So, welcome to the Pediatric Section Year in Review. I'm Anne-Marie Brown, and it is my pleasure to moderate this session. A few housekeeping items. So, for those of you that are tweeting out any part of the session, you're welcome to do that. Please remember to use the hashtag CCC49 and hashtag PedsICU on anything you tweet out, so all of us around the world can follow along. Remember to keep your phones on silent, please. And we are going to have all three of our presenters come up and give their presentations, and then if there are, excuse me, if there are any questions, we're going to hold them until the end. With that, I'm going to introduce our first speaker, who I'm proud to have as a colleague and friend, Dr. Ramey Hickel, received SCCM membership as a sign-on bonus to her first critical care NP position. And I was really excited to see her include that, and I just wanted to provide that as what a fantastic way to onboard a new colleague, regardless of discipline, into Peds Critical Care. She has been grateful for that early connection and tries to pay it forward as a clinician and an interprofessional educator in Peds Critical Care. She feels fortunate to continue to practice in both the PICU and the cardiac ICU, teach novice clinicians and mentor colleagues as an assistant professor at the School of Nursing and School of Medicine at Duke University and as a simulation educator for the Division of Pediatric Critical Care. Please join me in welcoming. started. I have no disclosures except for the fact that I could talk about quality and safety and education all day, and I don't have that much time, so we better blast off and get started, I guess. Florence Nightingale was one of the first quality and safety advocates. In her work, Notes on Hospitals, that she presented to the National Associations for Promotion of Social Sciences, she indicated that conditions within hospitals placed patients at higher risk of dying if they were in the hospital than if they were cared for at home. These days, hospitals report data and processes to national registries and consortiums in a collective and shared effort towards quality improvement. My colleagues are going to describe scientific papers based on the familiar background, hypothesis, methods, results, framework that we're all used to, but now I just wanted a reminder that we use a framework called Standards for Quality Improvement Reporting Excellence to talk about quality and safety. The background identifies the problem or variance in communication or systems or processes and the impact that that variance has on patients that our intervention is supposed to improve. We have dissemination when we recognize that these outcomes are generalizable to our populations and can make process changes in our populations. So let's look at a couple of these then, or as many as I can get in in the next 15 minutes or so. Evaluating the practice of repositioning endotracheal tubes in neonates in children is based on radiographic location. So if we're going to talk about variance in practice, I would suspect that this reflects a lot of variance in practice across, this is generalizable to everybody's ICU in here. If I were to ask all of you, I would probably get different feedback about how these processes are done in your ICUs. So let's see what these guys did to talk about this, and we want to identify opportunities to generalize their findings to our ICUs. So in 123 patients, which turned out to be 183 ET tube interventions, these patients ranged from one day to 21 years. They did a retrospective chart review about ET tube placement and orders written to change where the ET tube was positioned. And then they compared that to measurements of pre and post distance on the x-ray about how far the ET tube was moved. So they compared the absolute difference between actual measured ET tube distance changed and what the intended distance changed was asked for. So the intended distance change, for example, these are all in centimeters. If it was plus or minus two centimeters, the actual distance that was changed was 0.7 centimeters with an interquartile range of 0.36 to 1.1. It's a lot of centimeters, it's a lot of little distance, but what it really tells us is that there's a lot of variability in practice. They also had to adjust for head position and who was reading the x-rays. And so does this resonate to anyone about the impact that it could have on our patients about the processes that are involved in evaluating where the patient's ET tube is and sedating a patient or not? And do we have an opportunity in our ICUs to take this minimal change difference and apply it for quality improvement? Perhaps we do. Oops, sorry, I'm going the wrong way. What's happening? Who puts the left click as the forward? Hold on. What's that? Who did that? Okay. So that's in one ICU. So just as a reminder, I have flipped too many pages on my notes here. The NEAR for Kids is a national emergency airway registry to characterize and describe processes around intubations and make recommendations. It's multi-center data collection and this data has been used to make recommendations such as airway bundles and standardized processes. Or in this case, this data has been used to modify a bundle and evaluate apneic oxygenation as a quality improvement intervention in an academic ICU. Hold on, let's see if I can go forward. Nope. There we go. Okay. So what did they do here? They implemented apneic oxygenation for all primary tracheal intubations. They used their data from NEAR for Kids for 24 months prior to the implementation change and over 700 intubations in the 24 months after the implementation change. They used a NEAR for Kids toolkit to educate and adapt their airway bundle checklist with apneic oxygenation instructions and they determined that following adherence to implementation was 80% of the time. They continued to report data to NEAR for Kids and observed for moderate and severe sepsis. I'm sorry, hypoxia. These are the results and you can see that they had after intervention, after this intervention they showed less frequency of desaturation events. Wow, I am really far ahead in my notes. Okay. So that's one way that we can use, shoot, I'm sorry you guys. Relax. Talk. Collaborative QI. So that is a national database and collaborating and providing data to a national database. This is an example of collaborative QI done with five PICUs in the Children's Hospital Association talking about variability and transfusion practices within and among PICUs. When we talk about variability in practice that's also known as what we all talk about is that's the provider preference, right? And so this study or this paper describes evaluating a standard implementation plan would lead to fewer RBC transfusions outside of the evidence-based thresholds. It included interdisciplinary teams in five hospitals, a lot of description about the education, the reminders and the audits and feedback that they did. What were their outcomes? Well their outcomes showed a 7% decrease in transfusions outside the guidelines after implementations. But really what I took away from reading this was the challenges of trying to herd cats essentially and change process even with a standard approach because they wrote all about these challenges and they were very specific about the implementation plan. So this gave us an opportunity to learn from their challenges and successes. Speaking of that, so the Pediatric Cardiac Critical Care Consortium is a non-profit organization that collaboratives. So that was five ICUs that we just talked about. This is over 50 pediatric cardiac ICUs in a voluntary QI collaborative who report demographics, comorbidities, surgeries, procedures, systems, resources and other outcome data for all patients in pediatric cardiac ICUs. Let's see how they use this data to talk about extubation failure rates after pediatric cardiac surgery. Extubation failure was defined as a need for reintubation within 48 hours after the first planned extubation. Extubation failure rates and duration were found not to be inversely related to ICU factors, not associated with lower extubation fit, the lower extubation fit, sorry, less time on the ventilator. You can see here in this figure they had an observed to expected oxygen extubation failure ratio and rate and duration of mechanical ventilation. All the hospitals in the consortium are ranked here from lowest to highest in extubation failure. And then this compares hospital performance across those two quality metrics, extubation failure rate and adjusted postoperative mechanical ventilation. This is an example. So there were two outliers. There were two hospitals that performed better than expected and three hospitals that performed worse than expected of the rest of the group. And this is an opportunity for collaborative QI because they can disseminate practices from high performing hospitals for collaborative learning and improved outcomes. That is for, that data reflects postoperative patients. What about the patients in cardiac ICUs who are medical conditions, have medical conditions and not operative conditions? In fact, 37% of admissions to the PC4 hospitals are medical admissions. So this paper describes a PC4 CICU medical mortality model which is a case mix adjusted model to predict mortality in this more heterogeneous population of medical patients. Also describes calculating a standardized mortality ratio and again looking at hospital performance to examine differences in CICU outcomes for both surgical and medical patients. And the transparent reporting of this consortium allows for collaborative learning and quality improvement. It seems like every year when we talk about quality improvement, we talk about simulation. And we use simulation to improve processes around low frequency and high acuity events. These are two examples of similar simulations around these low frequency, high acuity events and both talk about providers' perceptions of how well we do in these events, specifically to chest compression rate and depth. Despite audio visual feedback and feedback from a coach in this study, those that are doing, running a code or doing the compressions have very little awareness about the accuracy of their compression rate and depth. Our outcomes from these studies demonstrate to us improved opportunity for teamwork, communication, and situational awareness. Simulation again for codes. This is a very small study, but wanting to try to improve communication, this group came up with these phrases that you see there. And could we teach a team, a just-in-time teaching of a team, to use these phrases in the midst of a code rather than closed-loop communication? Because they were finding that closed-loop communication to have the readback of a series of instructions sometimes took quite a bit of time and would increase time pauses in chest compressions. This was a feasibility study to show that it was a, and in this study they showed that it was, you could educate a team this way and it may also reduce chest compression pauses if you use standard language. So this is both a simulation study, a QI study, and describes education, which I was also tasked with trying to talk about in this short period of time. This one I liked a lot as well. So this talks about interoperative, a near miss in interoperative magnetic resonance imaging. So why am I talking about that impedes critical care? It's hard enough to take care of our patients in our units, let alone when we let them off the units and things can happen. So this was a use of simulation to evaluate a safety event that happened in MRI. A pediatric patient with refractory epilepsy was transported for an interoperative MRI after his head was stabilized in the OR, and there was significant artifact on his MRI. The MRI was stopped immediately, but he was found to have ferrosulfate pins in rather than the MRI safe pins. And this group hired a simulation specialist, did this high-fidelity simulation with a structured debrief to try to evaluate threats and where the breakdown occurred, and in multiple places, as is often the case. People, processes, equipment, or technology, and or technology. And the point here is not for you to be able to read this small print, but just to realize that simulation can be done for more than just teaching in mock codes, and to improve those processes, but also to identify threats for education, teamwork, communication, situation awareness, and process improvement. Again, not just mock codes, rapid cycle deliberate practice versus reflective debriefing for pediatric septic shock training. These are just two different types of simulations. This describes two groups, whether they went through standard simulation where you carry out a simulation case. In this case, these were residents in a septic shock case, and then they debriefed and got the learning points at the end. As opposed to the other group that did rapid cycle deliberate practice, where that's you start the case, you start to manage the patient, and you stop if you don't get it right. You get some intervention and some education along the way. You rewind and do it again until you get it right. And so both were effective, but maybe we're going to need to revise this, and we'll hear more about this study in an interprofessional education way, and with the new surviving sepsis simulation written into the case next year. So that brings me to finally about education. These two papers talk about simulation use and maybe the opportunities for simulation in education. So this paper from Pediatric Emergency Care was a survey done of all residency program directors to try to categorize how much simulation was used in the programs, what type of simulation, and the outcome is that it is variable. Many programs use mock codes, and there's a lot of variability in how these simulations are evaluated. The only thing that was constant was that residents wanted more of this and felt like this was a useful way to learn. Evaluation of Pediatric Cardiac ICU Advanced Practice Provider Education and Practice Variation is another survey. It describes a descriptive survey of 35% of institutions that use APPs in their cardiac ICUs, and no surprise, it also describes that practice and education before they enter into the intensive care unit is variable, with variable structure. These two, both of these two papers, I guess, support the fact that learners, novice clinicians in our ICUs come to us with variable experience, and maybe simulation is an opportunity to evaluate and mitigate risks and threats, educate teams, and influence quality and safety throughout. I hope that some of you got those same things that I thought I got from these papers, and we'll pass the baton. Thanks, Ramey. Next, I'd like to introduce Dr. Nilesh Mehta. Nilesh is a professor and director of Critical Care Nutrition, Senior Associate Medical Director of the Medical Surgical ICU, and Director of Quality and Outcomes, along with being a professor of anesthesia at Harvard Medical School. He's a leader in pediatric ICU nutrition research and is also past president of the American Society of Parental and Enteral Nutrition. It is certainly my pleasure to collaborate with him on a lot of our nutrition research work, and I welcome him to the stage for his talk on clinical research. That was terrific, Ramey. So exciting to see a lot of QI work. I now move to the next session, which is Pediatric Year in Review Clinical Research. It is such a privilege to present this. It's been an exciting year. I have no conflicts to disclose. In 2019, we saw a lot of interesting studies, many of which have and will continue to change practice in the bedside in the pediatric intensive care unit. I found four main groups. There were some impressive randomized control trials, which we will dive into. There were some pilot RCTs, which have become an instructive and a necessary first step, and we'll allude to at least two of those. There is a registry trial, and then a guideline, and some lessons from systematic review of prior trials that I will simply mention in brief. So let's move on with the RCTs that addressed some important questions in the field. The first question was, what is the clinical consequence of RBC storage? As facilities and technologies have allowed more and more red blood cells to be stored for many days, many blood transfusion units now are able to provide older blood, compatible blood, for use in the pediatric intensive care unit. The hypothesis that was tested in this next study that we'll talk about is, can older blood influence organ dysfunction in critically ill children? The concept was to compare fresh blood transfusion, which in this case is defined as less than one week of storage, or a standard issue red blood cell. And this is an elegant study. Philip Spinella and his group together, the Age of Blood in Children, ABCPICU group, conducted what is a trademark future study design for any question that we want to answer in pediatric intensive care unit. And that includes a multicenter, a collaboration between multiple trial groups. For example, here, it was not just the ABCPICU, but the BloodNet, the PELISI, and the Canadian Critical Trials Group. And together, an impressive 50 pediatric intensive care units from four large regions of the world contributed to over 1,500 patients that were randomized in the intensive care unit to receive one of two strategies. The first one being fresh red blood cells, and the second one being standard issue. The standard issue essentially meant that you could get the oldest compatible available blood for transfusion. The primary outcome was a composite of either a new multiple organ dysfunction or progressive organ dysfunction that also included death. As you can see on the figure to the right, there is very little attrition in this study. Of the 1,538 patients randomized, close to 700 in each arm were included almost towards the end of the protocol. And close to 700 included in both intention to treat and per protocol analysis. What was interesting is the treatment separation. If you look to the table on the left, the trial design achieved the separation between fresh and old, as in the duration of storage for the fresh red blood cell group was five days as a median, and the median duration of storage on the other group, the standard issue, was 18 days. The time to development of new or progressive mods, which was the primary outcome, was not significantly different between the two groups. The per protocol analysis showed no significant difference. This was a huge study, and the results certainly describe that fresh blood versus standard issue, no difference, and no demonstrated benefit of seeking out fresh blood for transfusion in the ICU. I hope you get a chance to read this study, both in terms of how it was conducted and what it takes to produce this kind of a large intervention and a result that is practically relevant to us on a day-to-day basis. The second question, along the same theme of blood product transfusion, alludes to pre-term newborns. Pre-term newborns, less than 34 weeks of gestation, are both at risk of head bleeds and at the same time, a high incidence of thrombocytopenia. So the specific question of how much is the platelet threshold so that prophylactic platelet transfusion may be given with the hopes of preventing fatal or large intracranial bleeds. This question was first studied or last studied 25 years ago. It was a study between 50,000 and 150,000 threshold, and it was an equivalent study. And as a result, newborn practitioners, neonatologists, have chosen the 50K. And over the years, they have been moving towards a more restrictive threshold, pushing towards 25,000 transfusion threshold for platelets. So in this study, a multicenter study that enrolled a large number of newborns born before 34 weeks of gestation, a median of 26, they were randomized to either having platelet transfusions prophylactically if the platelet threshold was 50,000 or 25,000. They excluded patients who had any major bleed within the previous 72 hours, either on imaging, which included ultrasound. If you look at the figures on the left, there is a decent separation on the top figure between median platelet counts in both the low and, between the low and the high threshold groups, so achieved what it needed to. Of the patients who were randomized to 50K, appropriately, 90% of them got at least one platelet transfusion. Of those in the lower threshold, as you could imagine, 53% got a transfusion. And the trial day versus the number of platelet transfusions is depicted in the figure below. The primary outcome here was high-risk major bleeding or death within 28 days of randomization. And if you look at the figure to the left, which shows the Kaplan-Meier for patients who survived without any major bleeding, it was interesting that the high platelet transfusion threshold group, the 50K, actually had a significantly higher odds ratio for both death or major bleeding. This was interesting. There are a lot of mechanisms that are being postulated. This study did not address them. But I'm quite sure that for neonatologists all over the country, or those of us who may take care of pre-dumb newborns, this must move us towards more restrictive thresholds for prophylactic transfusion of platelets in these vulnerable infants and newborns. The third question was addressed by some of our follow-up data coming out of larger trials. And this one is particularly the HALFPIND trial. The question here is the significance or the clinical consequence of hypoglycemia. As we've learned over the years, both with the tight glycemic control trials and more recently with some of the nutrition trials, that hypoglycemia is significant even in the pediatric ICU. And when insulin is used, it may be as high as up to 25% to 30%. So the HALFPIND trial very elegantly employed continuous glucose monitoring and a very tight control loop for monitoring for hypoglycemia and treating it. And therefore, in that era of this modern CGM-employed management of glycemic control, it seemed to be a perfect ploy to look at both the incidence of hypoglycemia and then follow up these patients for both short-term and long-term outcomes of hypoglycemia. The first study that I'll talk about is the 2019 description by the HALFPIND group on the short-term adverse outcomes of hypoglycemia. What they found in this trial is, which was, as you all recall, stopped prematurely after 700-plus patients that were enrolled for lack of benefit of tight glycemic control. And with the CGM, the continuous glucose monitoring being used, severe hypoglycemia, less than 40 milligrams per deciliter, was not common. And this was very interesting because of other trials without CGM at a higher risk. But there were a significant number of patients who had hypoglycemia defined as less than 60 milligrams per deciliter. So in this group of patients, they divided patients who had one exposure to hypoglycemia, and these were the cases, or controls that were matched and never had any hypoglycemia. What they found is, in terms of short-term outcomes, here in the table, you see hypoglycemia patients, almost 110 out of the 700-plus patients, of which 500-plus were included in this nested case control study. So 110 hypoglycemia exposures and 427 patients controls who did not have hypoglycemia. And there were significant differences with worse outcomes in terms of ICU-free days and hospital-free days in patients who had any exposure to hypoglycemia. And these were the short-term outcomes that they alluded to. In a subsequent study, which is not yet out in print, but is available as an e-publication by the same group, Catherine Bierguss was the first author, but again, the same half-blind group looked at one-year outcomes, including neurobehavioral scores. And we won't, for the purpose of 2019, I won't go into details, but it's exciting that at one-year outcomes, they used pediatric quality-of-life physical and psychosocial subscale scores. And both of them were worse in patients who were exposed to any hypoglycemia. So hypoglycemia is preventable, severe at least with a very tight glycemic control with continuous glucose monitoring, but hypoglycemia incidence remains high in these patients and is associated with both short and long-term outcomes. What's exciting is the shift in 2019, as we've seen for the past few years, of studies focusing on long-term outcomes. And it is this theme that I'll continue with the next few studies. So let's talk about the PAPANIC study. The PAPANIC trial, as you all recall, and we'll allude to just a few basic principles of the trial in a bit, was early versus late p.m. And once again, the group challenges us with their focus and commitment to long-term outcomes. And an exceptionally enterprising group, very much like HalfPine, pushing towards long-term outcomes as we should. So here's the PAPANIC trial. Just to give a quick overview, the PAPANIC trial published in New England Journal 2016 was in patients who did not achieve 80% of their goal by enteral nutrition, they were given supplemental parenteral nutrition and they were randomized into early, and if you, excuse me, I call it very early, which was P.N. started within 24 hours of admission, or P.N. delayed until day eight in the pediatric ICU. On the right, I just wanted to jog your memory, here is the group of patients. The enteral nutrition provided to both groups. And what's fascinating about this study is that enteral nutrition was achieved to such impressive numbers by day seven, almost up to 60 kcals per kg per day, which in many studies recently would be very close to the resting energy expenditure. And on top of that, the groups either get the blue bars, the supplemental P.N. for started in early group from day one to day seven, and the other group gets it only after day seven, which is on day eight onwards. I highlight two things. One is, look at the total nutrition delivery, almost up to 90 kcals per kg, which most nutrition folks today will tell you is very close or very likely overfeeding. The other thing is, unfortunately, the reality of PICU studies is half the patients left by day three, and by day eight, when they were randomized to the late part, just a fourth or less patients were left in the ICU. But an impressive study. What they did, in addition to the initial trial, is their commitment to the two-year follow-up. And I've been to this institution, and anybody, the half-pine, the papanic, who's following patients for this degree of outcome, it takes a huge effort and commitment. The detailed neurologic examination in the follow-up clinics in Netherlands, I've seen firsthand a huge battery of neurocognitive and motor testing, and a structured interview with the parents. As you can see, there is attrition in a study at two years. 391, 395 out of the 700 and 700 randomized to each arm were included in the neurocognitive testing at the end. What's interesting is the choice of controls was healthy controls, which is arguable. Rather than looking at the impact of nutritional strategies, this healthy control doesn't allow you to tease out the impact of critical illness itself. So it's a little debatable what would have been better controls. But what was clear is on the right, this figure shows both the early versus late parental group with the inhibitory control as reported by parents about the children, and it was significantly worse after adjusting for multiple comparisons in the late PN group. So there's no doubt that the late PN, sorry, significantly better in the late PN group. So there's no doubt that the late PN strategy, when compared to a very early or aggressive PN strategy, started within 24 hours, is better. What I want to show you is in addition to learning that the long-term outcomes continue to show the same benefits, the balance was that the late PN group, especially in the newborns and young infants, they reported this last year in Lancet Child and Adolescent Health, had high incidence of severe hypoglycemia. So it's an important balance in malnourished, especially newborns, where the late PN may have benefit on some accounts, but you need to watch the severe hypoglycemia side effect. And lastly, the long-term outcomes, once again, from the RESTORE study. As we know, a huge undertaking. The RESTORE study patients were randomized, as we know, to a structured nurse-led, goal-directed sedation titration algorithm versus what was commonplace, the standard of care for sedation. The concept of post-intensive care syndrome, Scott Watson, Martha Curley, so many others have conceptualized this phenomenon in pediatrics, as we've seen in adults, and it is this combination of physical, cognitive, emotional, and social toll that critical illness takes, and it has both the developmental impact as well as impact on the family as such. So using that construct and some of the trajectory that are described in that initial paper in 2018, the RESTORE study provided opportunity to follow up their patients to look at what are outcomes of patients enrolled in the RESTORE who were essentially acute respiratory failure patients, and what happens to them in terms of their functional status and here, the PCPC and the POPC, which is the cerebral and the overall performance category, as outcomes. What's interesting is, once again, some attrition, as one would imagine. 2,400-plus patients in the RESTORE study, 1,330 were eligible survivors, and these survivors, a large number, were able to gather interview data and health-related quality of life data, and what you see here is the schematic that was shown in the previous paper. Essentially, this is the group that had improvement in these scores, the cerebral and the overall performance category. This is a group that had somewhat consistent scores between admission, baseline, discharge, and then at two years, and then this is the group that had decline, a variety of different ways in which their performance declines, and this constituted almost one in every five childs, so the comment made by this paper and the conclusion is that post-discharge morbidity is common. What I love about the field, the way we are moving towards morbidity, focusing not just on morbidity in terms of numbers, but detailed neurobehavioral validated scores that come up as outcomes of interest. So in this study, PCP, CPO, PCP, Q, quality of life, and these are outcomes that are now becoming desirable for trials and of interest, both to us and families. One last study in the randomized control follow-up is the GAS trial. So most of us practice in pediatric ICUs where we take care of post-operative patients. If you're following the literature in the last five to 10 years, a huge amount of concern borne out by animal studies about the impact of short-term anesthesia on long-term development of infants, and this took a whole turn in social media and lay press and family concerns, and this was a question that we needed to answer. The GAS trial is an impressive effort, once again, like we've described in the previous trials of lots of centers coming together and addressing this question. So the GAS trial reported their five-year outcome after the initial conduct of the trial. So to remind you, this was an international multi-center equivalence trial. Large number of hospitals, as you can see, infants who were supposed to get hernior-raphy surgery, inguinal hernior-raphy, were randomized to receive either a regional anesthesia awake or general anesthesia. These patients had not been exposed to anesthesia before and did not have risk factors for neurologic injury. The randomization to intervention for awake and regional and sevoflurane-based general anesthesia, and of which 205 out of 363 provide the five-year outcome and 242 out of the 359 in the general anesthesia provided the five-year outcome. The main outcome was the full-scale intelligence quotient on the WPPSI-3, a validated tool, at five years. Of note, the intervention for the sevoflurane group was just close to one hour, so a majority of the patients were 50 minutes, 55 minutes. So we are talking about short-term anesthesia here. The five-year, there was no difference in WPPSI-3-FSIQ, the full-scale intelligence quotient between the two groups. The mean difference was 0.23, with not at all any meaningful confidence intervals. This was proven both on protocol as well as intention to treat analysis. It's a fascinating study. This study is probably beginning to close the debate and the concerns on a specific group of patients who are healthy at baseline, who get short-term anesthesia and come to the ICU. And I don't know how many of you have parents who have been aware of the literature, and it's a big relief for our anesthetists to begin to address this as they consent patients for surgery. But remember, it's short-term, and these are healthy patients, and at five years, no difference. At this stage, I'm going to show you a few papers in the pilot and then a registry and the guidelines just for you to go and read and think about what we learned. But the concept of pilot studies is here to stay. No one does it better than the PICS group, the Pediatric Intensive Care Study Group from the UK. The fever pilot began to explore not just the feasibility of conducting a large-scale temperature-restrictive versus permissive strategy, but they begin to start looking at what are the perspectives in the community, in the staff, and can we do this trial? So it's very impressive, and you'll see the objectives they're exploring. Are there patients that we can enroll? Will they complete the study? And will I achieve separation? And it's only after doing these initial pilot trials that they move on to the next stage of multicenter trial. And Marie Brown, sitting here, did an elegant nutritional study, not easily done, of continuous versus bolus feeds in ventilated children. A pilot study, small numbers, but described to us the challenges and how best to do these trials. It will require a lot of centers to come down to focus on the protocols and adhere to these protocols. What I learned from Anne Marie's study is that irrespective of what protocol you're using, a protocolized approach to enteral nutrition gets you faster to whatever your goal is. And I'm excited to see that nutrition studies will find a multicenter approach after some of this pilot work. I wanted to show you the registry trial, which is beginning to answer some of our questions about roller versus centrifugal. No way has this closed this chapter. It's just whetted our appetite. The ELSO registry provides useful data to begin to explore. And what we found is the centrifugal pump patients over a period of a decade, almost five, six years, had increased mortality. But this story's not over. Centrifugal pumps are coming in different flavors, and as you can see, the use of centrifugal pumps is actually overtaken and higher than roller pumps in infants less than 10 kg. Lastly, honorable mention for some very elegant guideline work, much harder than one thinks, provides a lot of bedside, stepwise guidance, accumulating a lot of literature. Pat Koenig, Robert Tasker, and the group that did the TBI guidelines. The cardiac arrest group, the resuscitation group, are so prolific. I found that some of the work over the years put together in this meta-analysis guided us in terms of looking at temperature management, targeted temperature management versus not. And in this particular meta-analysis published in Resuscitation this past year, there was inconclusive evidence for out-of-hospital cardiac arrest to do a targeted temperature management strategy following cardiac arrest in children. And similarly, the group also looked at is intubation versus supraglottic airway superior to bag and mouth ventilation? It turns out not. And I will leave it to Vinay and Bob Berg and the resuscitation gurus to translate this. And they have already into bedside messages that will help us as we run to a code and worry about intubating kids rather than focusing on chest compression. So to my mind, these were the messages I took. Long-term outcomes are here to stay. It's exciting to watch some of our colleagues in the field and the community getting there. The pilot feasibility trials, a necessary step before we embark into these large ones. And how it's fascinating to see single, well-conducted trials finally then coming together in a meta-analysis, giving us some confidence in describing change in practice. So I won't rehash all the messages in the interest of time. But we learned a lot in 2019. Kudos to all the investigators. Kudos to all that I have not shown your work here. There's a lot of ongoing trials. I'm excited to come back in 2020 to hear what else is developing. I thank you for your attention. Thank you, Nilesh. Finally, bringing us home at the end of the day is Dr. Jennifer Muszynski. She is a pediatric intensivist and clinician scientist at Nationwide Children's Hospital. Her translational research program focuses on the immunobiology of critical illness and blood product transfusion. Welcome, Jennifer. Great. Okay. So we're going to round out the afternoon with a little exciting basic and translational research. I promise. Well, I can't promise I won't put you to sleep. I can promise there's not a single Western blot here. So there you go. So what we're going to talk about this afternoon, a little bit about immune and inflammatory phenotypes in sepsis and sepsis-induced mods, a couple of papers on that, some mechanisms of clot formation related to flow on ECMO, as well as a little bit of inhalational anesthesia in the developing brain in the preclinical space. So the first study comes to us from Joe Carcillo and colleagues at the Capcorn Network. The reason I chose this study, most of us in the room are already familiar with the fact that morbidity and mortality in pediatric sepsis-induced mods remains high. And unfortunately, our current toolbox is still largely limited to antibiotics and supportive care. There are immunomodulatory treatments that exist and could be used, but in order to use them and to evaluate their efficacy, we really need to better understand the immunobiology of our patients, and we need to understand that not all patients have the same pathophysiology. So we need targeted treatments. So the PhenOM study was a prospective observational study conducted within the nine centers of the Capcorn Network. It enrolled children with severe sepsis, and here was defined as SIRS plus at least one organ failure using the organ failure index criteria. Longitudinal blood samples were evaluated twice weekly along with clinical characteristics in order to identify inflammation phenotypes defined thusly. So the first phenotype that was evaluated for was TAMOF, or thrombocytopenia-associated multiple organ failure. The clinical criteria used in this study for TAMOF was thrombocytopenia, of course, as well as renal failure and an OFI of three or more, so three or more organ failures, with a confirmatory biomarker of an ADAMTS13 of less than 57% of normal. A sequential liver failure-associated MOF was defined clinically as both hepatic failure and pulmonary failure on at least day three or later, with a confirmatory biomarker of soluble fast ligand. Immunoparalysis-associated multiple organ failure had a clinical criteria of at least two organ failures with a confirmatory biomarker of ex vivo LPS-induced TNF-alpha production capacity of less than 200 picograms per ml at any time on day three or later. It's taken me maybe 10 years to be able to get that to roll off my tongue. And then lastly, the macrophage activation syndrome, which could probably also be termed hyperferotinemic, hyperinflammatory mods, or some term yet to be determined, that was defined by clinical criteria of both hepatic failure and hematologic failure, as well as a ferritin greater than 500 on any study day. So in terms of results, there were 401 children enrolled, 289 of whom had multiple organ failure. Of those with multiple organ failure, 35% had at least one inflammation phenotype. Of those inflammation phenotypes, immunoparalysis was the most common and was present in 29% of those children with multiple organ failure. And what was striking, if you look at the Venn diagram over there on the left, is there was a lot of overlap among the inflammation phenotypes. And in fact, those with TAMOF, or sequential liver-associated multiple organ failure, the majority of whom also had immunoparalysis. And it's important to note here, when we say immunoparalysis, that is a severe immune suppression. So you have inflammatory phenotypes that are overlapping with severe immune suppression. What's also important to note is that the overall mortality in the study was 11%. But if you had multiple organ failure plus a phenotype, the mortality was 24%. And further, the mortality was different based on what phenotype you had and whether or not there was coexistent macrophage activating syndrome, as defined by the study criteria. And so if you had immunoparalysis with MAS, per the criteria, mortality was 40%. Without MAS, mortality was 16%. Now, I should note that the MAS phenotype was a relatively rare phenotype in the cohort, representing about 6% of the entire cohort, or 8% of those with MOF. For TAMOF, similarly, if you had TAMOF plus MAS, mortality was 50%, compared to TAMOF without MAS, where mortality was 40%. And for those with sequential liver-associated multiple organ failure, all of whom had MAS, mortality was about 43%. So my take-home points from this study is that not all sepsis is the same, and not all MODS is the same. So we probably ought not treat them all the same. Inflammation phenotypes are associated with higher mortality, and they overlap. And importantly, hyperinflammation and immune suppression can and do occur concurrently. And importantly, the immune suppression is often not apparent to us at the bedside. It's difficult to diagnose this right now with tests that are clinically available. So we need to keep this in mind when we're taking care of our patients, and we need to be mindful of this as we move forward toward immunomodulatory therapies, and need to be thoughtful about how we approach the patient who appears to be hyperinflamed by protein biomarkers, but also to be immune suppressed by functional assessments. Along the same theme is this really fantastic paper by Hector Wong and colleagues. One of the reasons I chose this paper in particular for the translational science section is that this paper in and of itself is a great example of translational science. So this really was a bedside-to-bench-to-bedside approach, and has been an iterative bench-to-bedside- to bench approach throughout the research program of Hector Wong and his colleagues. One of the things that this paper did was to validate the Persevere 2 biomarker-based risk prediction model in an independent cohort of children. So for those of you who aren't familiar with the Persevere research program, this is really work that's been done over many years through an iterative process of refining protein-based biomarkers that in previous cohorts have been associated with mortality differences. And aside from just validating in another cohort of critically ill children, the investigators also validated a similar risk prediction tool in a murine model of sepsis, which is important because it allows us to potentially use these murine models to investigate biomarker-based therapies and targeted future clinical trials in a preclinical model. So the patient cohort was a prospective multicenter study of children who are one week to 18 years of age who were admitted to the PICU with sepsis by consensus criteria. The Persevere 2 biomarkers are those that you see there. One of the differences between Persevere and Persevere 2 was the addition of platelet count, which improved model characteristics and ability to predict mortality. The animal model was C57 black six mice who were subjected to cecal ligation and puncture. It's a very common sepsis model in mice. And then regression tree analysis was used to derive the murine Persevere panel. In terms of the patient cohort results, there were 461 critically ill septic children enrolled with an overall mortality rate of 12.6%. And as you can see in the figure, those who had low risk per the biomarker panel up there in blue had a very high survival compared to those who were intermediate risk in black and high risk in purple. In terms of model characteristics, the area under the receiver operating characteristics curve was about 0.83 with a sensitivity of 86% and a specificity of 69% to predict mortality. In terms of the animals, when we look at the animals who were high risk by the biomarker panel, we can see that they had high inflammatory markers such as NPO activity, which is a marker of neutrophil activity, as well as circulating quantities of the inflammatory cytokine IL-6. They also had higher bacterial burden. And what I found fascinating is that when they took the high risk animals and treated them with corticosteroids in order to dampen the inflammation, there was no survival benefit. However, when the investigators took the high risk animals and treated them with higher dose antibiotics, there was a survival benefit. And this suggests that perhaps the excess mortality seen in the high risk animals might not be due to the inflammation itself, but to failure to clear infection. And so my take home points from this paper is that Persevere 2 performed well in an independent cohort of critically ill septic children, in that a risk-based biomarker panel that was very similar was identified in murine sepsis model. And this allows investigators to use murine models potentially to evaluate targeted sepsis therapies, as well as to model both prognostic and predictive enrichment for future clinical trials. Lastly, that the murine model suggests that the excess mortality may in fact be related to failure to clear infection as opposed to just the inflammation itself, raising the intriguing possibility that perhaps in this animal model as well, we may be seeing concurrent inflammation and immune suppression and inability to clear infection, though that was not directly evaluated here. So I'm going to turn the tables just a little bit onto ECMO. So ECMO, again, I'm sure everybody in this room is familiar with the challenge of ECMO. It's what we talk about for our ECMO patients every single day. It's that delicate balance between a circuit that's clotting that you don't want to clot and a patient that's bleeding that you don't want to bleed. So conventional wisdom would tell us that low flow causes stasis that causes clot formation. That's what we all learned. But it's possible that high flow through an ECMO circuit may cause sheer stress that causes release of pro-inflamm... I'm sorry, well, also pro-inflammatory, but pro-thrombotic mediators that could also contribute to clot formation. And so the investigators took ex vivo neonatal ECMO circuits. Basically, they took the neonatal circuit. They took healthy volunteer blood, circulated it through the circuit for six hours. They did use a heparin drip to maintain ECTs within 180 to 220 and then evaluated blood samples after either low flow or medium, what they considered normal flow through that size circuit, or high flow. And they, of course, hypothesized that low flow circuits would be associated with greater clot formation compared to medium or high flow. Well, what they actually saw was that, indeed, in the normal flow circuits, there were no occlusive thrombi identified, but that the low flow and high flow circuits both had occlusive thrombi. In fact, three out of the five high flow circuits had clot. Digging a little bit deeper, I found that the higher flow circuits were associated with a higher proportion of platelets expressing phosphatylserine, as well as a greater number of tissue factor expressing leukocytes. Now, this is important because phosphatylserine, particularly when it is expressed on the outside of a platelet, serves as a nidus for the coagulation proteins to come together and to initiate the coagulation cascade. And then tissue factor, since I'm sure everybody here remembers the coagulation cascade, will remember, the tissue factor plus factor seven is what initiates the extrinsic coagulation pathway. So, two potential mechanisms by which high flow may actually be prothrombotic. And yes, I did have to look up the extrinsic coagulation pathway. Even I. So, take home points from this paper. Mechanisms of clot formation on ECMO may, in fact, differ based on rates of flow. And it's possible that higher flow might not always be better with respect to thrombosis. However, it's a huge caveat in any of these ex vivo ECMO circuit studies, and that is when you don't have a patient attached, you don't have the patient's natural anticoagulant properties. And importantly, patients can, and probably at variable rates, clear some of these mediators. And so, you don't have that factor at play. And so, it is a little bit questionable how well this would translate to the bedside, but still an intriguing hypothesis that I think should be evaluated further. And then lastly, we're going to go to the preclinical space in terms of this same question about the effects of a single exposure to inhalational anesthesia on the developing brain. And Nilesh already gave sort of the background to this, that there are concerns based largely on previous animal models about the safety of inhalational anesthesia to young brains. And so, what this study wanted to look at was, well, we see that you can have death of about 10% of the neurons. What happens to the remaining 90%? Are they injured or are they normal? And so, they used a transgenic mouse model. Basically, this means that whenever the mice were exposed to tamoxifen, the cells in the portion of the brain they were interested in and the cells that came from them, their progenitor cells, would glow so that they could track cellular fate and structural development over time. And so, this is important because they gave the animals the tamoxifen at 14 days and then randomized them to inhalational anesthesia exposure or room air at 21 days. And so, the cells that glowed, that they were following over time, were 14 days old at the time of exposure. Why this is important is because a 21-day-old mouse is kind of like an adolescent, maybe 12 to 14 years old. A 14-day-old mouse, or I suppose mouse cells, would be more like a 5 or 6-year-old. So, if you're trying to model children, you need a younger mouse model. And I don't know if Ken Remy is in the room, but I have to thank him for that information. I don't do mouse work. So, I just have one slide of results here for you, and I'm just going to tell you the end of the story. Basically, they evaluated the brains two months after exposure and found no differences in neuronal structure or density or organization between the control and anesthesia-exposed groups. And so, the take-home points from this study were that perhaps single exposure to inhalational anesthesia might not cause long-term effects on hippocampal neurons. However, obviously, it cannot necessarily be extrapolated to younger animals. And, of course, have the same caveats about the fact that young mice may or may not be young people. And with that... So, we're at 5 o'clock right now. So, I think perhaps rather than open the mic for questions, the three presenters will be up here if you'd like to come and talk to them. Thank you for your attention at the end of the day.

Pediatric critical care

sepsis

multiple organ dysfunction syndrome

phenotypes

mortality rates

risk prediction model

biomarkers

clot formation

ECMO circuits

anesthesia effects