SCCM Resource Library-Late-Breaking Studies Affecting Patient Outcomes

Late-Breaking Studies Affecting Patient Outcomes

Video Transcription

So, good afternoon. My name is John Savransky, and I have the privilege of introducing this session and moderating this session of late breakers that might affect clinical practice.  We will have two parts to this. The first part is that the seven people to my left will be presenting simultaneously released journal articles, and then around 5 o'clock,  we will have a special section where Dr. Peng from the Wuhan province will talk a little bit about his experience treating patients with coronavirus, COVID-19,  and we'll have two U.S. experts also speak briefly. So, again, the first part, seven people, and then at 5 o'clock, some new information on coronavirus.  The first speaker will be Dr. Stephen Simpson talking about trajectories of sepsis.  Hi, good afternoon, and thank you for allowing us to present our data on the trajectories into, through, and out of sepsis over the seven-year period from 2012 to 2018.  My disclosures are these. Most important among them is that the views I'm presenting are those of the authors and not necessarily those of the U.S. government.  I'll be presenting data from a triad of papers that were presented or published on Friday the 14th, and I'll be presenting data from the trajectories of sepsis paper  and from the methods, models, and forecasts of sepsis paper. These were published in Critical Care Medicine. I need not tell this audience  that sepsis is a tremendous healthcare burden, both in the United States and across the world. This was a healthcare quality improvement study carried  out conjointly between BARDA, the Biomedical Advanced Research and Development Authority, and CMS, the Centers for Medicaid and Medicare Services, using CMS's data link project  and focusing on Medicare patients. The objectives of these studies were to explore patient sepsis-level trajectories before, during,  and after hospitalization, and to model the influence of clinical features on short and long-term outcomes of sepsis. We used Medicare claims data that were analyzed from January  2012 through December of 2018 in the largest inpatient epidemiologic study yet reported. Dr. Ken Sciretta reported yesterday on the burdens of sepsis in terms of lives and finances.  We're going to take a smaller subset of that data, 736,000 patients that had a unique sepsis visit in calendar year 2017, and no sepsis admission during calendar year 2016.  We will compare those patients to 6.8 million patients who had a non-sepsis admission in 2017 and no sepsis admission during 2018. In this slide and in all subsequent slides,  sepsis patients will be represented by dark blue and non-sepsis patients by light gray. We will look at a clean year going  in one week before the inpatient stay, one week and six months after the inpatient stay for these patients. On the lead-in to admission, we ask the question,  can we use chronic conditions to predict who is going to be septic? These are the top chronic conditions in both the sepsis and the non-sepsis group.  Notably, the top five conditions are exactly the same condition in exactly the same order. Additionally, among the next five chronic conditions down, three of the five are the same.  In essence, one cannot tell the difference. We then looked at conditions that sepsis patients expressed more commonly than non-sepsis to see if they could be discriminatory.  Two things were important. One is that sepsis patients did have a history of a previous sepsis diagnosis. And about three and a half times as likely as a non-sepsis patient.  And sepsis patients were more commonly had ulcers by about two times. The remainder of these conditions were present in fewer than 20% of the septic patients.  We then looked whether sepsis patients had seen a provider in the week previous to their admission, what the rate of those provider encounters might be,  when those encounters occurred, and where those encounters occurred. Sepsis patients saw a provider in the week prior to their admission over 50% of the time.  Sepsis, again, in dark blue, non-sepsis in gray. Sepsis encounters were slightly more prevalent than non-sepsis, seeing a provider in the week before.  And there was a slight declination in this over the period of study. When were the patients seen in that week? Going from right to left from seven days to one day before admission.  There was a gradual increase up until the day before. And there was a tremendous spike one day before admission, both in the sepsis patients and in the non-sepsis patients.  There was no discernible difference. Where did these encounters occur? They could occur in the outpatient setting. They could occur in nursing facilities.  The odds ratio for a sepsis patient being seen in the outpatient setting compared with a non-sepsis patient is .74. The odds ratio of a sepsis patient being seen  in a nursing facility versus a non-sepsis patient is 3.25. However, the large number of non-sepsis patient admissions obscures this difference.  So that would be difficult to see in a prospective fashion. In summary, on the incoming trajectory, we can't differentiate sepsis from non-sepsis in the week before admission.  And there is no discernible difference in either the number or the type of outpatient visit. For the post-sepsis trajectory, we ask can we predict the modeling longer-term outcome  in sepsis? We hope to do this with a simple or parsimonious model. First, let me show you that 30% of sepsis patients are only 30% of sepsis patients are discharged directly to home  with no further assistance. The data I'm showing you are obtained from a web-based application available at Critical Care Medicine. The links to this application are available  in all three articles. And all of the data backing up these papers is available for perusal of any interested party.  This slide we call our sanity check slide. The sanity check because we ask do our models predict the raw numbers and do they make sense? The answer in both cases is yes.  Non-sepsis admissions have the lowest mortality. Increasing levels of severity of sepsis have higher mortalities. Septic shock has the highest mortality at 38%.  Focus on this slide on the top two bars. We ask the question does developing sepsis in the hospital make a difference? And the answer to that is yes.  Hospital-acquired sepsis confers a double the risk of likelihood of an early death compared with present on admission. We also ask the question do cancers make a difference?  Major cancers we define here as hematologic malignancies, lung cancers and other metastatic cancers. Cancer increases the mortality of even the non-sepsis admissions  but the take-home point here is if you have a cancer and you get septic and septic shock, your mortality is above 50%. This slide shows the one-week, six-month  and three-year mortality of sepsis among the various severities of sepsis. The key take-home point here is mortality is high even in the lowest severity of sepsis  at approximately 60% at three years. It is decreasing in all severities of sepsis across the study period. And the question why will we do our modeling at six months?  The answer to that is of all the deaths that occur within three years of the sepsis admission, approximately half of them have occurred by six months.  Where are patients six months after discharge? Following a sepsis admission, just over half are in their family home compared to just under 80% of non-sepsis admissions.  In six months, one-third of sepsis patients have died compared with 13% of non-sepsis admissions. Roughly one-and-a-half times the number  of sepsis patients will be in a nursing home in six months. Now what happens if you are ill enough to be discharged to a SNF, skilled nursing facility, rather than directly to home?  In six months, you are under half likely to be at home, but your likelihood of being in a nursing home is tripled. Age affects the odds ratio for mortality at six months.  Each decade adds additional risk. And sex. Gentlemen, I apologize, but being male confers an odds ratio of 1.16 compared with being female. In summary, post-discharge,  having a sepsis hospitalization is associated with a markedly different post-hospital trajectory. Only about a third of sepsis patients go home with no additional services.  Only half are home at six months post-discharge. And an initial trip to a SNF worsens the trajectory. How bad is that trajectory worsened? Here's one example.  Having one episode of septic shock confers an odds of mortality at six months that exceeds that of being admitted with metastatic cancer. Our overall conclusions are  that we cannot predict sepsis versus not sepsis admissions based on these data. Sepsis markedly worsens the outcomes and the quality of life even in the mildest cases.  But we are left with one question. It is unclear to us based on our data whether acute sepsis is a marker for an already declining health status  or whether sepsis itself causes the decline in health status. Thank you very much. Please, for further information, visit one of these websites. Thank you to our co-authors.  And thank you for allowing us to present.  So we have time for perhaps one question if people want to step up to the microphone.  So very nice presentation, Dr. Simpson. Let me ask this question from you as a leader of HHS. If people with heart attacks were dying at such rate,  there'll be outrage in the country. Why is there not an outrage across all the medical society and medical professions with these results? Why is there air?  I'm sorry, I don't, I have wondered myself for many years, why is there not more outrage on this topic? Literally, approximately 1,000 people a day die from sepsis,  which exceeds a whole lot of other conditions. I'm sorry, I don't have an answer. It's a philosophical question. But I do appreciate the comment.  So the article that you just heard was simultaneously published in Critical Care Medicine. Our next article, which will be or is published  in simultaneously in Pediatric Critical Care Medicine, will be presented by Dr. Allison Nair on the clinical characteristics, major morbidity and mortality,  and trauma-related pediatric acute respiratory distress syndrome.  Hello. So my name is Allison Nair. I'm a pediatric intensivist at the University of California in San Francisco. And I'm really honored to be here today to talk to you  about the clinical characteristics, major morbidity and mortality, and trauma-related pediatric acute respiratory distress syndrome. So here are my and my fellow co-authors'  disclosures and funding. Traumatic injuries are a prevalent cause of morbidity and mortality in children. Specifically, they account for more than 45% of all deaths  in pediatric patients. That's more potential life years lost than due to sepsis, SIDS, and cancer combined. In those that survive, ongoing disability affects approximately 50%  of these patients. Eight million pediatric patients are injured severely enough to require hospital care each year. And the financial burden is significant, estimated  at $14 billion in lifetime medical spending and $66 billion in present and future work loss. Acute respiratory distress syndrome is a significant complication following trauma.  In adult trauma-related ARDS, we see a wide range of incidences with a median of 8.4%. I've listed a few important associated factors here.  And you can see that the mortality also has a pretty wide range, up to 60%. In general, we think of these patients as doing better  than patients with ARDS from other causes, but worse than patients who have just injury alone. Despite the importance in the adult population, there's very little literature  to describe this entity in our pediatric patients. There have been only a couple of studies to look at this. And they've all been helpful in starting the conversation,  but have a few important limitations, including that they were looked at retrospective data collection. They looked at, some of them looked at billing codes to define ARDS.  They've used adult definitions of, to define pediatric ARDS. And there's been an overall focus on mortality as an outcome. What we know so far is that the incidence is up to 4.2%.  I've listed a few associated factors, although these haven't been agreed upon in the current studies. And the mortality is significant, so up to 34%.  So with that in mind, our goal is to describe the presence, characteristics, associated morbidities, and mortality in trauma-related pediatric ARDS  in a large U.S. cohort over a 10-year period. So we turn to Benioff Children's Hospital Oakland, which is our local American College of Surgeons verified level one trauma center.  It serves most of Northern California. And through the trauma center, we have a trauma registry that prospectively identifies all patients presenting.  We were able to identify patients between 2006 and 2015. We included patients with a presenting diagnosis of or related to traumatic injury, with subsequent admission  as an inpatient to our hospital, and with a PAO2 and chest imaging within four hours of each other and within seven days of the initial trauma.  We excluded patients who were greater than 18 years of age, so not children, and who were transferred to another hospital within seven days of the initial injury  because we couldn't reliably follow their clinical course. We define pediatric ARDS by the Pediatric Acute Lung Injury Consensus Conference definition.  So specifically acute onset within seven days of the trauma. Hypoxia demonstrated by a P to F ratio of less than 300 or an OI of greater than four.  Chest imaging with a new pulmonary infiltrate consistent with parenchymal disease, and respiratory findings that were not explained by cardiac failure or fluid overload.  Our primary outcome was in hospital mortality, but we also looked at newer progressive multiple organ dysfunction syndrome as defined by PrEU and other morbidities.  We used univariate and bivariate analyses, as well as multivariable logistic regression. We set an alpha of .05 as a cutoff  for statistical significance and used the Bonferroni correction for multiple comparisons. So we were able to identify over 7,000 patients presenting in our 10-year study period.  6,725 patients did not meet inclusion criteria, so that's most of the patients. Sixty-five percent of these patients did not meet inclusion  criteria because they did not have a PAO2. That left us with 657 admitted patients with complete laboratory data and imaging data.  Eleven of these patients went on to meet exclusion criteria. Eight of them were adults, and three were transferred to a different hospital.  So ultimately, we had 646 included in our analysis. We identified 59 patients, or 9% with trauma-related pediatric ARDS by our definition.  And the remaining 91% of patients did not have trauma-related pediatric ARDS. So when looking at demographic characteristics between the two groups, what we saw was  that there was no difference in age, sex, race, or ethnicity between the two groups. In terms of injury characteristics, patients with trauma-related pediatric ARDS were less likely  to have injury due to falls, were more likely to have injury due to non-accidental trauma, and they also had a higher median injury severity score.  So that's shown at the bottom, 25 compared to 9. Another sort of interesting thing that we saw was that in patients with the most severe injury,  so that's an injury severity score of greater than or equal to 30, the incidence of trauma-related pediatric ARDS was 55.9%.  So this affects many of our most severely injured patients. This shows some associated factors that we've seen in adult studies. And what you can see here is that patients  with trauma-related pediatric ARDS, which are shown in the dark blue, had a higher percentage of severe injury and traumatic brain injury when compared to those  without trauma-related pediatric ARDS, shown in the light blue. In terms of organ dysfunction, patients with trauma-related pediatric ARDS had a higher incidence.  It's actually a very high incidence, so 86.4%. So really, most of these patients had this sort of global organ dysfunction. This is compared to 10.4% of patients  without pediatric ARDS, so they had eight times the organ dysfunction. Age, injury severity, and trauma-related pediatric ARDS  were all found to be independent statistical predictors of new or progressive multiple organ dysfunction syndrome.  And these patients also experienced greater morbidities in addition to organ dysfunction, including length of hospitalization, fewer PICU-free days at day 28,  fewer ventilator-free days at day 28 as well. There was no difference in readmission between the two groups. In terms of mortality, this was also higher in patients  with trauma-related pediatric ARDS, 18.3% compared to 3.1%. So they had five times the mortality. This graph shows percent mortality  across different injury severity score categories. So zero to 14 is mild injury, 15 to 29 is moderate injury, and 30 to 75 is severe injury. So we see a couple of things here.  First, when looking within injury severity categories, patients with trauma-related pediatric ARDS had higher mortality in the mild and moderate,  but not in the severe injury categories. When looking across injury severity categories, you see that patients with trauma-related pediatric ARDS had the same mortality,  regardless of injury severity. However, those without trauma-related pediatric ARDS had increasing mortality with increasing injury severity.  And finally, in our logistic regression, we saw age and injury severity, but not trauma-related pediatric ARDS were independent statistical predictors of mortality.  So in conclusion, trauma-related pediatric ARDS occurs in 9% of trauma patients and 55% of the most severely injured patients. It occurs at a higher rate in children  with traumatic brain injury or severe injury. It was associated with eight times the development of organ dysfunction and five times the mortality.  And it was also an independent statistical predictor of organ dysfunction. A few strengths to note with this study include that this was a relatively large sample size,  especially for a pediatric study, that we were able to use prospectively identified trauma registry data. We also verified that data in the medical record.  We used a pediatric-specific definition of ARDS, and we were able to focus on organ dysfunction in addition to mortality. A few limitations that will lead us into future directions.  So first, this cohort that we analyzed was actually a very small percentage of people presenting with trauma in general. These are likely the sickest patients.  So we don't really have a sense of those patients who maybe are a little bit less sick or maybe have more mild respiratory failure.  In addition, we were not able to look at certain biologic and clinical variables from the registry or from the medical record.  So for example, we wanted to look at saturation as well as non-invasive respiratory support to augment our definition of pediatric ARDS.  And we weren't able to get those from either of those sources. So this really speaks to the importance of prospective evaluation in this patient population.  We hope to focus on delineation of drivers in the development of trauma-related pediatric ARDS, identification of key biologic and clinical determinants, and a refined understanding  of how these determinants impact outcomes. Ultimately, we want to develop pediatric-specific diagnostic decision support and targeted therapeutic strategies  to help improve outcomes in our patients. So with that, I'd like to thank you all for your attention. I'd like to thank SCCM and the Critical Care Congress and PCCM.  Heidi and Mitch were my co-authors and have been very instrumental in helping mentor me. I have some other amazing mentors listed, and there's my funding and institutional support.  Thank you. So we again have time for a single question.  So first of all, congratulations on being selected to present and on a very, very nice presentation. There's a school of thought that ARDS is an iatrogenic disorder  and the two big implicated factors would be fluid administration and tidal volumes on the ventilator. Do you have any information about those and whether or not they contributed?  Yeah, that's a really important question. We unfortunately were not able to look at volume resuscitation and tidal volumes on the ventilator from these two sources.  So I think this really, again, speaks to the importance of having really targeted prospective evaluation of this entity. Thank you.  So our next speaker is Dr. Scott Weiss, who will be talking about identification of pediatric sepsis for epidemiologic surveillance using electronic clinical data.  And this, too, will be found in pediatric critical care medicine. Thank you very much, and thank you to the Editorial Board of Pediatric Critical Care Medicine for the opportunity  to present this study.  Here we go. Thank you. So here are my disclosures, none of which are directly relevant to what I'm going to present today.  So the most common strategy used to measure the epidemiology of sepsis involves applying ICD-9 and 10 codes to large data sets. Unfortunately, depending on the strategy used  and the combination of codes selected, you can get considerable variability in your estimate of both incidence and mortality shown here for adult sepsis,  as well as for pediatric sepsis. In 2017, Chanu Rhee and colleagues at Harvard and with the CDC Prevention EpiCenters Program similarly  found that using different strategies of ICD-9 codes, there were differences in both incidence and mortality for adult sepsis.  But they went on to use direct clinical data abstracted from the electronic health record to show that they could identify an incidence and mortality  that was in between the two most commonly used ICD-based coding strategies, and they went on to validate this algorithm against manual chart review.  However, a similar algorithm to detect cases of pediatric sepsis using clinical data direct from the electronic health record did not exist.  So we sought to develop a computational phenotype that would allow us to objectively, efficiently, and reliably identify pediatric sepsis cases using routine  clinical data, and then apply that computational phenotype to study longitudinal trends in pediatric sepsis epidemiology. We conducted a retrospective observational study  at a single academic children's hospital using all ED and inpatient encounters over an eight-year period, excluding patients hospitalized in the cardiac and neonatal ICU  who have very specific physiology and likely need a specified algorithm that's different from more general sepsis. The exposure were sepsis episodes  that we identified using our algorithm that we derived and validated over the course of this study, and we compared them to a reference standard that was identified  as suspected sepsis cases, which were flagged by a quality improvement alert that was embedded within our electronic health record, which we then manually adjudicated each  of those suspected sepsis cases for the presence of severe sepsis or septic shock based on 2005 pediatric sepsis definitions. So we specifically sought to identify cases of sepsis  that were defined as infection plus severe organ dysfunction, which models the sepsis three definition, and we sought to identify only those with confirmed rather  than suspected sepsis, and we specifically only utilized data variables that were thought to be routinely collected clinical data that would be accessible in a wide range  of electronic health records from both academic and community children's hospitals in order to optimize generalizability of our algorithm.  And so to model infection, we looked for blood culture orders or evidence of collection, or because many children are transferred from community hospitals to referral centers,  transfer from an external healthcare facility, plus treatment with antibiotics for at least four consecutive days, approximate to the time  of the blood culture transfer, or if the patient died or was discharged to the hospice before day four. The first antibiotic must be new in that it wasn't previously administered  in the prior seven days, which is relevant for hospital-acquired sepsis and must be parenterally administered. But the antibiotic could vary over the course of those four days.  We chose four days because two days is commonly used to rule out sepsis, and we were worried a third day would potentially, you know, if people were slightly delayed  in stopping antibiotics in those cases. So that's how we selected four days, which would indicate some confidence at the level of clinician that they were  at least treating a bacterial infection. For organ dysfunction, we turned to sepsis three, which requires a increase in SOFA score of at least two points,  which indicates moderate dysfunction in one organ system or mild dysfunction in two or more organ systems. So for pediatrics, the best validated scoring system  for organ dysfunction is the PILOD2 system, and there's also been published a pediatric modification of the adult SOFA score.  So we utilized variables and cut points from those scores, which have been previously validated, and were able to identify criteria that would indicate moderate dysfunction  in at least one organ system within two days of blood culture or transfer. Through a several-month period of several iterations,  we ultimately settled on seven criteria to define dysfunction in four organ systems, and these are listed here. We did not include neurologic dysfunction  because we could not identify a consistent variable within the electronic health record that would consistently and reliably identify the presence of neurologic dysfunction.  In addition, although many children with sepsis do have hepatic dysfunction, it is very rare, if ever, the single organ dysfunction defining problem  that causes a child to be labeled as having sepsis, although it often accompanies other organ system dysfunctions. And so ultimately, it was excluded  because it did not add to the sensitivity of this algorithm. So for our analysis, we did standard test characteristics with sensitivity and specificity  on a 10-month derivation cohort and a temporally separate seven-month validation cohort, and then looked at mortality and incidents using personal logistic regression controlled  for variables likely to influence changes in sepsis epidemiology over time. So over the course of the entire study period, we evaluated over 800,000 hospital encounters.  You can see the patient characteristics from the 10-month derivation and seven-month validation periods were very similar to the overall study cohort.  For the derivation period, there were over 93,000 hospital encounters from which 1,065 cases of suspected sepsis were flagged by our QI alert. We did a chart review on all of those  and confirmed sepsis in 187 patients. When we applied our algorithm, we were able to find a sensitivity of 78 and 76%. In our validation cohort,  we had a similar sensitivity and specificity. And this compared favorably to the adult study by RE-INT published in JAMA a few years ago with a slightly higher sensitivity  and a slightly lower specificity. When we looked at where our algorithm disagreed with our reference standard, we investigated all false negatives.  So these were cases in which our algorithm failed to identify sepsis when our chart review said they were septic. And the majority of these were in situations  where the patient received less than four days of qualifying antibiotics. And this was largely viral sepsis. None of these patients died.  And ultimately, we were okay with missing these patients since it's very rare to have a mortality from viral sepsis in kids without a secondary bacterial infection.  For false positives, again, these are patients in which our algorithm suggested sepsis, but our chart review said they were not septic. And this was primarily due to differences  between our algorithm, which was based on sepsis 3, and our reference standard, which was based on the 2005 definitions, which require the presence  of either shock or two organ system dysfunctions. And so single organs, moderate dysfunction in a single organ system was included in our algorithm  as having sepsis, and we were okay with that because we feel like it approximates a more contemporary definition of sepsis 3. We then applied this algorithm  to the entire eight-year study cohort. And we were able to show in the blue line that there was a slight increase in sepsis incidence among all ED and inpatient encounters,  and a significant increase in the orange line in the adjusted incident rate ratio of sepsis incidence among all inpatients. Mortality, however, was stable at around 6.5%  over the entire time period without a significant increase. However, with the doubling of incidence, this meant that twice the number of children were dying  of sepsis each year at our institution. We then, for face validity, looked at whether the number of organ dysfunction criteria was associated with a stepwise increase in mortality,  which is what we anticipated. And in fact, you can see that in the orange bar. And so that suggested that patients who had more organ dysfunction criteria based  on our algorithm were more likely to have mortality, which was what we anticipated. And then we looked at the different levels of organ dysfunction and how they related  to the level of mortality. And there were three specific criteria that were associated with the highest risk of mortality, including use of vasoactives, elevated lactate  and kidney dysfunction. Again, all consistent with prior literature of risk factors for mortality in children with sepsis. So again, supporting the validity of our algorithm.  Finally, we asked whether or not a change in the utilization of lactate, which was included as a measure of cardiovascular dysfunction, contributed at all to the rise  in incidence, because we've been emphasizing over the years to use lactate more and more in these children. And there was no increase in use of, or in the number  of patients who identified either only because of elevated lactate in the orange line, or as lactate included as one of several organ dysfunction criteria in the blue line.  So limitations are that this algorithm excludes viral sepsis. However, it should pick up bacterial secondary infections, which is the high-risk group for viral sepsis in children.  We excluded cardiac and neonatal intensive care patients, who, again, who have unique physiology and likely need a dedicated algorithm.  We had no measure of hepatic and neurologic dysfunction. The hepatic dysfunction doesn't bother me, because it didn't add much to the algorithm.  The neurologic dysfunction, I think, is a problem, but was not able to be, we could not find a variable to identify this reliably in the electronic health record.  We had the incongruent reference standard in that we tried to model sepsis three, but compared that to the older pediatric definitions.  This was used for retrospective case identification and cannot be applied for early recognition of pediatric sepsis, and this was done at a single institution  with generalizability that's yet unproven. So in conclusion, we were able to develop a computational phenotype that could objectively, efficiently,  and reliably identify pediatric sepsis using routine, electronic clinical data, and using this algorithm, at least at a single institution,  showed that there was an increased incidence with stable mortality, and for the first time, we think this was really able to be free from bias introduced  by variability in either diagnosis and claims-based coding practices. Next steps are to validate this algorithm in a multi-center fashion and compare them directly  to claims data, and then ultimately to apply this algorithm for the epidemiological surveillance of pediatric sepsis, hopefully across the country.  I'd like to acknowledge our collaborators, in particular, Fran Ballamuth, who is my partner on this work, and thank you very much. Appreciate the opportunity to present. Thank you.  So thank you. We additionally have time for a single question.  So in the absence of, you are using this or plan to use it for epidemiologic reasons, but it seems as though you could potentially use this for other things,  identifying patients, perhaps? Yeah, I think for research studies, I think this is great, but because we require four days of consecutive antibiotics, there's a slight delay.  So I don't think it's, it's not applicable for early recognition for clinical purposes, but certainly has potential other applications besides epidemiology, in particular,  identification for research studies. Quality improvement, as well.  So our fourth speaker will be presenting an article that is simultaneously appearing in JAMA on the effect of intravenous interferon beta 1a on death and days free  for mechanical ventilation among patients with a moderate to severe ARDS, a randomized clinical trial. Dr. Jeffrey Bellingham. Thanks very much.  I'm delighted to be here, and as you heard, presenting the result of the INTEREST trial, which is a multicenter study of interferon impact in moderate or severe ARDS.  Disclosures, the trial was funded significantly by a major grant from the EU, and it was also, the study sponsor was Farron Pharmaceuticals.  Myself and Marco Ranieri were the two PIs, and you can see the steering committee there representing each of the eight EU countries.  EU because Britain was part of the EU at that stage. So the background story is that in the muscles, ATP is fundamental. It makes muscles work, but in the extracellular fluid,  ATP is pro-inflammatory. The body knows that and tries to get rid of it by dephosphorylating the ATP down to adenosine,  which is anti-inflammatory, and it does that through the action of two enzymes, CD39, which is constitutive, and CD73, which is expression is increased by the action  of interferon, and it can be shed. So here we have the normal body situation with lots of CD73 expressed, lots of adenosine being made,  anti-inflammatory, and that keeps a non-inflammatory environment, and we believe contributes to lack of lung leak, and interferon is one of the agents endogenously or exogenously  that helps protect that position. In the inflamed situation, CD73 is shed, and in the absence of CD73, you have no adenosine, lots of ATP  in extracellular fluid, and we believe that to be part and parcel of the lung leak you see in ARDS. So ARDS shedding of CD73, and no current treatment for that.  We undertook a phase one, two trial across a number of centers in the UK, and quite a remarkable outcome, which was really a dose finding and safety study  where we found a mortality of 8% in the interferon beta treated patients, and 32% in our control group. This was published in Lancet and led to the work  up for our phase three interest study. The design of the study was a double blind placebo randomized control study across eight countries in Europe, 47 sites.  We wanted to get the drug in early, within 48 hours, which is when we thought the lung leak was paramount, because later you get all the inflammatory and cell destructive processes  that we didn't think this was necessarily going to be part of. Primary endpoint was a combined composite of survival and ventilator free days at day 28,  and multiple logical secondary endpoints, mortality 28, 60, 90, et cetera. Safety being studied, and you can see the powering of the study there.  To be sure that we had ARDS, we used two medical monitors who are on call 24-7, so huge thanks to the two of them, Nicola and Peter, and any site wanting to recruit a patient  had to match the x-ray from their patient with one of a number of x-rays provided by the medical monitor, only some of which had ARDS.  And that helped us focus down on ARDS in the population and recruiting the right patients. As you heard, we're very proud that our study is being  published in JAMA today, and I'm going to present the results here.  The consult diagram, we obviously screened a large number of patients with ARDS, and a significant number were excluded. The majority of exclusions were patients with ARDS diagnosed  more than 48 hours before. A number with ECMO, and another group who had significant exclusion factors, home ventilation, severe heart failure, et cetera.  We got down to a number here, 301 who were randomized, and you can see equal randomization then into the two groups, and the number here that we're able to assess out at 90 days.  The demographics, I don't want to spend too long here, very similar between the interferon-treated and the placebo-treated group.  You can see that they, pneumonia was the most common cause of ARDS, sepsis, the second most common cause. The patients were reasonably sick with a significant  Apache score, SOFA score, and the number with severe ARDS, similar in the two groups. It was worth noting that the number on corticosteroids  at the onset of the study was quite significant, with about 31 and nearly 40% in those two groups. It was a safety study as well as an outcome study,  and the safety groups between placebo and interferon for the most common treatment of emergent adverse events, no significant difference. I note the slight increase in pyrexia here,  that's what you would expect with interferon beta being administered, and it's just worth keeping an eye on rhabdomyolysis, whether this five was a random chance event  or not, we'll need to keep a watch on. Otherwise, no obvious difference in terms of most common safety events, and similarly, no obvious difference in terms  of serious safety adverse events. If you give interferon, how do you know it's doing anything? People don't go blue and turn pink with spots or whatever,  but you can measure a biomarker, and MXA is the best biomarker, and you can see MXA increasing and peaking here at the end of our six-day treatment course, compared to our placebo.  So, we were comfortable that the interferon was working. We were measuring CD73, recognizing that this is soluble CD73, so we don't know what's happening in the lung,  and we saw an increase in soluble CD73 in our interferon-treated patients, but there was a similar increase in our placebo-treated group.  So, whether this meant there was an endogenous interferon or other reason for shedding, including just the major injury, we have yet to find out. 28-day mortality, no difference  between either the interferon or the placebo-treated group. It's worth noting a relatively low mortality with both groups. When you look out at 360 days, again, no difference  in the mortality between the two groups. They were remarkably consistent over that time, and when you look at ventilator-free days, again, no real difference in ventilator-free days  between the two groups. We were interested in the impact of the corticosteroids. We had far more corticosteroids being administered  to our patient group than in our Lancet respiratory paper, and we did retrospective analysis of the impact of corticosteroids on the outcome, and we noted  that glucocorticoids at baseline tended to favor the placebo group versus no glucocorticoids at baseline, possibly a slight impact favoring interferon.  We've done further analysis on that, and that will be subject of a subsequent paper, which will be of real interest. So, in conclusion, we can say that for moderate  or severe ARDS, intravenous interferon beta for six days compared with placebo resulted in no significant difference in the composite score of death and number  of ventilator-free days over 28 days, and nor did it in any other of our significant secondary outcomes. There were no major safety concerns with the use of interferon beta.  The results, therefore, do not support the use of interferon beta in the management of ARDS. We have to say that that's where there's part of an SOP  that includes the use of steroids. Concomitant use of glucocorticoids may have interacted with results. Glucocorticoids use was associated with mortality,  both in interferon and the placebo group when adjusted for Apache. And, therefore, we think further studies may be warranted for interferon beta as a treatment for ARDS  without concomitant use of glucocorticoids. And I make that point because care is needed in the treatment of clinical conditions associated with interferon beta activation.  So, if you consider viral-induced ARDS as is happening right now, your own body is producing significant interferon beta, and the use of glucocorticoids is not recommended there  by the WHO, and I think we should be considering that as a different group of ARDS versus non-viral, non-interferon endogenous production, and that may need further look at.  I'd like to acknowledge a large number of study sites and a very large number of workers on what was a fascinating study to do, and I thank you all for your kind attention.  Thank you very much. So, we have time for a question. Any explanation for the relatively low mortality in your group, in both groups? I'd love to know.  We opened up the, well, we try to keep our exclusions very, very limited. We were allowed all forms of cancer. We had hematological malignancies. We did exclude liver disorder.  We did exclude people on, with significant lung, pre-existing lung injury, but nothing very different to other studies. So, we don't really know why the mortality was so much lower  and lower than our baseline launch study.  Our fifth manuscript, the effect of using a machine learning derived algorithm versus standard care on intraoperative hypotension during elective non-cardiac surgery  or the HYPE randomized clinical trial will also be released in JAMA by Dr. Van Biera. Thank you for the introduction, and I would like to thank JAMA  and the Society of Critical Care Medicine for the honor to present our results here today. My name is Marije Wijnberge. I'm an MD from the Amsterdam University Medical Center  in the Netherlands, and I will present the results of the HYPE trial. This was a randomized clinical trial where we assessed the ability of a machine derived early warning system  to actually reduce hypotension during surgery.  And as of today, published in JAMA. As a research group, we received funding from Edwards. The problem, intraoperative hypotension, it occurs frequently.  Eighty-seven percent of patients have at least one hypotensive event during surgery. So, why is this relevant? This is relevant because it's associated with postoperative outcomes.  This has been shown in both cohort studies and also in one RCT published by Fouchier. And if you summarize those studies, you see that it's associated with cardiac outcomes,  such as myocardial injury and myocardial infarction, with renal outcomes, such as ACCI, and even with mortality. So, therefore, the perioperative consensus statement  of 2019 concludes we should try to maintain a map above the 60-70 threshold during surgery. And they also state that injury is a function of both the time spent in hypotension,  but as well the depth, so the severity of hypotension. And current treatment of hypotension is reactive. Before we conducted the RCT, we did a short perspective observational study.  And we saw that the average time from onset of hypotension till first treatment is 1 minute and 30 seconds. So, actually, that adds up.  A potential solution would be to change this treatment from reactive to proactive. However, we as doctors are not always able to predict hypotension in a timely manner.  Luckily, a machine learning algorithm is able to do so. And the machine learning algorithm we used in this study is solely based on the arterial waveform.  It's developed by a company with input from intensivists and anesthesiologists. And it provides you with a value ranging from 0 to 100.  So, for example, 85 means there's an 85% chance of hypotension to occur in the following minutes. This algorithm has been internally and externally validated,  showing good results 5 minutes prior to hypotension. But as we all know, prediction is not the same as prevention. In order to actually reduce hypotension,  the prediction needs to be followed by a correct and timely intervention. So, therefore, we performed the HYPE trial, an RCT,  where we wanted to see if it actually reduces hypotension, measured at both the time and the depth. For this study, all adult patients planned  to undergo elective non-cardiac surgery, requiring general anesthesia and with an indication for an arterial line were eligible.  And we excluded patients in which the treating doctor wanted to maintain a different map, so a map higher or lower than 65. And he was also not allowed  to use a goal-directed fluorotip protocol or any other protocol. And the other exclusion criteria are based on the early warning system.  It's not validated in those patient populations.  So, what did we do? In the intervention group, we connected the study monitor with the early warning system, and we provided the treating doctor  with a diagnostic guidance and treatment flowchart created by the study team to enable timely diagnosis of the underlying cause of the impending hypotension.  And in the control group, we had standard care, and we connected the study monitor, but we blinded the early warning system. And this is the diagnostic guidance  and treatment flowchart. So, in this study, we asked the anesthesiologist to act when the chance was higher than 85. And then, based on hemodynamic variables visible  on the study screen, such as stroke volume variation and systemic vascular resistance, the treating doctor was able to diagnose the underlying cause of the impending hypotension.  And important to note, the treating doctor was free to choose the dosage of the drug he or she wanted to administer. A primary outcome was the time-weighted average in hypotension.  So, as I mentioned before, this is a combined endpoint. You take the time in hypotension and the depth of hypotension, and you divide it by the total study time  to be able to compare patients. And one of the potential negative effects we thought of using such an algorithm could be overtreatment. So, therefore, secondary endpoints,  we assessed hypertension and the dose of medication used.  We randomized 68 patients, and eight of those patients had to be excluded after randomization, but before surgery, because the treating doctor wanted  to maintain a different MAP threshold, or because the surgery was canceled, or in one case, because of technical failure of the device.  So, we ended up with 31 patients in the intervention group and 29 in the control group. So, as you can see, the median age of our population was 64,  and the majority of patients were gastrointestinal surgery. And I want to do the exciting part, I think. What we saw is that the time-weighted average  in hypotension was significantly lower in the intervention group. But as time-weighted average is a relatively new endpoint  and a little bit hard to grasp, we also noted a total time in other hypertension endpoints. And what we saw is that the median time spent  in hypertension in the control group was 33 minutes, compared to eight minutes in the intervention group, which I think is an astounding difference.  And actually, all hypertension endpoints were significantly different between the two groups. And then, onto hypertension, so the assessment of overtreatments.  We saw no difference between the two groups for all hypertension outcomes. And lastly, the dose of medication used. We saw no difference, no significant difference  between the two groups in vasopressor use and in fluids used.  This study has limitations. First of all, our results are only applicable to this specific patient group, entailing high-risk surgical patients.  It was a single-center RCT with a small sample size, powered on the primary endpoint. And we used a physiological, rather than a purely clinical endpoint.  And in this stage, that was necessary, so we chose hypertension. And I think this study is a really important stepping stone that future studies should aim  to assess whether this also actually leads to reduction in mortality and morbidity, and whether it's cost-effective.  So, to conclude, in this trial, we assessed whether we could use a machine-learning-derived early warning system to reduce hypertension.  And we were able to do so without inducing overtreatment assessed by hypertension or dose of medication used. And future trials should assess whether this also leads  to a decrease in complications and is cost-effective.  I would like to thank the HYPE study team. And I would like to thank you for your attention. So, we have time for a question.  Excellent presentation. Phenomenal. We've been trying to do some work with waveform data from the Arturo line, and one of the problems that we have is a lot of noise,  whipping and clipping and things like that. Was that a problem with the trial, or does the device already have a way to sort of ensure the quality of the waveform  so that the output is consistent?  So do you mean you have problems with developing the algorithm or with using it in clinical care?  The waveform data is not clean in the sense that there's a lot of movement artifacts that will affect the quality of the waveform, and hence any models that we build on the waveform  has to take into account the quality of that data. Has that been an issue or not necessarily? No. So for the development of the algorithm,  that data was cut out because it's not good to build your algorithm on, and in clinical practice we sometimes had that issue,  so therefore it's important there's a doctor to ensure the Arturo waveform is of good quality. Okay. Thank you.  So our next manuscript, which will be published in Critical Care Medicine, is Data-Driven Analysis Reveals Shared Patterns in Transcriptome Responses  and Immune Cell Compositions Associated with Distinct Mortality Rates Across Differing Ideologies of Critical Illness, and will be presented by Dr. Zadar. Thank you. Good afternoon.  I'm Joel Zadar, one of the fellows in neurosurgery at the University of Toronto, and today I'm going to talk about a conserved molecular signal  that is shared across different ideologies of critical illness, and I'm going to present some initial evidence that this molecular signal can serve potentially as a universal marker  of how patients respond to critical illness due to various reasons, and also serve as a molecular endpoint for patients' resilience.  So the main motivation behind our study was the heterogeneity of diseases. So we sort of inherited this fairly old classification here from the 1700s,  and gradually over time as we found out more and more about the diseases that we treat as doctors, including the molecular characteristics,  mainly the molecular characteristics here derived from, for example, the Human Genome Project, completed in 2001,  and it took a subtle 14 years for my field, neurosurgery, for example, to adopt it in the classification of brain tumors in the WHO grading.  The second thing that really captured our imagination is the other broad impact of disease heterogeneity is that oftentimes it underlies why trials are negative.  So this is a landmark example of the sub-phenotype that were identified from the reanalysis of the ARMA and the ALVA-like cohort  were using a machine learning technique called latent class analysis. They have identified homogeneous subgroups, phenotype one,  that responded better to low PEEP, positive end respiratory pressure in acute respiratory distress syndrome, versus phenotype two, where high PEEP was more beneficial.  This study mainly included clinical parameters, and our interest honed down on the molecular aspects, the molecular response that patients mount to critical illness.  And our research question was motivated by these two streams of studies, one by Zhao et al., who have identified a genomic storm,  as they rightfully called it, this robust response that can be traced in the whole blood transcriptomics of patients suffering from critical illness of various kinds.  And they've demonstrated that, for example, burns and trauma share a lot of the genes that are expressed in response to critical illness.  The second line of studies that inspired us was that when you take the gene expression signal from patients suffering community-acquired pneumonia,  there are homogeneous subgroups that can be identified by consensus clustering, and another unsupervised machine learning algorithm. And consistently, there is a subgroup  that displays the suppressed immune responses and is associated with poor outcome. So here's the backbone of our research.  On one end, we hypothesize that there is a molecular response that is conserved across etiologies of critical illness, represented by the space invaders over here.  And then there's also evidence in the literature that patients mount a diverse response to that. And we hypothesize that that unified response can be grouped  into endotypes, homogeneous subgroups that share clinical outcome and can serve as a fundamental marker for patient resilience to illness of any description.  So we carried out a meta-analysis of openly available datasets, mainly from the Mark Consortium. I'd like to thank the authors for depositing this data.  382 patients with critical illness due to trauma, community-acquired pneumonia, and abdominal sepsis. And the analysis we carried out  in a nutshell was a multi-step machine learning approach. On one arm, it was WGCNA-weighted co-expression network gene expression analysis. The crux of this multi-step technique is  that we establish a network of gene co-expression. And based on how close they are, the individual genes to one another, they get grouped into this module down here  as depicted by this dendrogram on the left. And these modules are, the intuition is that they serve a unified biological process, which can be annotated through a gene ontology.  The second approach was a technique called CyberSort, where from transcriptomics data, we can infer various leukocyte fractions, which would normally be differentiated  through using a number of molecular markers in clinical practice. So the first line of results showed, we tested something that we already knew.  We wanted to know whether gene expression patterns are preserved through trauma, community-acquired pneumonia, and abdominal sepsis.  And firstly, we annotated the individual gene expression modules. And it's quite, it's a fairly crude annotation. This will require some further refinement from our side.  But immune response and as expected, inflammation enriched in all the disease subclasses. And here, instead of comparing the individual genes  on this heat map, we show the similarity measures that are derived from the closeness of relationship for each biological mechanism based on the gene ontology hierarchy.  And as shown by the similarity measures, they are all very concordant. The more interesting line of results was the leukocyte composition of the individual disease groups.  These heat maps summarize the frequency of each leukocyte fraction, shown here in the columns. The rows are the individual patients.  And on the left, we have the dendrograms for the clusters that the patients can be grouped into. And there's a consistent pattern in trauma,  community-acquired pneumonia, and abdominal sepsis, that there's a neutrophil surge associated with a CD4 abundance in all three diagnoses. And the second element of the pattern is  that there's always a neutrophil suppressed group with some suggestion of further mechanisms of immunosuppression, such as abundance of dendritic cells, T-regulatory cells.  When we label these consistent subgroups with mortality, we find that the neutrophil-depleted group consistently has higher mortality rates.  And in some groups, the entire mortality is captured by the subgroup that has suppressed neutrophil counts. So we have the ability to go back and check what the molecular profile  of these patient subgroups are. And returning to our gene co-expression modules, we've consistently found that the immune mechanisms, biological mechanisms that map  onto immune response are consistently suppressed. So we have a group with neutrophil surge and gene ontology mechanisms that map onto increased immune activity with better outcomes.  And this holds the promise of the general universal marker that I was referring to. To explore what this truly means in terms of underlying biology,  we would like to conduct a set of studies which is focused  on the connectivity of these cellular fractions based on their co-expression with various interleukins. With this network-based approach,  we can derive the biological activity of each cell fraction as expressed here on these histograms and ultimately explore the biological mechanisms that underlies these phenotypes.  Another interest of ours is the concept of patient resilience. We've recently published this paper on critical care where multimorbidity configurations were studied  using latent class analysis in 36,000 patients from the MIMIC dataset. We've essentially identified lethal combination of morbidity states,  some of which were not immediately intuitive. So whereas we would expect increased mortality in elderly and multimorbid over here, the highest mortality  and the greatest susceptibility to sepsis was actually in more of a middle-aged group with liver disease, coagulopathy, essentially consequences of addiction.  So this is really the dream where we want to get to create complex phenotypes that captures the baseline susceptibility of patients  to disease, which would include categorical representation of multimorbidity states, gene expression data, imaging. There's an upgrade coming for the MIMIC dataset  that includes chest x-rays. We're eagerly waiting the release of that to carry out the study. And also some EHR parameters, vitals and the pathogens that cause the illness.  The technique we're proposing is a similarity network fusion which is a local development in Toronto. Anna Goldenberg's group has pioneered this technique.  A little bit about ourselves. We're a budding neuroscience data analytics group, Machine Intelligence Network Toronto. Mint is the name we've chosen.  I'd like to thank the team for helping the work. Some of us are from a physics background. Some of us are computer scientists.  The rest are neurosurgeons, but I guess nobody's perfect. Thank you very much for your attention. I'm looking forward to questions.  Thank you.  So again, we have time for a question.  So might I ask, do you think that these two different patterns are more related to the insult that people are getting differential insults  or whether it's related to the patient's genotype?  Sorry, could you repeat that question again? There's so much. So you've got, you showed two nice patterns or two different patterns that were associated with outcomes.  Do you think that the reason that people are having these two distinct transcriptomes are related to their underlying genotype or do you think that they're related to differences  in insults that cause differential responses? Thank you. That's a very relevant question. I think the limitations of the molecular signal  that we're capturing is that it is purely looking at the expression profile. So there's a lot of machinery that's superimposed upon that,  such as epigenetics and other patient factors, which we're hoping to capture with that complex phenotype that I described.  Thank you.  So our final manuscript in this session,  before our next part of the session, will be simultaneously released in Critical Care Explorations and we have Dr. Craig Lilly talking  about vaping-associated respiratory distress syndromes, state-of-the-art guidance for clinicians.  Thank you, Dr. Svansky. I've received no compensation as the chair of our vaping task force and have received nothing of commercial value from any entity involved in the sale  of vaping-related marijuana or tobacco products. Today we're going to go over the exposure, symptoms, testing, risk stratification, and management of folks  that have been exposed to vaping. This presentation and this review is not offered as a guideline. Unfortunately, the available evidence doesn't allow us  to create a evidence-based guideline. It's simply provided as guidance to clinicians who are confronted with patients that may have vaping-associated lung disease.  You can see from the variety of devices displayed here in this slide that there are a number of different generations, styles, and forms of vaping devices.  So one of the things that we need to do when we judge vaping exposure is to try to determine what sort of a device is being used. You can see that the people  that make international symbols have had a problem because there's no one symbol that covers all devices. The currently available You Can't Vape Here signs has nine  specific types of devices that it advises against use. All devices there are third and fourth generation vaping devices. In terms of the device, the things you should really pay  attention to are does the device use cartridges that prepare commercially available product and deliver aerosols of uniform or user-controlled size?  Or does it contain a chamber where users can put in whatever solution they desire to put in there? And is the user using a solution that's commercially prepared  and therefore free of dangerous adulterants? Or have they prepared their own solution using internet-obtained extraction protocols often  that contain vitamin E, terpenes, and other toxins? The key thing to exposure is dose, device, use, and the solution. So one of the key elements is have they been exposed  in the last 90 days? So the recency of exposure matters. The frequency is probably less important than the intensity. So are they a regular user that vapes according  to manufacturer's recommendations? Or do they get advice from social media? It turns out that by dabbing or dripping solution directly  onto a heating element, one can generate a much denser, much larger volume of aerosolized particles per unit volume and a much greater diversity of the size of particles.  This allows larger particles to impact around areas of the lung where gas flow is turbulent and the airstream changes. And this leads to central airway deposition  of both thermal energy as well as airway toxins. Particular risk factors are vitamin E-containing or any member of the tocopherol family chemicals.  And solutions that are prepared by non-commercial means. So after you've established whether or not your patient is vaping, and sometimes this is a challenge.  In my intensive care unit, I noticed a really sleek-looking Juul device similar to the one here on this slide and a patient that had told me she wasn't vaping.  And I said, well, what's up with that? And she said, well, Dr. Lilly, old people vape. Cool people Juuls. So part of getting the history comes down to understanding the language.  The clinical evaluation really comes down to unexplained cough, chest pain, weight loss, fatigue or dyspnea. This is part of the CDC case definition.  The key thing is a clinical judgment about whether these are caused by vaping or caused by some other exposure. The key questions that are helpful are,  was anyone else exposed to the vaping solution? Did you share your vaping device with anyone? Do they have similar symptoms?  And is there anybody who's sick with this kind of illness who wasn't exposed to vaping? Those help you sort of triage cases that are likely  to be vaping associated from those that are not. Another clue is whether or not there are GI symptoms. The same way that aerosols can go down to the lungs,  they also go into the GI tract. And when nausea, vomiting or cramping abdominal pain is tightly associated with vaping exposure, one should assume  that the patient's been toxin exposed. So what tests should we do? Well, there's really two types of tests. Some tests are looking for other causes.  Electrocardiograms, workup for thromboembolic disease and workup for infection should be considered, particularly to exclude and to help us determine  whether the symptoms are toxic exposure or some other cause. If there's no, if there's strong reason to believe that the association between the symptoms and the vaping is real  or if it's not possible to attribute them to another cause, one should do a rest and exercise oximetry and chest imaging. This allows us to perform risk stratification.  One of the key things that our public health officials in Massachusetts came to us and asked us to do was to develop a risk stratification model  that we call the Worcester model for doing this. It's very simple. If you just look in the top row of this particular table, if you're not exposed and you don't have symptoms,  you don't have any risk and you shouldn't be tested. If you've been exposed but you don't have symptoms that are attributable to vaping, we recommend against testing  and we classify you as low risk.  We recommend against testing at the time and testing should symptoms that are associated with vaping occur in the future. Moderate risk patients have vaping exposure  and they have symptoms. They have an abnormal imaging study that has a pattern that's been associated with vaping and these patients require special consideration.  The highest risk patients, the patients at risk for death and we've had one death in our community from this, have exposure, symptoms, an abnormal imaging study and hypoxemia.  Most patients that have died from vaping have been sent out of emergency rooms when they were noted to be hypoxemic. So our low risk patients, what should we do about those?  Should we do nothing? We think not. We think that every vaping patient should be advised against, advised to quit vaping. Secondarily, you should ask, have you tried to quit  with the help of your primary care physician? When that, when the patient's interested in quitting and they have not worked with someone in their healthcare team  that they already trust, we suggest referral to their primary care physician for counseling. If they've been struggling with addiction and unable to quit  and not able to partner with their primary care physician, we recommend referral to an addiction medicine specialist. For moderate risk patients, these are patients  who have been exposed and do have symptoms that are potentially attributable. We think that vaping cessation is absolutely mandatory. So when you can supervise the fact  that they will not be vaping, I think it's reasonable, we think it's reasonable to perform additional studies as an outpatient. The nuances of how to evaluate what subphenotype  of vaping they have are present in our larger article and are beyond the scope of this talk. But the important thing is that for moderate risk patients,  they need to be followed until their symptoms have resolved and their activity has returned to its baseline normal. For our high risk group, we advise supervised vaping abstinence  and continuous pulse oximetry. We believe the patient should be hospitalized. At the first sign of increased respiratory rate at rest,  dyspnea at rest, hypoxemia at rest, oxygen therapy and early intervention with noninvasive ventilation is really important. It's probably more important than any other class of ARDS  because these patients often rapidly improve when they abstain from further exposure. We recommend a conservative fluid strategy. We recommend corticosteroid therapy for those,  for any patient without rapid improval and who's not rapidly improving. And mechanical ventilation early when it's required with a low tidal volume strategy.  We recommend that bronchoscopic evaluation be considered when infection, hypersensitivity, pneumonitis, eosinophilic or lipoid pneumonia is suspected.  We also share in our report a little bit about the airway phenotype of vaping. So it turns out that not infrequently, vapors will have a vaping tongue, tongue discoloration,  just like this young woman here. This is from small particle aerosols. There's no burn here. Exactly the same coloration when it appears in a variegated pattern.  And particularly if there's depression in the lesions, suggests that there has been thermal injury related to vaping to the upper airway.  Those patients are at very high risk when you see that. Shahzad Khan and Kate Akwunski-Silva from our group have performed bronchoscopies on several  of these patients and have noted that the, not surprisingly, the burns go all the way down the airway. Shahzad, in this particular example, noted that there were pseudomembranes.  Pseudomembranes that actually covered the bronchial, the segmental openings for the airway. And three-dimensional reconstruction  of the airway demonstrated that distal to these occlusions, there was, there were radiographic infiltrates that cleared when he removed them.  So we believe that the airway phenotype is caused by impaction and is an important sign of vaping-related injury. I'll simply conclude with this Gary Trudeau comic,  where Butsy and Jewel are talking against one another and would encourage every intensivist when they leave their intensive care unit at night to ask,  have I advised against vaping today? Thank you. So we have time for a question. Do you have any suggestions on the dose and the duration  of the corticosteroids in patients with severe injury? We prefer one milligram per kilogram per day. We usually give it twice a day. We use enteral steroids when it's possible  and IV steroids when it's not. This is clearly expert opinion based. We feel that it's important to use enough steroid to have a therapeutic effect.  So prior to moving on to the next phase of the late breakers, I hope you'll join me in giving the seven presenters a round of applause and thank them for sharing their work with us.  So for the second part, and we will release the seven speakers, we have the pleasure using some technology  of welcoming Dr. Jiang Peng, who is Chair and Professor of Critical Care Medicine at Zongan Hospital and Vice Director of the Center for Clinical Trials at Wuhan University,  who is going to present a videotaped version of a talk on 138 patients who've been infected with coronavirus, COVID-19.  And if the technology allows, perhaps he'll be able to answer some questions.  Hello, everyone. It's my great honor to be here to share my work with you. My topic today is 2019 Nova Coronavirus Pneumonia in Wuhan, 138 case analysis from a single center.  I'm Dr. Peng, Intensive from Wuhan University. The Nova Coronavirus Pneumonia, which is called NCP, is widely spreading, highly contagious, and become public health emergence  of international concern. This figure demonstrates how this problem spread so fast within 20 days from one area to other areas. This also demonstrated the number on February 6th  of the patients. The confirmed patients is around 28,000. The suspected cases is around 24,000. So I try to summarize the clinical features  of the NCP patients from my current database and try to describe the disease process and the epidemiology features and try to compare and analyze the differences of the patients  in ICU and outside ICU. This is the retrospective cohort from a single center. All the patients in note from January 1st to January 28th.  All the patients were confirmed using both of the criteria, including typical chest CT scan and the positive real-time RT-PCR assay.  We collected the following data, including the demographic data, the underlying comorbidities, symptom size, NAP, chest CT scan, treatments, including medications,  mechanical ventilation modes, and other organ supports. Results. This is the basic features of the NCP, including the age, gender, comorbidities.  It was likely to affect old persons with comorbidities. And the patients in the ICU were relatively older and much more comorbidities than the patients outside ICU.  This table demonstrated the symptoms and signs of NCP. The most common symptoms, including fever, fatigue, dry cough. Around 10% of the patients presented initially with diarrhea  and subsequently developed fever and dyspnea.  Dyspnea, abdominal pain, very frequently occurred in ICU patients. This is the clinical process of NCP. The time from the first symptom to dyspnea was five days,  seven days to hospital admission, eight days to ARDS. The most common NAP changes, including the dyspnea, prolonged prescribing time, elevated lactate dehydrogenase.  And the patients in the ICU were relatively lower lymphocyte counts and relatively higher lactate dehydrogenase compared to the patients outside the ICU.  This is the dynamic changes of the NAP test in some patients, including survivors and non-survivors. The survivors labeled with blue color demonstrated  lower white count, lower neutrophil count, and higher lymphocyte count, lower d-dimer, lower urea, and lower creatinine. This is a very typical chest scan in one patient.  A, it's the time in day one. You can see the bilateral infiltration in this patient. B is the after two weeks treatment, the infiltration,  the opacity become better compared to the A. This is the general features of the ICU patients. They have higher aperture score, higher SOVA score,  and lower P4 ratio around 136, very, very low. These patients in ICU always develop the organ injury. The most common organ injury and the complications,  including ARDS, cardiac surgery, arrhythmia, and organic injury, arrhythmia, and septic shock.  Regarding ventilation supports, only 4, around 10% of the patients survived from the high-flow oxygen therapy. And 40% of the patients received non-invasive ventilation.  About 50% of the patients needed invasive mechanical ventilation. 10% of them switched to ECMO.  About the prognosis on February 8, 40% of the patients still in hospital, 50% discharged, and 5.8% died. The hospital stood around 10 days in survivors.  Here, we also presented the updated information. Today, on February 10, we took around 50 patients already in my ICU.  Most of the patients right now much more severe because they some transferred from other hospitals and 60% of the patients needed invasive  mechanic ventilation. 11 of them switched the ECMO. 4 win of ECMO already. Another issue about the transmission you can see the figure almost 40% of the  patients transmitted in the hospital including the medical care professionals and hospitalized patients. About 70% of the infected  healthcare workers from general world. Around 20% from ER, 5% from ICU. About the hospitalized patients, 40% infected in the surgery department,  30% from internal medicine, 30% from oncology. In conclusion the transmission of the NCP was frequent especially in the hospital. The  untypical patients were probably the main source of the transmission. Critical ER patients tended to be older with comorbidities, specific symptoms  and lab abnormalities. The most common complication was ARDS, arrhythmia and septic shock. Nearly half of the critical ER patients needed invasive  ventilation. Thank you for your attention. Thank you for inviting me. Thank you for your support to my city, to my country, to my patients.  So I believe if the connection works we may be able to to have a question or two for Dr. Pang. After all the speakers. Okay in a few moments. So our next speaker  Dr. Ryan Maves is going to talk about the 2019 or actually I guess the 2020 novel coronavirus, what we know so far. Good afternoon everyone and I want to  obviously thank Dr. Sebransky and Dr. Pasquale and Dr. Howell and the remainder of the program committee and all of you for the opportunity to speak to you today. So I'm  Ryan Maves. I'm an infectious disease and critical care physician in San Diego. I do work for the federal government. Just to point out that the opinions in this  presentation are mine and do not reflect the policy of the US government. I have no financial conflicts of interest. I will talk about some off-label and  investigational drugs. So coronaviruses are a large family of zoonotic viruses spread to humans through some usually mammalian intermediate, very often bats.  So human coronaviruses, there are seven including the new novel one which probably comes at us from a bat most likely. It's part of a group called beta  coronaviruses which also includes the other six human infections. The four usual human coronaviruses which cause a relatively mild respiratory infection in  most patients, SARS, MERS, and now this novel one. Very recently the International Commission on Taxonomy and Viruses has decided to name this SARS  coronavirus 2 due to a high degree of homology with the previous SARS coronavirus from 2003-2004. There is a little bit of controversy about that but  right now a NIAID is using that name on its website. So this virus is 79.5 homologous with SARS-CoV with the original SARS and is 96% identical at  the whole genome level to a previously identified bat coronavirus. Its target in the human respiratory tract appears to be angiotensin converting enzyme 2 which  is similar to the older SARS-CoV and is presumably although not yet 100% certain to be the root of infection. It's a pretty typical looking coronavirus  for structural proteins and a single single strand positive RNA genome there in the middle. All right so I updated this this morning and I have to say  this is interesting for a talk in that it is almost certainly out of date by the time I give it. But there are approximately 72,000 cases reported to  WHO. This is the Johns Hopkins CSSE website that tracks this on an ongoing basis. So about 72,000 cases 1,775 deaths reported so far. Overwhelmingly these are  in Hubei province in China. Handful of cases in the United States, 15 possibly going up and a variety of cases in the remainder of Asia, Europe, elsewhere in  North America. So far not quite 12,000 people reported as recovered so a number of these people remain hospitalized remain ill. So early on cases were linked  to the wholesale seafood market in Wuhan. There were some attempts to assess whether or not this was due to animal exposure in the market but it turns out  there were a number of cases that didn't have an epidemiologic link that to the market that occurred very early on. It is possible that the market which is a  crowded aggregate setting could have been an amplification for the outbreak where someone who was infected entered the market and then transmitted it  within the market leading to the outbreak within the city. That remains unproven but if you look at the one of the early Lancet series here blue are  the non-market associated cases, red are market associated cases. You see that these blue cases occur occur pretty early. So again the market may not be the  source but rather a source of amplification. All right so transition dynamics. We're still learning how this virus is transmitted in nature. So in  general coronaviruses and other respiratory viruses have about a two meter range. If you're coughing on someone the zone of maximum danger is  about two meters. It's like when you're on an airplane how many rows you have to be away from someone with influenza before you're at lower risk. Other  viruses other coronaviruses persist in the environment in low in sort of room temperature low humidity conditions for up to four days. What the implications of  that are going to be for this virus and for infection prevention with hospitals is still being determined. Another topic you probably have seen in the news this  idea of the R naught that is the number of secondary cases that arise from a primary infection. So the estimated R naught for this novel coronavirus is  somewhere between one and a half to three. That's still being sorted out and that can change over time as changes in a population with in terms of immunity  and exposure evolve. So for example H1N1 influenza had an R naught of about 1.5. Measles in an immunologically naive population is an R naught of 12 to 18 so  very very contagious. The original SARS from 2003-2004 had an R naught initially about 3.5 but as environmental and policy controls were put into place that  declined very rapidly down to 0.7 and that's part of why that epidemic did not propagate further. All right so there are a couple practical things for critical  care professionals to know about. One is how do we identify what we call a person under investigation or person of interest who is at risk for having this  infection having COVID-19 and determining whether they need further workup. So there's kind of a sliding scale of severity and epidemiologic risk  it will determine the risk. So the highest risk category is going to be someone who has been exposed to a confirmed coronavirus case or has had a  laboratory exposure to the virus itself. Those people can have fever or signs and symptoms of a lower respiratory tract infection. That would then be sufficient  to put them into being a person of interest who would require further evaluation for disease. In someone who has traveled to Hubei province in China  within the last 14 days the case definition is fever and signs and symptoms of a lower respiratory tract illness. And then in someone who's been  simply to mainland China than last 14 days its fevers, signs, symptoms of a lower respiratory tract illness, and illness severity requiring  hospitalization. So the scrutiny goes up the less firm the epidemiologic risk is. And this is continuously updated at CDC and this may potentially evolve over  time as our understanding of the epidemiologic risk factors and it's the nature of this outbreak evolve. So CDC also has a flow sheet and Dr. Evans will  go over this in greater detail but I think it's important enough to take more than one look at. This is something practical for application in the  hospital and this is again available on the CDC website. So a patient who has had travel to China or close contact with someone with a known case of COVID and  has fever or signs of a lower respiratory tract infection, that's a person who we should at least consider this as a possibility with further  assessment of the risks and the and the institution of some initial basic infection control measures that again my colleague will discuss in greater  detail. All right so this is Dr. Peng's paper and again with nothing but gratitude for him for the work that he's doing taking care of hundreds upon  hundreds of patients in in his home city. But just to reiterate the key points of his paper, 26% of his admissions required placement in the intensive care unit and  there was a median time of 10 days between onset of symptoms and development of severe disease requiring ICU admission. With 60% of them having  ARDS, half of them requiring mechanical ventilation and a large and well a substantial amount of patients requiring ECMO and or vasopressors. So their  overall mortality is 4.3% but a lot of their patients remained admitted at the time of that publication and markers of severe disease and risk of death  included lymphopenia and neutropenia that the non-survivors had sustained neutropenia and lymphopenia where survivors tended to have recovery of  their cell lines earlier on. But that 10-day period is important right because I think that may inform what we do for treatment. If this is a late for example  immunologic complication that could affect timing of antiviral therapy much like neuraminidase inhibitors in influenza where their efficacy is  greatest the earlier you start them in relationship to their symptoms. So diagnostics. So if you look at the report on the multiplex PCR in your hospital  there will be a line that will say coronavirus. That test does not detect this virus right. That detects the four conventional human coronaviruses. It  doesn't detect SARS, it doesn't detect MERS and it doesn't detect this. So all diagnostics have to be done in coordination with your local public  health authorities, United States that would be your county public health department or analogous organizations in your in your home of origin. Testing  until very recently was in an entirely at CDC that led to some time delays because of bottlenecks. That is now improving as other laboratories around  the country have access to the primers and probes and can perform this closer. But again all of that does need to be coordinated with county public health. So  when we identify those PUIs that we described before the next step is to call after taking care of the patient calling the county and finding out is  this person suitable for screening should I be sending assays off to one of these reference laboratories. All right so clinical care. There's not a lot to  say about clinical care yet and certainly nothing that I could add to Dr. Pang's presentation. Care is principally supportive and staff safety  when you saw those high numbers of health care workers infected staff safety is always going to be at the top of the list right next to care of the  patient and safety of the patient. So potential interventions that are under at least consideration. Glucocorticoids are used in a large proportion of the  cases in China and in Japan. They are not presently recommended for SARS or  MERS. Whether they will have a role in this disease I think remains to be proven. Certainly if someone has another indication for glucocorticoids like  vasodepressor, vasopressor resistant shock that would be another story. Bacterial and co-infection is reported in about 10% of patients and there may  be a role for empiric antibacterial drugs. Investigational therapies. The most ones gotten the most press is remdesivir which is a viral RNA  polymerase inhibitor under evaluation by Gilead. This is available as an  emergency IND but use of this drug should generally be done in coordination with a registry such as the MEURI framework for the use of  investigational emergency agents. So that we're getting good data and can have an idea of what is the actual use of these drugs. There are a number of randomized  trials that are in development and moving towards approval very rapidly and if you look in clinicaltrials.gov there are 50 RCTs right now for novel  coronavirus and a equivalently large number on the WHO website. There is some interest in Asia looking at lopinavir rotonavir which is an older  antiretroviral drug. Whether that has any benefit remains to be proven. Another thing just to make a push for is the idea of standardized data collection. One  of the challenges in other epidemics was difficulty with comparing different case series. We'd have cohorts out of multiple different institutions around the world  and comparing those cohorts in an organized manner was difficult. So within the Society of Critical Care Medicine, the Resilience Intelligence Network under  the leadership of Dr. Perrin Cobb and a number of people in this room is looking to develop networks to permit the standardized data collection in public  health emergencies. Internationally and under WHO auspices there's an organization called ISARIC, the International Severe Acute  Respiratory Infection Consortium, which has a standardized protocol with standardized case report forms that you can download off the internet and are  available to use only requiring attribution. And that sort of thing will help us make our data easily interpretable between each other. Then  there's other comparable protocols in development. So just for a little bit of perspective, obviously this is a public health emergency. Right now as of this  morning about 72,000 cases of COVID in the world, about 700, a little more than 1,700 deaths. Each one of those deaths a tragedy. In the United States in a  typical influenza season there are 30 million cases of influenza and there are 30,000 deaths. So while this mortality is troublesome, we should remember that  this is flu season. Remember to look out for those ones. Remember to take standard precautions for other comparable respiratory diseases. Right? So in summary  this is a dynamic and rapidly evolving outbreak and the direction this is going to take in terms of our hospitals and in terms of global public health remains to  be determined. Current case fatality rate two to three percent. The impact of mild or subclinical diseases unknown. There are going to be a number of  undiagnosed cases and how that will affect the case fatality rates unknown. Care is primarily supportive. There are some investigational therapies that will  be under study. Priorities for us in our hospitals, early detection, infection prevention, staff safety, and obviously taking care of sick people. Thank you  very much for the opportunity and we'll be available for questions afterwards? Afterwards. Thank you very much.  So again at the end of this second half of the session there'll be time for questions I hope with Dr. Pang available. Our next speaker is Dr.  Laura Evans who will be talking about preparing for COVID-19. Thanks John and thanks to SCCM and all of you for the opportunity to be here. I'd also like to  take the opportunity in the moment to thank Dr. Pang and all his colleagues in Wuhan and in Hubei province for the incredible work that they've done as  well for sharing all of that information for us so that we can help to prepare as this virus potentially spreads to our communities. So with that I'm going to  take, this is not a research presentation, I'm going to take a step back and try and be really practical with some guidance about what steps we might be  able to take now. And I think I'll echo a lot of points that Ryan may have just made in terms of the central nature here of really, really tight adherence to  infection prevention and control is the root of how we're going to approach this. I don't have any financial conflicts to disclose. I do get some salary support  from the National Ebola and other emerging diseases Education and Training Center which is funded by HHS, ASPR, and CDC. So this numbers need to be  updated after today. This is as of before the weekend actually where we had 15 positive cases in the United States just to put that in perspective of what we're  looking at in terms of global numbers and a total of 443 persons under investigation in the United States. Those numbers have clearly changed now with  the repatriation of American citizens particularly from the cruise ships around Japan. So those numbers will be updated later tonight. I wanted to  start with kind of the take-home messages in fact which is I think the keys to how we're going to prepare here. And that is really a central focus on  some of the ideas that Ryan already brought to bear in terms of early identification, prompt isolation of persons under investigation, and then  activating systems to inform public health colleagues and our colleagues in infection prevention and control to mitigate the spread to other patients or  to health care staff within our institutions. I want to talk briefly about the hierarchies of control with a focus a little bit on PPE but  recognizing that PPE is not the central method, not the central way that we control spread of infection. Tight adherence to infection prevention and  controls, how to use PPE resources that are available to you. The first thing I want to call your attention to is what's posted online at the CDC which is a  hospital preparedness checklist. And this is more geared towards institutional preparation but I think us as individuals also can take points from  this home in terms of helping our institutions prepare but also helping our units and ourselves become ready to deal with any emerging pathogen. This  is clearly not COVID-19 specific. This applies to any situation and as Ryan already mentioned as well you can use many of these for responding to  seasonal influenza as well which clearly has a major impact on us every year. So we need to know what our procedures are for rapidly identifying isolating  persons. We need to make sure that we know where our airborne isolation infection rooms are and if they're functioning. So developing a close  relationship with our facilities management folks. We need to understand what we have in terms of PPE availability, supply chain and begin to  develop some contingency plans for what PPE might look like in a shortage of supply. We need to think about what our surge capacity procedures might look  like for not just a single individual patient with COVID-19 but rather a small cluster of patients, a family or even a potential surge of a larger number of  patients. Where would your overflow areas be? How would you apply environmental infection control in those situations? And review procedures for how  we would get specimens to the lab particularly since the diagnostic testing as we already learned is done in the regional public health labs or at  the CDC. So these are not testing confirmatory testing for COVID-19 cannot be done at the hospital level at this time in the U.S. And then we need to  assess the effectiveness of environmental cleaning procedures. Some more highlights from this checklist include ensuring that our health care  staff has been medically cleared and fit tested and knows how to safely use an N95 respirator. Now is a great time to provide education, refresher training to  our staff regarding the diagnostics, some of the clinical information as we're learning how this disease behaves and evolves, how to get specimens, what  specimens are necessary for testing, how to appropriately use PPE and where these patients would go within our hospitals. And then critically around this we also  need to log who's involved in the care of these patients so that we can enroll them in symptom monitoring programs as needed. It's also really important to  think at now how you would approach visitors for patients with COVID-19 in your institution. What kind of restrictions might you impose upon  visitors in that setting? We saw this, I broke this down, this is sort of that first critical step. If you think about where our vulnerabilities are in the  health care system, it's predominantly most likely to be before these patients are identified as having COVID-19 or seasonal influenza or any other readily  transmissible pathogen. So facilitating processes to make sure that we can identify these patients promptly is clearly our first and probably our best  line of defense. And this is directly from the CDC again in terms of the schematic that they represent for the COVID-19. I would actually recommend that  perhaps we've sort of flipped the script here and because the CDC algorithm starts with travel first and we just have to remember back to Ryan's slides  about the impact of seasonal influenza and think perhaps what we should be doing at sites of entry is not talking about travel as the first question but  rather fever or symptoms of lower respiratory illness is the first question and use that as the opportunity to implement risk mitigation at that  stage. So anybody with a fever, cough, respiratory symptoms should be given a mask to try to control transmission within our facility sort of regardless  of diagnosis. That also gives us the time then to safely pursue questions about travel. And these are just some examples from different medical centers and  different systems around the U.S. in terms of signage that can help folks to identify patients early. And you can see at least at the Emory system here, John  is probably quite familiar with this, is that it starts with fever and face mask and then it proceeds to travel. And I think that's a really good way to  approach this in general from an infection prevention and control standpoint. So if I identify somebody who has some epidemiologic risk, that  doesn't mean they're a person under investigation. It means they have some epidemiologic risk and some symptoms. I'm going to put that person in an isolation  environment so that I can take the time then to ask further questions and determine if they may meet the case definition to be a person under  investigation. So I need to have that area identified where I'm going to put that person. I'm going to then proceed with my evaluation wearing appropriate  PPE and we'll talk about what that might look like. Following those additional questions and your individual hospital procedures may vary as to what  point you involve your local infection prevention and control team. But following that, if you think a person meets the case definition, you're clearly  going to activate notification of your local public health jurisdiction and decide if testing for COVID-19 is indicated. We saw this already. I just  want to bring it home that these criteria may change over time. They've already changed in the last few weeks as the outbreak has spread from Wuhan City  to Hubei province in a larger sense and then to other areas within mainland China. So keep your eye on these sites on the CDC because the definitions of the  case definitions of PUI are likely to evolve as if we develop significant community-based transmissions in sites outside of mainland China. Hierarchy of  controls. Some of you may or may not be familiar with this. This is sort of a NIOSH pyramid for this that describes how we can mitigate risk from it. And  many of these don't readily apply to us in the healthcare setting, but they're I think a good construct for us to think through how we can help protect  patients and ourselves from that. And I'll call your attention that PPE is down at the bottom here. So it's not the first thing we do in terms of mitigating  risk. We want to think of can we eliminate the risk? Can we substitute it? But I think when we're talking about healthcare, we're really talking about  these three triangles down at the bottom. Engineering controls. So that would be examples of our airborne isolation rooms. Administrative controls. So having safe  work practices already in place with people properly trained, knowing who's going to go in the room, and making sure that untrained personnel do not enter  the room. And then using PPE of course as an additional supplement. Where we are with PPE guidance right now is we're seeing a lot of different images come  across certainly media as to what people are wearing for PPE. CDC PPE guidance is quite clear. And they have some baseline language about that.  That we should be receiving training and demonstrate how to use it, what PPE is necessary, and most pertinently I think how to properly don and doff PPE to  prevent contaminating ourselves during that process. What the limitations may be are as well. Clearly PPE that's reused needs to be properly disinfected before  reuse. And we should all have policies and procedures in place to ensure that we have safe practices for that. Current CDC guidance is gloves, a standard  disposable isolation gown, a fit-tested N95 respirator or equivalent, and eye protection in the form of a face shield or equivalent. That can be goggles, that  can be, for example, I just have a standard face shield here. So that is and when you think back to what Ryan just told us about what is known about the  mode of transmission that this appears to be droplet based transmission, this is appropriate PPE. So this is current CDC guidance. CDC also offers some  guidance on proper donning and doffing and I would encourage you to review that. I would also encourage you if you want a video of it, if you're more of a visual  learner, NETEK, the National Ebola Training and Education Center, has produced a PPE donning and doffing video here. It's available free on YouTube. So  if you search YouTube for NETEK COVID-19 PPE, it'll give you a video for how to don and doff this recommended CDC PPE for COVID-19, as well as I think take you  through some of the safe behaviors in PPE. From that you'll notice everything starts and ends with hand hygiene, which should be our normal practice anyway in  the ICU. Aerosol generating procedures I think provoke, you know, a certain degree of anxiety and in people when we think about respiratory transmitted pathogens.  And so the guidance is to basically be cautious about this. And if you're doing something like bag valve mass ventilating a patient, intubating them,  developing nebulizers, inducing sputum, I think there's an unanswered question about the use of non-invasive ventilation in this patient population  and what that may do in terms of infection control. But there are certain things that you can do to mitigate that. You can put HEPA filters on the expired  heirloom. You can do that to a bag valve mass ventilator. You can obviously do that on a standard mechanical ventilator. If you're performing procedures like  this, they should take place in an airborne isolation room. And you should limit the number of personnel to only those numbers that are absolutely  necessary to do that. And obviously the personnel who are doing that should be trained in proper PPE. I want to spend just the last couple minutes sort of  drilling down on some of this because it's not, people focus a lot on PPE talking about equipment. And it's not just about equipment and supplies. It's  about how we use them and how we behave in them. So I really want to drill down on this and we'll kind of go around this clockwise here. So the way we behave in  PPE is one of the central elements of how we keep ourselves and our colleagues safe and our other patients safe. So trying, reducing the environmental  bioburden. So trying to keep the room environment clean. We don't know exactly how long this virus may persist in the environment. So using things like while  we're in the room, taking the opportunity to disinfect and wipe down high-touch surfaces may mitigate the environmental bioburden. Trying to keep our PPE  relatively clean. If I'm doffing cleaner PPE, I'm less likely to contaminate myself during the doffing process than if I'm doffing PPE that's grossly soiled.  Obviously don't touch your skin, your hair, your eyewear. We see a lot of like reach up under the face mask to adjust your N95. That's a big no-no. Again, use  additional barriers when you're going into an area that you think may be highly contaminated or doing something that you think may result in gross  contamination of your PPE. Know what your breach protocols may be. Don't  hesitate to clean and disinfect or change your gloves. You can do, you can use alcohol-based hand rub on exam gloves to disinfect them. You can do that.  You can also change them during that. Be aware of your condition. Don't get yourself into trouble. So if you're having trouble with an N95, feeling hot,  feeling excessively sweaty, that's the time to call for help and get out of that room so that you don't have to be extricated from that. If you're using a  trained observer, which is not required by CDC in this situation, if you opt to use a trained observer and you follow the instructions from your trained  observer, and obviously check your PPE for contamination before doffing. So to summarize again, the situation is rapidly evolving. There's, I think, very  substantial uncertainty about what the impact may look like in terms of we think that Ryan gave us some great information about current data about  case fatality ratios, but I think we also know that if we're in the United States, if we experience a significant influx of seriously ill patients, that  will perhaps put stress and strain on our healthcare delivery system. None of us have tons of extra capacity in our emergency departments, inpatient units,  or ICUs, and I think we need to be prepared for that. We need to know what our process is for identify, isolate, and inform, and we need to be testing that  now. Now is also the time to review and train on the proper use of PPE. This is the window of opportunity to do that. It's a good idea to do in general, and  this is a great opportunity to do that. We talked a little bit about already about preparing to make sure that we have the capacity to collect clinical  data, and I think that's one of the things that Dr. Peng illustrated to us really nicely, is that his ability to collect and report clinical data has  informed us, and so we need to be able to do that as well, and Ryan gave us some resources to do that in terms of there are already pre-prepared case report  forms. You can submit and get those through your IRB now, so that if you start getting patients with COVID-19, you can start pulling data on those  patients, and again, the keys to preparedness across any event is always communication, collaboration, and coordination, and I will stop there by  just putting some resources up there. This CDC website, of course, the WHO website, this Johns Hopkins Center for Health Security is publishing a  fantastic daily newspaper. We have some neat resources on the NETEC repository, and then WHO just released an update to their severe acute respiratory infection  course, which is a fairly lengthy online course. It has about 12 modules, takes about 10 hours to complete. It's more geared towards low and middle-income  countries, but I think still a great resource for us, and I'll stop there.  So, while it's at 515, we do have this room for a little bit longer, and we have both Dr. Maves and Dr. Evans to answer questions. Unfortunately, we had a  technological kerfuffle with Dr. Peng, and he will not be available for answering questions.  This is for Dr. Maves. Tim Buckman, you mentioned that there is a significant, or at least 10%, estimate of super infection with this viral pneumonia. Is that a low  estimate, a high estimate? And the reason I ask is that in the data that Dr. Peng showed, it appeared that among the survivors, there was actually a  significant rise in the total white blood count. That is, there was a significant neutrophil leukocytosis. Would you comment?  So, actually, in Dr. Peng's series, it was less than 10%. He isolated a handful of cases of Acinetobacter and some likely commensal  yeast. So, his experience is less than that. The 10% comes from some other series, and those are among hospitalized patients, not from the full 77,000 or 72,000,  I'm sorry, number of cases. So, whether a post-viral bacterial pneumonia is a major part of the immunopathogenesis of this, I think is completely unknown, and  that 10% is entirely speculative off of a few small series. Again, Dr. Peng's experience doesn't reflect that, and I think it is premature to say at this  point. But it's a great question, and that's one of the, you know, like flu and metapneumovirus would be two major comparably severe viruses that would have a similar syndrome.  Wow, he's a crowd.  Thanks very much. Very timely and informative. Any information about the effect in children? I think we don't know. Certainly the number of cases reported  in children appears to be low. We have to remember that the case definition that was publicly reported has changed in recent days.  And previously it was a laboratory confirmed diagnosis. And so you wonder about underreporting of milder disease. So if the clinical disease is milder in children  than in adults, they may not have been tested. I think we'll have to keep an eye on that now that there's also a clinical definition  for case reporting, particularly from Hubei province, to see how that changes. But I think we don't know. There does appear to be different signal in children rather than adults.  Do you want anything? No, that was great. Jeff Bellingham, two questions. First one is if you could comment on the story where some patients had test negative potentially  over a number of times and then positive. And the second one is the choice of 14 days and some of the stories suggesting that might be out to 20 or 24 days before you can move  from being asymptomatic to symptomatic. In regards to the negative testing, you know, there was, this was reported in the media in my home city  in San Diego, cases of initially negative PUIs were later found to be positive. That appeared to be the result of an administrative issue rather than a technical issue.  Certainly goes to reinforce the importance of administrative controls in addition to technical controls. But any test has sensitivity and specificity, even anything  as remarkably sensitive as PCR. In terms of the 14 days, it seems that the median time between exposure and clinical symptoms is about five to six days.  So the 14 days is actually at the longer end. I think there will certainly be handfuls of patients who shed for more prolonged periods of time or have a longer period.  Like anything, it's a bell-shaped curve and there's going to be a tail at the other end. But I think with the 14-day period, that's keeping us  within a few confidence intervals of the great majority of cases, at least as described thus far.  When I read like in the news about the cardiovascular complication that there is a specific predilection of the virus to coronary arteries, so what are the clinical implications?  Do we expect like STEMIs, cardiac ischemia, and what are the treatment implications? So I don't think we have an honest answer for you about that, sir.  I will say that influenza is also clearly associated with myocardial infarction and myocarditis as well. You know, it's called hemagglutinin for a reason.  It's a prothrombotic state. Whether that is also true of this coronavirus I think is unknown.  I think I heard you say that there's about 80% genetic identity with the previous SARS or SARS type 1. Correct, sir. For which there was a vaccine developed, I understand,  but was never used. Is there any thought that that old vaccine would be potentially effective against this new virus? So kind of a two-part answer. So sort of a two-part answer.  So number one is I am certain folks are thinking about that. I'm not directly involved in that, and I can't speculate. I can say, and this is in the press, that Dr. Fauci  at NIAID anticipates having phase 1 human trials of an investigational vaccine likely using, and I'll ask Dr. Evans to correct me if I'm wrong,  the RNA vaccine platform that they've developed and tested for Zika and some arboviruses previously. So that should be hopefully in the next few months. In terms of cross immunity,  there is some interesting literature on the effects of cross immunity between different coronaviruses. There doesn't appear to be cross immunity between, say,  like the other four routine human coronaviruses, right? So I don't think we would necessarily anticipate that in this case.  And there is this interesting signal from animal coronaviruses that secondary infections may actually be more severe, kind of analogous to dengue. So I would be hesitant to assume  that an existing SARS vaccine would be useful in this case, although it certainly does provide a model that vaccine development could use going forward  for developing a specific vaccine.  So please join me in thanking Dr. Maves and Dr. Evans for their timely presentation. And we'd like to thank Dr. Peng in absentia for his wonderful summary and thank all of you  for sticking around well past the two-hour mark. Have a fine afternoon.

Video Summary

In this video, Dr. Ryan Maves and Dr. Laura Evans discuss the current state of the 2019 novel coronavirus (COVID-19) outbreak and provide guidance on how to respond to it. They stress the importance of early identification, isolation, and notification of potential cases, as well as adherence to infection prevention and control measures. Proper use of personal protective equipment (PPE) is emphasized, and detailed instructions on how to don and doff PPE are provided. The importance of standardized data collection and collaboration among healthcare professionals is also highlighted in order to effectively manage the outbreak. Dr. Maves gives an overview of the virus and its transmission dynamics, while Dr. Evans discusses the hierarchy of controls and offers practical advice for hospitals and healthcare workers. Both speakers acknowledge that the situation is rapidly changing and urge healthcare professionals to stay informed with the latest guidelines and recommendations from reputable organizations such as the CDC and WHO.

Asset Subtitle

Quality and Patient Safety, Research, 2020

Asset Caption

The Society of Critical Care Medicine's Critical Care Congress features internationally renowned faculty and content sessions highlighting the most up-to-date, evidence-based developments in critical care medicine. This is a presentation from the 2020 Critical Care Congress held February 16-19, 2020, in Orlando, Florida, USA.

Learning Objectives:

Outline the increasing incidence and impact of sepsis 2012-2018
Identify uses of electronic data in epidemiologic surveillance of pediatric sepsis
Describe the findings of the HYPE trail and the INTEREST study
Identify state of the art and guidance for vaping ARDS
Relate what is known about the 2019 Novel Coronavirus (COVID-19) to date and discuss preparations for the spread of the virus

The Trajectories of Sepsis, 2012-2018

Steven Q. Simpson

Clinical Characteristics, Major Morbidity, and Mortality in Trauma-Related Pediatric Acute Respiratory Distress Syndrome

Alison Nair

Identification of Pediatric Sepsis for Epidemiologic Surveillance Using Electronic Clinical Data

Scott L. Weiss

Effect of Intravenous Interferon Beta-1a on Death and Days Free from Mechanical Ventilation Among Patients with Moderate to Severe Acute Respiratory Distress Syndrome: A Randomized Clinical Trial (the INTEREST Study)

Geoffrey J. Bellingan

Effect of Using a Machine-Learning Derived Algorithm Versus Standard Care on Intraoperative Hypotension During Elective Non-Cardiac Surgery: the HYPE Randomized Clinical Trial

Marije Wijnberge

Data Driven Analysis Reveals Shared Patterns in Transcriptome Responses and Immune Cell Compositions Associated with Distinct Mortality Rates Across Differing Aetiologies of Critical Illness

Zsolt Zador

Vaping Associated Respiratory Distress Syndromes (VARDS) State of the Art and Guidance for Clinicians

Craig M. Lilly

Coronavirus - COVID-19

Zhiyong Peng

2019 Novel Coronavirus: What We Know So Far

Ryan Maves

Preparing for COVID-19

Laura Evans

Meta Tag

Content Type Presentation

Knowledge Area Quality and Patient Safety

Knowledge Area Research

Membership Level Select

Membership Level Professional

Membership Level Associate

Tag Mortality

Tag Outcomes Research

Year 2020

Keywords

2019 novel coronavirus

COVID-19 outbreak

early identification

isolation

notification

infection prevention

control measures

personal protective equipment

don and doff PPE

standardized data collection

collaboration

transmission dynamics

hierarchy of controls

healthcare professionals

guidelines

recommendations

	Society of Critical Care Medicine 500 Midway Drive Mount Prospect, IL 60056 USA Phone: +1 847 827-6888 Fax: +1 847 439-7226 Email: support@sccm.org	Contact Us About SCCM Newsroom Advertising & Sponsorship DONATE	MySCCM LearnICU Patients & Families Surviving Sepsis Campaign Critical Care Societies Collaborative	GET OUR NEWSLETTER
© Society of Critical Care Medicine. All rights reserved. \| Privacy Statement \| Terms & Conditions The Society of Critical Care Medicine, SCCM, and Critical Care Congress are registered trademarks of the Society of Critical Care Medicine.