Hello again. My name is Michael Connor. I'm an Associate Professor of Critical Care Medicine in Nephrology at Emory University School of Medicine. I want to thank Dr. Raj Basu and the SCCM for inviting me again to speak at this SCCM Master Class on Saving the Kidneys. This is the second of my talks. The first was regarding fluid overload, and in this talk we're going to discuss deresuscitation. Here are my disclosures. Importantly, I do receive some royalties from UpToDate for a chapter regarding CRT prescription writing. However, those topics will not be covered necessarily in this lecture, although we will be discussing CRT fluid removal. Additionally, my wife is on Emory faculty, and she has no conflict of interest relative to this talk. As I've said before, I always think it important that we have our take-home message up front. First is a quote. A long habit of not thinking something wrong gives it the superficial appearance of being correct. This holds true from the 1970s to 2010s. We very much believed that we had to swell to get well in the ICU, but more recently we questioned our long-held dogma from the 70s to the 2010s, and now we need to deresuscitate our patients as soon as possible. Let's begin with the case. 45-year-old previously healthy male with no past medical history presented with dyspnea, cough, purulent sputum, fever, and depressed mental status. He was hypotensive in the emergency department, tachycardic, and febrile. His chest X-ray showed multifocal infiltrates. Lactic acid was elevated. Keratinin was 1.6. We have no prior baseline premorbid keratinins, and his white blood cell count was 25. Cultures were collected, empiric broad-spectrum antibiotics were started, and IV fluid boluses were begun. By day three, he remains on the ventilator requiring 60% FiO2 and Ipipa-12. He has marked a total body edema and is net positive approximately 10 liters since ICU admission. He remains dependent on low-dose norepinephrine, but the vasopressin is off. He continues receiving vancomycin and cefepime, and his spontaneous urine output was 950 mLs in the last 24 hours, and his keratinin is 1.8. The question I want you to think about and consider as we go through this talk is, what is the most appropriate next step in management of this patient's volume status? A, bolus volume expansion to allow for more rapid liberation of vasopressors. B, diuretics to target a net even daily fluid balance and continue vasopressors. C, beginning deresuscitation efforts to target a net negative fluid balance. D, no diuretics until off all vasopressors, or E, start acute dialysis for fluid removal. As we discuss fluid management and deresuscitation, it's really important to think about fluid administration based on the phases of critical illness. Increasingly, it's recommended that we have a volume status that follows this curve, where we increase our volume during resuscitation, we eventually plateau and endeavor to accumulate no additional volume, and then eventually we need to remove volume and return our patients back to eulemia. And we can divide this time period into multiple different points. Resuscitation, optimization, where we sort of make sure that all the I's and T's are crossed. This whole process of resuscitation and optimization should take no more than 12 to 24 hours. In the overwhelming majority of patients. Then we enter into a maintenance or stabilization phase of illness, which can be variable in length from anywhere from a few hours to a few days. These are the times that the patients are still on vasopressors, they're still on the ventilator, they're not getting worse, they're in the process of sort of slowly getting better, but we have not been able to take away an extensive amount of life support, and we have not been able to necessarily begin to remove fluid from the patient. Then finally, we enter this recovery phase or what's called the de-resuscitation phase. In my first talk, you recall that I displayed this graphic, which is that fluid overload prevents patient recovery and survival, prevents organ recovery and causes organ and kidney failure. During the first talk, we discussed that fluid overload remains very common in hospitalized patients. And as I said, fluid overload impairs organ function, prevents organ recovery, prolongs hospitalization, and increases mortality. We discussed the mechanisms by which all of that occurs and all of the studies that support those statements. In this talk, we're going to discuss about de-resuscitation, but in order to de-resuscitate, we have to focus on recognizing these phases of fluid management that we discussed and how the patients are progressing through their hospital course from resuscitation to stabilization or maintenance, and then eventually active de-resuscitation. So if we're going to de-resuscitate, then we need to ask ourselves, can we actually improve organ function and ICU outcomes by returning our patients to uvulemia? The largest study to date, and perhaps the most important study, was something called the FACT trial, which was a comparison of two fluid management strategies in ARDS. And in this study, patients were randomized to a usual care versus a conservative fluid management strategy in the setting of ARDS. And the usual care group, they just received the standard of care. And in the conservative group, they were prescribed diuretics based on a very specific protocol. And once the study was done, you could see that the mean fluid balance was drastically different between these two groups of patients, such that the conservative group quickly returned to uvulemia, whereas the liberal arm continued to escalate volume throughout the first seven days. Ultimately, the trial proved not to be powered sufficiently to determine its primary endpoint, which was death at 60 days. There was a strong trend towards a decrease in death in the conservative fluid arm, but the magnitude of that difference was smaller than what they had powered the study to see. But there was a clear statistically significant decrease in time on the ventilator and time in the ICU with the conservative fluid arm. You'll recall in my first talk that we also discussed about how volume overload causes acute kidney injury. And so therefore, it stands to reason that if we take volume away and we decrease interstitial edema and we decrease pressure within the nephron and Bowman space, that will improve filtration at the level of glomerulus and thereby improve renal function. We also know from evidence-based studies that in order to maximize or improve our chances in the ICU, that there are a number of quality metrics and things that we know from evidence-based trials do improve outcomes in the ICU. These include things like providing nutrition and absorbing nutrition in your gut, liberating from the ventilator, resolving or hopefully preventing delirium, early mobilization, preventing infections and other ICU complications such as DVTs, and fostering wound healing in surgical patients. The unifying thing between all of these is that every one of these processes and important things that we have to focus on is easier when the patient is not hypervolemic. So in summary, fluid overload impairs recovery and we need to return our patients to euvolemia. The problem is that the patients don't do this on their own and it really requires active management on our part. When left to their own devices, patients accumulate volume almost invariably in the ICU. So then how do we manage fluid overload in these patients? In preparing for this battle, in managing this fluid overload and discussing how we're going to address this, the first part of this is recognizing that our patients are fluid overloaded. G.I. Joe used to say that knowing is half the battle and so half the battle of dealing with fluid overload in the ICU is at recognizing and knowing that it exists. And this really requires strict I's and O's and we really have to measure all urine output and other fluid losses as best as we can. The intervention then or the battle is the other 50%. And while these numbers may not be 100% accurate, we know that about 15% of patients end up requiring dialysis in the ICU. The ratio of patients who receive diuretics to no interventions required could be varied from ICU to ICU. But importantly, an overwhelming majority of patients end up requiring diuretics in order to accomplish this. Now again, we spoke about these phases of fluid management and as we look a little bit closer into this, we can of course also say that this isn't really necessarily the only curve that patients can be on. They could be on a curve like this where they initially get resuscitated and then their volume status continues to very quickly increase. They could be on a curve like this where, similar to the FACT trial, as the patients get resuscitated, they then have a slower but inexorable increase in fluid, or they could be on sort of this greener curve like this. At the red curve, that sort of almost invariably ends in death and the orange curve ends in trait, long-term acute care facilities, post-intensive care syndrome, and disability. So then the question is, is we need to pick which curve are we going to be on? Are we going to be on the red curve, the orange curve, or the yellow curve? Or, and I don't really like either of those two, or are we going to even accelerate this de-resuscitation curve by making decisions earlier on that it's time to take fluid away from the patient? So when we think about managing fluid overload, the first step is that we have to really stop excessive IV fluid administration. We need to avoid hypervolemia as much as possible. This means a goal-directed expansion during the resuscitative phases of shock only, while constantly assessing whether the patients need more fluid during that phase. Move away from the concept of maintenance IV fluids, convert to a conservative fluid strategy after resuscitation, and then realize that third spacing of intravascular volume is definitely less than estimated, and the third space is not infinite. We need to have a more liberal use of vasopressors, especially after the initial resuscitation, and we probably need to have a lower threshold for inotrope support, although the studies that support this are a little bit inconclusive. Then we need to think about, does my patient have symptomatic hypervolemia, and if they do, I need to remove volume safely. This means giving diuretics in a goal-driven fashion, not just saying give 80 of Lasix, but give Lasix to keep this patient net negative two liters by tomorrow, for example. We need to think about extracorporeal fluid removal anytime patients are symptomatically fluid overloaded and failing management with diuretics, and then we need to remember that pressor use is not a contraindication to fluid removal. In fact, the phase of illness, whether they are in the stabilization or deresuscitation phase, is sort of what determines whether we can remove fluid, and if that means that we need to keep our patients on pressors a little bit longer to allow us to deresuscitate, then we should weigh those risks and those benefits. So fluid management sort of proceeds like a circle. It starts with strict I's and O's and measuring urine output. We need to discuss and establish our daily fluid balance goals. If we don't on rounds actually discuss what we want our fluid balance to be, then we won't make any plans and the patient will undoubtedly continue to accumulate volume. We need to stop excessive IV fluid administration. We need to stop entirely maintenance IV fluids or significantly limit them, and then we need to ask ourselves, is this patient fluid overloaded? If they're not, then we can just continue to go back to sort of the strict measurements, discussing daily, stopping unnecessary fluids and sort of continuing on this pathway. However, if they are fluid overloaded, then we need to consider a more liberal use of vasopressors. In other words, don't prioritize liberation from pressors. Prioritize potentially getting fluid off, and then we need to remove volume while monitoring the tolerance of the patient as we remove that volume so that we can do that safely. And when we think about fluid removal options, we obviously have diuretics and we have renal support therapy. Diuretic use is extremely common in the ICU. The furosemide stress test, which many of you have heard about, really has great prognostic value, not only for acute kidney injury, but for those who will end up requiring dialysis. And we should remember that our patients are not just water overloaded, they're salt and water overloaded because of the solutions that we use. And that combination diuretics may actually enhance sodium loss and therefore actually help us resolve total body fluid overload faster. And then when we look at dialysis therapies, modality actually makes a big difference. We know from many different studies, this is just one, that when patients get intermittent hemodialysis, even daily, it can be difficult in ICU patients to resolve volume overload. And in fact, many times they continue to accumulate volume because they get more between each dialysis session than they're able to remove on each session. And in CRT, or prolonged intermittent renal placement therapy, patients tend to lose volume more consistently. But there are some important areas of uncertainty as we talk about fluid removal. The first is, what is the rate of fluid removal? Should we be fast or slow? Can we be too fast or too slow? And then, is there really an upper limit of pressors that can be tolerated to facilitate fluid removal? Does this actually differ from patient to patient? And what is more important, the absolute amount of pressor or the relative change in pressor dose when we're taking fluid away? And then, are diuretics better than kidney replacement therapy or dialysis? And this probably is so because it's obviously less invasive than doing dialysis, and I'm not really sure there's any equipoise here. So I doubt we'll ever get any real studies that look at randomizing patients to diuretics or dialysis for fluid removal in patients who are actually diuretic responsive. That seems just to be a bridge too far. And then the question, final question is, is there really a preferred diuretic regimen? Which is the best to use, whatever it takes, or some sort of combination? So let's zero in for the rest of this talk, specifically on CRT fluid removal, because I think there's some important questions here that we need to sort of address. The one is, how do we do fluid balance management? And then, are there any important tools that can help us determine how much or how quickly the speed of fluid removal should be targeted? So when we talk about fluid management options on dialysis, there's really sort of two ways to do this. There is the fixed patient fluid removal rate, or otherwise called ultrafiltration, or dynamic patient fluid removal strategies. In the fixed strategy, the CRT orders specify a fixed patient fluid removal rate, or PFR rate. Patient fluid removal is set in the machine and not adjusted hourly. This relies on the ordering provider to determine the desired fluid balance, closely estimate the daily fluid intake over the course of the next 24 hours, and then calculate the hourly patient fluid removal rate to achieve the desired fluid balance over those 24 hours. And so basically, the way you have to do this is estimate a daily intake plus the desired net negative fluid balance, and then you divide that by 24. So if they're getting two liters in and you'd like them to be one liter net negative, that would be 3,000 in the numerator, divide that by 24, and that would be your set rate. Obviously, things can change a lot in the ICU from hour to hour, and so many of us actually recommend more of the dynamic patient fluid removal strategy in which the CRT orders specify the desired hourly goal fluid balance. In other words, I want this patient to be net even, or negative 100 an hour, or negative 100 to 150 an hour, and then the bedside nurse will adjust the patient fluid removal rate on the CRT machine hourly based on the hourly intake to achieve the desired fluid balance. And in this situation, they would take the goal net negative and add it to the intake, and that's what they set the machine for. So if you have, you want your patient to be net negative 100 mLs an hour, and they're getting 200 mLs that hour, then the machine needs to be set for 300 mLs per hour. Now, turning to the speed of fluid removal. So with the known negative consequences that patients have from fluid overload, the important questions that then follow is how fast can we remove volume, and is there a safest rate for fluid removal? For the rest of this talk, I'm just going to go through a number of studies that have actually been published over the last few years that have attempted to address this question. The first two trials are two very related trials. They are actually published by the same group using the same data set, but asking different questions. The JAMA study wanted to look at fluid removal rate and mortality in ICU patients, and this blood purification study used the same data set to look at fluid removal rate and recovery from acute kidney injury in those who are requiring CRT. So these two studies both use a retrospective cohort study from the RENAL trial. The RENAL trial was a randomized controlled trial of renal replacement therapy intensity in acute kidney injury in New Zealand and Australia. It looked at high-dose versus low-dose renal replacement therapy, and in these studies, they looked at a retrospective cohort of that trial cohort with the two endpoints for the two different trials being time to independence of dialysis and then risk-adjusted 90-day survival, and the exposure variable was UF net rate in terms of mLs per kilo per hour, and that was calculated as cumulative fluid removal by the CRT machine divided by the admission weight times the total duration of CRT. So in other words, if in 96 hours they removed 48 liters of fluid, they would take the 48 liters, divide it by the patient's weight times 96, and get mLs per kilo per hour. Importantly, this is not the net fluid removal of the patient. This was the fluid removal that the machine itself removed. The patients were divided then into tertiles, a small amount of fluid removal, less than one mLs per kilo per hour, one to 1.75, or greater than 1.75 mLs per kilo per hour. Now, after they did the analysis, it's important to recognize that the time to randomization in this trial, in the original trial, and to initiation of CRT, and the percent of patients who were actually started on dialysis for fluid overload as their indication to start dialysis, increased as the tertiles increased. So these patients were more fluid overloaded than these patients, and these patients had a longer time to start dialysis. And what did they show? So they showed that compared to the lowest tertile of fluid removal, independence from dialysis decreased as more fluid removed. In other words, as they went along, if you had higher fluid removal rates, you were more likely to remain dependent on dialysis. And this held true across the spectrum when you compared the high towards the low dose. And you can see in this table here, and then death was also marginally higher at 90 days in the patients who had more fluid removed. Now, I want you to compare that to several other trials. This was a trial, retrospective trial, looking at about 1,100 patients with acute kidney injury requiring dialysis. And importantly, to be included in this study, everybody had to be at least 5% fluid overloaded at the time of dialysis initiation. And they divided patients, again, based on the UF net rate of mls per kilo per hour. Again, this is the total amount of fluid that was removed by the CRT machine, not the net balance, but the total CRT fluid removal. And again, not accounting for fluid intake, divided by the weight and the number of hours. And again, patients were divided into less than 20 mls per kilo per day, 20 to 25, greater than 25 mls per kilo per day. And what did they show? They showed that compared to the low intensity arm, those in both the moderate and the high intensity arm had an unadjusted odds ratio that showed that it was protective to have more fluid removed. And when it was adjusted for a lot of statistical variables, then the high dose group still had a protective effect for having more fluid removed. And you can see that here. Again, this is now mortality on this side, but you can see that the patients that had more fluid removed had a decrease in mortality. Two more studies to go through. The first was, again, a retrospective study of the MIMIC4 database. This is a big database of ICU patients out of the Harvard system. All patients had to be on CRT for greater than 24 hours, starting within 14 days in the ICU. They found about 900 patients after the exclusions. And in this situation, they again looked at the cumulative fluid removal by CRT only within the first 48 hours that the patient was actually on CRT, and then divided by the weight at the duration. And the primary outcome was 28-day mortality. In this group, the tertiles, again, were less than 1.6, 1.6 to 3.1, and greater than 3.1. And obviously, these tertiles are much higher than the tertiles in the first study I showed, first two studies I showed you that were out of Australia from the Renal database. The lowest tertile here was the highest tertile, essentially, in that other, in those first two studies. And in this group, the percent fluid overloaded, so in other words, were they 10% fluid overloaded, and the percentage of people who were fluid overloaded before CRT increased, and the duration of CRT decreased as the tertiles increased, duration to initiation, excuse me, of CRT decreased as the tertiles increased. And what did they show was that in the 28-day mortality, there seemed to be sort of a U-shaped curve where too little fluid removal, where low fluid removal increased your risk of death, and very high fluid removal increased your risk of death. Then the final study to look at was a study of net ultrafiltration rate and impact on mortality out of Mayo Clinic. Again, it was a retrospective single-center study looking at patients from 2006 to 2015. They had about 800 patients in total, excuse me, about 1,400 patients in total, excuse me. And again, they looked at the cumulative fluid removal divided by the weight and the duration, and the fluid balance at initiation was clearly higher in those that had the high-intensity arm. What did they show? They showed that 30-day mortality was significantly decreased in patients who had more fluid removed, and that the MAKE 90, which is major adverse kidney events at day 90 defined by these things here, was also lower at 90 days in the patients who had more fluid removed. And when you look at both mortality and MAKE 90, and you looked at the adjusted odds ratio, having more fluid removed was protective in this situation. So are you confused about what we've seen? Because these studies have shown different things. So how do we explain these discrepancies, and what other issues do I have with these studies? Well, first of all, the location of the study. Studies in which the higher fluid removal was bad were performed in Australia and New Zealand, where they tend to be very fluid conservative. Studies where higher fluid removal was good were all done in the United States, which tends to be a very fluid liberal place. Additionally, I'm not sure that fluid removal, net fluid removal of ICRT, is really the right metric, because all of these studies use quote-unquote net fluid removal, but when you looked at it, this was actually total fluid removal, and all of these studies pay no attention to how much fluid was going into the patients, and therefore they can really say nothing about the overall total body fluid balance for these patients. And I would argue that in a lot of these studies, the low arm of the fluid removal probably wasn't actually making a lot of these patients net negative. And so there's tons of bias in this. Specifically, if we know that fluid overload itself causes bad outcomes, then how can we differentiate then if fluid, too much fluid removal is bad when we know that we, that fluid overload itself is predictive of bad outcomes? Additionally, why would I prescribe a lot of fluid removal to a patient that is not fluid overloaded or someone who's getting a lot of fluid? In other words, if they're not fluid overloaded, why would I put them on a high fluid removal strategy? So therefore, I'm already concerned as the prescriber that this patient's having deleterious effects of fluid overload, and that's why I'm prescribing a high amount of fluid removal. And so therefore, is it really fair then to say that fluid removal is bad when it may just be the actual fluid overload itself that's bad? Okay, so what do we do for fluid removal at my institution? I try to remind people that fluid overload is bad and prevents recovery. In the maintenance phase of illness, at minimum, we try to target a net even daily fluid balance to avoid further accumulation. We start de-resuscitation fairly early, and we monitor the tolerance closely. We use tools to reassure us that the patients are tolerating the fluid removal, things like point-of-care ultrasound, stroke volume variability, passive leg raise, and then if they're tolerating the fluid removal and a net even fluid balance, we start increasing the speed of fluid removal, and we tolerate low doses of vasopressors with a race to liberate from sedation in the ventilator, not a race to liberate from pressors. And in summary, to end this talk, I just want you to remember that fluid management does require a lot of work by the ICU nurses, clinician, and staff documenting intake, measuring urine output, and it requires active participation. The phase of illness should guide the fluid administration and the fluid de-resuscitation rather than how much pressors necessarily the patient is on. Volume overload is common and hampers recovery. We need to discuss fluid balance strategies daily, and we need to actively attempt to de-resuscitate by fostering net negative fluid balances in the later phases of critical illness. That's the end of this talk. Thank you very much for your help, and if you have any questions or comments, you can reach me by email or Twitter.

deresuscitation

fluid overload

ICU patients

organ function

conservative fluid management

diuretics