Hello and welcome to today's Journal Club Critical Care Medicine webcast. This webcast, hosted and supported by the Society of Critical Care Medicine, is part of the Journal Club Critical Care Medicine series. In today's edition, we feature two articles from Critical Care Medicine. The recording of this webcast will be available to registrants on demand within five business days. Log in to mysccm.org and navigate to the My Learning tab. My name is Thomas Zagmani and I'm a Professor of Intensive Care at Cardiff University in the United Kingdom. I will be moderating today's webcast. Thanks for joining us. Just a few housekeeping items before we get started. First, during the presentation, you will have the opportunity to participate in several interactive polls. When you see a poll, simply click the bubble next to your choice. Second, there will be a Q&A session at the conclusion of both presentations. To submit questions throughout the presentation, type into the question box located on your control panel. Third, if you have a comment to share during the presentation, you may use the question box for that as well. Finally, everyone joining us for today's webcast will receive a follow-up email that will include an evaluation. Please take five minutes to complete this. Your feedback is greatly appreciated. Please note, this presentation is for educational purposes only. The material presented is intended to represent an approach, view, statement, or opinion of the presenter, which 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. And now, I would like to introduce today's two presenters. Kimberly Rangel is an assistant professor of anesthesiology and critical care medicine at Vanderbilt University Medical Center. She received a BS from Southern Methodist University in Dallas, Texas, and her medical degree from the University of Texas Medical School at Houston. She completed her anesthesiology residency and critical care medicine fellowship at Vanderbilt Medical Center, where she joined the department's mentored research training track, the B. H. Robbins Scholars Program. Her research interests include understanding and preventing functional decline after major surgery or critical illness, particularly in the aging population. Our second presenter is Guido Falcone, who is a neurointensivist and epidemiologist with expertise in neuroimaging and population genetics. Dr. Falcone's research lies at the interface of clinical neurology, neuroimaging, population and medical genetics, and big data. His lab is interested in identifying new treatment targets and developing precision medicine strategies for stroke. Dr. Falcone's work is funded by the NIH, the American Heart Association, and the Neurocritical Care Society, and has been recognized with the Michael S. Pessin Stroke Leadership Prize, American Society of Clinical Investigation Young Physician Scientist Award, Robert G. Seeker New Investigation Award, and the Paul B. Beeson Emerging Leaders Career Development Award in Aging. Okay, I will now turn things over to our first presenter, Dr. Kimberly Rangel. Hello, and thank you to the Society of Critical Care Medicine for the opportunity to participate in today's journal club. I will be discussing our recent paper, Motoric Subtypes of Delirium and Long-Term Functional and Mental Health Outcomes in Adults After Critical Illness. I have no conflicts of interest to report, and the work I'm presenting today was supported by the National Institute of Health and a grant from the Department of Veterans Affairs. As critical care providers, we frequently have to address delirium in the ICU, this acute brain dysfunction that is characterized by a fluctuation in mental status, inattention, altered levels of consciousness, and disorganized thinking. It's extremely common and can sometimes be difficult to manage, and we know that it can cause prolonged ICU stays, ventilator times, and is associated with a higher mortality. But we also know that experiencing delirium during critical illness is associated with a prolonged or worse recovery. In one study, delirium in the ICU, each additional day of delirium in the ICU was associated with a higher probability of disability in the basic activities of daily living, things like bathing, dressing, and eating. And in a different study, there was another association between developing delirium in the ICU and having a higher mortality in the hospital six months after discharge from the hospital, and then having prolonged disability in those basic activities of daily living six months after leaving the hospital. Delirium can be further classified into motoric subtypes or the psychomotor presentation that patients experience, where hypoactive delirium is defined as having lower motor activity. Patients may have slowed speech. They may be more confused, somnolent, or withdrawn, whereas patients with hyperactive delirium are often confused or agitated and very restless, pulling at lines, tubes, and drains, potentially hallucinating and having high levels of motor activity, like trying to get out of bed. Many patients will also experience a mixed type of delirium, where they have symptoms of both hyperactive and hypoactive delirium in the same day. And when we start to examine these different subtypes of delirium, in previous work, it has been shown that hypoactive delirium is actually far more common than hyperactive delirium, and it's associated with longer periods of mechanical ventilation, longer stays in the ICU, and a higher mortality. But what remains unknown is whether or not hypoactive or hyperactive delirium is associated with a difference in recovery, in long-term function, and mental health outcomes. There is some literature that suggests that hypoactive delirium might be considered acute apathy syndrome because of the presentation that is associated with the depressed affect, with the immobility and the sort of decreased activity of patients overall. And so, in thinking about the effects on long-term outcomes, we hypothesized that hypoactive delirium, as opposed to hyperactive delirium, would be associated with worse recovery in functional outcomes, including basic activities of daily living and the instrumental activities of daily living, those activities necessary for independent living after a critical illness, like managing medications, keeping the home in order, and coordinating transport to medical appointments. We also hypothesized that it would be associated with worse depression and post-traumatic stress disorder in adult survivors of critical illness at 3 and 12 months after their ICU discharge. In order to test this hypothesis, we performed a secondary analysis of the bringing to light the risk factors and incidence of neuropsychological dysfunction in ICU survivors, or brain ICU study, and the delirium and dementia in veterans surviving ICU care or mind ICU data sets. We looked at patients who did not withdraw their data and completed a 3 or 12-month follow-up. Patients who were enrolled in the original parent studies were adults admitted to either a medical or surgical ICU with respiratory failure or either cardiogenic or septic shock, and this included a cohort of about 1,040 patients. Patients were excluded if they'd already had a substantial recent ICU exposure totaling more than five days, if they had an inability to be assessed for delirium, including blindness, deafness, or an inability to speak or comprehend English, inability to complete the long-term follow-up, unlikely to survive the coming 24 hours, or at a high risk for pre-existing cognitive deficits because of pre-existing neurodegenerative disease, cardiac surgery, anoxic brain injury, or severe dementia. While patients were enrolled in the study, trained research nurses would visit and perform daily delirium evaluations twice per day using the confusion assessment method for the ICU or CAM ICU, as well as a measure of their awareness with the Richmond agitation sedation or RAS scale. We then classified patients as hypoactive delirium if they had a positive CAM ICU screen and had a corresponding RAS score of negative three to zero. If the CAM ICU was positive and the corresponding RAS score ranged from plus one to plus four, then patients were classified as having hyperactive delirium, and patients with a RAS score of negative four to negative five were deemed as having a coma and unable to participate in the CAM ICU screening. Our outcomes of interest were collected by trained neuropsychology professionals who are completely blinded to the baseline data and a hospital course, and were conducted at three and 12 months after discharge from the Hospitalization for Critical Illness. The four outcomes that we looked at for this study were the activities of daily living measured by the CAHPS Index of Independence and Activities of Daily Living, the instrumental activities of daily living, those higher-level functioning activities that were measured by the SAFIRS Functional Activities Questionnaire, and then we tested for depression with the Beck Depression Inventory 2 and PTSD with the PTSD Checklist Specific Trauma Version. We then performed proportional odds and linear modeling in order to study the association between the total number of days of hypoactive delirium and the total number of days with hyperactive delirium during the hospitalization, and looked at the association with ADLs, IDLs, the BDI-2 score, and PCLS scores, both at three and 12 months. We used separate models for the three and 12-month outcomes. We also examined the potential interaction between hypoactive and hyperactive delirium days, but this was removed from the model because it was insignificant, and then adjusted for a handful of covariates. Of note for this particular study, participants could be assigned as having a day with hyperactive delirium and a day with hypoactive delirium, both in the same day, in order to increase the number of exposures, and we elected not to use the mixed subtype of delirium that is commonly used in some other studies. The covariates that we included in our model were baseline factors of age, gender, education, a Charleston comorbidity index, the IQ code SF, which looks at baseline cognitive function, disability and enrollment, the location of the hospital, and a history of depression. We again looked at the interaction term between hypoactive and hyperactive delirium days, and that was later removed from the model. We also controlled for factors associated with the ICU and hospital stay, including days of severe sepsis, days of coma, days of mechanical ventilation, the mean modified daily SOFA score, and a mean 24-hour daily doses of benzodiazepines, opioids, haloperidol, and a SOFA fall index metatomidine. So, in our cohort, there were 556 patients who survived their hospitalization and completed either of the three or 12-month outcomes. They were around 62 years old, a median of 62 years old, 60% were male, 90% were Caucasian, had completed about a high school education, had a Charleston score of 2 and an IQ code of 3. They experienced a median of 2 days of severe sepsis, 1 day of coma, 2.2 days of mechanical ventilation, were in the ICU for around 5 days and in the hospital for a total of 10 days, and received a fair amount of opioid, but very little benzodiazepine, haloperidol, and a SOFA fall or dexmetatomidine. At baseline in our cohort, 72% of patients were completely independent in their activities of daily living, and 69% were independent in their instrumental activities of daily living. 34% of our patients did report having a history of depression, and this was either by patient or proxy report, and only 4% had a history of PTSD. Of the patients in this cohort, 71% developed delirium. Of those, 69% had days with hypoactive delirium, and 17% had days with hyperactive delirium. The hypoactive delirium lasted a median of 3 days compared to the hyperactive, which was 1 day. 9% of the cohort did have both in the same day, which lasted for a median of 1 day. Looking at the 3-month outcomes overall, 37% of the cohort at 3 months reported some dependence in their basic activities of daily living, and 63% reported some dependence in their instrumental activities of daily living, meaning that they needed some assistance with at least one of the tasks included in the assessment. There were around 36% of patients reporting somewhere from mild to severe symptoms of depression, and 5% of the cohort had probable PTSD. When repeated at 12 months, only about 6% had recovered, and so 31% still had some dependence in their basic activities of daily living, and 56% still had some dependence in their instrumental activities of daily living. The numbers for depression and PTSD were about the same at around 35% and 5% respectively. When we performed our regression analyses, we found that there was no significant association between days with hypoactive delirium and activities of daily living at 3 months. We did find an association between hypoactive delirium and the instrumental activities of daily living at 3 months. Each additional day of hypoactive delirium was associated with a 0.24 point increase in the instrumental activities of daily living FAQ scale. While this was statistically significant, it would require eight additional days of hypoactive delirium to develop one new partial dependence in the instrumental activities of daily living, so this may not be a clinically significant finding. We did not find any association between delirium subtype and mental health outcomes, and we did not find any association with any of the functional activities of daily living or the instrumental activities of daily living at 12 months. Looking at the mental health outcomes, there was no difference between either subtype or no association with either subtype of delirium and any of the mental health outcomes at either 3 or 12 months. Some of the strengths of this particular study were that we chose to classify the motoric subtypes of delirium per assessment and performed two assessments each day, so while we elected not to use the mixed subtype of delirium, this allowed for an increase in the granular data and a better assessment of hyperactive delirium, which in multiple studies of motoric subtypes of delirium is far less common and thought to be far harder to catch on assessment. There may be multiple reasons for this, one being that hyperactive delirium is more likely to be treated, and patients may no longer display the symptoms of hyperactive delirium at the time of their assessment, and it may just be much less common overall than hypoactive delirium. Additionally, this was a large cohort with diverse diagnoses across major medical centers and Veterans Affairs hospitals, and we had a large sample size that did allow for multiple covariates to be included in our final analysis. Some of the limitations of this study, as I have already alluded to, are that delirium assessments were completed twice daily by trained research personnel, but this does not give us the real picture of the 24-hour course of delirium throughout the day that the bedside nurses and providers are seeing, and so future studies could be enhanced by really trying to capture all of those episodes, but we did, again, look at individual episodes of hypoactive and hyperactive delirium to try to increase the granularity of the data. There was also decreased power to detect association between hyperactive delirium and any of the outcomes because of its low prevalence, which is a common limitation of many of these studies. We were unable to perform functional status assessments prior to the hospitalization, but we did provide questionnaires to patients who were able to interact and also collected data by proxy in patients who were unable to interact to try to have a better understanding of patient's function at baseline. There is a potential survivor bias given our interest in long-term outcomes and evaluation of only survivors, and this was an observational study, so we cannot determine causation. So, in conclusion, we did find that longer durations of hypoactive delirium during critical illness are associated with statistically worse disability in the instrumental activities of daily living at three months after discharge, so this finding may not be clinically significant. But hyperactive delirium does not appear to be associated with any difference in functional disability or mental health outcomes in this cohort. We think that these studies are important and that this conversation continues to be important because while hyperactive delirium is quite easily recognizable, hypoactive delirium, because of its features of sedation and lethargy, may be very difficult to differentiate from a patient who is just somnolent and often goes unrecognized and leaves delirium undiagnosed. But it is associated with worse outcomes, particularly in the short term, and so increasing the vigilance in making sure that we're screening daily for delirium and then thinking about some of the long-term outcomes of these conditions is important. Looking at future studies of delirium, we should consider incorporating some types of delirium into the analysis and also, as a community, consider finding ways to improve capturing the episodes of hyperactive delirium as it is still often hard to catch in research studies and we could gain some much better data if we could capture all of those incidents. So I really appreciate your time and I'm happy to answer any questions, but as a kind of – I would first like to acknowledge my mentors, Christopher Hughes and Pratik Sanjara-Pande, as well as the B.H. Robin Scholars Program, for giving me this opportunity, the great work of the Critical Illness, Brain Dysfunction, and Survivorship Center, which conducts these large studies and all of the folks involved that make this work possible. And I'd like to open up a poll question just to get sort of a survey of the group. How often do you perform daily delirium screening? Your options are never less than 25% of the time, 25 to 50% of the time, 50 to 75% of the time, or 75 to 100% of the time. And we can discuss those results more in the question and answer session. I look forward to talking more with you at that time and thank you for listening. I will now – and I will now hand the presentation over to my colleague, Dr. Falcone. Thank you, and thank you to the Society of Critical Care Medicine for the opportunity to share our work. Also, thank you to Dr. Zagmani for such a wonderful presentation and introduction. I'm going to be discussing our recent paper, Admission of Hemoglobin Levels are Associated with Functional Outcomes in Spontaneous Variegated Hemorrhage. I wanted to, first of all, recognize the work of Julian Acosta, who is a postdoc in my lab who has led this work, and also to our 14 co-authors. The study is the result of pulling data from three large studies, as I will describe in a minute. So, it really took a village. We are very grateful for the trust and the opportunity to work with so many colleagues. I have no conflicts of interest, and we're very, very thankful to our funding sources, including the American Heart Association, the NIH, and the Neurocritical Care Society. So in terms of background, spontaneous intracerebral hemorrhage is a type of brain bleed that happens without any obvious reasons, including tumors, vascular malformations, or other kinds of anatomical problems. So spontaneous intracerebral hemorrhage, that I may refer to as ICH during the presentation, and I apologize in advance. We use that acronym so often that I may forget to not use it. Spontaneous intracerebral hemorrhage is the result of long-lasting small vessel disease in the brain, and we usually categorize intracerebral hemorrhage as lower when it compromises the superficial areas of the brain, and we, you know, in general tend to attribute that kind of ICH to cerebral amyloid angiopathy, and then the second type of ICH are brain bleeds that happen deep within the brain, compromising the deep brain nuclei, and we, in general, we think that this second type of intracerebral hemorrhage are associated with long-lasting hypertension or, more specifically, to long-lasting cerebral small vessel disease that is caused by hypertension and many other vascular risk factors. One important feature of intracerebral hemorrhage is that it has a very poor prognosis with roughly 40, 30 to 30 percent of patients achieving functional independence, and, you know, one of the driving factors of this poor outcome is that we don't have effective treatments, so this study was motivated by the goal of finding new targets and new pathophysiological pathways that we can target to improve the outcome of our patients, and, you know, with that goal in mind, it's always a good strategy to try to evaluate pathways for which, first, we already have extensive experience, and second, there are already interventions that we can use to modify that path. So along those lines, observational evidence suggests that low hemoglobin levels are associated with poor outcome in intracerebral hemorrhage. Importantly, this evidence is coming from small single-center analysis and small single-center studies, and, of course, we don't want to diminish the value of this prior evidence. To the contrary, these prior studies were the ones who—the ones that motivated this larger study. Importantly, low hemoglobin levels are known to be an important target for acute treatment interventions in other diseases, for example, acute myocardial infarction, and to a given extent, subarachnoid hemorrhage. So there is some support to the idea that optimizing hemoglobin levels will lead to improved oxygen delivery and may result in better outcomes. And again, in terms of possible mediators, as I was explaining before, impaired cerebral oxygen delivery could be one possible pathway, and there are some other prior studies indicating that coagulopathy could also be related to low hemoglobin levels or could be associated with low hemoglobin levels and could mediate the expansion of the hematoma internally to worse outcomes. So the aim of this study was to evaluate the relationship between admission hemoglobin levels and outcome in spontaneous intracerebral hemorrhage in a way that acknowledges the limitations of the prior studies, meaning the small sample size and the single-center nature. And we were also interested in evaluating whether the mediators of any possible association were well-known neuroimaging markers of poor outcome in ICH, and there are two in particular that we were interested in, the size of the hemorrhage, which is the most important predictor of outcome in this disease, and then hematoma expansion, which is another important predictor of poor outcome in this condition. For methods, our study design was to combine individual patient data from three large studies. We pulled individual-level data from the ATAC-2 clinical trial, from the FAST clinical trial, and from the ERIC observation study. The ATAC-2 clinical trial evaluated the role of aggressive blood pressure reduction in the acute setting for patients with spontaneous intracerebral hemorrhage. The FAST clinical trial evaluated the administration of recombinant factor VII in these patients. And the ERIC study, it's a very important and landmark study in ICH research, an observational study that enrolled, in a very systematic way, 3,000 ICH cases with an equal proportion of whites, blacks, and Hispanics. In this study, our exposure of interest was admission hemoglobin levels, defined as a continuous variable, and also anemia, meaning the modeling of admission hemoglobin levels in a dichotomous way, following the framework that we use in clinical medicine of calling some patients anemic using specific thresholds. In this case, we're going to use 12 grams per deciliter for females and 13 for males. The outcome of interest was the modifiant ranking scale three months after the ICH, and following what is usually done in this area of research, we dichotomized the modifiant ranking scale as 0 to 3. That represents a good outcome, and this 0 to 3 cutoff is basically representing the idea that these patients are independent, versus 40 to 6, which we labeled as poor outcome, representing the idea that either the patient is not independent or the patient died when evaluated three months after the ICH. And following our secondary goal of evaluating possible mediators, we also evaluated the volume of the hematoma on the baseline scan, so the scan that was used for the diagnosis, and we also evaluated the expansion of the ICH, defined as more than 6 mLs or 33% growth on the follow-up CT. For association testing, we performed numerous statistical analyses, but the idea, or the main idea, was to leverage access, full access, to individual patient data. That means that even though we are calling this a meta-analysis, we had full access to every single covariate in the datasets for each of these three studies. So we implemented two types of analysis. In the first approach, we combined all the individual patient data in one dataset, and we used mixed-effects regression. And on the secondary analysis, we ran study-specific analysis, meaning that we evaluated each study separately, and then we combined the results, we pooled the results, using inverse-variance fixed-effects meta-analysis, or random effects, when we found significant heterogeneity. Important additional analysis included evaluating what happened when we looked at the location of the bleed. As I was explaining in the introduction, the pathophysiology of these hemorrhages is very different for lobar and non-lobar. Lobar bleeds are related to cerebral amyloid angiopathy, and deep bleeds are mainly related to long-standing hypertension. And finally, and just to confirm that we ran a numerous statistical analysis, we're very interested in implementing those response analyses, looking at whether there is a ceiling effect, whether there is a threshold effect, and, for example, hemoglobin more than X is associated with better outcome, or whether improvements in clinical outcomes can be seen across a wide range of hemoglobin levels. This figure summarizes the patients and the data that were included in the study. From a total of 1,000 ICH patients included in the ATAC2 trial, 22 did not have hemoglobin data, and 73 did not have outcome data, so we included 905 patients. From the FAST trial, out of a total of 841 patients, 73 did not have hemoglobin data, and 23 did not have outcome data, so we included 746 patients. And from the ERIC study, the observational study that we used in this meta-analysis, there were 47 patients without hemoglobin data and 432 patients without outcome data, so we included 2,521 patients. So this slide, I think, emphasizes one of the most important strengths of our study, which is that we were able to evaluate these questions in a total of 4,172 ICH patients. That's sort of strength number one. Also, another important strength is that, as I was explaining before, we had full access to the full dataset of each of these studies, so even though we call this a meta-analysis, and indeed it was, we were able to run standard regression models, adjusting for any variable that we wanted to model to account for confounding. And then a third important comment to make related to this slide is that these were very different studies on one hand, because ATAC and FAST were clinical trials and ERIC was an observational study, but they all defined ICH in the same way. All three studies were interested in looking at spontaneous non-traumatic intracerebral hemorrhage, meaning these bleeds that are not related to trauma, they are not related to tumors, vascular malformations, or other anatomical problems. So it's a nice combination of studies that are slightly different, clinical trials, observational studies, but also they defined the qualifying event, the ICH, in a similar way. In addition, as you can see here, the exposure is very well defined. Hemoglobin levels is one of the routine measurements that we draw when patients come to the ED with an acute stroke. And you can see that there was minimal missing data for the exposure. There was some missing data for the outcome for the ERIC study. And this is where this individual participant meta-analysis carries a very important weight, because we can counterbalance the missing data in ERIC with the almost inexistent missing data for outcome in ATAC2 and FAST. And this makes sense, because missing data, especially for outcomes, it's something that is dreaded in clinical trials. This is the table one of the study describing the demographic characteristics of the overall cohort, and then the specific characteristics of each of the included studies. And again, the study evaluated a total of 4,172 patients. As you can see, the mean age was very similar across study. These were people in their mid-60s. There were more men than women, 60%, 40%. And this is something that we usually see in ICH. And then importantly, white participants only represented one-third of the population. So it's important. We usually say for many studies that results cannot be extrapolated. For this particular study, we have good reasons to think that these results could be applicable to other race and ethnic groups. The distribution of hypertension, as expected, is very high with values, prevalences of hypertension in the 75% to 80% range. And these people tend to have other vascular risk factors, including diabetes, hyperlipidemia, atrial fibrillation, and other risk factors. Importantly, we included roughly 23% of lower hemorrhages with, to a given extent, allow us to fully evaluate what happens in lower and non-lower locations. Although, as you can recognize by this 23% number, most of the bleeds were deep. And this has to do with the design of the clinical trial, or the clinical trials included in this meta-analysis, ATAC2 and FAST, that were mainly designed to include deep hemorrhages. Although some lower hemorrhages were included as well, the inclusion criteria were geared towards deep ICH. On this second table, we are describing both the exposure and the outcome. It's interesting to see that the hemoglobin levels were pretty similar across all studies. And it makes sense that they were a bit lower on the ERIC study, which was, again, an observational study. Clinical trials tend to capture healthier sub-cohorts of a given disease. So, you can see that the hemoglobin for ATAC was 14.26. It was 14 for FAST and 13.64 for ERIC, which, again, makes sense considering that two of these were clinical trials and ERIC was an observational study. The baseline, the mean or median baseline ICH volume was also different, and that also has to do with the inclusion criteria. It's interesting that the FAST study was conducted 10 years ago, and by then, it was perhaps not immediately clear that we had to select the patients in terms of the ICH volume. So, the FAST study basically included any size of bleed. So, there were some catastrophic bleeds included at trial, and that's sort of represented here because the mean admission ICH volume for FAST, 22.88, it's pretty similar to the 21.39 that we see in ERIC, which, again, is an observational study that just wanted to enroll ICH patients without very strict selection criteria as we would normally apply in clinical trials. ATAC-2, which is a more recent clinical trial, already captured this idea that we would want to test interventions in a healthier population or in a population that has better prognosis, so the mean baseline ICH volume for ETAC2 was 13.7, substantially smaller, reflecting the inclusion criteria that was geared towards small and intermediate-sized hemorrhages. The prevalence of hematoma expansion is roughly as expected. There have been descriptions between 20 and 40 percent, so the overall 24 percent prevalence of hematoma expansion is expected given what we know about the disease. And then the poor outcome, again, reflects the inclusion criteria for each study. For FAST, which had no limits to the size of the bleed, and ERIC, which was an observation study, there was a poor outcome in 50 percent of the population. And then in ETAC2, which restricted the study to small and intermediate-sized bleed, the intermediate-sized bleed's poor outcome was observed in 38 percent, so again, substantially better outcome as for the other two studies. And then that also is reflecting the death proportion of deaths. You can see it was 20 and 22 percent in FAST and ERIC, and only 7.3 percent in ETAC2. So here we summarize the main analysis of this paper. And again, the main question was whether the admission hemoglobin levels was associated with outcome three months after the bleed. And we are presenting the results for all ICH patients, and I'm hoping that you can see my arrow moving. Here in the first column, we have all ICH patients. The second column is lower ICH patients, and the third one is non-lower ICHs. And then each row represents each of the studies, ETAC2, ERIC, and FAST. And then in the bottom half of the table, we have the one-stage IPD meta-analysis. IPD stands for individual participant data, the two-stage IPD meta-analysis, and then the two-stage IPD meta-analysis utilizing random effects. And what is, I think, most important to convey here is that the point estimates were very consistent when evaluating each study in particular, and then when looking at different types of meta-analysis. So we can roughly say, if we take, for example, the first line of the meta-analysis results, that each additional gram per deciliter of hemoglobin on admission was associated with a 7 percent decrease in the risk of poor outcome. So higher hemoglobin levels were associated with better outcomes, and that was particularly true if you go to the third column for non-lower intracerebral hemorrhage for deep ICHs, and we couldn't find a positive association for lower or superficial hemorrhages. So again, the main result is that as hemoglobin goes up, the chances of having a good outcome goes up. And what this odd ratio represents here is that as hemoglobin goes up, the odds of having a poor outcome goes down. As I was explaining at the beginning, one important additional goal of the study was to evaluate the dose-response curve, trying to see if there was a ceiling effect, if there was a cutoff at which, you know, this association became non-important. And basically, what we found is that this association remained consistent across a wide range of levels. So here on this figure, on the x-axis, you can see different hemoglobin levels, and then on the y-axis, the proportion of patients that achieved good outcome and poor outcome. And as you can see, the light gray bar that represents good outcome becomes wider and wider as you move up in the hemoglobin range. And when tested from a statistical perspective with a test for trend, these results were, you know, very significant with a p-value, with a very small p-value. And then the final question of our study was to see if, you know, the mediators of this association were at the known neuroimaging risk factor for poor outcomes, large hematoma size and and hematoma expansion. And again, here you can see on the first column, ICH volume, on the second column, hematoma expansion. On the first line or on the first portion of the table, each line represents one study, and then the second portion of the table, you have different meta-analyses, one-stage IPD, two-stage IPD using fixed effects, and two-stage IPD using random effects. And although we did find an association between hemoglobin levels and ICH volume in Eric, these results were not sort of equally significant in ATAC2 and FAST. And this was reflected on the meta-analysis. If we go to the one-stage IPD meta-analysis, there was no significant association between hemoglobin levels and either ICH volume or hematoma expansion. We did find a positive result in the two-stage IPD meta-analysis for ICH volume, but of course that was driven by the results that we saw in Eric that was a study that contributed 60% of the sample size. So in terms of discussion and important points to make, the main conclusion of our study was that higher hemoglobin levels are associated with lower risk of poor outcome in ICH, and these results held across three different studies that included two clinical trials and one large observation study. We did not find a consistent association between hemoglobin levels and either hematoma volume or hematoma expansion. And again, these are two well-known neuroimaging risk factors for poor outcome. And as I was explaining before, the dose response analysis suggests, I wouldn't even want to use the word indicate, the dose response analysis suggests that this association between hemoglobin levels or higher hemoglobin levels and better outcome could be important across a wide range of values and not only the extreme values that we usually use to transduce, for example, a hemoglobin cell. There are important limitations to the study. The main one is, of course, that this is an observational analysis, an exploratory analysis that used existing data to provide some support to the observation that perhaps hemoglobin optimization can lead to better outcomes. There is an over-representation of non-lower or deep intracerebral hemorrhage, and only 25% of the plates were lower. There were different distributions of outcomes, especially when comparing ATAC2 to ERIC and FAST, and the reason was that the inclusion criteria for ATAC2 was geared towards small and intermediate-sized bleeds. And we don't have data on transfusion, so unfortunately, we cannot test the opposite question of, okay, what happens if you optimize hemoglobin levels if you intervene on this pathway? So, in conclusion, you know, we think that in combination with prior reports, our results point to hemoglobin levels as a potential target to explore interventions in ICH, and that's sort of if we can find more evidence that there is a causal relationship between hemoglobin levels and outcome in this condition. And then, irrespective of whether there is a causal relationship between hemoglobin and outcome, low hemoglobin can or may be able to be used as a biomarker of a poor outcome in ICH. So, aside from the discussion of the causal relationship, we may think of using hemoglobin levels, for example, for selecting patients for clinical trials if we are interested in selecting a healthier subgroup of the population. In terms of future directions, it will be very important to confirm this observation, these observational results in other cohorts, and, you know, it will be very important to perform a similar study looking at transfusions. And again, we're not advocating for a clinical trial, but perhaps as a first step to consolidate data on transfusions and outcomes in these types of patients as a first step towards an intervention. Another important direction is to try to extend these findings to other populations. Our population had only 30 percent whites, so there were other race ethnicities included in these studies, but because 40 percent of the patients came from clinical trials, we know that younger people tend to be overrepresented in these studies, so it's important to look for effect modification by age. And with that, I would like to recognize very specially Julian Acosta, who has led this work, all our co-authors, our colleagues in the Divisions of Stroke and Neurocritical Care at Yale, and our funding sources. Thank you all, and I'd be happy to take any questions. I'm going to turn the microphone to the moderator. Thank you very much, both, for the fabulous presentations. There are a number of questions coming, and the first one that I read is that, to Dr. Rangel, that do we have different delirium for different diseases? For instance, the hyperactive delirium, can it be more frequent after certain diagnosis? And I would like to allude to the recent paper on COVID-19 and delirium. Yes, that's a great question. Generally, when looking at the different subtypes of delirium, they're studied by sort of baseline risk factors. There have been some studies that have associated higher Apache scores at admission with a higher risk of developing hyperactive delirium. But then, overall, hypoactive delirium is more common in patients who are older and have a higher level of baseline frailty. And so, there's actually more of a thought of, are there sort of subsets of patients who at baseline are more predisposed to developing a subtype based on their underlying conditions, and that those who are more frail and less able to sort of mount a response tend towards hypoactive, and those who are perhaps more robust are able to mount this hyperactive response. But it's not clear at this time. Thank you very much. Another question to Dr. Falcone about targeting specific hemoglobin levels. Have you found a correlation that would show better outcomes with a specific hemoglobin level? And as you have alluded to previously, what would you propose to as a target if you would want to do an RCT in this field? Yes, thank you for the question. Our results at this point, again, show that the association remains important across a wide range. So, it's difficult to identify a single cutoff. In terms of how to use this information, of course, by no means we're advocating for using at this point hemoglobin levels other than the ones that we usually use. So, I think that a hemoglobin goal of more than seven still applies. But if we were to think about, you know, a trial, a clinical trial, I think that it would be reasonable to learn from the literature from cardiology and other fields that are targeting hemoglobin levels of eight and nine. And again, you know, the effect modification by gender plays an important role. So, it would be another important discussion to think whether we need to tailor the hemoglobin level goal to males and females. Thank you. Another question for Dr. Rangel is, have you seen any effect of delirium treatment? Have you seen any movement from one type of delirium to the other type? Yes, there is actually some thought that giving treatment, especially giving antipsychotics for hyperactive delirium may cause a transition to hypoactive delirium. And there's some thought that whatever the treatment may be, that's part of the reason why it's harder to catch hyperactive delirium in these studies as well. And the concern being that as hypoactive is associated with these were short-term outcomes. And then, should we reevaluate our treatment strategy? So, at this point, it is still for the safety of the patient recommended to treat hyperactive delirium as needed, but recognizing that it may not be transitioning patients out of delirium, simply converting them from hyperactive to hypoactive. Thank you very much. We are going back to Dr. Falcone, and we are going back to the hemoglobin levels. Is it possible that the lower hemoglobin levels in the patients actually just mark more serious illness? Because I haven't really seen that data in the slides, but I might have missed it. Yes, thank you for the question. And, you know, this same item came up during the review process, and we're fully aware that this is certainly a possibility. And this is why, you know, on the final slide, I was sort of indicating that, you know, these results could be interesting from a causal perspective, if there was one. And if this association is perhaps only representing the comorbid status of the patients, then the results would sort of support the idea that we can use hemoglobin to identify patients who may have poor prognosis due to pre-existing conditions. Having said that, we ran a number of secondary analyses to try to address this question, and one of the sensitivity analyses in the paper that, unfortunately, we didn't include in the slides, was geared towards including only people who had a great baseline status as identified by very low levels of the Modified Rankine Scale, and by excluding patients that had serious comorbidities like cancer. So, we tried to address this question from an analytical perspective, but we recognize that residual confounding is always a possibility in observational studies. So, at this point, our results still support the idea there could be a causal association, but we need to recognize that, you know, there could be bias related to the pre-existing conditions and the pre-existing or the pre-morbid status of these patients. Thank you very much, and that concludes our Q&A session. Finally, thank you to our presenters and the audience for attending. Again, everyone who joined us for today's webcast will receive a follow-up email that will include an evaluation. Please take that five minutes to complete this evaluation, because your feedback is greatly appreciated. Thank you so much. Thank you, and on a very final note, please join us for our next Journal Club Critical Care Medicine on Thursday, June 24th. And this concludes our presentation today. Thank you.

Journal Club Critical Care Medicine

motoric subtypes of delirium

long-term effects

functional health outcomes

mental health outcomes

hypoactive delirium

instrumental activities of daily living

admission hemoglobin levels

spontaneous intracerebral hemorrhage

Webcast

Select

Professional

Associate

2021

Neuroscience

Behavioral Health and Well Being

Delirium

Behavioral Health

Intermediate

Advanced

Behavioral Health and Burnout