the privilege for about the next 15, maybe 16 minutes, talking about what Mike was just kind of framing for us, this concept that we look at repeated clinical trials that are not always reaching the same uniform conclusion, and how do we reconcile those as clinicians and apply those to our practice. So that part I was asked to do really focuses on protocolized resuscitation and sepsis, looking at the seminal 2001 EGDT article, and then integrating that with the process promise and ARISE data. In terms of disclosures, I did serve on the scientific advisory panel for a company called Endpoint Health in 2022. And like Emily was saying, I serve now as a member of the Surviving Sepsis Campaign Steering Committee. And in terms of potential intellectual conflict of interest, I really do think sepsis protocols specifically focused around early recognition and early management have improved patient care. And I will distinguish those protocols from specifically early goal-directed therapy. So I think to sort of start this conversation, we need to go back in time a little bit. So we're in early 2024. But in 2001, the landscape of sepsis was quite different than what it looks like now. And so this is some administrative data that you're probably familiar with from other talks that you've been to and papers that you've read that looks at sepsis mortality over time. And it uses kind of different ways of defining sepsis, whether it be administrative coding data or data extrapolated from randomized controlled trials reported at that time. But when you look at this, as I circle 2001 here, and the dark line here is the regression of multicenter trial reported data for the mortality of septic shock. And it's 2001, and the reported mortality was right around 40% in most randomized clinical trials. And I'll just let that sink in for a minute. And then I'll ask you, do you think that's really changed over time? And I see some nods in this. And I think it has changed over time. The landscape is changing. And I will argue to you that this particular study from Manny Rivers in 2001 is a large part of how that landscape changed how we think about sepsis and how we think about resuscitating patients with sepsis. And so this came out in the New England Journal in 2001, and there was tons of buzz about this. This was a positive trial in septic shock, a floridly positive trial in septic shock that showed a massive reduction in mortality associated with something specific that we could do. And those of you who have followed clinical trials in critical care for a long time will know that we don't see very many of these trials come along. And so it generated a huge buzz. And so the design of this trial is this was a prospective randomized controlled trial that was partially blind. It was done in a single center in the emergency department at Henry Ford Hospital in the United States. And to be included in the trial, you had to have two or more SIRS criteria. You had to have a systolic blood pressure less than 90. After about 20 to 30 milliliters per kilogram, fluid bowl is given over 30 minutes, or a lactate greater than 4 millimoles per liter. So a pretty sick patient population in this trial from that. And I'll call your kind of mind to the design of the trial for those of you who have been a minute since you thought about this or looked at this trial. And so patients were randomized. You're randomized to standard therapy or early goal-directed therapy. The standard therapy arm looks kind of different than what we probably do for standard therapy now, targeting a CVP of 8 to 12, a MAP greater than 65, and a urine output greater than 0.5 mils per kilo per hour. And then early goal-directed therapy sort of added on these six-hour goal of achieving a central venous oxygen saturation greater than 70 percent, a oxygen saturation greater than 93 percent, and a hematocrit greater than 30 percent for six hours. They were then transferred to the ICU as needed. And I put one of the points that I want to come back to later in this talk of antimicrobials were given at the discretion of the treating clinicians. Antimicrobial timing wasn't protocolized. It was not a criteria for randomization to have received antimicrobial therapy. And in this landscape in 2001, there was no sort of protocolized screening for sepsis upon triage to the emergency department. And I think most of us will say, that probably doesn't reflect what we hope current practices in our emergency departments now. So I already told you this study was floridly positive, and it was, right? So the 28-day mortality was the primary endpoint for this. In the standard therapy group, the 28-day mortality was 49 percent, and the early goal-directed therapy arm had a 28-day mortality of 33 percent. So a massive reduction in the risk of death associated with the early goal-directed therapy as reported in this trial. So that's all well and good, right? And then hewing to our typical scientific process, you know, we thought about this for guidelines. And so when you look at the 2012 surviving sepsis guidelines, there was a recommendation for protocolized quantitative resuscitation with sepsis-induced tissue hypoperfusion. And in the six hours, hitting all of those same targets that you'll recognize from the targets of the early goal-directed therapy trial. And we then sort of continued along this process and said we should, this is a single-center, partially-blinded trial. And part of our scientific method, of course, includes can we replicate results? And so appropriately, there were three large multi-center trials of early goal-directed therapy that the design, we won't go into in great detail today, but the design of this was really, really cool in the sense that folks putting these studies together met a priori. One was based in the U.S., one was based in England. The ANZICS trial sites were mostly in Australia and New Zealand. And the a priori decided we're going to design these trials similarly enough that we can do an individual patient-level meta-analysis combining results from three large international trials. And that was a really novel concept that they were able to do this. And all three trials enrolled patients to be randomized in early goal-directed therapy versus usual care. And then the process trial, which was a U.S.-based process trial, had a third arm of protocol-based standard therapy added to that trial design as well. So you probably already have jumped to the punchline. It's 2024. We kind of know what these studies show. This is the individual-level patient meta-analysis that looks at these three trials. And you'll see the line of identity, right? And you see the, I have the mouse here, process, arise, and promise. All three of the point estimates are pretty close to one, and certainly all the confidence intervals there are kind of focused right on that line of identity. Whereas the RIVERS trial is up at the top here, and that clearly has results favoring early goal-directed therapy. And so this, I think, helps trigger Mike's question, right? So how do we reconcile this? And how do we go about thinking through these disparate results that we're seeing, or these discordant results that we're seeing? So let's unpack some of these trials a little bit more before we kind of think about how we can contextualize that. So when you look at the process trial here, and this is, again, published in 2014, so 13 years after their initial early goal-directed therapy trial. On the left-hand graph here, you have the hourly volume of intravenous fluid received in the first six hours after randomization. And you can see that the red line here is early goal-directed therapy. The green line here is usual care. And those are the two groups that we're going to focus on here. And the fluid lines are slightly different. Protocol-based early goal-directed therapy got a little bit more fluid hour by hour than the usual care group did. But they all got fairly substantial volumes of fluid. You can see here early goal-directed therapy got a total of 2.8 liters. Usual care got a total of 2.3 liters. When you look at the right-hand graph here, you can see the use of vasopressors, the use of packed red blood cells, endobutamine, those hitting those other elements of achieving a central venous oxygen saturation of greater than 70%, a peripheral saturation of greater than 93%, and hemoglobin greater than 30. So you can see early goal-directed therapy got a little bit more vasopressors, substantially more red cells, and more dobutamine, although overall usage was not terribly high of both red cells and dobutamine. And in the process trial here, the receipt of intravenous antibiotics prior to randomization was over 95% in both groups. And so I want you to focus on that because we'll come back to this discussion point in saying, how similar are these studies and how different are these studies? So again, how do we reconcile these different findings? What's different about them? So I've taken the liberty of pulling out some data for some key points about each of these studies for your consideration. So the RIVERS trial, patients were randomized in less than an hour on average, so 0.8 hours on average to randomization from ED to randomization, which is extraordinary when you think about the logistics of doing that. In process, it was an average of about three hours. In ARISE, it was just under three hours at 2.7, and PROMIS was about two and a half hours. Looking at the fluids received prior to randomization, RIVERS doesn't report it, but I would argue to you in that first 0.8 hours, it's probably not the same volume of fluid as received in the average of two and a half to three hours that process ARISE and PROMIS has. And so when you look at the volume of fluid there, it's kind of focused probably around this roughly two liters-ish range. In terms of a weight-based average, we can calculate that for process and ARISE, and it's about in that roughly 30 mils per kilo range of fluids received prior to randomization. And then you can look at the volume of fluids received in that early goal directed therapy period at six hours after randomization there, and you can see that in that RIVERS trial was about three and a half liters, which is substantially more than in the other three early goal directed therapy trials. The timing of antimicrobials is probably fairly different amongst these trials. So ARISE required as an enrollment criteria that the first dose of antimicrobials be initiated prior to randomization. PROMIS required that the first dose of antimicrobials be initiated prior to randomization. PROCESS didn't report that this as an enrollment criteria, but when you look through, like dig through all the supplemental materials, it was over 95% got antibiotics prior to randomization. And RIVERS, again, I'll bring you back to that graph we looked at at the beginning, where antimicrobials were given at the discretion of the treating physician. And again, I'll bring your eye to the timing of randomization here. So in process ARISE and PROMIS, we have a group of patients who are treated prior to randomization, and as part of that treatment prior to randomization includes fluids on average in the roughly 30 mil per kilo range and antimicrobial therapy. That's not the same study population as in the RIVERS trial. The RIVERS trial patients are enrolled very early. The pretreatment is probably substantially less than what we're seeing in the process ARISE and PROMIS trial. And I think this baseline SCV02 data, when it's available in these studies, reflects that these are probably very different patient populations at that point because of the treatment that they received prior to randomization. So in RIVERS, we're familiar that the sort of first average central venous oxygen saturation was 49% in process ARISE and PROMIS. It was all in the 70% range. And you'll remember that that's one of the targets of early goal-directed therapy, which is kind of awesome that we already hit it, but it also makes your randomization to a treatment where you're targeting that thing and you've already achieved it, which makes, I think, the sort of intervention less likely to show benefit. And again, once again, just sort of refreshing our memories about the time to randomization. So I think there are real caveats to the interpretation of early goal-directed therapy process ARISE and PROMIS in that I hope I painted a picture to you where I think the overall management of sepsis in this intervening 13-year period changed substantially. In process ARISE and PROMIS, these patients, prior to randomization, received fluids. They received antimicrobial therapy, which we think are obviously some of the most important drivers of outcomes for patients with sepsis. And I think we need to be very cautious about over-interpreting the results about protocols in sepsis versus should we be doing early goal-directed therapy based on the negative trials of process ARISE and PROMIS compared to early goal-directed therapy from that. So I'll continue and basically say that the mortality rates have also changed very substantially. So you remember the timeline we looked at sepsis mortality over time. Process had 60-day mortality as their primary endpoint. ARISE and PROMIS both had 90-day mortality. And this is a randomized trial background. This probably is not generalizable to real-world data. And I think we see that from epi data that our results largely don't look like this. But these are exceptionally low mortality rates for septic shock, particularly compared to when you look at that historical control from the 2001 randomized controlled trials of average mortality around 40%. And so this really probably does reflect, I think, a sea change in practice over time. And when you look similarly, even if you're saying, well, that's all randomized trials, that's all an idealized setting, this is data from the New York State mandatory statewide sepsis quality reporting initiative where New York State was required to submit data on each and every patient in New York State with severe sepsis or septic shock to the State Department of Health. This is in the same time period as the process ARISE and PROMIS trials. And so it's in the sort of mid-2014, 2015 range. And you can see that New York State mortality for septic shock was in the mid-20% range. So again, I think representing a decline compared to that 2001 time period. So once again, circling back to reconciling these different findings, we have three large, well-conducted, really cool methodologically combinable individual patient-level data, multi-centerized statistics that didn't demonstrate a benefit to protocolized resuscitation for septic shock in the context of recognition, fluids, and early antimicrobials. In that setting, I think we can say that we have not been able to demonstrate an incremental benefit from further protocolizing resuscitation after that point. And I think the danger for us is to conclude that there's no benefit to sepsis protocols full stop rather than putting it with this context of in an environment where we're effectively recognizing, giving patients appropriate, hopefully appropriate, but antimicrobials up front and a fluid or some fluid resuscitation volume TBD, that additional protocolized benefit may not have additional benefit on that. And I would just conclude with those of you I've ever talked to probably know this is like my perennial mantra that sepsis screening, early identification are essential and really make a difference for our patients. So I will stop there. And once again, thanks our moderators.

sepsis treatment

clinical trials

EGDT study

resuscitation protocols

patient outcomes