one.

We know that these factors are elevated in acute COVID. We know that there are markers for more severe disease in non-COVID ARDS. We also know that these biomarkers are also implicated in the pathogenesis of age-related sarcopenia and frailty in longitudinal studies of aging.

We moved on to measure GDF-15, which is a stress-related biomarker that's released in response to hypoxemia, inflammation, and even aging. We had done some prior work that related this to physical function, higher levels associated with worse physical function in a pre-COVID acute hypoxemic respiratory failure cohort. This biomarker is also emerging as a potential sensitive, but perhaps not all that specific biomarker of mitochondrial dysfunction.

And then we went after the biomarkers of endothelial dysfunction, such as ADAMS-TS13, angiopoietin 2, ICAM-1, VCAM-1, because there was some really nice literature published by Dan Leisman out of MGH and Harvard that really showed that these biomarkers in days three through seven in COVID-19 ARDS really identified those who are much more likely to have worse outcomes and more prolonged dysfunction, at least in the first 28 days. And then finally, we added on the compliment in response to emerging literature at that time that had suggested that thromboinflammation was very distinct and important in the pathogenesis of COVID ARDS.

We know that histopathologically, COVID ARDS is distinct insofar as there's microthromboses in the lung. And there was one lung COVID study that came out of Europe published in Science that suggested that compliment-related thromboinflammation perhaps persisted and was associated with PASC-like symptoms. And for that reason, we picked the compliment biomarkers.

Dr. Enfield: That is a well thought out hypothesis. I guess along the same lines, you have patient-supported outcomes. And how did you measure the tension when the objective measurements and the patient-reported outcomes differed, at least in your planning phase?

We'll get to the interpretation and outcomes in just a second.

Dr. Baldwin: The audience can understand, we were designing this study while we were kind of fighting back COVID. And what we did is we had a committee at Columbia University of pediatricians all the way through geriatricians, many of whom had participated in longitudinal studies in the outpatient setting. And we just created a symptom survey of 43 symptoms with a seven-level ordinal categorical scale just through kind of multidisciplinary meetings after rounds.

And that was the best thing that we could come up with at the time, kind of pulling from expertise across the lifespan of both health and disease. And luckily enough that this 43 symptoms, seven-level ordinal categorical scale questionnaire was quite similar to what the recover cohort ultimately decided to go with. And perhaps that's not too surprising because some members on this multidisciplinary committee ultimately were part of the recovery study, especially on the pediatric session.

Other members of that committee who helped devise this questionnaire ultimately became PIs of the C4R NIH initiative of longitudinal measurements of COVID outcomes in existing NHLBI cohort studies. So how do we handle the gap between the biomarkers, what patients actually feel? These symptom questionnaires is really the best thing that we could come up with.

We tried to be really thoughtful about it. We tried to capture every organ system and some degree of severity. And then we tried to basically ask ourselves, well, are we capturing something that's real and that's legitimate?

And that's why we tested associations between the past score based on these symptoms and objective measures of physical function with regards to frailty.

Dr. Enfield: Your findings, one that starts and is sort of a headline finding is that one in four of your participants, about 26% still met past criteria at three years. For the critical care audience, put that number into context and tell us, did that surprise you?

Dr. Baldwin: Yes, it did. I thought it would be a little bit lower. What's interesting about our cohort is we specifically sought to enroll patients who were community dwelling and ambulatory without any reported significant mobility disability at baseline.

What's also important for the audience to know is these patients who we enrolled during the first wave of COVID, many of whom were ARDS survivors, our mean age was 56 years. Our standard deviation is 13. So if you just kind of think about that, there's a lot of 40 and 50 somethings in our cohort with no preexisting disability, therefore they don't have a lot of other confounding reasons or comorbidities to feel this way because they didn't have much comorbidity prior to their COVID illness.

So the fact that there's a 25, 26% prevalence at 15 months and then three years, nearly the same prevalence that was noted in the NIH recover cohort, only six months out in a population that was mostly outpatient really means that this is a population of patients, the acute lung injury, the ARDS patients who are survivors who we need to focus on. Their problems aren't going away. This truly is an infection associated chronic illness.

Dr. Enfield: So in those lines, many of the patients that I see in our long COVID clinic are influenced by the influence out there promoting the thromboinflammatory and endothelial damage biomarkers and how that's important in treating post acute sequela of COVID. But that's not what you really found. Talk about the findings as it related to thromboinflammatory and endothelial biomarkers and how that has changed your thinking about this disease.

Dr. Baldwin: I think it's important to note that even though we have negative findings with our biomarkers of endothelial activation and complement, the ones that we chose, we thought were the best candidates, but it's not necessarily, it could miss, it could miss the mark. It could miss endothelial activation dysfunction to some extent. Angiopoietin-2, ICAM-1, VCAM-1, you know, I was a little surprised that angiopoietin-2 didn't have any signal because that's probably one of the best, or at least most widely used biomarker of endothelial dysfunction.

We know that endothelial glycol calyx dysfunction is important in the pathogenesis of sepsis, likely COVID-19, ARDS. But how we measure that over time in a stable way is challenging, and that's still an area of active research. With regards to complement, I think it's very important for us to know that measuring complement is really difficult.

We have to be very, very careful that we get the sample and that it's measured consistently the same way, that there aren't delays in processing the serum. Terminal complement, which has been implicated in the pathogenesis of COVID, is incredibly difficult to measure accurately and very unstable, and in fact, we tried measuring that, but our coefficients of variation were so wide when we did repeated analyses, and we did it four times, that we actually didn't include it in the study under the idea that any poorly or inconsistently measured biomarker is not worth reporting because it's too likely that your results may be biased. So I do think thromboinflammation could still be pathogenic. There's a wealth of literature out there, especially in the translational space, in places like science, science translational medicine, nature and nature immunology, that really suggest that these are pathogenic using omics technologies.

How we get from that omics measurement to a biomarker with a threshold that requires some type of pharmacologic action or intervention, that research is still, I think, in its infancy.

Dr. Enfield: So if I, if I interpret what you're saying correctly, while you had a, essentially a null hypothesis when it came to these biomarkers, there's still some questions out there and some need to further evaluate both how we draw the biomarkers and which biomarkers might be the most sensitive as we relate the omics research that has been done to our patients and how that might be used as an intervention.

Dr. Baldwin: Yeah. And that's especially true with regards to complement measurement in endothelial biomarkers. I'd like to also extend my comments to inflammatory markers.

I think the measurement of inflammatory markers is in cytokines is fairly well-established. I was surprised to not find elevated IL-6, IL-8, TNF receptor inhibitor, or GF-15, because prior research has really suggested that these might remain elevated, particularly at hospital discharge and in a couple months after COVID. But what I think our negative findings might explain here is that inflammation may be prolonged after COVID or other critical illness for at least a few months, but then after one year, it may resolve to some extent.

But what we don't know is how that prolonged inflammation in the early post-acute phase, let's call that months, ultimately affects longer-term function of things like hormones, metabolomics, muscle function, so on and so forth. So I have a feeling that our negative inflammatory findings really could be that there's only a subset of prolonged inflammation, we were underpowered to detect it, or inflammation is more important only in the first year after hospital discharge.

Dr. Enfield: So not surprisingly, and you just comment on this, there were links to frailty and reduced physical function in these patients. And I wonder if you might opine a little bit about the directionality and also what your findings really show about how these patients become more frail after they have this acute disease.

Dr. Baldwin: Yeah. So we had really two goals in testing associations of frailty and function in this study. First is we really wanted to see, one goal of this study is to see whether the PASC phenotype based upon this scoring had construct validity.

What's that mean? That means, does it really measure something that's clinically meaningful beyond what patients grade on a seven level categorical ordinal scale? And I think the answer is yes, because those with PASC were more likely to have a higher frailty phenotype category.

They were more likely to have a lower short physical performance battery score, which is a measurement of lower extremity function based on balance, gait, speed, and chair stands. And they also had a lower six minute walk distance. This also proved true for the simple questions that suggested that patients might be suffering from post-exertional malaise, as well as physical fatigue.

What is interesting though, is that the brain fog group didn't have as many consistent associations with the physical frailty category, which really suggests that the brain fog of COVID is something that might be fundamentally different, both in terms of its pathogenesis. And it kind of makes sense that the physical function of frailty really correlates with the more physical symptoms and less so with the brain fog. The thing that kind of surprised us was we didn't, besides GDF 15 at three years, we didn't find much association of higher levels of inflammation with physical frailty.

And we know from the aging literature, and we know from some critical care outcome studies that to some extent, frailty is associated with higher levels of inflammation over time. But I also want to note that the prevalence of the frailty phenotype in our population was just under 10% at 15 months and three years. And that speaks to the fact that we enrolled a middle-aged cohort without disability.

And so whether inflammation is important in frailty after critical illness, well, it might be, but it might be more related to those with greater degrees or prevalences of frailty or older age.

Dr. Enfield: That finding is unique, but contextually, if your cohort expanded off across all the patients who had ARDS or critical illness from the first wave of COVID, it's still a large population that is now an older middle age or even younger middle age having high frailty scores. And I know this was not the focus of your research, but what do you think that means for us as clinicians moving forward as that population ages?

Dr. Baldwin: That's a really good point. We know that the prevalence of frailty in the 65 to seven-year-old population is somewhere only around, I think it's a little under 10%. And here we are finding a similar prevalence of frailty in a population mostly in their 40s and 50s with a low burden of comorbidity.

This goes back to the idea that in some patients, COVID is going to change the trajectory of someone's health and longevity significantly. Even if a couple of years out, we don't see a huge signal, if they were already on a trajectory towards more rapid aging and physical decline, that gap between someone who doesn't have PASC and has PASC and frailty may be many fold greater in hugely clinically significant at maybe one decade or two decades. An analogy that I like to draw on is this idea of we can see the beginnings of emphysema now that we're doing CT scans so often, but we know that emphysema may only become clinically significant two decades or three decades later, sometimes even longer.

And so I wonder whether we'll see a similar effect with long COVID.

Dr. Enfield: Yeah, I think my mind was going to the same place and also wondering what it really means for the rehabilitation of these patients, since we don't have a true understanding of the cause of this increased frailty, how do we approach recovery for them? And that may be a talk that you and I as critical care physicians are ill-equipped to answer, but I do think it's something that the healthcare community has to wrestle with as we go forward. Do you have any thoughts?

Dr. Baldwin: Yeah, I think there's a wealth of literature that suggests that more exercise is better, especially as you get older, that, you know, not only cardio training, but functional strength training is crucial. And so anyone who's willing to do rehab or is interested in it, I tend to make referrals to them for a physical therapist to help them to continue to either function well or to build back function that they lost. I was a little disappointed in this study with these biomarkers because perhaps naively, I came in with the hypothesis that, you know, so many of these biomarkers, as well as other cytokines now have monoclonal antibody therapies that were originally derived to treat rheumatic disease.

And could there be a way to repurpose existing monoclonal antibody therapies to inflammatory cytokines that are FDA approved or do have an IND to potentially knock out high levels of persistent inflammation in patients with long COVID? And I think I'm not the only one with that hypothesis. I know that there are a couple of long COVID studies that are looking at that.

But I think there was a lot of, in a weird way, hope that we'd be able to achieve some success with that. And I think to me, this article tells me that that idea is not going to work out as much as I had hoped to really kind of find some degree of inflammation that might be treated with monoclonal antibody repurposed from another disease.

Dr. Enfield: We know also that the first wave of patients who had COVID was different than the subsequent waves. And part of that may be us learning how to care for these patients differently. But also we know that the virus has a high propensity towards mutation and that has led to changes in the disease course as well as changes in sort of, you know, health, the global herd immunity.

What do you think your findings suggest for those patients in subsequent waves? Or is it that we're going to need to repeat these kinds of studies for each wave of the illness to really understand how PATH presents in future patients?

Dr. Baldwin: So I agree with you. This cohort of patients who are critically ill in 2020, I think are likely distinctly different because they got sick in the pre-vaccination era. And we know that vaccination is protective against long COVID.

We also know if we kind of dig through the data of the recover cohort, that once they excluded patients, which might've had selection bias, and they just looked at the prevalence of PASC six months out amongst those who were infected from December 2021 onwards. So after the vaccine was widely available, that the prevalence of PASC was only about 10%. Between the virus losing its virulence, widespread vaccination, I think this cohort is distinctly different.

So should we repeat this study in every wave? I think this study tells us that if we're going to be repeating studies, we need to kind of focus on what I'm seeing in long COVID clinics. And I'll be curious to what your experience is, is those with kind of the chronic fatigue syndrome or the more than neurogenic etiologies, the orthostatic hypotension, the post-exertional malaise, the cognitive impairment, and whether there's a completely different set of biomarkers that focus on nervous system dysfunction that might be able to characterize these patients.

I think that's where I'd like to see the field go now.

Dr. Enfield: Yeah, I do think one of the challenges that we've been stuck with, with our definitions is that we've created a syndrome that probably lumps too many different things together. And it makes it hard to really do great research. One of the things that I enjoyed about your study was the patient-centered questions and sort of this way of trying to phenotype people into different groups, because I do think that is the future.

And I think the recovery trials have headed in that direction of trying to distinguish between those with predominantly brain fog versus those with predominant physical symptoms, predominant other symptoms, and hopefully that trajectory will continue. I guess the other question I have for you is, and you've mentioned this about omics, where is your research heading in this area? And how are you going to apply what you've learned from this study to whatever you're going to do next?

Dr. Baldwin: I have moved towards omics. I currently have an NHLBI R01 funded study where we're examining ARDS survivors and doing omics both at hospital discharge in three months and correlating that with several objective measures of physical function. We're also doing some single cell analyses to try to figure out what the source of the dysregulated immunity might be, especially whether it's coming from peripheral blood mononuclear cells.

A couple of my comments on omics is usually the way omics works is it gives you a log fold change or how much different the biomarker is relative to the average level of biomarker within your own study population. Increasingly certain companies are now able to provide more healthy and age related population norms for omic measured levels of biomarkers. And I think that's a real advancement in the field.

My initial hesitation with omics was always around the idea that, okay, so this value is higher in individual number one in my study versus individual number four relative to the mean. What does that mean? What is that level?

I can't measure it and then react to it if I'm in a clinic because that level is only specific to my study population. I think omics is going to move into a new era where we'll be able to kind of have a better idea of what population norms are, and then potentially we'll be able to kind of understand where our concentrations in our study population kind of measure up to more healthy individuals. I think that'll be a huge advancement.

And second, I think beyond with the omics technologies, we've looked at so much inflammation collectively as a scientific community, and we just haven't found as strong of a signal as we thought in terms of long-term outcomes after critical care. And so I wonder whether we should start to think about the other omics. And that would be things like metabolomics or certain single cell studies that could potentially say, hey, in a subgroup with inflammation, what is the source of inflammation?

Maybe there's a cellular therapy or in the metabolomics, maybe we'll start to better understand changes in hormones and metabolism and mitochondrial function, which would then elucidate potential pharmacologic therapeutic targets.

Dr. Enfield: That sounds like a great career moving forward. And I guess as I think about the trainees and early career investigators who might be listening to this podcast, recognizing that running a three-year longitudinal study is a logistical and emotional rollercoaster. I wonder what advice you would give them as they begin designing their own studies so that they can go the distance.

Dr. Baldwin: Yeah, well, first I want to just say thank you for acknowledging that. We put in tremendous effort into recruiting and retaining this study. One of the things that I always do is I make sure my research coordinators are as professional and charming and as polite as possible.

We really thank the patients for their participation and thank them for the greater good. There was a couple of PIs on this project. Each one of us would try to at least shake hands with each patient as they came in for a follow-up visit.

Any unaddressed medical concerns, sometimes if they were having problems with a referral within our health system, which we know how difficult it can be to sometimes access specialists, we try to kind of go the extra mile and help pin down a referral for them. So I think if you're doing it yourself and you're prospectively following patients, it's a multidisciplinary, multi-pronged effort to really keep these patients coming back because they enjoy participating and feel like they're doing something good, both for themselves as well as for society. I think the other thing we have to really think about is joining forces.

And I know the NHLBI ARDS Pneumonia Sepsis Consortium is kind of an idea of that, that they're trying to really kind of create a 4,000-person longitudinal measures cohort, and they're trying to then follow a subgroup of them out six months. Any time you can get your hands on existing data and an existing biobank, and increasingly I think the NIH is trying to make efforts towards that, it creates incredible efficiency because you can start with data analysis and with biomarker analysis rather than spending years kind of boots on the ground enrolling and following these patients. That being said, there's no substitute that if you have a novel hypothesis, you might just have to recruit a little bit of a cohort to measurements you need to answer it.

To early career trainees, I would say examine the body of work that's out there and see what's publicly available for secondary analysis while you pursue perhaps a smaller project with very focused measurements to answer your hypothesis.

Dr. Enfield: That is great advice. On the other end of the spectrum, as clinicians walk away from this podcast and reading your article, which I would highly recommend they sit down and read, is there anything that you would like the listeners to walk away thinking about?

Dr. Baldwin: For the clinicians, I'd like them to walk away recognizing that COVID survivors, critical illness COVID survivors from 2020, one in four still reach a threshold of symptoms to meet the current definition of long COVID three years out in that their symptoms really should be acknowledged. And even if we don't have a treatment for them, we should at least help normalize how they are feeling, especially in this age where we don't have a medication to get them just yet. Sometimes just helping critical illness survivors know that they are normal, that to feel the way they feel is normal, can actually create a sense of comfort in them and perhaps motivation to kind of move forward and try different things, whether it be rehab, cognitive behavioral therapy, or other therapies.

Dr. Enfield: That is really well said. And I know that the patients I care for would appreciate you saying that and I hope the other critical care physicians and even the primary care physicians who might have stumbled onto this podcast will take that to heart. It's been great talking to you.

This concludes another episode of the Society of Critical Care Medicine's podcast. If you're listening on your favorite podcast app and like what you heard, consider rating and leaving a review. For the Society of Critical Care Medicine's podcast, I'm Kyle Enfield.

Announcer: Kyle B. Enfield, MD, is an Associate Professor of Medicine in the Division of Pulmonary and Critical Care at the University of Virginia. He received his undergraduate degree from the University of Oklahoma.

Join or renew your membership with SCCM, the only multi-professional society dedicated exclusively to the advancement of critical care. Contact a customer service representative at 847-827-6888 or visit sccm.org/membership for more information. The SCCM podcast is the copyrighted material of the Society of Critical Care Medicine and all rights are reserved.

Find more episodes at sccm.org/podcast. This podcast is for educational purposes only. The material presented is intended to represent an approach, view, statement, or opinion of the presenter that may be helpful to others.

The views and opinions expressed herein are those of the presenters and do not necessarily reflect the opinions or views of SCCM. SCCM does not recommend or endorse any specific test, physician, product, procedure, opinion, or other information that may be mentioned.