false
Catalog
Current Concepts in Pediatric Critical Care
Panel Discussion
Panel Discussion
Back to course
[Please upgrade your browser to play this video content]
Video Transcription
Now we'll jump into some questions. So we've just been talking about some of the work around AI and machine learning, so I'm going to jump into you first, Adam. How do we ensure that our validation sets are representative of the population that any given model may be used on? And is that something you mentioned, it's important to monitor this over time, is it something where we should be doing revalidation with any sort of cadence or other things like that? Yeah, it's a great question. So I think there's a couple of schools of thought on that. So my school of thought generally tends to be in the developed models that are generalizable so that can be used across institutions, across sites. There is another school of thought that really focuses on the idea that your model is your model and it can be built internally, used internally, and it's more focused on the population that you have in your own particular institution. I think there's pluses and minuses to both. I tend to like the idea that you can take a model and generalize it across populations, and in order to do that, to your point, you need a representative data set. And so you need to think about the places that you're going to be bringing this model to and are we including all of the different permutations and possibilities of patient populations that we have. We all have different units in terms of the patients that we see. And so how do we generalize that? Well, we use a large representative data set. And again, I highlight a couple of the ones that were on there. The VPS data set that many of us are familiar with, it's more administrative, but there are some features that can be used for machine learning models in there. There's a number of data sets from Capcorn that are useful for machine learning model development because they are relatively granular, and then the PICU Data Collaborative and PEDSnet is another EHR extract from multiple sites. It's important to look at a number of different parameters within that data set to ensure that you're seeing a representation from different populations, and you can subdivide the data in ways that make physiologic sense for what you're looking at. And then, absolutely, a recurrent cadence of checking the model is important, as well as honestly taking the model that was developed externally and validating it in your own local data set to ensure that it does actually apply to your data set. One of the reasons I think generalizability is important is because not every institution has the capabilities or opportunities to extract, define their own model locally, and run it. So by providing some of those from a single centralized source, we can actually provide opportunities for sites that might not have those resources to do that to make use of some of these tools. Thanks. We've got microphones in the room, or you can pop up here to the front, either way. Yeah. And those microphones in the room are available for any of you who have questions. Hello. Good morning, everybody. My name is Adeyinka Adebayo, one of the co-chairs of The Current Concept. I have a question for Dr. Harry from Beacon Children's Hospital. We all know how challenging sometimes, or for the most part, when you are dealing with a very obese patient. So you talked about higher PEEP that patients that are obese need. So my question to you has to do with, one, is there an optimal PEEP for alveolar recruitment for patients that are obese? Because it's a double-edged sword, the higher you use also, sometimes it can compromise our pulmonary blood flow. So in your talk, I didn't hear that. So maybe you can kind of guide us if you are having an obese patient that has the RDS, that you are going higher on the PEEP. I mean, is there an optimal PEEP that you can use? So that's my first question. And then my second—actually, I have three questions, if you don't mind. So if you can kind of talk a little bit more about the apneic oxygenation that you talked about. Apneic oxygenation. And lastly, if you remember all these questions, it's about is there or should there be a standard protocol that you want to develop in—or do you recommend that once you have an obese patient, you have to develop a standardized protocol for managing those kind of patients? I know you talked a lot about that, but I kind of want you to elaborate a little bit more on that. What was the last question? Protocol on what? Yeah, like, do you—is it recommended or will you recommend, like, if you—because of the prevalence of obesity that we see right now, seeing a lot of more obese patients ending up in the ICU, will you recommend for ICU directors to develop a standard protocol for managing those patients, or is it on a case-by-case basis? Yeah. All those are great questions. I anticipated the high PEEP. So high PEEP, there is no optimal number. Like when we looked at the data, like different articles, they did not mention anything about, like, a certain number for patients, but always use it to actually, like, you know, do it along with your patient, like, you know, how your patient is doing. That's why I talked about trans-reserve visual monitoring to see, like, you know, where your—like, there were PEEP was 10, and the—when we adjusted the PEEP to 20, that's when actually the lungs were actually recruited, and you see the exhalation completely in positive phase. So using trans-reserve visual monitoring to actually, like, you know, optimize PEEP is recommended, but there is no certain number I can give you. It's all, like, individual and patient-dependent and, like, you know, dependent on, like, how severe the disease is. Same thing with NIV as well. Like, if you look at it, like, you need—like, when I talked about—I didn't talk about this in the presentation, but, like, when you look at the high-flow nasal cannula as well as the CPAP or BiPAP, you need to actually, like, my—when I looked over the literature and everything, so it looked like—so majority of the time, wherever you go, like, you start on high flow, then you end up on, like, CPAP, BiPAP, or then you transition is to go for invasive. So if you have an obese patient who starts on high flow, so now things are not working, probably quickly transition to BiPAP or CPAP, whichever one, and—but I would titrate it to actually see the work of breathing, titrate the PEEP enough to actually, like, give—overcome the pathophysiological effects, upper airway obstruction or airway collapse effect, like, this is to recruit more so that, like, you are not marching towards, like, you know, intubation. So you have that, like, you know, there is no data to support that CPAP or BiPAP is going to decrease the number of intubations in pediatric obese patients, but the literature actually suggests for, like, you know, BiPAP or CPAP would be better than high-flow nasal cannula. That's the first question. Then, well, I mean, I had a lot of questions. Apnea coxygenation. Apnea coxygenation, can you kind of enlighten us a little bit more on that? Yeah, so this actually comes from a near study. So when they did, like, they almost had, like, you know, 20 percent of obese kids, like, they had, like, 30,000—close to 27,000 to 28,000 intubation events all over the nation. So when they looked at it, like, obese patients had, like, you know, severe peri-intubation hypoxemia. So a lot of this hypoxemia is due to medication errors, and also, like, you know, due to pathophysiological effects of obesity. So they—and they ended up having a lot of—so when you look at difficult, like, not only on near study, like, when you go back and look at difficult algorithms from, like, different studies—so apnea coxygenation actually helps you, buys you the time. We all know that, like, we use, like, nasal cannula or high-flow nasal cannula. So if you use high-flow, if you are on BiPAP or, like, high-flow nasal cannula, like, make them 100 percent—like, do apnea coxygenation, that pre-oxygenation actually buys you time. We also know that the obese patients are difficult to intubate. They're difficult to bag mass. So you need that few more seconds for—to get the tube in, and the most experienced person actually goes with some video laryngoscopy or, like, you know, fibro-optics. So that's where the apnea coxygenation is going to help you and, like, you know, make you and, like, rest of the team lives better. So if I can pause, let me ask the room, how many of you are using apnea coxygenation in your units for intubation? Okay, so now I'm going to say keep your hand up if you're using it consistently for pretty much all intubations. And so we lost a few hands, but for the most part, it looks like the people who are starting to use it are using it pretty consistently. And that's actually been our practice at Duke, I'll say. That's our ongoing quality improvement project that's tied to our incentive plan is the use of apnea coxygenation. And we've had pretty good success with it that supports the data that came out of the Near4Kids database. And that's obviously in all comers. We haven't looked at it specifically in obesity, but we've seen benefits. So coming, like, you know, coming back to apnea coxygenation, apnea coxygenation, like, if you go back and look at the literature, not only in obese, but also, like, you know, the normal kid, we think, who's going to get the tube passed through also gives BICU the time. So apnea coxygenation literature actually, like, supports apnea coxygenation, not only in obese, but also, like, you know, non-obese children, like, who are going for intubations. Okay. And then my last question, sorry for kind of bothering you too much, Dr. Harry, is, is it when you're managing patients with their different condition, when you are managing obese patient with their different condition that are obese in the ICU, do you recommend or suggest to develop a standard protocol for managing those patient in terms of intubation? I know, you know, I have an experienced person that is skilled in advanced area management available. I'm talking about developing a protocol that it also pertains to drug, the drug infusion that you're going to give this patient, factoring the fact that the lean body mass is not as much as you'll have in a regular patient, and most of the drugs we give in the ICU are lipophilic, as you said. So yes, that is a very great idea, to be honest with you. So like most of us actually practice, like when we get this obese kid, like we actually like, you know, look into a lot of these details, like unknowingly, like they are difficult to, they have difficulty where they are hard to back mask. And like we are preparing the most experienced person or someone comes, comes to intubate, we have all the difficulty caught, caught up like outside the, outside the room or inside the room. So it's, it's a very good idea to actually like, you know, get, make all those things come together and like, you know, have, like we have advanced airway card. So possibly like, you know, if you have in the advanced airway card, if you have an obese kid, then actually like, you know, have these steps. So it'll, like a lot of times we have seen, and like we have practiced like protocolized medicine standardized, standardized plan actually makes things much better and like much smoother so that the team is also anticipating what to get and like what to expect from it. And if this person is not able to, not able to intubate, who's coming next, are we calling ENT or like you have your own difficult, difficult airway algorithm. So it's better to have someone like in a small, small sidearm, like you have an obese kid, like where you are, like, you know, where you are high risk of having these, these complications. So that is something like a very good research project, I would say. But actually it's a very good idea. I would suggest to have some of these, which we are doing unknowingly to put them into a basket and say, this is what we have. Thank you so much. And the reason why I brought this up, I'm going to leave the podium right now, is because like I've had instances where you'll, you know, you'll order medication from pharmacy. Case in point, let's say like about 150 kilogram, you know, like 14, 15 year old. And then the electronic medical record and the CPOE, one microgram per kg of fentanyl, you will just spew out 150 microgram, which is technically an overdose. If you really look at it, I mean, you should, you know, you should base your dosing based on the ideal body weight or lean body mass. If I'm not, I think the pharmacist or whoever is very well-versed in pharmacology can correct me here. So that's just what I wanted to add, but thank you so much for answering my question. I really appreciate it. Thank you. Thanks. Hi. Good morning. I'm Mike Uche from Monarch Medical Center. A question back to the obese patient. A problem that we have faced on occasion with very obese patients, particularly if they have type 2 diabetes, is the amazing amount of insulin resistance that we see in terms of trying to control their blood sugar. And we're using amounts of insulin which are basically insane. And is there anything in the literature about that or any advice on that? That's a really good question. Very, very little on literature on like it just says like it causes insulin resistance and there is no literature I have found in DKAs. Like I went back and looked up to see any obese, like, you know, any literature on like obese management of obese DKAs, but there is hardly none. So the only literature I could find was a lot of literature on like, you know, management of diabetes in obese patient, obese kids, on that. Like not on like, you know, acute illness, obese DKA management or like obese diabetic management in critical illness. There is very little, none actually I would say. This isn't even DKA. This is just trying to. No, no. I'm just saying like there is very little, very little literature on that. Great. Thank you. Yeah. I think most of the literature that I've seen has basically described exactly what you said of yes, there's resistance and you have to use really high doses. So it increases that. Go ahead, Dr. Donnell. Hi, I'm Bridget Donnell. I work at a big community hospital in Raleigh. I had questions about the transition to adult care. We struggle with that a lot in our group, I think, because as you pointed out, there's a lot of disjointed. Like I have a vested interest in it in the ICU, but then there are different providers may keep the child a little bit longer, a little bit less. But one of the things we have struggled with is children who as they transition to adults are not going to be large if they're under a certain weight. So do you know any of the guidelines address this? Because we've run into that a lot where we have children who have congenital diseases that cause them to remain at 20 kilos and the adult units are freaking out because they're so small. And the other question I had was just about one of the other bigger challenges is our children with developmental delay and children who are not neurotypical, when they go to the adult world, it's not as family friendly. The parents are expecting to stay. Some of the beds are not private. So we've struggled with that a lot, too, and I didn't know if you had any advice or had seen that in your practice. And if I can, as you go to answer that, I'm going to add my question that I was going to ask, which is very much in line with that, is a lot of the data that you showed or the guidelines that you spoke to talked about ages of transition, and should it be age-based or should it be based on some of these other factors, developmental stage and other things like that. So I think that will tie right in. Yeah. Thank you so much for your question. Very, very good questions, actually. There is not a lot of literature out there on transitioning yet. Eva, I'm just going to go to your first question first. I'm not surprised that different people in your ICU want to do different things. But my suggestion to you would be, if it's possible for your division to make the six core components application as one of the QI projects, that is actually what the guideline is saying. Make it a QI project, make it a goal of the division that that's what you want to do. What that is going to help you do is to help the division to form a guideline development so that everybody keys into the same thing. Then in terms of weight, there is nothing in literature that talks about weight. It's all about the age of the patient. And we know that the adult medicine, the way that I describe adult medicine is like a whole cell package. They do more whole cell, and in Peds we are more retail packaged. And what we've noticed actually is that most families are reluctant to actually move to the adult care. And we understand that. We are very detailed. We provide everything more comprehensively. But I think that if you started early, from 12 years of age, actually this is what the six components keeps advocating. If you started early, from the age of 12, before they get to the age of 18, you would have had a good process that would have familiarized them with the next providers that you're going to send them to. You know, so starting early is key. But there is nothing that talks about their weight. They are all timelines based on ages. Although I'll validate we have those same issues with weight sometimes. And if I can follow up one time before I get to Dr. Irving. So how many of you are in true freestanding children's hospitals, i.e. there are no adults cared for on your site? Okay. So maybe a quarter to a third of the room. So the rest of you I'm assuming are in a place where there are some adults. And so our experience has often been we start that transition process and we make the appointments with the adult providers and we send the patient out. And then they either bounce back to the hospital before that appointment has happened or they miss that appointment. And then they come in and well, the last time they were here, they were cared for by PEDS. They haven't been transitioned yet. So they're back onto the PEDS floors or the PEDS ICU. Is there ever a place, and I'm happy to actually hear from any of the three of you or you can weigh in as well, Sharon. Is there a place to ever do that transition during a hospital stay or is that ever appropriate or should it always be done in between hospitalizations? So what the guideline says is to do the actual transfer, which is the fifth component when the patient is more stable. But even what is more important is to start that process early enough because if they are not comfortable, they are going to come back. But let me also say this. As good as the transition process sounds, there are patients who will still have to, they are sick. If you transition them to adult, but they will come back. I'll give you a real life story. Allison knows what I'm talking about. We were in the same hospital. We had a 30-year-old kid that has some muscular dystrophy that went to an adult hospital. She was in an adult ICU for a couple of days. She was deteriorating respiratory-wise and they had to, 30 years old, they had to insist that she comes back to Nemours. And we had no option to still accept her to come back. But I also know that kids like that probably did not go through a structured healthcare transition. This is not yet popular, but that's what we are beginning to ask people to do. Start it early enough, get them comfortable. Even if it means at some point getting everybody around the table, the adults, it doesn't have to be physical, it could be remotely. The adult providers, the ICU team, the social workers, the family, getting people together to talk about this patient over time just to get them comfortable to go to the adult healthcare system. Hi. Thank you very much. Sharon Irving, Children's Hospital of Philadelphia and University of Pennsylvania. A question and a comment about the transitions of care. So there are popping up around the country these programs specific to that because it has been such an issue. And when you talk about early age, there are some models in asthma where you start teaching the child at the time that they can learn their meds. So six, seven, because they're going to school, some of these children are going to school and they have to be able to identify when they're in trouble, right? So they have to know their meds and know their symptoms. So there's a model that's beginning to spring up in the GI population, kids with IBD, kids with Crohn's and such, younger and younger, that they can understand what their symptoms are and that has shown anecdotally to help that transition process. So just, and there are centers around, for the person that asked that, there are centers around the country that you can reach out to. Children's Hospital has one, but there are other programs around the country that can help to formulate exactly what you said in terms of putting together what does that look like, when do you start doing it? But my question to you surrounding that is more on the side of the adult providers accepting and how do we help them? Because you say something like, you know, double outlet right ventricle and they're like, I don't think so. I'm not doing that, you know? Or a weight of 25 kilos or significantly developmentally delayed. And so the things that I have encountered and the things that we have more and more begun to struggle with, even having the guidelines available, is how do we help our adult colleagues recognize that there's opportunity for collaboration here and that they don't have to be afraid. We've had adult colleges say, nope, not taking them. You know, and it's like, okay, they've hit 30. They cannot be in the pediatric ICU anymore. So that's my question to you surrounding that. No, thank you so much. Those are very good questions and observation. I'm happy that you mentioned that some of the chronic illnesses like asthma, that there are some institutions that are beginning to start that early. And that's good. But for the purpose of the six core components, it starts at 12. But part of that component is actually to be able to extract what their self-care needs are. So if you have a child who started at seven, eight, to understand what their needs are, it makes it even smoother when they get to 12, and you get them into that process. So for asthma, for diabetes, if they started early, fantastic, but for your guideline development, you want it to actually start at 12 and end around 26 years of age. Does that make sense? It does, it does. And your second question? Is how do we help our adult couples? Excellent, excellent. So I think one of the things, again, if you start a normal, formal transition program, one of the things that we advocate is to get the primary care physician to also understand that if you have a kid that has a particular organ system issue, let's say a cardiac kid, you're not sending that kid to the primary care physician to take care of the heart conditions of this kid. So what the primary care person needs to know is that these kids are gonna come for you for your routine care, but by the way, we have also connected this kid with an adult congenital heart specialist. They need to know that. And if it is possible, they all need to be talking. The adult specialty clinician, the adult primary care provider, and us will need to talk so that it can make them comfortable. And finally, finally, we do not lay blames on how people are reluctant to accept these kids that have childhood onset conditions. They're not prepared for that. Their training does not prepare them for that. So one of the ways to set them up for success, like we advocated, is to also begin to teach healthcare transition, make it part of the curriculum, part of the ACGME curriculum that people need to understand healthcare transition. I think if you put it as a part of the training, with time, folks are gonna start getting comfortable with doing this. This is not a completely new concept. You know, it's been there, but there hasn't been any active, intentional support to do this. And I think part of the reason why we're having this discussion is for people to go back home and think actively about this and actually start doing it. My last comment to that, I agree with you 100%, is to think about interdisciplinary and multidisciplinary groups in terms of doing that, because this is some of the work that case managers, which are oftentimes nurses, this is some of their groundwork that that's what they do. And excellent, so finally, so in the stage one, which is guideline development, make sure that all the stakeholders are involved at that point. You know, get their buy-in, let them make their imputes. If I want a social worker to help me make calls, to help me link these patients to subspecialty clinicians, if they are not part of that guideline development and you don't have their buy-in, it's gonna make it a little bit difficult. If you have the case managers part of forming the guideline and the policy, it makes it easier. So at the point of developing that guideline, please, if it's possible, all stakeholders should be involved. Thank you. You're welcome. Yeah, I think, Dr. Irving, actually, part of what I wanted to say or question you about, Dr. Odira, she kind of mentioned it. So your presentation was excellent, thank you so much, because this is kind of like a new field to most of us, and also for our adult colleague. So I will want to think that true advocacy for the technologically dependent kid that we see, a lot of them with tracheostomy, with G2, with chronic lung disease, a lot of patients that have transitioned to adult with single ventricle physiology that have been repaired. I think the goal here is like collaboration with education of our adult colleague, and perhaps maybe ACGME can create a specialty, which I think would be better, actually, for training people to take care, like a sub or mini fellowship to train people to really take care of this patient as they transition from pediatric to adult. One last question, role of nursing home in taking care of this category of patient, if you can just kind of elaborate a little bit about, because I didn't hear a lot about that when you were doing your present. The role of nursing home in taking care with the transition for this category of patient. I don't know that I saw anything about nursing home in particular, but what we talk about is medical home for these patients. And by medical home, we're talking about a multidisciplinary approach, making sure that their primary care providers understand what their specific organ system issues are, making sure that they also know that we're not leaving them out there to hang out alone. That for those specific organ system issues, we'll have their subspecialty clinicians who are also gonna take care of them. So it's more of a medical home and not a nursing home issue, yeah. Carolyn. Hi, I'm Jocelyn Grenwall at Emory and Children's Healthcare of Atlanta. This is a question for Adam. So there's a huge gap and gulf between people who use this retrospective databases to develop these models, and then there's a whole nother pocket of people who control the actual implementation of that. And I find that it's very difficult to get people to implement a lot of these fancy methods, whether they're clusters. Not only do you have dynamic data capture that has to happen, and the models need to run in real time, which they are not doing when you derive and validate them typically. But then you have to then have the people that are really behind the scenes in Epic or CERN or whatever health electronic system you're using to actually be able to run the models in real time. And typically it's not their priority. So my question to you is, how do you start to bridge that gap and make the cohesive argument that this is actually cost effective and beneficial and that people ought to start doing that? Because we need to get, people are trapped in this logistic regression models, and it's like ancient history. And sometimes they're good and they're good enough, but there's a lot more sophisticated and perhaps better technology that we can be using. A hundred percent agree. I think there is a huge gap between development and implementation. And that's one of the big reasons why we've seen so much out there in literature about development and so very little about the implementation. And you're right, that they're often very different groups in the hospital infrastructure, and they're very siloed. I wish I had great answers for you on how to do this. I think it's a work in progress at many institutions, including yours. I think one of the key features is you need buy-in from key leaders and key stakeholders who bridge that gap. So as you mentioned, demonstrating the ROI, I think one of the ways that your leadership in particular has been able to do it has been to show kind of small wins in different pockets. So you develop a model that's focused in a particular content area that has one very contained focus and demonstrate its usefulness in prospective, maybe silent running, so it's not being shown to clinicians, and then you can demonstrate how it has changed practice and that there's some ROI associated with that. And that builds to the next model, and you do that a couple of times, and then you can show there's value in these. And then there's an infrastructure that needs to get built. But this infrastructure is not cheap, unfortunately. The people who do this work tend to be highly technically skilled, and that requires typically an out front lay of capital because they can get jobs in industry for a lot of money. And so there has to be this buy-in between the leadership and the groups. I think one of the things we've struggled with, and you guys as well probably, is what's research and what's operational, and when does it kind of make that switch, right? A lot of the work that folks have done is in the research space, understanding the accuracy, the validity of these models, the usefulness. But then we'd like to get them into the operational space, and so somewhere along the line, it has to transition. And that means we give up control of that, and it goes into the folks that keep the lights on, keep the EPIC and Cerner's running. And that can be really frightening, I think, for them as well, because they might not understand that. And so working with those teams to figure out how do you build that partnership of, we're still going to be involved, but this is really becoming an operational tool now. And then there's a whole arm of discussion around the technology. Like you mentioned, EPIC can probably do logistic regression well because it can multiply and add numbers. But once you start thinking about how do you run deep learning models inside EHRs, I think that gets really complicated. And so there's a variety of tools to extract data in real time and operate on it and return it back to the system. And those are certainly more coming down the pipeline. And I'm just going to follow up on that as we move to our next question. But is that infrastructure that's needed, is that the biggest barrier between moving from predictive models that we're really using now to more true clinical decision support? Or are there other barriers that really are preventing us from taking that next step? Yeah, I think it is definitely one of the biggest barriers. The other piece is sort of tied together is the usability and usefulness of these. So simply building a tool and putting it in place does not make it a great tool, as many of us who work with electronic health records know. And we're interrupted in our workflow all the time with things that are relatively meaningless to that workflow. So there has to be an element of usability that goes into that. So it's not just designing the model, designing the tool, putting it into the system, but it's understanding how does it fit into the workflow, doing that analysis of the contextual environment and figuring out where it's going to be useful. These are very research focused things. These are things that your ISD folks that keep Epic and Cerner running are not doing and are not interested in. And so it really takes a partnership to understand how do you follow that process through and get it to a point where it can be used. And then there needs to be an evaluation afterwards. So I think it is one of the barriers, probably the biggest, but it ties in with some other barriers as well. All right, thanks. Okay, this is Teresa Camacho, South Texas. Looking for some tips. Transition care. The issue we know weighed on also complex patients. How do you deal with the specialists that don't want to give up their patients? You know, they end up anyway to the ICU really critical. You know, it doesn't matter if you do the transition. Sorry, I didn't hear what you asked. Yeah, how do you deal with a specialist that they don't want to give up the care of those patients? I think that's one of the major issues. We can already sign off with the patient, but the patient comes to the ER anyway. You know, we're gonna continue taking care of that patient. Yeah, so that's actually a very good question and a very good observation. So healthcare transition process is a new field. It's a new field. And what we advocate really is that if you have a formal process and you have your guideline that is developed already, those sub-specialists, at least in the hospital there, need to understand or you need to communicate with them that this patient is now part of this process. And we're not asking them to release this patient when they are 12 years old. We're starting this process around 12 and we're expecting that by the time they hit 18, 21, they can now go to their adult care services. So if you bring them early enough, I think they would key in. But if you leave it and you don't do anything till they are like 21, some of the sub-specialists may find it difficult to release them. And more so, even the family themselves might not want to go. So I think the key really is to have a formal process and let everybody know right from time to time that you are putting them into that formal structured healthcare transition. Thank you. I'm looking also for a tip for obesity. You know, I'm in South Texas. We have obesity really high. And I find it kind of surprising every time I have obesity patient, I have some limitations on the settings of different hospitals. Beds not available, doors are not wide open, MRI not fitting for obese patient. Is this the time that we should push some standard for obesity patients for the children hospitals or facilities that deal only with pediatric patients? Yeah, so with the challenges of the equipment that we need for obese patients that are often are not available in children's hospitals, does that change? So from an obesity side, does that change when we transition? And also you were saying like, do we need to push for that equipment in children's hospitals differently? Exactly. Do we have any kind of, this is the limits now. Like the chairs, for example, bathrooms, you know, they are equipped for pediatric patients, but it's the time to start setting up. We need also for obesity patients for all the children's hospitals. This is the time. I think that's easy for you to fit in there because it fits both of you. Did you find any guidelines, recommendations? The equipment you're talking about, like what kind of equipment for, I didn't understand the question. When I have obesity patients, many times I have to get a bed from the adult side because I don't have it. Sometimes I have to look for a room that's a wide open to fit the bed. And I found out one time the MRI wouldn't fit the patient. I think you're absolutely right on the money. So we do have like, you know, a lot of obese kids like coming to ICU. So we need to make sure that like right now, like if you have only a few rooms with the forklifts or like lifts needed to actually, if they need to change linen or like, you know, put waffle mattress or anything like that, we need to have this equipment. I think you're right. So we should have like, you know, an institute like in the ICU itself, have like more rooms available with forklifts to actually address these issues, even with like any equipments or anything like that. But I think to your point, that's a potential advantage of being in a hospital that is a children's hospital within an adult hospital that you can pull some of those equipment from adults, whereas the true freestandings hospitals I think are increasingly have to put thought into having that equipment available. Or does this adolescent who is more, you know, significantly obese need to be moved to an adult center earlier than they would have been otherwise? And just to also add, that's where it becomes really important that these components of six components of healthcare transition gets customized and gets specific to the individual patient. Because at the point they are transitioning in their lifestyle, those are the discussions that needs to happen. If he's a morbidly obese kid, then workplace accommodations has to be made is a discussion that you have to have with them and have with, you know, wherever they pick a job, wherever they go to, then that discussion has to happen and it has to be individualized. Thank you. We'll take one last question and then we're gonna go to break after that because we promised we'll get you there on time. Sorry, I do have a lot of questions. Oh, okay. Well, it'll be a compound question. My first question is, so I come from a PICU in a small state of Maine. In our PICU, we have one doctor, two nurses, and there's no unit secretary. We answer our own phone calls, we open our own doors. How do we transition care in a place where we have no resources to have, you know, secretaries following up with patients or talking to them? It's quite jarring. And the second is more like a, maybe just a philosophical question. Why do we have to transfer them? Because most of them have, like, for instance, the patient I have in mind probably doesn't have a long life expectancy. She trialed the adult ICU, she coded there, and then we realized she just does well with us. So why is this push to transfer them, especially when there's such limited physiology and they act like it's basically for the rest of their lives? Then the other question I have is surrounding data. So I did, I've done VPS research, and I did a study on A-lines, and one of the complication for A-lines listed by the VPS data entry people who are actually certified to enter data was death and seizures for a radial A-line. So the point I'm trying to make is a lot of data is crap data, especially as a wife of a data scientist. So how do we ensure that data is better? Because coming from a small hospital, we can't make our own data. We rely on bigger institutions. And the second thing around data is alarm fatigue, because a lot of times a nurse will call me, hey, the sepsis calculator went off again. Just wanted to let you know. Because they do go off a lot, and how do we make it refined enough that we are not getting fatigued by this alert that keeps coming? Sorry, those were my four questions. Let's field transitions first. So transitioning without resources, and might it be the right decision is to not transition at all? And then we'll hit our data questions. All right, thank you so much. And those are really very good questions and good observation. When you start developing your policy and your guideline for transition, it's always advised that it's multidisciplinary, because there is so much that goes into it. I do not honestly think that that's what just one physician and a few nurses will be able to do. So I would say that for your own situation, your own institution, that there has to be a way of getting people who make hospital-wide policy to see how they can resource your units. I wouldn't sit here and expect you to be able to do all the work out of what goes into this process. That would be asking for so much. So I would just advocate for some sort of resource reallocation to help you guys, because I'm not sure you'll be able to do it just by yourself. Making all the calls, doing all the readiness assessment is not what I think the physician has to be doing all the time. You need help. And then why do we transition? That's a very good question. We had mentioned earlier what happens when kids do not transition. I want you to know that these kids have varied multiple needs, that even if the ICU, just the ICU, does not feel like transitioning these kids, the sub-specialists that are seeing them might want to transition them. The pediatric primary care might want to transition them. And if you don't have a structure, it becomes haphazard. And that's where problem comes. We talked about the study in Lancet that transplanted kids, about 35% of them lost their functioning kidney in the first 36 weeks. We don't want those haphazard approach as this kid's transition. So we always advocate for a structural transition that brings all the stakeholders around the same table. That's much better to improve the patient's overall life outcome. So if I can add one thing I've kind of heard there is, if you have this young adult who has a very limited life expectancy and you feel the right thing for them is to continue their care with their providers who know them best, there may be special adult needs that we don't think of. And so there may be some value in at the very least engaging some of your adult providers to make sure that there's things that are being met that we just were not used to thinking of as pediatricians. And then Adam, I think we had quality of data going in and alarm fatigue. Yeah, perfect. So certainly very important issues, garbage in, garbage out, right? If you don't put the right data in it's gonna come out looking like who knows. And real world clinical data is messy. Your specific comment was on a more administrative data, I think more data retrospective. But certainly when you're thinking about data feeding into models in a real time setting, there needs to be guardrails or parameters around that to understand what errors in data entry or data feeding in can look like. I mean, I think we've all looked at CR monitors that have respiratory rates that are clearly not physiologic or normal or reflective of the patient in front of us. So how do you balance that? Well, you have to again, you have to incorporate that error into the model and predictions. Administrative data is particularly challenging, but often is not something that's likely gonna be used in a real time predictive model because it's often retrospective, right? And then sort of to your second question, the alert fatigue, alarm fatigue absolutely is a huge deal. There's a number of strategies that people can use to mitigate that. I think one, the biggest mistake is forgetting that it exists and that it's real. And part of that challenge is the people who are receiving the alarms often are not the people who are making the alarms, right? These are being put into place by administrative organization or administrative leaders or information systems, IST folks who don't recognize the challenges that we face as clinicians and providers in that space. There are a number of different, again, strategy or opportunities. One can use two phases approaches. This has been more popular recently where there's an initial cutoff that might have a lower precision potentially, but capture more patients. And then that would escalate to a higher second phase model or a second level with maybe more data, for example. Unfortunately, the downside of those models and those two phase approaches is that the first phase often falls to nursing and the second phase then moves to the providers. I don't personally think that's right, but that's often how they're implemented. It'll be a lower threshold and then it'll ask for additional data like cap refill or something like that. That'll feed into another layer model, which then has a higher PPV. Another approach is to use non-interruptive decision support. There's a ton of decision support that we interface with daily that's not interruptive. I'm thinking of columns on your list that are stoplight, red, yellow, green type things, highlighting of laboratory values, order sets that provide some decision support within them. These are things that are in our workflows that we can incorporate that are not interruptive alerts that stop what we're doing and force us to have that cognitive burden of changing what we're focusing on. So a lot of opportunities if working with people who understand the contextual factors of the environment, the workflow, and the physicians and providers that are using these, and obviously our multidisciplinary experts as well, nursing and others, that's really important. So it's not just about building the model and putting it in place, it's about understanding where it's being used and who's using it. It gets to that usability aspect. And if I can add to that, I think using the expertise of the clinical informatics folks, at our institution, they use the same techniques and skills that Adam spoke of to reduce the number of best practice advisory alerts that were popping up by 80% in our electronic medical record and made a huge difference in terms of those interruptions and workflows.
Video Summary
The discussion focused on ensuring AI models' validation sets are representative of diverse populations. Two approaches were outlined: creating generalizable models for use across various sites and building internally-used models tailored to specific institutions. Emphasis was placed on utilizing large, representative datasets and regular revalidation to ensure model applicability.<br /><br />Regarding challenges faced with obese patients, no optimal PEEP level for alveolar recruitment was identified, as it's patient-specific and disease severity-dependent. The use of advanced imaging and apnea oxygenation for intubation in obese patients was recommended to manage severe hypoxemia. A discussion on the need for a protocol for managing obese patients was raised, suggesting a standardized approach could improve outcomes.<br /><br />The challenges of transitioning pediatric patients, particularly those with chronic conditions, to adult care were discussed, emphasizing early and structured transition processes starting as early as age 12. Including adult care providers in transition planning and advocating for specialized training on childhood-onset conditions for adult care providers were suggested solutions to improve transition outcomes.<br /><br />Addressing data issues, garbage data input in AI models was noted as a concern, needing guardrails and understanding of data errors. Alarm fatigue due to AI predictive models was addressed with strategies like two-phase approaches and non-interruptive decision support to reduce workflow interruptions.
Keywords
AI model validation
diverse populations
obese patient management
pediatric to adult care transition
data quality in AI
alarm fatigue
advanced imaging
standardized protocols
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.
×
Please select your language
1
English