First of all, thank you so much for joining us for this session. Aloha to everybody. And we know that it's a beautiful day outside, so we'll keep it short and sweet. But I think this is a problem that needs discussion. My name is Siddharth Dugar. I'm a staff physician at Cleveland Clinic. The only disclosure I have is I have a grant from Chest Cenocyte and has nothing to do with hospital-acquired sepsis. So what is the goal of this lecture? The goal is just to go over the definition, make sure that we know how to define hospital-acquired sepsis, what are the outcomes associated with it, and why are we even discussing this topic? Because as we will go over the lecture, you will see that it uses a lot of resources, and the outcomes are still pretty bad. So I think this slide, everybody in this room has gone over it multiple times, but just to reiterate, the sepsis is one of the biggest challenges that we face in our 21st century, especially working in pulmonary critical care or critical care. We have close to 1.7 million people who develop sepsis annually in this country, close to 350,000 deaths of people going to hospice annually. But the cost, it costs us $57 billion annually to take care of patients with sepsis. So what is a CDC definition of hospital-acquired sepsis? So the way they define it is, if you have a patient who has been in the hospital for three days, did not have sepsis on admission, on day three or later, if you are drawing blood cultures, giving them antibiotics, and they have signs of organ dysfunction, that's how they define hospital-acquired sepsis. Other literature have used 48-hour time zone from hospital admission to development of sepsis, with the sepsis definition same as sepsis 3 or CDC guidelines. Now why are we even talking about this? So just to put it in perspective, all the patients that we see with sepsis in the hospital, one out of those eight people with sepsis has hospital-acquired sepsis. And one out of the three deaths that we see because of sepsis is from hospital-acquired sepsis. So it's a huge problem. So if we say that, hey, there are close to 1.7 million admissions with sepsis, and you're like, one out of eight is hospital-acquired sepsis, that is a big problem. And their hospital mortality, when we look at it, has an odds ratio of 2.5. So if you don't have infection on admission, but you develop infection during hospitalization, your chances of death is 2.5 times higher than if you presented to the ED with sepsis. And length of stay, almost three times or 2.5 times to what you will have if you present it to the hospital with sepsis. And this slide is very interesting. This is a study that looked at Cerner Healthcare System data. It had more than 200,000 patients that they were looking at. And what is the interesting part about this is, if you can see, the people don't present the same way. The black line is the people with hospital-onset sepsis, and the bar is the one with community-onset sepsis. The lactate, which we use so commonly for diagnosing sepsis, that lactate never crossed to. Only 30% of the people actually had lactate elevation, compared to close to 40% in community-acquired sepsis. Their creatinine is also not that high. But their use of vasopressor, almost 50% of the people who develop hospital-acquired sepsis will require vasopressors, compared to only 30% for people presenting with sepsis. And even the use of mechanical ventilation is way higher compared to community-acquired sepsis. They also found the people who developed healthcare-associated sepsis, they were actually younger, by two to three years median. They had much higher SOFA score. The amount of them requiring ICU admission was 60%, compared to only 40% for people with community-acquired sepsis. And the mortality, two times higher, 33% versus 17%. So these are the younger group of patients who get more sicker, they use more resources, they come to the ICU more, and their death is still very high. So this is another study that was published in 2015. Again, that was more of a nationalized data set. This is a retrospective analysis, but still included more than 300,000 patients with severe sepsis. Again, not looking at sepsis, but severe sepsis here. So they had three groups. One is community-acquired severe sepsis, which are people presenting with an infection and having severe sepsis on admission. Healthcare-associated severe sepsis is an interesting one. These are the people who are coming from nursing home, who are getting dialysis, who live in a long-term facility, and they have infection on admission, they just don't have, they don't meet the sepsis criteria, but they later develop sepsis. The one that we are interested in is the hospital-acquired severe sepsis. These are the people who don't come with infection on admission, but two days later, or three days later, they develop sepsis, or severe sepsis. Now interesting part is, we think that these two will be closer to each other, while this will be a separate group altogether, but look at the cost. Even if you come with infection and you develop sepsis later on, the cost is not much higher compared to community-acquired sepsis. But if you develop sepsis in the hospital, which was not present on admission, the cost is almost four times higher. The length of stay, again, two to three times higher. The hospital mortality, yeah, it's bad if you come with infection, if you're living in a nursing home, we all understand that, but the mortality of people who develop infection in the hospital is almost close to three times higher than if you have present to the ED with infection, or sepsis. Again, the adjusted odds ratio is 3.2, suggesting, compared to presenting to the hospital with sepsis, if you develop sepsis in the hospital, your chances of death is almost three and a half times higher. That's a pretty significant increase in mortality. Now we took a different approach. This is a study that we published, looking at our data set, since this was close to 2,500 patients, and we did not look at sepsis. We looked at septic shock because the time was more clearly defined, and what we wanted to see is, is the mortality changing at any time period? Because then we know, what is the pathophysiology of them having a higher mortality? And interestingly, what we found is, there is no transition point that we could find that says that, hey, at this point, you transition from community-acquired septic shock to hospital-acquired septic shock. What we found is, the longer it took for you to develop septic shock from hospital admission, the mortality just kept getting bigger and higher and higher. And the first time we see the difference is 7.4 hours, suggesting that there is not one transition point. Whatever we are struggling with, which is early identification, early use of vasopressors, getting them early antibiotics, all the comorbidities that we are dealing with, they become additive as the patient stays in the hospital longer. So it's not that if you are thinking about, how do I control it, when should I put surveillance? You should be surveying them as soon as they get into the hospital system. So to summarize it, this is a systematic review and meta-analysis that looked at 51 studies, close to 14,000 patients. And just to put it in perspective, one-fourth of the patients who have sepsis in your hospital actually has hospital-acquired sepsis. Mortality, close to 50% of those patients with hospital-acquired sepsis need admission to the ICU, and their mortality is close to 52.3%. So again, a lot of patients with sepsis actually has hospital-acquired sepsis. The prevalence is increasing. They require ICU a lot, they require resources a lot, and their mortality is still high. And which other patients should we be worried about? Again, everybody thinks it's a patient with dementia, heart failure, but the patient that really caught my attention were the patients who have stroke. These are the people who have a high risk of developing hospital-acquired sepsis. And liver disease. If you have a patient who has moderate to severe liver disease, those are the patient very high risk of developing hospital-acquired infection. Again, no doubt, the use of mechanical ventilation in hospital-acquired sepsis, again, is 21% compared to 17%. In this patient population, compared to 7%, higher use of vasopressors, and still higher mortality. So this presentation was just to tell you why are we discussing about this. It's a big problem that is getting bigger and bigger and taking an epidemic form. And we will have Dr. Dougal first to go over what are the challenges on identifying them in our hospital system. Thank you so much. Thank you. Thanks, Ed. All right, so I'm going to be talking about the challenges associated with the identification of hospital-acquired sepsis. So my name is Avi Dougal, I'm a vice-sheriff of critical care at Cleveland Clinic. I work with Seth Dougal, and we've been kind of working on these. These are my disclosures, nothing to do with my talk at this point in time. So the objectives for the talk today are going to be, you know, we're going to discuss why is it difficult to really identify hospital-acquired sepsis, discuss the diagnostic challenges with this entity, and then go into the fact of automated sepsis alerts and what are the problems with them. And we'll end with machine learning and artificial intelligence, and is that the way to go for us as we move forward in things. So I guess the first question for us is, is it really difficult to identify sepsis in a hospitalized patient? So let's do a quick case. A 58-year-old, underlying history of diabetes, coronary artery disease, AKI, on CKD, stage three. The patient's got a heart rate of 105, respiratory rate of 18, temp of 36.8, white count's gone up from 11 to 15, lactate's 2.8 at the time you evaluate this person. Is this sepsis? Who knows, right? Now if I tell you this is day six of his hospital stay, the patient had received antibiotics on presentation for a presumed UTI that was stopped after 72 hours, and they've been off of antibiotics for two days and have deteriorated now. Is it sepsis? Okay. Echocardiography reveals that there's a new decrease in ejection fraction. It's 30% from a baseline of 50, and you're seeing some regional wall motion abnormalities. Is it sepsis? The patient's four liters positive since they came to the hospital. Sepsis? The creatinine went from 2.1 at admission to 4.6 in the last 96 hours. Is it sepsis? Could be, right? So we see this all the time, right? It could be sepsis, it could be decompensated heart failure, it could be worsening AKI with volume overload, right? All of these situations will cause this person to develop signs of not really doing well. So this is the problem when we think about diagnosing hospital-acquired sepsis, right? So until and unless we have a way of really distinguishing these factors in a patient, a lot of the downstream effects are always going to be lacking in terms of really having a good answer about things. So why is it so difficult, right? So if you think about it, there's a bunch of factors that you have to be mindful of. Significant confounders that usually will mask things that you would use to identify sepsis. Tachycardia, tachypnea, multiple things that will cause that in a hospitalized patient. The hemodynamics can be very, very different in these patients, right? So Sid just talked about the fact a lot of times you're not going to see super high lactates. Why? Because they're not coming in preload dependent. You have most of the times optimized their volume status to some degree. Your associated organ failures and the trajectories of how these organ failures are progressing is very different. Like this is not someone who's coming in from home and seeing like a frank multi-organ failure at the time of presentation. A lot of times this is going to be a lot more indolent and difficult to really parse out. The severity of illness scores that we usually use, right, are all validated really at the time of admission. So when we really look at the downstream effect of someone who's been in for seven, eight days, it becomes very difficult for us to really apply those factors. And then the big factor is microorganisms themselves, right? Like when you're saying that I have a patient who is very sick, right, the fact of multi-drug resistant organisms, the fact of, you know, indolent slow-growing organisms becomes realities for us. And then those are factors that we have to be mindful of as we're thinking about these patients, right? So if you think about like, you know, most of the literature on sepsis, septic shock is driven by the surviving sepsis guidelines, right? Like so I love this study from Auden, this is like 2015 or so. Basically what they said was, you know, if you were to think about the recommendations that come from the surviving sepsis guidelines, how many of those recommendations are really focused towards patients who are on wards, right? Like, you know, these are not coming from the ED. So this first column is no ward patients in any supporting evidence. And what you can see is most of the things that we really take for granted in terms of sepsis care were not involving ward patients, right? These recommendations come from patients who have come into the ED or there's been like things that have been looked at patients who directly come into ICUs, right? But the usual patients are sitting up on the floor, gets decompensated and comes back down. We really don't know because we don't have any recommendations. So I looked at the latest sepsis guidelines to really see how they improved. So the word ward is mentioned four times in the entire guidelines. The hospital acquired ward is mentioned 96 times, okay? Out of those 96 times, more than 90% of the times it is in the context of hospital acquired infection. All they're talking about is, hey, be mindful of the bugs being different, that's it, right? So this is a data-free zone really in terms of when we look at guidelines, right? So a lot of the things that we have looked at and tried to translate have been only studied in that initial part of sepsis and we need to be very mindful of that, right? So why is it important, right? So this is a very interesting study where they basically looked at patients that came in into the inpatient wards or ED admissions that were put on the wards. And they followed these patients consistently over a period of time to really see how many of these patients develop sepsis. So this basically is patients who were alive versus patients who were dead. What they basically found out was that out of these patients, if they were to take the definition of sepsis based on either sepsis one criteria or sepsis three criteria, there is a very minimal overlap in terms of how sepsis would be looked at if you were using one or the other criteria. So the three boxes are basically looking at SOFA score, looking at QSOFA score, looking at SIRS as an aspect, right? And they basically, those are the key factors that were being used to define sepsis one and sepsis three at that point of time. Only 17% of patients who were alive met both sepsis one and sepsis three criteria, okay? 15% of the patients met neither. Patients who died did better in terms of, you know, only 1% met neither. So yes, if you're really sick enough, like, you know, these will capture. But again, even patients who died, only 22% had met criteria for both sepsis definitions, right? That becomes really important. Why? Because Sid talked about the fact that we need triggers, we need to think about triggers, right? So if you're utilizing these factors as your triggers for your automated alerts, right, there is a very big possibility that you're gonna be not capturing a good cohort of these patients, right? And that becomes a very important factor to think about, right? So when we look at the factor of diagnostic performance in these situations, so for me, like, you know, a lot of times people will spend a lot of time on sensitivities and specificities. That's not really the key, right? Because I can really have something that's very sensitive because I'm capturing everything. I can have something that's sensitive to capture things. I can have something very specific and I can really make that specificity high. But that's not how real life works, right? So for me to really say this is a test that's going to give me actionable knowledge and not just give me fatigue because of multiple firings of BPAs, I need a good positive predictive value. And what I want you guys to look at is that every diagnostic criteria that is used to look at sepsis in the hospitalized patients have a very, very low positive predictive value, right? So what that really means is that you're going to have a lot of fires in terms of patients who are not going to have the disease process, right? So it becomes very difficult to maintain those factors for you guys. So when we think about screening for these patients, right, like there is two factors to think about. It could either be a manual abstraction for these patients or it could be an automated abstraction for these patients, right? But what I will tell you is that if you look at both manual and looking at automated aspects of this thing, the average positive predictive value that you get is between about 30 to 35% and it ranges between 20 and 56%, right? So what, in a best case scenario, half of the alerts that you're going to get are not going to be related to sepsis, right? Why is it important? So this is a very interesting study. It looks at the early identification of sepsis in wards by nurses, right? So basically the carry home message was that if you're using nurses to really do this, you will be looking at sepsis recognition in a much timely manner and it will have downstream effects potentially for the care that these patients are getting. But it's a fascinating thing and I want you guys to look at this, like the things that I've highlighted here. The mean patient load across all staff registered nurses is 0.24, right? So you have up to five nurses taking care of a one patient. Daily dialogue between the ward physician and the ICU doctors was done to make sure if there was a possible need for ICU transfer. And it's 124 bed hospital with 15,000 admissions, right? Now take a thousand bed hospital, take that same nurse, right? All of a sudden the load's gone up significantly higher, right? So the key factor for us to remember is that these are resource intensive things, right? So you can have some very good results if you've got dedicated people doing this. How many hospitals here are going to allow you to have 10 nurses looking at just best practice alerts and not do anything else, right? That's just not going to happen. So this is a nice systematic review that looked at, you know, if we were to just look at automated like electronic sepsis alerts, what's the data on that, right? And I think it's really kind of just captures the same thing. It may improve the processes of care, but has very poor positive predictive value. This is a different systematic review that came out around the same time. And they basically looked at what were the key findings from all these key automated stuff. And what I will tell you is if you read the first thing, no difference, no difference, increases receiving some degree of fluids, right? Increase in lactate collection. So these are not hard outcomes that we're changing. You know, you will change some processes of care, but you're not going to make a lot of difference if you're just using automated, like, you know, sepsis alerts. So going back to the good positive predictive value, we can look at this, like, you know, very poor positive predictive value for most studies. The AMLIN study has a very high positive predictive value. But when you look at the study, like, they were just selecting these patients very, very specifically. And that's the only reason they have that. Now, when we do have these automated alerts, right, the next question becomes, does it change things for these patients? We've already established, like, you know, these patients are not being, like, there's a lot of, like, false, like, positives that are coming through. So does it change the process of care? The answer is yes. Like, you know, you will change some things, like, such as initiation of fluids. You'll give antibiotics a little bit more, like, faster. But when we start looking at, like, you know, does it change the overall heart outcomes? Does it change the number of patients that are coming to the ICU and timing of change? Does it change mortality? There are no studies that have really shown that. So I think this is a really good study because, you know, once you look at these, then the question really becomes, hey, listen, the process of care changed whenever we looked at these systematic reviews. Now, the next logical step really is, let's do an RCT. Let's do an RCT and show that these processes of care are changing. So this is a study from the UK, where they, no, sorry, this is a study from Stanford, where they basically developed a QI project to look at, you know, how, like, you know, the processes of care for, like, water cord sepsis changes in a randomized trial. And what you can see is that there was no change when this was studied in a randomized fashion. Similarly, this is a study, it's an older study from Hooper et al., like, about, like, almost a decade old right now. When they looked at the same question in the ICU to say, you know what, do we do better in terms of processes of care if we are standardizing best practice alerts? Again, there was no difference in these situations. So the next logical step becomes, hey, listen, we are not very good at doing this, right? So in today's day and age, everyone's going to be like, you know what, AI, machine learning, that's your friend, it's going to make things easier, and then we're going to move forward with this. So this study basically looked at that question, it's a randomized control trial using machine learning algorithms to see. And what they did was they looked at machine learning algorithms and compared it to the usual triggers that we use. And basically the answer was, hey, listen, machine learning algorithms are amazing. Look at this. My area under the curve is 95, like, you know, I'm not going to miss a lot of these patients. The sensitivity and specificity is pretty decent. Much better than all our traditional things. And look, like, you know, when we did this, the length of stay for these patients dropped, like, you know, both for the ICU and the hospital. And you know what, we've changed the hospital mortality for these patients. So great, like, you know, machine learning is the way to go. But again, I will always ask you guys, when you study these, go to the methods and see exactly what was done. A machine learning algorithm was developed, but in that case, when the machine learning algorithm fired, a phone call was placed to the charge nurse on duty. The charge nurse then followed the UCSF standard sepsis evaluation and intervention process, went to the bedside nurse, and went over the triggers for that. And basically, the trial was designed to demonstrate the superiority of using an algorithmic predictor relative to the hospital's current HR. But this is not, like, you know, electronic health records versus machine learning. Why? Because it's machine learning plus a senior provider going and now talking about this with the bedside nurse again, right? So when we look at the literature around this, like, you know, these are, this is a nice systematic review that really talks about all the different kinds of prediction models that have been developed using machine learning. If you look, there's a common thread there, most of it is to mimic data. And basically, it shows that, like, you know, they did not do a PPV for this, but what they did was a positive likelihood ratio, thank you, sir, a positive likelihood ratio. Now, a positive likelihood ratio of 3, 1, 3, great, it's all positive, right? But in clinical medicine, what we say is a positive likelihood ratio is clinically significant if it's closer to 10, all right? So what this is really saying is, eh, it's not really going to change things, you're going to have a lot of false fires there too. So what are the take home messages here? Like hospital acquired sepsis is definitely challenging to diagnose, existing diagnostic interventions are time and resource intensive, right? Automated alerts are always going to be having a low diagnostic yield. Machine learning and artificial intelligence are possible solutions, but I don't think they can exist in isolation. I think what we need is human interactions with these to really have a good system to parse out the complexity of the disease process. Thank you very much. Excellent lecture, Dr. Dougal. Next we have Dr. Namita Jayaprakash. She comes to us from Henry Ford System, and she will be talking about sepsis one bundle. Everybody thinks about it. Does it apply to hospital acquired sepsis as well? Thank you so much, Namita. Thank you. As you heard, I'm Namita Jayaprakash. I'm an emergency medicine and medical intensivist at Henry Ford Health, and I also serve as the Henry Ford Health System's sepsis program lead. So hospital acquired sepsis and sepsis in general and applying the sepsis bundles is a lot of the work that I do from that quality perspective. I'm a researcher as well, and one of the things I didn't put on there is I am the site PI for an NIH-based implementation trial for sepsis, but it's very much focused on the emergency department and not the inpatient setting. So I'm gonna review the conceptual drivers of the sepsis bundle here. I'm gonna examine some of the challenges of implementing the bundles when it comes to hospital-acquired sepsis, and we'll illustrate the applicability of these bundles to the hospitalized patient. There has been a large evolution in terms of sepsis care over the last 20 years, since the publication of EGDT, the introduction of the surviving sepsis guidelines. There's been a huge change in terms of protocols, policies, regulatory, and most recently, the CDC has published recommendations for core elements for hospital-based sepsis programs. And a lot of the focus here has been around introducing the concept of bundled care. So in 2004, the Surviving Sepsis Campaign first introduced that concept of bundles, and in their most recent iteration in 2021, they focus on the importance of timely interventions here, and in fact recommend that once you recognize sepsis, really start thinking about trying to implement these bundles within an hour. The Surviving Sepsis Campaign bundles include measuring lactate, obtaining blood cultures, starting broad-spectrum antimicrobials, rapid administration of a crystalloid fluid bolus for signs of hypoperfusion, whether that's a lactate greater than or equal to four or hypotension, and applying vasopressors for persisting hypotension. Another version of these sepsis bundles are the CEP1 bundles, sponsored by CMS. But again, the concept here is really that early recognition and treatment. So the CEP1 bundle promotes recognizing two or more SIRS criteria for that host response, but then pausing and thinking, is this from sepsis? Is that the driver of that SIRS response? So identifying and considering infection. Determining if there's an end-organ failure present, making that diagnosis, labeling it a severe sepsis or septic shock, and then moving forward to implement the treatments. Within three hours, obtaining an initial lactate level, obtaining blood cultures, thinking about broad-spectrum antibiotics, proceeding to administer crystalloid fluid bolus, and then within six hours, reassessing the patient, repeating the lactate if it was initially elevated, checking the blood pressure, understanding if there's persisting hypotension, swiftly move over to implementing vasopressors. So we've heard about these sepsis bundles, we know about the sepsis bundles, and practicality, how are these applied when it comes to hospitalized patients? You've heard today that hospital-acquired sepsis is an emerging and evolving spectrum of illness that's important. Roughly about 20% of cases end up being hospital-acquired sepsis. In this retrospective study of across four university hospitals in California between 2014 and 2016, over 4,000 patients were included to assess a primary outcome of the adherence to sepsis bundles as defined by SEP1. What they found was that hospital-onset sepsis occurred in this cohort at about 36.6%, fairly high. The overall SEP1 adherence was low, it was 30.9. But look at the difference between SEP1 adherence for community-acquired sepsis, which was 39.9%, versus hospital-onset sepsis, which was only 13%. So clearly an opportunity does exist. We've seen that evolution of our awareness around sepsis bundles, about timely interventions, but how we implement it is not quite the same. It was even more interesting to look at the breakdown of the bundle elements in this study. So the top curve that's in dark blue is actually patients who were in the emergency department. And this is the proportion of cases that received each of those bundles. The lower three curves are the wards, the ICU, and the perioperative regions. And each of these bundle elements, you can see the difference between the time and interventions for each bundle. So we've heard about the challenges in recognizing hospital-acquired sepsis. And in 2019, the Surviving Sepsis Campaign, sponsored by the Betty Moore Foundation, launched a sepsis on the wards collaborative to really dive into what are some of the challenges and barriers for implementing sepsis bundles in the hospital-acquired patients. The top five barriers that were identified for nursing included delays in recognition, poor communication regarding that change in patient status, a failure to measure lactate, delayed or no antimicrobial administration, and inadequate fluid resuscitation. Looking on the clinician side, the barriers mirrored each other. Again, does not recognize sepsis. Sepsis treatment is not prioritized or lacks a sense of urgency, perhaps because it can be a little bit challenging. There is a little bit of what's causing it. What is the driver? What is the cause of the change in the patient status? Delay in or failure to measure lactate levels, delay or no antimicrobial therapy, and inadequate fluid resuscitation. But let me ask you this. If you have a patient who's on day three, four, five, and there is a change in the status, what are we typically doing as clinicians? Well, we're probably all going to the bedside, hopefully going to the bedside, and asking, there's a change, why? We're probably asking, how sick are they? Maybe what's going through your mind is, is it an infection? Maybe they're already on antimicrobials and you're starting to wonder, is the source controlled? Maybe they're hypotensive. What's the next go-to intervention? Probably gonna be fluids, IV fluids. And then, if things are evolving, the next question is probably, do they need to go to the ICU? Well, I would propose to you that the sepsis bundles are really that evaluation. Measuring the lactate, risk stratification. Obtaining blood cultures, broad-spanning antibiotics, asking, is it an infection? Is the source being identified? Is it being controlled? Rapid administration of a crystalloid bolus. Well, if there's hypotension, why? And what's the intervention? Applying vasopressors. If there's persisting changes in the patient, it's not a response, there's not improvement, you're probably saying, well, this patient needs to go to the ICU. And in the ICU, we're likely starting vasopressors. So recognizing sepsis, as you heard, is really driven by both the healthcare worker and also a combination of physiological signs. We might recognize these, utilize systems and scoring systems, they're ranging from SIRS, QSOFA, SOFA, NEWS, MUSE, NEWS2, and even proprietary models. And they range from sensitivity to specificity quite significantly. Is the future of machine learning? Maybe, we just heard a great perspective on some of the challenges there. As of now, though, the surviving sepsis campaign would suggest that if you want to use a single screening tool, stay away from QSOFA and utilize either SIRS, NEWS, or MUSE. But that early screening tool is very much in line with the approach for sepsis and septic shock of early recognition to prompt intervention. As you recognize sepsis and you're thinking about the interventions, the next question is certainly gonna be, is there organ dysfunction? What is that organ dysfunction? A lot of hospitals, a lot of institutions now have point-of-care lactates or even whole blood lactates. So it's often a quick test. So is there a quick organ dysfunction identification test? Well, I would say maybe it's lactate. The seminal studies in lactate go back to 1964 when Max Harry Wheal described excess lactate as an indicator of oxygen debt. It's been studied for that metabolic index of severity in shock state. And that marker of four, in going back to his studies, really marks that rise in mortality as you get a lactate level greater than four. This holds true even in more recent studies. The ARISE investigators looked at the mortality in patients with isolated hyperlactateemia. And what they found is that there is 90-day mortality. It's a higher 90-day mortality. That these patients with isolated hyperlactateemia were less likely to be discharged alive from ICUs in the hospitals. They had longer ICU and hospital admissions, were ventilated more frequently, and required vasopressors longer. So lactate is a good risk ratification tool. When it comes to treating your infections and controlling the source, timeliness does matter. If septic shock is present, you wanna really try and administer those antimicrobials immediately. Immediately means within an hour. And especially when you're thinking about sepsis. So if sepsis is possible, the timeliness to antimicrobials is key. If there is a question of is this sepsis versus another cause of the dysfunction, there may be a little bit longer, but surviving sepsis campaign would certainly suggest that we implement antibiotics within three hours. But in terms of being the balance of early identification and treatment, but also good antimicrobial stewards, the approach for our hospital-acquired sepsis patients may certainly be initiate those antimicrobials, evaluate daily, look at your results, look at the cultures, with an aim to de-escalate as soon as feasible. I would argue that source control needs to be top of mind when it comes to antimicrobials, especially in hospital-acquired patients. So if they've got a patient who's already on antimicrobials, and the first question is, is it appropriate? Does it need to be broadened? Does anything need to be added? But source control can also, is evolved from a surgical-only approach to also include a lot of minimally invasive non-surgical or radiological procedures. But the things you're looking for is, is there anything that needs to be drained, debrided, does a device need to be removed, and what are the definitive measures to remove that focus of infection? And while doing all of that, this includes the continued physiological support, restored at aiming and reducing the disease burden. Hemodynamically, the optimization is for end-organ support. So this is the continuous balance of oxygen demand, oxygen delivery, to optimize the oxygen extraction at a cellular level. When it comes to oxygen delivery and demand, the balance between the inciting factor of bulimia, is it vasoplegia, is it myocardial dysfunction, often represents itself as hypotension at the bedside. And in determining the best interventions, we go through our mental algorithms that probably we're not even thinking of, but fairly frequently falls into this question of balancing the oxygen content of the blood versus the cardiac output, and thinking about fluids. Is there a need for contractility support after, of course, vasoactives perhaps have been started for afterload support. So do these bundles work? The New York State Sepsis Initiative was a mandate that was introduced statewide to highlight the importance of early recognition and early intervention. And when this mandate was studied, in the breakdown by Mitchell, Levy, and Tell, about 20% of cases were actually hospital-acquired cases. The implementation of the three-hour bundle increased after the statewide mandate, and this had a risk-adjusted mortality improvement from 28.8% to 24.4%. This held true, even when you compared the New York State Initiative's control states, there was still an impact, a significant impact. So after the introduction of the New York State Sepsis Initiative, there was a significant relative decrease seen in the adjusted risk of 30-day in-hospital mortality relative to those control states. So sepsis bundles do seem to have a role. 20%, about 20% of patients may end up being hospital-acquired patients, but that's a valuable population because they tend to be younger, they tend to be a little bit more challenging, and we're not doing as good of a job. So now what? Well, I'm gonna hand it off to Dr. Reckhoff, who will take you through optimization of delivering the sepsis bundles. Aloha. Thank you for the opportunity. Thank you for the invitation. I'll get to close up the session. The title of my talk is Care Path for Hospital-Acquired Sepsis. Is it same or different from community-acquired sepsis? This is me. I have nothing to disclose. And other than my wife maybe shaved my mustache earlier. So we'll review the key considerations of hospital-acquired sepsis. This should be review for my colleagues earlier today and actually earlier this past week. We'll reinforce the principles of care for new-onset sepsis irrespective of the site of onset. And then we'll go over the components of care and the team approach required to survive hospital-acquired sepsis. Again, we know this stuff. We know what we have to do. And this should be a little bit review. We know sepsis is a medical emergency. Requires immediate attention irrespective of the site of onset location, wherever you diagnose it. However, how we recognize somebody with sepsis needs to be context-dependent. The patient presenting from home to the emergency department, that patient is different than someone who's been hospitalized for seven days or is post-op day seven or has been on the BMT unit for 21 days. And that patient is different than somebody who's in our ICU who has a full complement of services. CRT, the vents, ECMO, RVAD, LVAD. And so it's the tools and approach we need to diagnose sepsis need to be tailored to the respective individual and the environment. Context is everything in medicine. And we need to start understanding that. We are starting to understand that. We're just the tools haven't kind of caught up to us yet. Again, as alluded to my colleagues, again, hospital-acquired sepsis presents unique challenges on multiple levels, on the patient level, provider, and on the system level. Again, these patients are sicker. By definition, they're in the hospital. They're increasingly fragile, right? We're dealing with different types of infections, potentially different types of bugs, different sources of infection. And often the presentation is more insidious, right? And it's not fully formed. That is compounded by provider issues. I'm also administrator. There's a lot of provider issues. But so, again, because patients are in the hospital, there's a lot more information that we have to deal with, a lot more kind of signal to the, or noise to the signal, if you will. And so a lot of times we can explain away perturbations in the vital signs. And that leads, again, these are the cognitive errors, premature closure, anchor on existing diagnosis. That leads, at least on the provider level, to delays in recognition and identification, which ultimately lead to delays in intervention. And this is amplified, these delays are amplified by system issues. On the floor, on, you know, the timescale of how things operate on the floor is different than it is in the emergency department. It is different than it is in the ICU at times, right? So sometimes the wards are on island time compared to the emergency department. And then, again, everybody's busy. We've all dealt with throughput issues, the idea of capacity strain. Again, throughput issues. We have our ED borders. Again, this also just adds to the overall complication. So, you know, it seems trite to say this, but again, hospital-acquired sepsis, to the point of all this morning is, yes, it's complicated, and the stakes are higher for our patients. Yet, there is clarity for this, though. And when I go, the clarity for this, I refer back to the Gospel of Evans, 2021, points one and two. Irrespective of hospital-acquired sepsis or community-acquired sepsis, early identification and appropriate management in the initial hours improves outcomes, number one. And number two, one of the most important principles of managing complex septic patients is the need for a detailed initial assessment and ongoing re-evaluation to the response to treatment. And I underline that because that is, I think, the crux of hospital sepsis care, is the re-evaluation. Because hospital sepsis evaluation management, it's iterative, it's ongoing, from the point they hit the hospital to the time they're discharged, right? In fact, one of the hats I wear back at the University of Minnesota is I direct the fourth-year clerkship for critical care, and every month when I give the orientation to the medical students, the fourth-years come to the ICU. I mentioned there's really two mantras of ICU medicine. The first are the phases of care, and you've seen iterations of this. We resuscitate, we stabilize, we optimize, and then we de-escalate. But within that is the true mantra of ICU medicine, which is we make our assessment, intervene, until the patient leaves our unit. The frequency of our reassessment depends on the acuity and the severity of the patient's problem, right? So there's times that we're at the bedside for hours because our ongoing reassessment is continuous. There's times, and again, we can back off where it's every three hours, six hours, et cetera. Depends on the trajectory of the patient. So applying this to the tenets of hospital sepsis, again, we know we need to identify the source, we're identifying severity, and hopefully in the future we'll be identifying the phenotype or the flavor of sepsis that we're dealing with. Again, broad spectrum antibiotics as soon as possible, and then also narrowing as soon as possible, as soon as we can. Again, we stabilize and mitigate organ dysfunction, and then anticipate to mitigate complications, because it's not only what the patient's coming in with, but what we do to the patient that makes, we don't want to make things worse. But not only at the provider level, but we want to set up systems that enable this to happen fast. And again, you've seen the two-by-two table, so the care path in terms of antibiotic timing, the 210, for antibiotic timing for hospital-acquired sepsis and community-acquired sepsis, it's going to be the same. And I've kind of flipped the question, you know, because often patients, we're dealing with patients who are maybe already on antibiotics, so we need to be looking for subtle signs, symptoms of shock or hyperperfusion. Otherwise, patients develop new onset shock, we can need to make sure, damage sure, that there's no new or undertreated infection. So if we talk about the PICC corporate wants to see the difference between the two photos here, community-acquired sepsis and hospital-acquired sepsis, they're the same photo. It's the same principles of care, once you've made the identification. That's in terms of antibiotic timing. We'll talk a little about resuscitation, the fluids versus pressors. Again, it depends on the context. We have a couple of randomized control trials, we're all familiar with the CLASSICS trial, again, looking at a restrictive versus a usual approach in terms of do we give more fluids or not, no difference left or right in terms of 90-day outcome. The more recent CLOVERS trial, this was published this past year, you know, there was a separation in terms of the amount of fluid, but early pressors versus, and restrictive fluids versus liberal, again, no difference in outcome. So if we go to the, we can anticipate if another randomized control trial in the wards, would that provide us any further information? Probably not. I think the big picture from this is that we talk about, you know, studies are for populations, but we need to treat the individual, and that's really the crux of it. If we need, the question, does my patient need fluid or pressors, you need to go to the bedside and make that decision, make that determination. I think the one piece of information that is beneficial, that came from the CLOVERS trial, is the safety of peripheral vasopressors. So, again, the total of 1,500 people are included in the trial, 500 patients got exposed, one-third, less than 1%, three individuals had a complication. So I think that's a fairly robust safety signal. So I'll get out of the weeds now, I want to kind of step back, and how do we survive hospital-acquired sepsis for our patients? Well, it's both on the provider level, and it's on the system level. On the provider level, it's simply having the, it's been present to mind, having the awareness and recognition of the emergency that it is, and it's having providers learn to embrace the diagnostic uncertainty, recognizing SERS isn't perfect, SOFA isn't perfect, news isn't perfect, you need to go to the bedside, it's a multimodal assessment, but also embracing that, again, it's an iterative process. So you see the patient, you make your best guess, I think they're septic, I'm going to treat them that way, but you need to make a reassessment, right? And depending on how sick they are. On the system level, again, relates to how well we can deliver complex care, how well we can mobilize and coordinate our resources, our personnel, and how well the system can provide the support, logistical, IT, but also personnel support, and promote a culture of sepsis improvement, or a culture of process improvement. I'll talk about that in just a little bit. Towards that end, again, some of us, a lot of us are familiar with this, this was published in late August, the Hospital Sepsis Core Element Program, or the elements for hospital sepsis programs, these seven kind of elements. I come from a coaching background, and when I read this, this just resonated with me, because, to be honest, having organizational commitments, having your leaders or captains, having your kind of playbook or strategy for the year, your action plans, your stats, your tracking, your ongoing skill development, this is the same, these are all the same concepts that I'm going to be telling my 12- and 14-year-olds next week when I go back to Minnesota about the upcoming hockey season. In fact, you know, we can learn quite a bit in terms of programmatic development and team development from other industries and other organizations. One of the things is that, yes, we have teams, but if we want lasting success, we want, we need to focus on culture, because team members, they're going to come and go. They're going to move in and out of the program. And so, I'm a Canadian who lives in Minnesota, so I'm legally required to have a hockey slide in here, so I included this. But I really, this one, this quote by John Cooper, he's the head coach of the Tampa Bay Lightning, he's one of the most successful hockey coaches in the game right now. He's a lawyer by training. He's never played professional hockey, and he fell into coaching by, as a favor to a friend, like 20 years ago. He started coaching high school kids, and they ended up winning the state championship, and he just continued coaching along the way. He's won his way up into the National Hockey League. He's the second highest winning percentage. And asked along the way, what is the secret to success? What is the secret sauce? It's culture over strategy, and have one hell of a goalie. But what does that mean, culture over strategy? Well, that's exactly it. We can have all the outlines, but if you don't have the culture, if you don't instill that common mission, and how do you instill that common mission? It's by repeated engagement and communication. Not just with your players, not just with the coaches, but with everybody from the front office of the organization, all the way down to the trainers. That persistent message, day in and day out, that is how you develop that. That's how you get buy-in, and that's when you get buy-in from the entire organization, that's when you get execution of plan. So yes, we have sepsis days, we have our sepsis month, but if you really want to move the needle, we need to have constant engagement regarding sepsis. Awareness for our providers, for our nurses, for our phlebotomists, for everybody. It needs to be pervasive. But we also need one hell of a goalie. And the corollary for that on the hospital boards is the rapid response team or the sepsis response team, yes. You can play hockey and play without a goalie, but your odds are vastly improved of a good outcome if you have a goalie in that. So, and this is the truth of the case for sepsis. So this is a study by Choi and all in 2021. They looked at, again, this is over the course of 10 years, but they looked at the benefits of having a rapid response system. And so you see the compliance in the red, the bundle compliance rate went up from 27 up towards the 70%, and more importantly, inpatient mortality, 20-day mortality went down from 50 to close to 30%. So patient-centered outcomes by having a rapid response team. But just like, again, you can, like any position, there are certain characteristics of higher performing versus low performing kind of roles. And so for our rapid response team, there are features, and this is a study by Dukers and all looking at characteristics of high performers. This is looking at cardiac arrest, so we can extrapolate for our sepsis patients. Number one is that they should have a dedicated, your sepsis response team, your rapid response team should have dedicated responsibility, right? They shouldn't have competing clinical responsibilities. They can't, if you have a fire, you want your firefighter to respond to the fire, to put out the fire. They don't want to be doing something else. Number two, you want your teammates, your team members to be experts at resuscitation, experts at the bedside. For sepsis, that means we want our providers to know what sepsis is, know how to diagnose it, know what the bundles are, know what the order sets are, and importantly, know what the documentation is. And lastly, and this goes with kind of the clinical responsibilities, is be actively engaged at the bedside, right? If they're called to the rapid response, they're there on the front line with the nurses, with the RTs. They're there to make the initial assessment, but they're also available to come back and make a reassessment. And so, again, and having that empowerment, the staff, that comfort level, that trust, that knowing that if I get in trouble, I can call the rapid response. They're going to be at the bedside with me, and they're going to help me figure this out. That leads to that trust and collaboration, and that moves the needle towards developing that sepsis culture. So, you know, in summary, hospital-acquired sepsis, again, we know presents unique challenges in recognition and diagnosis. The hospital-acquired sepsis evaluation and management is iterative and context-dependent, and that's for us to figure out. Early recognition and intervention improves outcomes, but, again, reassessment needs to be factored into the sepsis response. And when I look at the 300 and 600 bundles, that's kind of, but that's the intention of the bundles. You just can't, someone's septic, you can't just call it in, give them 500, start antibiotics, and walk away. You need to go to the bedside, make your assessment, and come back and make a reassessment. And lastly, again, if we're really going to move the needle, a team and a system approach are required to survive hospital-acquired sepsis. Thank you.

hospital-acquired sepsis

early recognition

appropriate management

community-acquired sepsis

hospital-acquired infections

care principles

lactate measurement

blood cultures

broad-spectrum antibiotics