Latest / CRITKNOW / The ICU Dialysis Debate Just Got a New Trial - Here's How to Read It

Transcript

The ICU Dialysis Debate Just Got a New Trial - Here's How to Read It · CRITKNOW

• Ramesh Venkataraman: Hi everyone, welcome back to our crit no podcast. This is our 11th episode. Look forward to no doing this again. we have over 650 subscribers now, and I think the subscriber list is growing, Bharat. So I'm happy about it. So thank you for everyone who has been following us. and you can follow us on YouTube, Spotify, and and Apple. So Bharat, I thought this week there are Lot of good trials that came out recently, and I think this week I thought we'll talk about the ICR AKI trial. I think this compared intermittent hemodialysis with continuous renal replacement therapy in patients with acute kidney injury, published in the most recent issue of New England Journal of Medicine. So before I go on to the be we go on to the trial Bharat, I thought we'll just quickly overview the existing literature, right? if somebody needs renal replacement therapy in the ICU, the question always has been: is one modality better than the other? Traditionally, there has been three modalities that have evolved over time. The first is obviously being intermittent hemodialysis, and then the exact opposite of that spectrum is continuous renal replacement therapy. In intermittent, we do short duration, high-efficiency dialysis. CRRT is kind of a slow 24-hour. a run. And people generally believed that CRRT would be better if patients were hemodynamically unstable. So, because the solute fluctuations, the fluid removal rate are all gentle over a period of time, and that would basically help the patient's hemodynamics and therefore the outcomes. There have unfortunately not been great many studies looking at this question. And very few studies have compared them head to head. In fact, there is only one major RCT, the hemodiaf study that was published in Lancet that looked at this question almost about two dec two decades ago now, and where they actually looked at 360 patients across 21 ICUs in France, and they randomized patients to intermittent dialysis versus continuous renal replacement therapy. That study unfortunately did not show any outcome differences between the two modalities. There were a lot of limitations methodologically to the study. There's a very, very well-written editorial by Dr. John Callum for that study there. But the most common limitations were that the dose was variable. And it was not like a controlled dose across both the modalities. So both modalities, the dose varied over a period of time. And therefore, one of the criticisms is that is it an effect of variable dose versus variable modality? What are we looking at? And that was one of the major limitations. Second one, I think they had to change the inclusion exclusion criteria halfway through because of the slow enrollment rate, and that has also been criticized. but in effect, that is the only randomized control study we had. It is also 360 patients, so maybe did not have power enough to show differences between the two arms. So those are all the limitations in the hemodiaf study, and after that, unfortunately, there has not been any major randomized control trials looking at these two modalities head to head. the reason probably being evolution and the widespread use of The the hybrid therapies, the sleds and stuff like that, where most of the places which could not afford CRRT go into sled, and sled give provides a compromise between the two different techniques and takes over the advantages of both. And I think many centers, including ours, have transitioned a lot of the patients from CRRT to sled, and maybe the incentive to do a randomized controlled trial was not there. In 2021, there is a Network meta-analysis that looked at IHD versus CRRT. There was a lot of heterogeneity in terms of how the data was pooled, but again did not show any significant differences between the two groups. So that's pretty much it in terms of evidence that that compares intermittent versus continuous renal replacement therapy. And this study, ICRAKI, basically said that we would like to do a large multi-center randomized controlled trial. Precisely to answer the question and say whether the choice of the initial mode of dialysis, which is the day zero to day four, the first four days in the ICU, the choice of modality, does it impact outcomes? And that's what these authors looked at in this particular study, Bharat. So, with that background, I'll I'll hand it over to you to quickly go over the study. No question and how they went about doing it before we discuss the results and our our thoughts on Bharath Kumar: Absolutely. And before I start, I I want to say that I'm very excited about this trial for a number of reasons. Of course, the science and the methods, the implications of the findings at the bedside, which we're going to talk about. But also because you are here with us, you are one of the most reputed voices in critical care nephrology. So I think this is this is perfect and I'm really looking forward to the discussion. Coming back to the background, my own reading of the literature Ramesh was also similar. The earlier trials were very small, right? Prior to even the even the study, the Himodayaf study that you mentioned. And and as you said, very different era, different, I think, care processes. So a lot of things that have changed. so I think that nicely lays the foundation for this study and also brings up the issue that there is equipoise for testing this question. So in this trial, what they have I'll dive into the PICO and the methods of this trial. this is essentially a French trial from it's an investigator-initiated open label non-inferiority trial, and we'll come back to the non-inferiority question in some detail across 42 French ICUs, 38 of whom eventually enrolled patients. The population is adult patients with acute kidney injury having an indication for renal replacement therapy. And most importantly, they they're either currently on or have already received invasive ventilation or catecholamine infusion or both. They also specified criteria for initiation of RIT based on a lot of the recent literature on you know indications for dialysis and comparisons of early versus late start and so on. These criteria were not mandatory, but sites were encouraged to you know follow these criteria. when initiating dialysis. Their hypothesis was that an intermittent mode of dialysis, which is diffusion based, would be non-inferior to CRRT. And the non-inferiority question, I think, is primarily because of advantages, certain advantages that IHT may have, and again we will revisit that. So the intervention is RRT by intermittent methods, which is primarily diffusion-based. The comparator or the control was continuous. Vino Venus, either HD, hemodialysis, hemofiltration, or hemodiafiltration. And the assigned technique had to be followed for at least four consecutive days before sites could switch to the alternate mode if they wanted to. The primary outcome is a composite outcome, and we have had several discussions of composite outcomes on this podcast. and the composite outcome they chose is also very familiar to our listeners and and and viewers. Which is the major acute kidney events at day 90. And this is essentially death from any cause, dependence on renal replacement therapy, or an increase in serum creatinin that is greater than 25% from the baseline, which is a standard NAC 90 definition. The range of secondary outcomes, including mortality, at different time points, they randomized using permuted blocks and they use stratification. And it's important to note that stratification was by center. It was by the fluid overload or fluid balance status, more than 6 liters versus less than 6 liters, and the dose of catecholamy, more than or less than 0.5 mics per kg per minute. And these made sense, right? Because of the hemodynamic implications of the therapy that we are choosing. I'll very briefly talk about the sample size and the analytical approach. They assume the incidence of make to be 75% at baseline. And they postulated that you know they would show non-inferiority within a margin of 7.5%, meaning the upper limit of the confidence interval in the IST arm should not cross the 7.5% threshold. This is at 80% power and a one-sided alpha of 0.025, which is generally a standard approach with non-inferiority trials. The analytical techniques were robust. there's a hierarchical approach, so they first test for non-inferiority. If it is proven, then they would test for superiority. Again, a very standard approach. And they had the same hierarchical approach of testing for each of the components of the composite outcome. There are multiple subgroup and sensitivity analysis. I won't get into the details of all of them. They had an independent DSMB that was looking at you know the as a trial progressed at the 28-day mortality during different interim analysis. So they had planned for two, eventually they had three rounds of interim analysis, and the trial, of course, had to be stopped. because of the findings in the interim analysis. So I know I will come to you now for the results and what they found and no yeah. Ramesh Venkataraman: Thank you, Bharat. So just to add on to what you said, just for the audience's sake, I think the IEHD they had given some standards to how IHD should be done. Like you said, they use diffusive techniques. They wanted at least three intermittent dialysis sessions to be done per week, each session lasting four to six hours, and a biocompatible membrane with a bicarbonate buffer dialyside solution. they try to maintain a blood flow of more than 200 ml per minute and a dialysate flow of more than 500 ml per minute, and try to maintain sodium in the dialysate of 145 on under temperature of the dialysate of around 35 degrees. They aimed for a urea reduction ratio of 65 percent. So, technically, this is the author's recommendation of how IHD should have been Bharath Kumar: Should be that. Ramesh Venkataraman: done across the two arms. In terms of CRRT, like you said. It can be either a diffusive or a convective method or a hybrid method. So C V V H D, C V V H F or C V H D F could could have been done. The dose prescribed was 20 to 25 ml per kilo per hour. And they also used a biocompatible membrane, and they left the anticoagulation to the local practice patterns of the institution, although they recommended regional citrate anticoagulation across those two patients. So I thought. this is what they aimed at, so it's worthwhile just mentioning for the sake of audience because we may come back to it a little later. Now, going on to the results, Bharat, as you said, they originally aimed for about 1000 patients and they had to stop somewhere around 554. And we'll start with the patient population. The age was around 65 across both groups. 75% of the patients were were were men. Baseline serum creatinine was 1 milligram per deciliter. About 75 to 80% of the patients had medical problems leading to the acute kidney injury. 22% in the intermittent arm and 15% in the CRRT arm had chronic kidney disease. Time to randomization was about 3 hours, and once they were randomized, the median time to initiation was 2.3 and 2.7 hours across both the groups. So pretty quick essentially. Septic shock patients 47% in the intermittent arm and 43% in the CRRT arm. ARDS is 44% and 50%. SAP score is 51% and SOFA score is 12. So, like you earlier mentioned, these patients were mandated to have either mechanical ventilation or vasopressor initiation or both. As you can see, there were a lot of septic patients and ARDS patients with high severity of illness. Ventilation at randomization was 84% in both groups, and vasopresses were received in anywhere between 77 to 85 percent of patients across both groups. So this was a very, very sick patient population that you and me would see in the ICU, appropriately randomized, and and the therapy started. So, quickly going on to the outcomes that we talked about, like you said, the primary outcome was a composite outcome, make outmake at 90 days. the intermittent arm at 78.1 percent who met criteria for the make 90 outcome versus CRRT at 71.6. When they adjust the risk difference between the two groups, the confidence interval was anywhere from minus 1.22 to plus 13.8 percent. So you had earlier meant that the up Limit of the confidence interval of the intermittent arm should be less than or equal to 7.5 points of the CRRT. Here, clearly, the upper limit is 13.7 and therefore it does not meet criteria for non-inferiority. We will delve into it a little bit later. But what was important here is, like you said, they had originally planned interim analysis at 300, 600 patients, and then finally a thousand patients. But when they looked at 300 patients, they looked at 28-day mortality. There was substantially substantial difference between the two groups, with one group doing worse compared to the other. So rather than 600 patients, they moved the next interim analysis to 450 patients just to make sure. And the same signal of one group having a higher mortality persisted. And therefore the Third analysis was done at 550 patients, and this signal continued to remain with a high 28-day mortality in the intermittent hemodialysis arm compared to the CRRT ARM, and therefore the study was terminated in view of this signal towards increased mortality in the intermittent hemodialysis arm. So, just going into the numbers, the death at 90-day, intermittent arm at 63% versus CRRT 53% with a Has that ratio of 1.3 and a confidence interval between 1.04 to 1.63. So clearly a signal towards that side. And if you look at 28 day, 60-day mortality, all of this have pretty much similar trends across Bharath Kumar: Yeah, signal. Ramesh Venkataraman: the two groups with IHD having a higher mortality. The other two components of the make outcome include RRT dependence and Creatinin of more than 25% of baseline were pretty similar. RRT difference dependence was only 2.1% across both the group, and the creatinine criteria was met in about 39% in IHD versus 38% in the CRRT group. So clearly, most of the make 90 outcome differences that we saw in the study came from the mortality compared to the dialysis dependence and And and the creatinin that that was kind of used in this particular study. So those were the overall results, Bharat. And I I am curious to to get your thoughts. And before we do that, I think there's a lot of confusion in the way the results was actually written among common readers, it's not non-inferior, right? So people some people have missed Took it for not non-inferior is a double negative, are they meaning superior, right? So, can you take a step by step about what non-inferior means? And the authors here clearly say that if the non-inferiority criteria is met, then they will look for superiority of intermittent hemodialysis. In this case, the non-inferiority was not met, and therefore they are looking for superiority of CRRT, right? that's what the authors have done here. Could you please elaborate on how this is done and what the audience should take out of all this? Bharath Kumar: Yeah, for sure. So non-inferiority Ramesh Venkataraman: Mm-hmm. Bharath Kumar: trials, they they primarily are asking a question: Is treatment, is treatment A not worse than treatment B within a certain margin, right? So why would we ask this question? Why would we even frame a non-inferiority question? Typically we frame this when we have an intervention, a new intervention or a com an intervention that we want to compare that. We think offers certain advantages. Those advantages could be related to cost, they could be related to toxicity, they could be related to route of intervention, they could be related to the logistics of delivering it. so mean you know, those are the reasons why you may want to test for the non-inferiority. And we know that the previous trials in the space haven't really shown one to be better than the other. IHT, you know, offers obviously some advantages, logistic advantages, right? The cost of delivering therapy in most. Settings is lower for IHT than for CRRT. And then you have the fact that the therapy is delivered over four hours or six hours, and the patient and the the room is free of the machine, which as intensivists we obviously would like, right? Because having so many machines, interferes with our care, it can interfere with mobility. There are so many other challenges. Of course, the downsides of CRRT, which could be the risk of anticoagulation, I the continuous anticoagulation that we need to do. So there are some potential advantages for of IHT. that I think made it attractive for them to ask it as a non-inferiority question. Now, when you're asking a non-inferiority question, the next decision that you have to make is by how much margin can I be worse off, can intervention be worse off, to claim that we are not inferior, we are within the non-inferiority margin. Now, this more Ramesh Venkataraman: So just to pop just to pause you there, just to elaborate on it and from my own understanding on the audience. So just by pure chance, there could be a difference between the two groups. And if the difference is not beyond a certain limit, then you could say that it's probably, you know, relatively close enough that it is not inferior. is that a reasonable way to to explain it for for the audience here? Bharath Kumar: Yeah, so you have to first say, let's we'll take a different example. Let's say an intervention, in an existing intervention has a mortality of let's say 10%. You now have Ramesh Venkataraman: Mm-hmm. Bharath Kumar: a new intervention that offers certain advantages. so you want to test it out. It's a maybe it's a low cost or a route is better or whatever. So you will now say that I'll test this intervention. As long as the mortality with this new intervention is less than 12.5%, right? I'm happy to accept that. as a margin. So if it is 2.5% worse than the existing intervention, but it is within that whatever decision that I made, that as long as it's within 2.5%, because of the advantages it offers, I'll accept that trade-off, right? so that is the similar logic here. And when I'm talking about 2.5%, I'm not talking about the effect estimate, I'm talking about the confidence interval, the entire confidence interval, and specifically the upper limit of the confidence interval. So here if 75% is the expected. Outcome of make 90 with CRRT. They're saying that as long as it's not worse than 82.5%, the upper limit, right? then we will accept it because of the advantages that it offers. Now, is 7.5% the correct number? That is obvious, always up for debate. And then the non-inflative margin is always a debatable, a contentious one. You typically choose it either based on previous literature or you talk to clinicians and find out what is an acceptable limit for them. Right. So there's lots of discussions. Often there may not be a right number, and you have to pick a number. So we can we can argue whether 7.5% is right or not. And as you already said, in the when they finally did the analysis, what they found is a conference interval goes from minus 1.2, meaning the make 90 can be better by 1.2% in with IHT to plus 13. Meaning it can be worse by 13%. Which is obviously exceeding the 7.5% threshold that we set. Hence, the trial is not able to prove non-inferiority. So when the when they have written the conclusion as not non-inferior, they are definitely not saying it's superior. In fact, all the signals point out within this trial that IHT is actually doing worse. But when they formally test the superiority for CRRT, they were not able to demonstrate superiority, possibly because the trial stopped early. Right, there isn't enough power for the trial at that stage. And the question was also not powered for a CRRT superior to IHT. Right? It was never framed that way. It was actually framed as is IHT non-inferior. I'm then looking for whether it is superior. In any case, the trial stops just beyond the 50% enrollment. So it isn't really power to pick up that. But certainly the signal suggests that there is harm with the IHT you know arm in this particular Trial. So that's how we would look at non-inferiority trials. yeah. Ramesh Venkataraman: So I know I I think this is very important for people to understand what you explained right now. Not non-inferior does not mean superior. So that's a message that I think we should be very clear about. The second question to you is I have noticed that a lot of these non-inferiority trials have a one-sided alpha of no point zero two five rather than what we normally do as a p-value of 0.05 looking at both sides. Do you want to maybe just No, elaborate on that so that no I know it's a nuance, but no, I'm sure some of our audience would like to understand these nuances too in terms of how this is analyzed. Bharath Kumar: Sure. So in a traditional superiority trial, your alternate hypothesis, your null hypothesis is that there is no difference between the two arms, right? And your alternate hypothesis says there is a difference. And now the difference could be on either side. It could be that it's more effective. It could also be that there is harm. And you want to ideally pick up both sides of the effect. So that is why you typically would do a two-sided test, right? A two-sided alpha, which usually is 0.05. Now, in a non-inferiority, your hypothesis is slightly different, right? Your null hypothesis says that the new treatment is inferior, and the alternate hypothesis is saying no, it's not inferior, or the null hypothesis saying it's inferior beyond a certain margin, and the alternate hypothesis saying it's inferior but within a certain margin, right? That's how. So once your hypothesis is already framed with a directionality, it makes sense to do a one sided test. So this is actually related to how the hypothesis is being framed. Now, when they do their obviously when they move on to do their superiority, ideally you would like a two-sided test. But since we have already done the non-inferiority, we can go with the one-sided alpha is acceptable. And that's generally the traditional approach because your hypothesis is specified with a directionality already. That's that, yeah. Ramesh Venkataraman: That's fantastic. Now I I think that makes a lot of sense, and I think hopefully that will hel clarify a lot of these issues for our audience who have been listening to our podcasts. So I I was quite intrigued Bharath Kumar: Yeah. Hmm. Ramesh Venkataraman: by the mortality differences between the two groups, right? I mean, the make 90 outcome, yes, there is a difference, and we talked about that. The fact that it's non-inferior, it's quite interesting that the power was not the mortality was a secondary outcome. The study was not powered for a secondary outcome, but the study was stopped. Because there was a signal in mortality. Right. So the question is why does that there is a mortality difference in the intermittent versus the continuous arm? Is the first question that popped to my mind. So normally you would argue that intermittent hemodiam dialysis causes more hemodynamic stress or more hemodynamic instability to an acutely sick patient population, like this. Like we said, these patient population no. Are expected to die because of the acuity of the illness they presented with. But if you look at their table and look at the incidence of hemodynamic instability or hypotension, it is 68.8% in the intermittent arm versus 66.5% in the CRRT arm, which is not substantially different. Even though hemodynamic quote-unquote theory or explanation seems plausible. Hypotension incidence is relatively similar between the two groups. And if you dig a little bit further, it depends on how you define hemodynamic instability or hypotension. There are two ways it could have been done: one, I have an arbitrary BP drop or a or a 10% drop, 15% drop from baseline or whatever, or the interventions that are necessary to counter the hemodynamic instability. Did I add or initiate Bharath Kumar: Mm. Ramesh Venkataraman: vasopressor or did I increase my vasopressor? Did I give a fluid bolus? Their definition of hemodynamic instability or hypotension was based on interventions rather than a cutoff. So if you started at a vasopressor, increased the dose of vasopressor or gave more than 250 ml of bolus for any particular blood pressure for whatever reason that was counted as hypotension. It doesn't quantify how much wase of pressure I should have increased. For example, I have increased from 0.02 to 0.04. May not be a big deal, but if I increase from 0.02 to 0.2, then that may be a big deal, and both would be count counted as hypotension if you go by their definition. So one, my thought process says that yes, maybe the hypotension is is the same, but the degree to which The magnitude of the intervention done between the two groups could have been no different. Maybe no, obviously, 0.02 to 0.2 mics of norepinephrine would hurt you more than 0.02 to 0.04, and both would be counted as hypotension. So that is something, especially in an open label tri trial where you know that one arm is getting IHD and other one CRRT. I am more inclined to say, okay, let me prevent hypotension, bump up the waste oppressor by. it's a bedset clinician commonly say bump it up by 5 ml, bump it up by no 6 ml or whatever, right? And then then they could have continued the dialysis. That's number number one. Number two is that if you become hemodynamically unstable and unable to tolerate dialysis, there are technically two choices, right? One increase vasopressors, add vasopresses, and still continue dialysis because you think it's worthwhile, or terminate the dialysis. Right? And even if you got one session of dialysis, you were analyzed in the outcome. So we don't know if if the proportion of patients, I mean, I couldn't find this information, if the proportion of patients where the dialysis was interrupted is more in one group than the other. So my f I would still put my eggs on the hemodynamic basket this because of these subtleties that I I think had go on in the backside. Apart from that, there are few other things that I I would like to Point out here. The first thing is the quality of dialysis, right? If you look Bharath Kumar: Dance Yeah. Ramesh Venkataraman: at the intermittent dialysis, initially I mentioned what blood flow, what dialysis flow, what all of that has been done, but there is no mention of how many patients and how many sessions actually achieved that. So that could be one. So, and in intermittent dialysis, we always look at the dose of dialysis as KT over V. Although we don't know what the right value for KTORV in acute kidney injury is, the arbitrary value people choose is 1.2 per session. In this study, they clearly mentioned that the first session they could achieve only 0.9, and the subsequent sessions they achieved only 1.1. Both of them Bharath Kumar: Right. Ramesh Venkataraman: are much below what the KTORV is recommended. And this brings back to my point: is maybe there are several interruptions in dialysis because of hemodynamic instability that. Bring down the dose of dialysis that was actually delivered to those patients. In contrary, if you look at the CRRT group, there was CRRT ran for a median of 22 hours, which I thought was very Bharath Kumar: Abbas. Ramesh Venkataraman: impressive. And the dose delivered was 26.5 ml per kilo per hour. Again, a good dose. normally we would say 20 to 25, giving some room for interruptions that and 26.5 seems reasonable. So one group probably did not get good. quality of dialysis, other group, the CRRT group got good quality of dialysis. Third, if you go back to the baseline characteristics, there are more emergent surgical patients, more CKD patients, and more patients with shock in the IHD arm compared to the CRRT arm. So there have been some baseline differences that could have contributed to the increased mortality, or that could have also contributed to the inability to provide good. Quality dialysis between the two groups, and any or all of that would have been responsible for the outcomes that we see at this point of time. So I feel that I I think rather than just saying that IHD has killed more, I think it also makes sense for us to look into how IHD is delivered versus how CRRT is delivered in our ICU and what are the trade-offs. To achieve that kind of effective dialysis is something that I would think of at the bedside, right? Rather than this the study probably will make us wary of putting people on IHD anymore. But it's not only that IHD is bad or good, but it is also how you deliver it that probably impacts some of the outcome that we are not able to capture in this particular trial, Bharat. So those are my thoughts Bharath Kumar: So Ramesh Venkataraman: about the mortality differences. And what what do you think, Bharat? Bharath Kumar: So, Ramesha, I'm I'm hearing you saying that I mean the top theories that you're thinking of that explain this is the there is a hemodynamic and physiological stress in the IHT arm that is not fully captured by the data that we have in the manuscript, right? There are probably other nuances that are not getting picked up. That is potentially one reason. In fact, I saw some data that supports that in the supplementary appendix. So I saw that there was there are greater number of RRT days with these so-called hypotensive events in the IHT arm, even though it doesn't meet the Ramesh Venkataraman: Mm-hmm. Bharath Kumar: the threshold that they defined or whatever. The second explanation that you had is the inadequacy of dialysis in the hemodialysis arm vis a vis the delivery in the CRRT arm, which was potentially the cause. And the third thing is of course some differences in baseline characteristics and their sort of effect on the outcome. I also want to check with you. there's always been this thought that CRRT can clear, all kinds of inflammatory mediators. And of course, from what I know, it's not a proven or well proven hypothesis. Is that also contributing to this issue, or do you think that's not the major Ramesh Venkataraman: No, to answer your second last question first, yes, the traditional CRRT at the traditional doses we do do not you know convincingly or substantially clear cytokines to provide clinical benefit. Cytokans are have molecular weight that are much, much larger than the middle molecules. And to eliminate them you are going to need very, very high volume emo filtration or filters with pore sizes that are much bigger than what they are traditionally used in C RT to be able to Do it. There is going to be some amount of adsorption that happens on the membrane, but the rate of production of cytokines and septic shock is so high that I think we'll be probably adsorbing a minuscule amount by any mechanism. Although there are people who believe adding convective clearance to diffusive clearance may have benefit, and the intermittent harm had only diffusive clearance without any convective clearance. I don't think that would have played a major role in this particular study. Going back to the quality. I also found two numbers here. The median BUN in the IHD arm was 57 versus 31 Bharath Kumar: Hm, yeah. Ramesh Venkataraman: in the CRRT arm. So you have a very steady solute and fluid control. over a period of time versus a fluctuating up and down and and and a relatively higher BUN and poor azotemia control. And net ultrafiltration on intermittent hemodialysis was 1.7 liters versus Bharath Kumar: Mm. Ramesh Venkataraman: only 1.1 liter per day in CRRT. So again, in a patient population, this sick. No, the rate of ultrafiltration at this rate would have also been contributing to some of the days with hypotension that you mentioned, Bharat. So I have three things. One, IHDM was slightly more sicker than the CRRTM. Two, there is a hemodynamic Bharath Kumar: Stress. Yeah. Ramesh Venkataraman: stress that that IHD imposed that is not fully captured in this trial by just the number of patients with hypotension, and three, the quality of dialysis. Was probably not able to be achieved in the IHD arm purely because of the other hemodynamic stress or other things that the IHD causes in this patient population. So all of this put together, I think, would have led to the mortality differences. So, what Bharath Kumar: Yeah, yeah. Difference. Yeah, right. Ramesh Venkataraman: do you think about the study overall? I mean, in in general, do you have any other thoughts on the methodology and how the strengths and limit limitations are? Bharath Kumar: No, I thought from a trial standpoint, from an internal validity standpoint, right? It's a well conducted trial. it has robust methods, there's actually great protocol adherence. I was again digging through the supplementary appendix. the switch, you know, they were mandated that people should get the same therapy for four consecutive days. Most patients actually end up doing that. Only 10 patients in each group switch before day three, the unauthorized switch. Everybody else seems to switch when it was actually allowed. The bigger problem is that is the trial stopped early. They had to stop early. there's an ethical imperative to stop early. but that always creates a problem with respect to the power, right? Because you were originally going to enroll 1000 patients and now you stop early. It also can magnify some of these mortality differences and so on, right? When you're stopping ahead of the planned stop point. So those I thought were some additional considerations. but n nothing other than you know what we've discussed so far. I I guess that then leads us to your you know take-home perspective, Ramesh on on you know where be where do we go from? Ramesh Venkataraman: So so so first of all let's be very clear, the study looked at the first or initial strategy of RRT, not the continuous the during the entire ICU stay. So as of now in patients who are on vasopresses or mechanical ventilation, there is a signal towards harm. Although the magnitude of the signal is something that I would be skeptic based on no what you told me right now, right? The number of patients was small and the magnitude can be exaggerated. But the direction has been consistent across all of that. So although I am I am convinced that the direction makes sense, the magnitude I don't know whether this has given us a precise estimate. It may be higher or lower if you had continued continued the study, but obviously it's unethical. Considering that stick signal, I think my first modality of RRT in an ICU, if I am able to have facilities for CRRT, would be CRRT in a patient population akin to this with chalk on ventilator for sure, because of what we have seen. So that would be my My take home based on this study. The question that still remains unanswered though are two things. One, what is the right strategy to transition after the first four days? This study cannot answer that question, obviously. And probably, you know, most of us would say if the acuity gets better, then transitioning to you know hybrid or intermittent makes sense. And from an economics point of view, I think it's probably the best thing to do. Second thing I'm going to say is if you were able to intensively optimize your intermittent dialysis delivery, which Bharath Kumar: Mm. Ramesh Venkataraman: they were not able to do in this particular study, would the outcomes have been different? And the other important point I want to make is there has been observational data to say that if your initial modality was CRRD, you seem to have better renal recovery. Not mortality, better renal recovery in previous observational studies and subgroup analysis. Unfortunately, we can't make the decision in this Bharath Kumar: Talk about gene, yeah. Ramesh Venkataraman: particular study because you have to be alive first and then recover your kidney functions first. So the number of patients who do that becomes so small that it is too meaningless for us to try and make any sense out of renal recovery between in these two groups, and that is a very, very important valid point that. the editorial brings up if you read it. I think that's something that that the authors should be looking at Bharat. So that those are generally would be my my kind of take-home messages from this particular study. And my practice in the ICU has always been in a center that I work that if somebody is on vasopresses beyond a certain dose or ventilated, then the default mode has always been CRRT for the first at least 48 days, 48 hours. not necessarily ninety six hours like the study. And then based on the clinical progress and the hemodynamic improvement, we either transition to hybrid or intermittent therapies or continue CRRT. So that's always been our practice and that's probably what I would do. Bharath Kumar: Yeah. I I I wanted to just follow up on one or two of those points, Ramesh. So the you were talking about renal recovery. I think the challenge here is also that with the trial stopping early and the fact that there are very few patients left or dependent on renal replacement therapy and the mortality is so high. Obviously, this trial doesn't tell us anything about renal recovery as as you were pointing out. The second issue is the trial also doesn't tell us anything about these prolonged in intermittent dialysis modes, right? Like sled or extended Right. So that question is not answered, which often is a mode that a lot of ICUs in resource limited settings sort of use right, that it doesn't really answer about that too. Ramesh Venkataraman: Absolutely, you're right. And they have just studied the two extremes, right? So to just to get separation of what whatever happens and and things like that. But then obviously nobody's going to do a three arm study looking at it. So maybe a sled versus C R R T study maybe at some point would come up, but The tricky thing is most centers who do CRRT don't bother with sled and most centers who do sled do not bother with CRRT, right? So b they don't have enough enough numbers, right? So Bharath Kumar: Yeah, yeah, numbers. Ramesh Venkataraman: that becomes a a tricky thing w when you design studies like that. So it is going to be challenging, but that's something that we need to evaluate going further. Bharath Kumar: Perfect. So on that note, I think we will conclude this episode. that you know, and I think the final message that I have is that if you are at a center that is having sick patients that are ventilated on hemodynamic support, the initial modality perhaps if you have the access to should be CRRT. And Ramesh Venkataraman: Mm. Bharath Kumar: we we obviously need to learn a lot more about many of the issues that we have identified. So we will stop there. Thank you for listening. Thank you for watching and thank you for all of the support so far. Thank you. Ramesh Venkataraman: Thank you, bye bye.