SCCM Resource Library-Pragmatic Trials in the Digital Age

Video Transcription

Please welcome Dr. Greg Martin. Good morning. This morning, it's my pleasure to introduce Dr. Gordon Bernard, this year's Lifetime Achievement Award winner.  As you heard yesterday, Dr. Bernard is Executive Vice President for Research and the Chief Research Officer at the Vanderbilt University Medical Center.  He joined the faculty at Vanderbilt in 1981 as a physician scientist, and he became the Medical Director for the Medical Intensive Care Unit and Director for the Pulmonary and  Critical Care Research Programs in 1983. Dr. Bernard's research is focused on improving the care of critically ill patients, and particularly  those with sepsis and the acute respiratory distress syndrome. He established the Vanderbilt Coordinating Center in 1987 to support the large, multi-institutional  and international clinical investigations that his team was leading. In 1994, he was asked to lead the NIH ARDS Clinical Trials Network, the principal NIH  team focused on clinical research in intensive care. His longstanding leadership was recognized by his colleagues in asking him to continue in that role for the next 20 years.  As one of Dr. Bernard's many fortunate mentees, I can attest to his collaborative style, his  insightful intellect, and his persistent focus on improving the care of critically ill patients.  Dr. Bernard has authored or co-authored more than 350 original articles and book chapters, including practice-changing articles in every major journal.  He was the Director of the Division of Allergy, Pulmonary, and Critical Care Medicine at Vanderbilt  from 2001 to 2007, during which time he began in his roles in the Vice Chancellor's Office for Research at Vanderbilt, arriving at his current role as Executive Vice President for  Research at the Vanderbilt University Medical Center. In these roles, Dr. Bernard oversees the extensive, multifaceted research programs at Vanderbilt.  He is the Director of VICTOR, the Vanderbilt Institute for Clinical and Translational Research,  and the Principal Investigator for Vanderbilt's Clinical and Translational Science Award, an NIH-funded program that's supported by one of the largest single grants in Vanderbilt  history, more than $140 million in its first 15 years. Dr. Bernard is also currently the Co-PI of the Duke and Vanderbilt Trial Innovation Center,  one of only three such NIH-supported centers in the United States, and his expertise and commitment to research has led to the creation of an entirely new platform for conducting  clinical studies in critical care. Please join me in welcoming Dr. Bernard as he presents Pragmatic Trials in the Digital Age.  Thank you for that very nice introduction, Greg. It's a really great honor to be here this morning and talk to you about some things  that have been going on in critical care research over the last four decades, and bring you up to date on some technology that we're using to actually improve the efficiency and  speed of clinical trials and clinical research. I just want to start off by just pointing out that it takes an army to do this kind of work.  This slide doesn't do justice to the dozens, hundreds of people that have been involved in clinical research that I've been involved in over the years.  You can see Greg Martin there is in the list of collaborators. The study groups, the ARDSnet group, the Prowess Study Group, NIH Biomarkers Network, the Victus  Vitamin C Study Group, and mentors. The first two doctors, Brigham and Snapper, were my scientific mentors as a fellow and early faculty member.  And William DuPont is a biostatistician and who taught me what I know about biostatistics. And Alan Morris was the voice of constant standardization and principles in clinical  trial conduct in the ARDSnet program that I'll talk more about later. And I'd like to provide a special call out to Art Wheeler.  There was nothing I did in clinical research over the last four decades that Art wasn't directly involved with.  Art had an untimely death from pancreatic cancer in 2015, and he's a very important part of all of this. So the story begins in 1979.  This is in Ken Brigham's sheep lab. I don't know if you can actually see. There's a sheep there in this plethysmograph, whole body plethysmograph.  And he's a very instrumented sheep, has a tracheostomy, has a PA catheter, a carotid line, a central venous line, pleural balloons.  And a very unique part of the anatomy of the sheep is that you can cannulate their caudal mediastinal lymph node and drain pulmonary lymph so that you have an almost real-time  readout of what the interstitial fluid is looking like in the animal. And so we were studying ARDS models, and the edema fluid was important to get a better handle on.  The other thing to notice in this picture is I have a scrub suit on, a scrub shirt on, and underneath is a tie. And of course, I still had hair then, too.  And this scrub shirt was just something I'd throw on to go and work with the sheep and then back to the ICU and was doing that daily during these eight and 10-hour experiments.  And it took me a while to realize that my real laboratory was in the ICU, not in the sheep lab.  And actually, this kind of crystallized with my clinical mentor, Roger Dupre, who one day when we were rounding in the intensive care unit and discussing whether steroids were  useful in the acute respiratory distress syndrome, we were going back and forth about whether that was a good idea.  And finally, he said, well, nothing ruins a good argument like real data. Somebody just needs to go and study this question and try to get the answer. And we did.  We launched the high-dose corticosteroid in ARDS study back in the early 80s, published in 87 in the New England Journal of Medicine.  There was no effect on survival in these patients. There was no effect on the percent that reversed ARDS or the time to reversal of ARDS.  And there was no difference in chest radiograph, static compliance, and oxygenation in these patients either. So there wasn't a hint of any benefit from steroids in this work.  So fast forward 10 years to the ARDSnet. This is the timeline for ARDSnet1. The program started in 1994, but we didn't get our first study going until 95.  And then it carried on for about 10 years. We studied about 2,600 patients in that 10-year period, 2,600 patients in 10 years. That's not a lot.  And that's the point I'm making here. And even though we did some work, like with the lower tidal volume trial, we did a two-by-two  factorial with ketoconazole and lysophilin that increased the efficiency of what we could do. And so we were able to actually factorialize the N, so it got up to 4,000.  So about a 53% increase in the N in those studies because of the factorial design. And to my knowledge, this is the first time that factorial design has been employed in  critical care. The other studies that were done, the late steroid rescue study. It's interesting that this study was planned to be about two years long.  It took about eight years to complete to get only 180 patients randomized. And what we were running into is that in order to be eligible for the late steroid rescue  study, you had to still be having ARDS criteria at day seven, at the end of a week. And with the low tidal volume trial, that wasn't happening.  These patients in the low tidal volume trial were improving substantially, and so they just didn't meet those criteria anymore.  And then we did the higher versus lower PEEP study that showed there was no difference. And then the FACT fluid trial, where we compared the pulmonary artery catheter to the central  venous catheter to manage liberal versus conservative fluid approaches in patients with ARDS. And these programs were all published here in this citation box.  So then the next 10 years, ARDSNet2, going from 2006 to 2014, about eight years, we enrolled about 2,500 patients.  So about the same total number of patients over this next eight-year period. And we did have one two-by-two factorial omega feeding with omega fatty acids.  And then the EDEN trial, which was early trickle feeding, if you will, compared to full feeds. And then there was a registry when H1N1 came along.  And then SAILS was studying statins in ARDS and showed no effect. So we were able to increase the N here with a little bit of that factorial work to around 2,800 patients.  So a 13% increase, not a huge increase, but somewhat. So then that brings us to, so where were we in the 80s before those studies began?  And a number of other studies that I obviously didn't go through. We frequently used the PA catheter.  We felt to see the central venous line was just not a reliable way to evaluate volume status in patients.  We used arbitrary left ventricular filling pressures, 15 to 18 millimeters of mercury. I remember titrating to that number religiously.  Tidal volumes were in the 12 to 15 milliliters per kilogram. And high airway pressures were largely ignored. There was a strong desire to normalize PCO2 and pH.  There was frequent use of deep sedation and paralysis. TPN was the order of the day. Routine use of renal protective dopamine was universal, at least in our units.  There was rare use of DVT prophylaxis. And there was really no effective GI prophylaxis. Malox was all we had.  Intricate arterial blood gas driven non-protocolized ventilation and weaning trials were also the way things were done. And there was sporadic use of steroids.  So if we fast forward now to 2019, 2020, most of those things are gone. The only thing left is sporadic use of corticosteroids.  I've been studying steroids for 40 years and I've given up on that they're going to go away. They will never completely go away.  So this brings us to the Arthur Miller quote, there are no passions quite as hot and pleasurable as those of the deluded.  Compared to the bliss of delusion, its vivid colors, blazing lights and explosions, whistles and liberating joys, the search for evidence is a deadly bore.  I wouldn't say that the search for evidence has been a deadly bore, but it isn't as exciting as the delusions.  And so the work that's been going on with the ARDSnet and other areas has been a search for evidence and I believe a lot of it has gone to those previous two slides where we  have changed everything we do in the ICU. And I think probably one of the biggest trial survival gains observed from randomized trials  in critical care came from the ARMA trial, the low tidal line ventilation trial. I was practicing in the ICU for a good 15 years or more before that study was completed.  And so I had a pretty good idea what an average ARDS patient looked like, how long they would stay on the ventilator, how bad their gas exchange would be and what their survival  would be. And when we went to low tidal volume ventilation, it was a striking change. There's been nothing in my experience in critical care management that has been so dramatic  as that. I still can't get used to telling families when a patient goes into the ICU that they're going to be on a ventilator for a few days instead of a few weeks.  I still tell them, well, they're going to be on a ventilator for a few weeks. And of course, we don't see that hardly at all anymore.  So this was a big move and it went from the big screen to the small screen in your living room when an episode from NBC's ER played.  Lower tidal volume of six cc's per kilo decreased mortality from 40% to 31%. You twirl a knob, you're going to save a life, okay? There you go.  How many times has the conclusion of one of your papers been touted on ER? So the challenge then was to enroll more patients in clinical trials. We just weren't moving fast enough.  And so if you wanted to enroll 10,000 patients in a study, that's unthinkable. It's just unthinkable in the way we traditionally do clinical trials.  Even 1,000 took something like the ARDSnet, a major investment by the NIH in clinical research in the ICU.  And you want to control the delivery of a time-sensitive intervention, and you want to collect patient-level data and outcome data in an efficient fashion.  So the traditional approach would take half a decade, hundreds of personnel, dozens of centers, millions of dollars, and what we wanted to move to was an alternative approach  where we had novel study structures, informatics-enhanced intervention, randomization, and data collection. And I'll show you some ideas about that in a minute.  So we're calling this a learning healthcare system. This term is thrown around quite broadly and has lots of different meanings to different  people, but to us it means conducting research embedded in operations in a scientific way,  meaning either randomization of some sort, whether it's cluster randomization or patient-level randomization.  So we learn from what we do, and then once we finish learning, we do what we learn. And the IOM in 2013 published this report that said we're kind of a mess in medicine  because we have science that generates all kinds of good findings, and the insights are poorly managed. We create evidence, and the evidence is poorly used.  We undertake care, and experience is poorly captured in how well we're doing with care. And so there are these missed opportunities that cause waste and harm.  So the learning healthcare system is a Venn diagram that looks something like this. It starts with research and operations working together to try to address questions that  are of common interest to both, and then bringing in implementation and dissemination because that is a huge obstacle that we all face in everything we do.  And so the intersection of all of that would be the learning healthcare platform, as we call it.  And this platform tries to do certain things, and I won't go through all the details on this slide.  But as I pointed out, the question is, does the intervention work in real-world practice? You're using a real-world clinical setting as well, and the population of patients that  you're bringing into the study is a highly inclusive population. And in addition, the costs need to be minimal.  You're running our program on less than a half million dollars a year, which, when you see it in a moment, gets a lot done for that investment.  And then duration to conclusion is usually pretty short. We consider pragmatic trials to be those that are in the 12- to 18-month range, sometimes  two years, but not five years, like a typical randomized clinical trial. And the implementation conceivably could be intermediate, and I'll show you an example.  So this is the platform. We create a reusable fixed infrastructure down here with institutional and regulatory  support, master protocols, clinical decision support, registry information, real-time EHR data.  We use REDCap for almost all of the data collection, and standing DSMB and other things that are just part of every study.  And then there's some study-specific tailoring that goes on top of that. And then what that allows us to do, then, is to bring interventions in and have all  of this platform ready to go so that we can run these studies sequentially or even in parallel across the platform.  And if we have an improved outcome, we publish it and try to implement it. If it's no improvement, we either de-implement if it's already something in place, or we  just publish and show that it's not something to consider. And that's how the program runs. So the intake process is pretty straightforward.  We have content experts, operations engagement, project navigation, literature review, study  design, biostatistics, informatics, IRB support, all the kinds of things that can stymie even a seasoned investigator in launching a project.  And we help them with this so that they can focus on the study that they want to do. And so we have some workshops where we further refine things, and we have studios, which  is a tool that we've developed over the last 10 years for a lot of purposes, where we put experts in a room, usually 8, 10, 12 experts in a room, with an investigator who can spend  15 or 20 minutes talking about where they are in their research and what problems they need to try to solve to get their research going or their project going or implemented  or whatever phase they're in. And the experts then work with the investigator in a collaborative way to improve the project that they're working on.  And these have been highly successful. So part of this that makes it pragmatic is the consent issue.  Obtaining informed consent is a very time-consuming process, and appropriately so, and is a costly  process in the design of studies because you have to screen and then talk to many more patients than actually enroll in the study.  But waiver of informed consent is allowed in regulatory parlance and ethically. When there's minimal risk, the waiver does not adversely affect the rights and welfare  of the patient. The consent is impracticable, and that's a word the government uses, impracticable. It's not a word I would use.  And then participants are provided with additional information after participation when that's possible. And then there is another approach, which is education exemption.  And that's when the intervention is delivered through a provider. And so in essence, you're providing information to a care provider, like a physician or nurse  or whoever is taking care of the patient, suggesting, would you consider doing this? Because this patient meets criteria X, Y, and Z, they would be eligible for this sort  of treatment that's available. And these would, of course, have to be treatments that are routinely FDA-approved or routinely available.  And so the provider is filtering that and either can say, no, I don't want to do that, or yes, I want to do that. And so they're free to ignore the prompt.  And then there's consent embedded in normal workflows. We initially were saying we weren't going to do any studies that required person-to-person consent.  But then we began to realize there are folks in operations that are so interested in answering  a particular question that they're perfectly willing to embed consent in their normal workflow. We had a particular study that came to us where the cardiologist wanted to determine  whether two different approaches to atrial ablation for atrial fibrillation was one better than the other. And that would obviously require consent. And they said, not a problem.  There are only four of us that do this procedure. We consent all the patients already for clinical consent.  We'll just layer in a research consent and randomize them to one of the two procedures. And we'll hopefully, in a year or two, know which one's better. So that's how that works.  So I'll give you some examples. The first was chlorhexidine baths in the ICU. This was a cluster randomized multiple crossover study.  The setting was the five ICUs that have been collaborating for some time now at Vanderbilt in these kinds of studies. The study went from July of 2012 to 2013.  So it was 12 months long. Inclusion, all ICU patients. The only exclusion was Stevens-Johnson and chlorhexidine allergy and burns. And the IRB approved this with waiver of consent.  So all patients were bathed daily with no rinse cloths, either using chlorhexidine in the cloth or not. And all infection control procedures were performed as usual practice.  So nothing else was changed other than the chlorhexidine question. And the composite primary endpoint was the known, the typical health care associated  infections, CLABSI, CAUTI, C. diff, and ventilator-associated pneumonia. And all of the data were extracted from existing hospital systems.  And the infection-related outcomes were adjudicated by the infection control personnel. And this is a job they do every day, all the time.  So there was no new effort that they had to put in other than just turn over the information to us so we could include it in the study.  And so here are the ICUs that were involved, cardiovascular, medical, neurosurgical, and trauma. Over that year, there were almost 11,000 patients admitted.  No patients were excluded. So there were five units randomized into this bathing sequence. So in this first cycle, there were three control ICUs and two chlorhexidine ICUs.  And then there was a 10-week intervention for each of these cycles, and then a two-week washout period, no pun intended, between the cycles.  And at the end of the day, there were about 1,400 patients that were in those washout periods. So those were excluded.  And then that left us with 9,300 that were in the tend-to-treat population. So a little over 4,000 in the chlorhexidine and 4,000 in the control.  And this was published in JAMA in 2015. And in short, there was no benefit to daily bathing with chlorhexidine in these patients.  Now after the study was completed, the infection control group in our hospital was not happy with this finding.  They were certain that chlorhexidine made a difference and that we just didn't find it. There was something wrong with the study.  And then they began to think, well, maybe we didn't teach the nurses how to use the chlorhexidine sponges just right. And we said, maybe you didn't.  Our job wasn't to come in and tell you how to do chlorhexidine bathing. You were going to do chlorhexidine bathing. We just wanted to see if it made any difference.  So this is the sort of thing that resonates in many of the kinds of studies that we do like this. And that's that the idea, the perfect idea, might actually be effective.  But in the way that it's carried out or in the population that it's being carried out in, it doesn't work.  So that's the kind of thing you find out in a study like this that you may not find out in multi-center clinical trials as they're traditionally done.  So the next example is the PREVENT trial. We know that hypoxia during intubation in critically ill patients could result in cardiac arrest and even death.  The question is whether bag-mask ventilation during the intubation process prevents hypoxemia without increasing the risk of aspiration.  This is a raging controversy by people who study this issue. And so we conducted this patient-level randomization to bag-mask ventilation versus no ventilation  between the point where a rapid sequence induction has begun and the laryngoscope is placed.  And we found that the degree of oxygen impairment was worse in patients that were receiving no ventilation.  And if you plot it out this way, this is proportion of patients and then oxygen saturation at its lowest point.  So if you go down here to less than 70%, 10% of the population had a SAT of less than 70% during their intubation process if they were in the no ventilation arm and about 4% if  they were in the ventilation arm. And similar differences were found with less than 80% SAT, less than 90% SAT.  So it was very clear that we were preventing hypoxemia with this approach. And this was published in the New England Journal just last year. And we moved on to other questions.  What about laryngoscopy? We have video laryngoscopes now. For decades, forever, we used direct laryngoscopes to intubate critically ill patients.  So we wondered, is there an advantage to video scopes? They're much more expensive. They're 10 times more expensive and they're fragile and have to be maintained and so forth.  So in direct laryngoscopy, we have here a mannequin where a direct laryngoscope is being used. The operator is looking directly at the larynx.  And then here is a video scope where the image of the larynx is projected on a small screen here from a camera down deep in the throat of this mannequin.  And so the first question is, does it improve glottic visualization? And what we're able to show is that yes, video laryngoscope use did improve glottic visualization  substantially. This is grade one, the best visualization of the larynx. And so that question was answered clearly. But did it increase the success on the first attempt?  There's no difference on success on the first attempt. And did it decrease the time to intubation? No difference in the time to successful intubation.  So this was information that we could actually use in our hospital to say, we don't have to put a video laryngoscope on every crash cart and think we're doing a favor for our  patients or for the hospital in so doing that. It's quite conceivable that a video laryngoscope in certain very select circumstances, maybe  in the hands of anesthesiologists, intubating a very hard, difficult to intubate patient with neck abnormalities or what have you, might be an appropriate tool.  But just looking across the board with the way that it was used, at least in this intensive care setting, it wasn't beneficial.  So this was published in Critical Care Medicine in 2016. The last example I want to talk about is the Isotonic Solutions in Major Adverse Renal Events trial.  This is the SMART trial. And the hypothesis was that physiologically balanced isotonic crystalloids compared to  9.9% saline in ICU patients will decrease the incidence of death, dialysis, and persistent renal dysfunction.  Animal studies have shown that the high chloride content of saline is responsible or can be responsible for acute kidney injury.  So here we have saline, the familiar bag of saline, and then balanced crystalloids that come in the form of either lactated ringers or plasmalite.  And you can see down here the content of these. And saline is 154 milliequivalents of chloride, way greater than lactated ringers or plasmalite.  And to remind you that human plasma contains about 100 milliequivalents. So this is 50% more chloride than your plasma has.  And we all know that in the ICU we infuse sometimes liters and liters of saline into patients. So that's a lot of chloride.  And it's reasonable to wonder if that was causing some kidney injury. So the study was structured this way. Across the top is time, 2015, 16, 17.  It was about an 18-month-long study spanning those years. And these are the ICUs that were involved, MICU, neuro, CVICU, trauma, and SI.  And then X and Y was when the months were saline versus when the months were balanced. And so the way this worked is that when you ordered saline in a saline month, nothing  happened in the computer ordering system. If you ordered saline in a lactated ringers month, you would get a prompt that would say, hey, it's a lactated ringers month.  Would you consider ordering lactated ringers? And you could do it or not do it. And it was very easy to switch the order if you wanted to switch it.  So this is how it went across these various units. And at the end of the day, the primary endpoint, which was MAKE30, which is a well-accepted  endpoint in the renal world, persistent renal dysfunction, which was a doubling of creatinine at hospital discharge, new renal replacement therapy, essentially dialysis, or death.  And you can see that the increase was about 1%. So you can imagine that the difference, if there was going to be a difference between  these fluids, it would be pretty small, or else we probably would have figured out by now that one of them was worse than the other. So about 1% difference.  And what was surprising to us is that we thought we would see a lot of persistent renal dysfunction, relatively large amount.  And then some of that would translate into need for dialysis, and some of that would translate into mortality. But we didn't see that, actually.  The big signal was coming from mortality directly. So about almost 1% difference in mortality, 0.9% difference in mortality, in favor of balanced crystalloids.  So we're still trying to figure out what, you know, it could be that the chloride or some other components of saline are causing mortality that we were not aware of.  At the same time that study was being done, there was a similar one being done in the emergency department for patients admitted to the floor, so non-ICU patients, non-critically  ill patients. And of course, the event rate in the non-ICU patients is much, much lower, much less renal failure, much less death. So these columns are much shorter here.  But still, a 1% difference between the two, again, in favor of balanced crystalloids. So that was published.  Both of these papers were published back-to-back in the New England Journal in 2018. So then that brings us to implementation.  Since we were conducting this through our medical electronic health record, we had a lot of control over things.  So here you can see, this is now the medical ICU, just pick medical ICU because we have more complete data there. And this is proportion of fluids up to 100%.  So before all of this work, the medical ICU was using saline about 80% of the time. And then when the trial began, you can see the oscillations back and forth as the prompts  were guiding use of either saline or balanced fluids. And so in a saline month, about 90% of patients were receiving saline.  And then in a lactated ringers month, about 20% of patients were receiving saline. So there was some crossover between the two arms.  But given the size of the population, we were able to overcome that particular design problem. And then after the study was turned off, usage went right back up to about 80% saline.  This dotted red line is when Epic was installed. So this gave us a few hiccups, as you might imagine, when we're trying to use data and  prompts and things like that in electronic health record. And then when the trial was published, the usage of saline plummeted to our goal, which  was less than 20% normal saline in the ICU. Remember that we use saline to dilute antibiotics, to administer blood, and other things.  And we're still in an ongoing process of figuring out when we really ever need to use saline. Maybe never.  But there are a lot of drugs that are cleared for dilution in saline, but are not cleared  for dilution in lactated bringers. So we'll never see a zero use of saline, we think. So this now looks at the whole hospital.  And I just really want to make the point that the study was published in the first quarter of 2018.  And a year later, again because of the Epic installation, the prompts didn't begin until the first quarter of 2019.  But you can see here that balanced crystalloids reached usage of about 80% in the entire hospital, compared to about 20% to 30% before this all started.  Some of this represents some early changes in piloting work that we did. So that's about as fast as an implementation as you likely see. It's a year. It could have been a day.  Had we had the same computer system available, we would have just turned off one of the prompts and left the other one on. And the prompts would have said, hey, you ordered saline.  You should be ordering lactated bringers. And we think we would have gotten exactly what you're seeing on this slide.  The other projects that are coming down the pike, this is just a, we call it a high level tracker for each of the projects. We have about 30 or 40 of them.  This is just the top of the list. This first one here, Future 30, there's a phone call three days after discharge, reduced readmission rates.  We had finished that work and it's being submitted for publication. Pilot trial is, does contact isolation for MRSA carriers benefit either that patient  or the unit that the patient is in? Not talking about infected patients, just people who have at one point in their lives had a positive MRSA.  Those are by many hospital rules standard contact precaution isolated and so forth. So I won't go through all of these.  And the kind of things that we look at when we have a new project coming to us for refining and implementation is something around dollars.  Will it cost us a lot of money to do the study? Will it cost us a lot of money to implement whatever we find at the end of the study? Will it save money at the end of the study?  Will it impact quality improvement efforts? Will it impact patient or healthcare provider satisfaction? And then safety, of course.  And then lastly, I just want to talk about how to migrate this to a multi-center platform. So this is called, the effort underway is called FLOWERBED.  It's kind of a weird acronym for Workflow for EHR Embedded Clinical Trials. And this is work being done by the PEDDLE network, Vanderbilt, and these institutions listed here.  And the idea is that the goal is to have real-time cohort identification and randomization. Real-time.  And then real-time data flow of relevant subsets of clinical data, whatever it is that you're trying to collect from the patients.  And then a method for verifying the data that's used to ensure that the right cohort has been identified, that the data are accurate, and that the study is being done as expected.  So this is the tool or the system that's being used. Up at the top is the EPIC or Cerner electronic health record. And there are silent provider alerts.  I mentioned alerts that, like the saline study, the provider got the alert about saline. Well here, this would be silent.  And it would tell the REDCap system through a web listening service that this patient meets criteria for a study or may meet criteria for a study.  Then that data is transported to a screening database where that's confirmed that they do, they are eligible.  And then once they're eligible, the data to look at that patient that would normally be collected in a case report form is extracted using FHIR techniques and continuous data  pulls to get into the study database. So this is all happening at the local level, at the local hospital level.  And then in an asynchronous, intermittent fashion, data are drawn or extracted from the study database at the local site into a data lake wherever, wherever the coordinating  center is. And so you could have three hospitals. You could have 1,000 hospitals. All of this would occur almost in real time.  You could have data extracts every one hour if you wanted to. But the real point is that you're getting real time data in a way that allows you to  critique the quality of the study and the rate of patient enrollment and so forth. So this is very much a real system.  We piloted it now in several hospitals, including our own, of course. And it works as advertised.  So the question is, what are the questions you're going to be, what cohorts are you going to be selecting and what questions are you going to apply to those cohorts?  So Learning Healthcare takes help from leadership, leadership from operations, leadership in research.  It takes expertise, operational expertise, as well as scientific and pragmatic research expertise. And we have community engagement with all of these projects.  We have community members that sit in on all of our committee meetings and our workshops and our studios to provide input from the user. And it's kind of interesting.  Their response to some of this is, it falls into two categories. One is, you mean you didn't know the answer to that?  You didn't know that saline was not good and lactated ringers would be better? All the way to, oh, we just know you do that all the time. That's what academic hospitals do.  They're constantly doing these kinds of experiments. So I thought that was interesting.  There's expertise, of course, in biostatistics, data and informatics, and the platform support team.  We always are appreciative of the volunteers, the patient volunteers that come to us in these kinds of studies, in standard clinical trials across the globe.  And then finally, I'd like to just thank my family for sticking with me through all of this. This is the picture we took on the occasion of my wife's graduation from Pediatric Acute  Care Nurse Practitioner Training at Vanderbilt. This is my son, Ben, the pink. My son, Ben, daughter Claire, daughter Aline, and son Jacques.  And then thank you all for the honor of your time. Thank you.

Video Summary

Dr. Gordon Bernard, the Executive Vice President for Research and the Chief Research Officer at Vanderbilt University Medical Center, was awarded the Lifetime Achievement Award for his work in improving the care of critically ill patients. His research has focused on sepsis and acute respiratory distress syndrome (ARDS). Dr. Bernard established the Vanderbilt Coordinating Center in 1987, which has supported large, multi-institutional and international clinical investigations. He has also led the NIH ARDS Clinical Trials Network for over 20 years. Dr. Bernard's research has led to changes in clinical practice, including the use of low tidal volume ventilation for ARDS patients. He has also conducted studies on chlorhexidine baths in the ICU, bag-mask ventilation during intubation, and the use of balanced crystalloids compared to saline for fluid resuscitation. Dr. Bernard's work has shown that balanced crystalloids reduce the incidence of major adverse renal events and mortality. He has also discussed the importance of implementing research findings into everyday clinical practice through a learning healthcare system.

Asset Subtitle

Research, 2020

Asset Caption

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

Pragmatic Trials in the Digital Age

Gordon R. Bernard

Learning Objectives

To understand the concept of pragmatic clinical trials.
To illustrate the ways that the electronic health record can be used to greatly facilitate critical care research.
To highlight the importance of randomization is developing causal inference in clinical research.

Meta Tag

Content Type Presentation

Knowledge Area Research

Knowledge Level Intermediate

Knowledge Level Advanced

Membership Level Select

Membership Level Professional

Membership Level Associate

Tag Clinical Research Design

Year 2020

Keywords

Gordon Bernard

Research

ARDS

Vanderbilt

Clinical Trials

Ventilation

Crystalloids

Mortality

	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.