Thank you, and thank you all for being here on the afternoon of the last day. We appreciate it. And you'll hear from me some similar topics to what you heard from Dr. Evans, and maybe a little bit of a slightly different flavor also. Like Dr. Evans, the biggest conflicts that you should think about for me giving this talk is that I have intellectual conflicts. I am a senior author of one of the trials that I will spend a reasonable amount of time going through, and if you think you can be heavily involved in a trial and then step back and be objective about the way people interpret that, you're kidding yourself, right? I at least think I can not take it personally when people critique my trial, but you should know that I may have some rose-colored glasses on when I talk about that trial. Secondly, at the end, I will talk about meta-analyses of a number of these trials and show you how they combine, and I'm also an author on those. I'm not a lead as much as I was on SMART, but I'm an author on those, so take that with a grain of salt, too. Funding is mostly funding for the trials that I'm going to talk about, and then you can see in the consultant at the bottom, none of that is actually pertinent. Those of you who know, Baxter does make IV fluids. I don't consult for Baxter with IV fluids. I do nutrition consulting for them, so none of my consulting is with regards to IV fluids. Here's what we're going to do. We'll talk about the evidence for and against balanced fluids in large RCTs in critically ill patients. We'll talk about the differences in methodology in those trials and how that may help us understand why they have slightly different results, and then at the end, we'll summarize kind of where I think we're at. So this all started in 1832. This took a little digging to actually figure this out, but this is the first report of a patient actually receiving IV fluids. This patient had cholera and got IV fluids, and in the report, they actually claim IV fluids are a success because only 75% of the four patients that they treated died, and this was the lady that actually survived, and that 25% survival from cholera in 1832 was a huge success. Here's where we're at now. We give, in the United States alone, about 200 million liters of saline to more than 30 million patients, so IV fluids are pretty ubiquitous in our care of the patients, and I think they may be the most common therapy given to our critically ill patients, although you may argue with me that oxygen may actually be above that, but if you're talking about patients in the hospital, then in general, not just the ICU, this is the most common therapy. I don't have to tell you guys. They differ in composition. You can see in this slide, the far left or the second left column is plasma and what our plasma composition looks like, and then saline, lactated ringers, and another balanced solution, plasmalite is a trade name. You can use normosol if you'd like. They're identical, and you can see in this that there's a big difference in the fluids from what's in plasma with the balanced fluids being more similar to our chloride levels in plasma than saline, and what does that mean? What do we know before any of these large trials? We know from some preclinical data that saline causes a hyperchlorimic metabolic acidosis. We know that saline causes renal vasoconstriction. This is in healthy volunteers, which is code for medical students that were enrolled in the study. We know that saline may cause an acute injury, acute kidney injury. There are some observational data that we're pretty convincing of that, although you'll see as we go through some randomized trials that may be less convincing now, and then there was a suggestion that maybe saline even increases mortality. All of these were from non-randomized trials. They're preclinical, and they're what prompted lots of people to have an interest in this area. So let me first talk about SMART. SMART is a trial that I will look at through rose-colored glasses. So SMART was done as a cluster crossover trial in a pragmatic way, and here's the setup. Every ICU in my hospital, we have five subspecialty ICUs that you can see that participate in this trial. They were all randomized to either balanced fluid or saline for a month at a time, and we knew the first of the month, every month, that you change to the other fluid, and then you can see starting about seven months in, we actually coordinated this with our pre-ICU locations, specifically our emergency department and our intensive and our operating rooms. At my hospital, these fluids don't actually get assigned to a patient, so they sit in this drug delivery. It's called a Pyxis machine, and they don't actually have a sticker on them that says a patient's name. This was fortuitous for us because it allowed us to just say, fill this machine with whatever the randomized fluid is for that month, and don't worry about assigning it to a patient. And it was important because when I'm standing in the room, and the patient's blood pressure is low, and I say to the nurse, can you please go get a liter of fluid to give this patient, the nurse would get the liter that the patient was randomized to in the study because that was the option that was available. I did have the option of ordering the other study fluid, and here you can see how that happened. If I ordered the off-study study fluid, I would get a prompt in our electronic medical record that would say, hey, this patient's in the trial. They are supposed to be randomized. This is actually to a balanced fluid. Pick one of the balanced fluids, please, one and two, or pick a reason, three, four, and five, why this patient needs saline even though they're in the balanced fluid month. So you could pick three, four, and five. Five is sort of like because mom told you so. Three and four actually had some objective criteria, although it won't surprise you to know that we got a lot of people that picked three when the serum potassium was 3.4. So here is what ended up happening in the trial. On the left-hand side, you can see fluid administration, about two liters of balanced crystalloid. This was in the balanced fluid arm, about two liters of balanced crystalloid over seven days, but we also got about 500 cc's of saline on average in those patients. And that was either mom told me to or IV fluid carriers like antibiotics, those sorts of things that were saline. Same thing, very similar in the saline group. There's a little bit less off-study use of balanced crystalloids in the saline group because most of the carriers, the antibiotic carriers, are in saline and not in balanced crystalloid. Here we go. So what do we find? So our primary outcome was major adverse kidney events, a combination of death, new dialysis, or doubling of your carotene. And here you can see there's about a one percent difference in the cumulative outcome of this. So a little bit different for each of the components. That's the top part of this, but not statistically significant for any of those components. But then in the composite outcome, a statistically significant difference. And you can see the only organ failure difference we saw was in RRT free days or kidney replacement therapy free days, which was lower or higher, more free days in the balanced group. Here is the subgroups, and you can see in the subgroups it's a pretty consistent signal. There's a couple of them that stand out that you go, I wonder if that's real and a benefit for saline. But in general, mostly balanced. So this got people excited, and we said, all right, maybe we should all be using balanced fluids. But at the same time, there were two other large studies going on, one by the Brazilian group, BrickNet, called Basics. And you can see the characteristics here. Basics is 10,000 patients enrolled when they got to the ICU, 75 percent of them within a day of their ICU admission. They actually had a blinded trial. I told you how ours was set up, and it was not blinded. It was open label. They had a blinded trial. You can see the breakdown of their patients there. They excluded anybody who already was on dialysis or you could pretty much predict was going to get dialysis. So they took kind of ongoing renal dysfunction out of the picture, and their outcome was 90-day mortality. And here's their Kaplan-Meier curves. They actually do separate a little bit, and cumulative mortality is slightly lower in the balanced group. But the p-value here is 0.47, so it's not statistically significantly different. Here are the numbers that you can actually see. There's a 0.8 percent difference in mortality at 90 days, and then the bottom is kidney injury, no differences in kidney injury. The second trial is called PLUS. It was done by the Australians. This was actually a hard trial for Anzacs to do, and it was a hard trial because of all of the places in the world, Australia had already bought into balanced fluids. They had kind of convinced themselves that balanced fluids were the way. So they had a hard time enrolling. They did individual patient enrollment, double blind trial, but they had a hard time enrolling because they didn't have equipoise in all of their clinicians, and they'd have clinicians say, no, I don't want this patient to potentially get saline. They enrolled 5,000. They stopped early because they weren't enrolling fast enough and knew that they were going to have problems getting to their almost 10,000 goal. Again, enrolled at ICU admission, double blinded, 90-day mortality is the outcome. Here you can see no differences in any of theirs. Their mortality is 0.2 percent difference between the two, which is obviously not statistically significant. Here is their subgroups, no subgroups that it looked like, wow, balanced is really winning or saline is really winning. So why the differences here? I think all of these are actually consistent, and I'll show you a table that actually has a mortality odds ratio, and when you look at them, you'll hopefully agree that the signal is actually more consistent than we get on first blush. But remember, two of them, BASICS and PLUS, are blinded. SMART was unblinded. BASIC and PLUS are both individual patient randomization, and SMART is a cluster randomization. And then size. SMART is actually 15,000, almost 16,000 patients. BASICS was about 10,000, and PLUS was about 5,000 patients. Here's your summary of the trials, and I put in here SPLIT, which was done by the New Zealand's Paul Young prior to PLUS, and SALT, which was actually done by our group as a pilot for SMART. So all of those two are included in here also. But if you look at the mortality outcome, you can see these odds ratios are not that different across all five of these studies. One of the other things that I think comes into play here is heterogeneity of treatment effect. What do you mean? Well, specifically, TBI, traumatic brain injury patients, are included. They actually aren't included in PLUS. They were excluded from the Australian trial, but they're included in both SMART and BASICS. And when you look at this population, here it is for SMART, and it's not impressive with SMART, but it does, the point estimate is on the side favoring saline. And here it is for BASICS. In BASICS, there's a statistically significant interaction that says the effect of fluids is different on TBI than it is on the other patients that are in this trial. And they actually saw a mortality difference favoring saline in their TBI patients enrolled in BASICS. In the overall trials, all of those TBI patients are included, and they obviously will move the signal towards the null a little bit when you include those in the analysis. Here's TBI in a meta-analysis. This meta-analysis is actually the BASICS investigators, the PLUS investigators, and Matt Semler and I from the U.S. side putting all of our data together and doing a meta-analysis to see when we combine these into 35,000 patients, what do we get out? And here you can see TBI on a gross meta-analysis. I'll show you an individual patient meta-analysis in a minute. The last one, which I think is the important one, and honestly, it scares me a little bit. Not really scares me, but scares me, and I'll explain why in a minute. Timing of enrollment and pre-enrollment fluid. So here you're looking at BASICS, and BASICS collected the fluid that was given to the patient prior to enrollment. And you can see here, this is the number of patients that got balanced fluids in the two groups, and it's about half of the patients that got balanced fluids. And if you say, well, but did they get any volume? Who got over a liter? It's about a third of the patients in both groups that got at least a liter of balanced fluids before randomization. Similar with saline, but smaller numbers. It's only about 20 percent that got a liter of saline prior to enrollment. Why do I care about this? Well, this is the contradiction and the contrast between the traditional randomized trial and our pragmatic trial that we did in SMART. And remember, in SMART, we got lucky in the last 15 months because we were coordinated with emergency departments and operating rooms, so they were given the same fluid pre-ICU that we were randomized to in the ICUs. And that allowed us to have the same fluid throughout the patient's hospitalization, which is the bottom. As opposed to the traditional RCT, where somebody gives fluid in the emergency department, then the patient comes up to the ICU and somebody else decides on what fluid they want to give. Then we approach them with a study coordinator, and they get randomized, and then they get a randomized fluid, and you get this mixing of fluids. And you say, well, Todd, does that really, really matter? This actually happened. There are good things that happen at 2 a.m. This thought process actually happened at 2 a.m. in a hotel room between Wes Self, Matt Semler, and I when Wes said, hey, what if we compare the first seven months where we didn't have coordinated fluid and the results with the last 15 months where we did have coordinated fluid? And here's that comparison that's published. When you can see the first seven months where we only controlled fluids in the ICU, we don't see any signal at all between balanced and saline. They're pretty much what I call the purple line effect right on top of each other. When you look at the last 15 months where we controlled the fluids pre-ICU, this is where you see the actual separation and where balanced fluids are better than saline. And that's why I think pre-enrollment fluids matter. And the reason that scares me is because it means for all I do in the ICU, the most important fluid that may be given to my patient is that first liter of fluid that they get maybe in the back of an ambulance, to be honest with you, let alone before I get a chance to decide what fluid they're going to give. Because we had these data out there when Basics did their analysis, they actually planned on doing this analysis. And here is they're all patients. And you can see balanced only is patients who only got balanced fluid prior to enrollment. And saline only is patients who only got saline. If it's pure balanced, they also start to see a signal for balanced fluids. All right, what about our individual patient data meta-analysis? Here are the results from it. And you can see there's some heterogeneity of treatment effect. I've specifically left sepsis. TBI is now significant when we actually, for saline, when we actually do an individual patient data meta-analysis. And then in this, we were able, PLUS doesn't have pre-ICU fluid. So between Basics and SMART, we were able to combine those and say, what's the signal if you got pre-enrollment saline? And you can see if you got none, that's where you actually see the signal that balanced may be better. Here is an actual putting all of the studies together as the meta-analysis. And even when you include the TBI patients, which this does, you see this effect. And I think it's important because I think it tells us this is a small effect. This is a 4% relative reduction in mortality. But it's highly likely that this is real, someplace between 9% and 1%, 9% reduction and 1% increase in mortality of balanced fluids. IPDMA, same thing, 4% reduction in mortality, about a 90% confidence that this is actually real and balanced fluids improve mortality. So here's where I think we're at. I think differences in methodology may explain some of the at least gross differences in balanced fluids across the outcomes across the trials. I think overall, the effect on mortality is actually pretty consistent across the trials. I think it's small. And I think you have to have a ton of patients in order to actually see it fleshed out. I think it's also, despite being small, the meta-analysis suggests that it's highly probable that balanced fluids are actually the better fluid for our all comers in the ICU take out the TBI patients. And we had a statistician that when we got done with SMART said to us, why do you care about this? It's a 1% difference in SMART. And I said, well, the reason you care is if there's 30 million patients getting these fluids every year, 1% of 30 million is a lot of patients that we could potentially be changing the outcome of in this arena. And then the thing that makes me a little bit anxious, but I think the data bear this out, is that what really probably matters the most is the composition of the first fluid that the patient gets. And unfortunately for us in the ICU world, that's often before they get to us in the intensive care unit. Thank you for your time. I think I'll still be here at the end. So if you have questions or want to throw tomatoes or whatever, we'll be happy to take that. So thanks. Thank you, Dr. Wright.

fluid composition

critically ill patients

randomized controlled trials

balanced fluids

ICU settings