This transcript is auto generated and unedited.
>> Fernando Zampieri: Thank you very much once again. So this was a five-year effort to make this trial from the beginning to the result. So it's a great pleasure to present the results today. My name is Fernando Zampieri. I'm a critical care physician from São Paulo, Brazil. And the BaSICS trial was sponsored by the Ministry of Healthcare in Brazil. It was conducted by HCor together with the BRICNet which is the Brazilian Research in Intensive Care Network. And we are very proud to be here today. Of course, the trial also received some logistic support for the fluids, and logistics from Baxter Hospitalar in Brazil, which was not involved in trial design, data collection, analysis, manuscript elaboration, et cetera. And we are very thankful for all the efforts that patients and their families. It's all about them. And trial sites and HCor team to make this trial come true. So this is a really great achievement, and I finally would like to thank the network for doing such amazing job. And so, it was a very large trial and we were able to do that despite all emergencies so we are -- I'm very thankful for being part of it and for being trusted to present the results today. The steering committee is on the screen. I have some conflicts of interest that are presented below. Mostly for investigator-initiated trials from the industry and public funding. And this presentation is divided into three parts. So I'm going through background and methods for our BaSICS trial. Then Alexandre will present the results and at the end of the session, I'm going to discuss a little bit what we learn when we deal with such a large trial. OK. So for the background and methods. So both the study protocol and the study statistics come out of this plan have been published before. They have been published -- the protocol paper three years ago. And the statistical [inaudible] plan earlier this year. So they are already freely available. But mostly of course, the question on whether balanced solutions may be helpful for critically ill patients, it remains open, right? So we have some data that suggests it can be related to -- can cause lower mortality, and lower acute kidney injury occurrence. But we are not actually that sure, as we will discuss today. And of course, the idea in this trial was to check another aspect of fluid use that is frequently not considered in other trials, which is infusion rate, or how fast the fluids are infused in the patient's veins. 4 We got in the first arm of the trial, we had some evidence from the split trial which was neutral. The trial compared buffered crystallized versus saline for -- and the primary endpoint was need for renal replacement therapy. It was a neutral trial. But we also had a very large trial, cluster-randomized trial, from Vanderbilt from Matthew Semler and team, where they checked whether balanced solutions would improve the composite endpoint of acute kidney injury, need for renal replacement therapy and mortality up to 30 days. And this trial was positive for reduction in the composite endpoint. So evidence was still not complete, let's say. So we needed yet another trial to check whether this was generalizable to other studies and not only to one scenario. We did a meta-analysis for the trial that include all the recent trials, and this meta-analysis that contributed to the equipoise for this. And we have this meta-analysis from Cochrane too, that also contributed to the equipoise for the BaSICS trial. And the second aspect of the BaSICS trial was infusion speed. So this was mostly based on Dr. Maitland's paper in the FEAST trial where fluid bolus was associated with worse outcomes in critically ill septic children. And when we discussed the BaSICS trial, we thought that perhaps we should consider a second arm where the infusion speed should be assessed inside a randomized trial. We know that slower infusion speeds are probably -- or physiologically they are more related to a prolonged or greater endovascular expansion. And it can also reduce tissue edema and perhaps reduce organ function. So we have some evidence that, for example, we should give two liters of saline to healthy volunteers, you are going to decrease oxygenation, and reduce exercise tolerance. And we also have some evidence that on septic patients, that when you give fluid to patient, you may actually decrease afterload through a decrease in arterial [inaudible] could be related to lower blood pressure and perhaps different outcomes in those populations. So both questions are probably both very relevant and they are part of the everyday job for all critically ill physicians and nurses and respiratory therapists. We were especially interested in this secondary analysis of the FEAST trial where the fluid bolus was associated with perhaps more cardiovascular collapse in septic children. So we thought a lot about this when we designed the trial. So BaSCIS was mostly a factorial, randomized, multi-center trial in 75 ICUs in Brazil. We planned to randomize 11,000 patients to either Plasma-Lyte 148 or normal saline, 0.9% saline. And in a factorial way to two different infusion speeds when patient needed fluid challenges. So one was 999, because that's the fastest most infusion pumps in Brazil go. And because that will result in 500 mLs over half an hour, which is the average fluid bolus. As by the [inaudible] study by Maurizio Cecconi. And we set up these lower infusion rate as a third of the control of the [inaudible] fusion rate, because that would be an infusion speed that would also be considered fluid bolus by most physicians, but is not simply [inaudible] fluid therapy. The trial was -- of course, this was randomized. This was blinded for fluid type, but it was not blinded for the infusion speeds. And blinding was made by bags labeled A through F that were provided to the site. So patients could be randomized for example, to receive Letter B fast or Letter C slow, et cetera. So all the possible combinations between the letters and the infusion speed. We included all critically ill patients that required at least one fluid expansion at the discretion of the attending physician. Patients thabwre not expected to be discharged the next day after enrollment. And patient had at least one risk factor for acute kidney injury or mortality. Age above 65, hypotension, presence of sepsis, need for mechanical ventilation or non-invasive ventilation for more than 12 hours, abnormal renal function, or previous diagnosis of liver cirrhosis or acute liver failure. So we got those criteria by querying some large data sets, and we had in Brazil the [inaudible] study. And we selected these risk factors to get an expected mortality that would reflect our [inaudible] calculations. We excluded patients that were already on renal replacement therapy, regardless of whether chronic renal replacement at home or in the hospital; or patients that were expected to require renal replacement therapy within the next six hours after ICU admission. We excluded patients that had severe electrolyte disturbance like very low or very high sodium levels. Patients whose death was considered imminent and inevitable within the next 24 hours. Patients with suspected or confirmed brain death. Patients receiving palliative care only, and patients that were already enrolled in the trial. So this is the main diagram of how the intervention worked in BaSICS. So the attending physician indicated fluid expansion, for example. So if it was a fluid challenge, then the patient should receive the required [inaudible] by the physician at the designed infusion speed. So if [inaudible] or they said 500 mLs, then it could be [inaudible] speed for infusion [inaudible] if the patient was randomized to a contrary infusion rate. For maintenance, the -- all maintenance were also requested to be done [inaudible] fluid. But at the designed speed by the attending physician, like one liter a day, two liters a day or something. And other fluids were free, like bicarbonate, albumin, blood, colloids, et cetera as needed. Regarding dilution, so we also asked the sites to dilute everything that was compatible with both saline and Plasma-Lyte at the specific letter of the study fluid. So this was valid only for dilutions that were above 100 mL. If it was below or up to 100 mLs, the sites were free to infuse at their usual care. We asked them to use glucose as a perfect food whenever possible. But if it was compatible with both fluids, then it was diluted in the allocated study fluids. So we did this to increase how much [inaudible] fluids that the patient will receive that will be the study fluid, right? So this was to avoid or reduce contamination. We knew that some contamination would occur in a way. But we did that to make patients receive as much as possible the study fluid. So this was how the fluids looked like. So it's written "Group D" here. So we had six letters, and bags are supplied boxes for a site. So each site had received up to two boxes with 15 liters each for each fluid letter. So all sites received at least one batch of all possible letters. And then that require a lot of, you know, organization and logistic support, because we had to make sure one site would not ran out of fluids when they were enrolling patients, right? So this was -- it was visually identical, so it was just written the group here. It was visually identical. The primary endpoint was 90-day survivor. It was assessed essentially at HCor by a dedicated team who had a lot of work. If the phone calls were unsuccessful, we did everything we could to obtain a primary endpoint, which was mortality, including hospital returns, whether the patients got a blood sample drawn at a given time. We sent telegrams. Sometimes we sent personnel at some patients' houses to check on them. And when everything failed, we use the National Database of Disease but the Brazilian Minister of Health. Our secondary endpoints included need for renal replacement therapy up to 90 days, occurrence of acute kidney injury defined as a KDIGO 2 or 3 within Days 3 and 7. And SOFA score and its components also on Day 3 and 7. We [inaudible] the score to higher or lower than 2 for some secondary analysis. And mechanical ventilation three days at 28 days. So the primary endpoint analysis was evaluated by our frailty survival model, which was adjusted by age. The SOFA score at enrollment, at admission time, and enrolling site as a random intercept for that, and admission type, it was point at admission, and point at admission with sepsis, and point at admission without sepsis. We had several subgroup analyses, that Alexandre is going to show the results, but mostly patients with or without sepsis, with or without [inaudible], urine in the background, surgical and non-surgical patients, TBI or non-TBI, patients with high or low APACHE score, and whether patients receive more or less than one liter of saline before enrollment. So this file, the power calculation was made. So we had 89% power to detect reduction in 90-day mortality from 35 to 31.5%. So that will be a hazard of 0.9 or less for 90-day mortality with an alpha value of .05. So I believe that that was my part of the presentation. So I'm very happy to pass the word to Alexandre. So Alexandre now is going to present the results of the trial. It was a pleasure to be here. Thank you very much.
>> Alexandre Biasi Cavalcanti: Hello everyone. I am Alexandre Biasi Cavalcanti, and I am going to present the results of BaSICS. I will start by presenting the results, and the comparison balanced solution versus .9% saline or by comparison with the fluid [inaudible]. So a total of 11,052 patients were randomized in this study. Patients were basically randomized to the four following groups: patients with the balanced solution and slow infusion; patients in the group with balanced solution and the faster infusion, the control infusion; patients receiving saline and with lower infusion; and patients receiving saline and the control infusion. A number of patients were excluded after randomization, mostly because consent was withdrawn. We had -- we used deferred consenting with 30 patients could be randomized in the trial. And whenever they would [inaudible] consent to us was asked for the patients in some cases. There were refusal. So there were 139 patients excluded in the group of balanced solution low infusion, 153 excluded in the balanced solution control infusion, 123 in the saline low infusion, and 117 in the saline and control infusion. When we consider all the groups, the groups that received balanced solution, we have 60 patients lost to 90-day followup, although half of those were followed to hospital [inaudible] so we have data [inaudible] data at cost of these [inaudible] without followup. For those we have data intuited. In the saline group, we have 10 patients lost to 90-day followup. Three with available hospital discharge. So [inaudible] data. And 7 [inaudible] had to be included. So we have then 5,230 patients included in the analysis for the [inaudible] balanced solution group, and 5,209 patient included in the analysis for the saline group. These are the baseline characteristics of the patients. They are basically very well-balanced between groups. Age of the patients was 61 years old, both groups. Female sex was 44% in both groups. In both groups, the number of patients they're going elective surgery was 47.8%. Non-elective surgery was 12.5%, with balanced solution group at 12.3 saline group. And medical patients were 39.7, the balanced solution; and 38.6 in saline group. SOFA score was, the medians of the score was 4, both groups. Around 1/3 of patients had KDIGO acute [inaudible] rescore of 1 or more. [inaudible] .5% of patients had sepsis at their baseline, the balanced solution group; and 19% in saline group. Four point seven, four point 5 were traumatic brain injury at baseline. And 60.7% at 60.5 had hypotension at baseline. And 44% were on invasive mechanical ventilation at baseline. Here we can see how was the use of other solutions, while the patents were in the [inaudible] room or operating room before they were in the [inaudible]. So the 24 hours before randomization, you can see that 48% of the patients both groups received any balanced solution. And 31% of the patients received more than one liter of balanced solution. Regarding use of .9% saline before randomization, in the 24 hour before randomization, you can see that 38% and 37% of patients, the balanced solution, saline group respectively received saline. And 17.9 and 18.8 received more than one liter before randomization. Here you can see the volume of infused fluids by group on Days 1, 2, 3 and Day 7. Most Day 1 patients received a little bit more than two liters median fluid in that day. And most of it was study fluid in red. The rest being open crystalloids and other fluids. This was seen [inaudible] groups, and of course the amount of fluids decrease as the Days 2, 3 and so on, as we expect. Anyway, most of the fluids used were study fluids. In total, patients used 1.5 liters of study fluid on Day 1, and from Day 1 to Day3, 2.9 liters. This resulted in differences in serum chloride levels so that in patients that received .9% saline, serum chloride levels increased as compared to the balanced solution group. And the difference was about 2 milliequivalents per liter along the base. Here we have the results for the primary endpoint. Ninety-day mortality was very similar between the two groups. [inaudible] 26.4% the balanced solution group as compared to 27.2 in the saline group. The survival [inaudible] is as you can see, very similar, has a ratio .97 with a 95% confidence interval, .9 to 1.05, and a [inaudible] P-value of .47. There were also differences for most of the second [inaudible] assessed. Acute kidney injury with need for renal replacement therapy within the first 90 days was very similar between two groups. Their rate of need for renal replace therapy was 199 per thousand patients they -- the balanced solution group as compared to .93. The incidence of acute kidney injury as measured by KDIGO of 2 or higher at Day 3 and 7 was very similar between treatment groups. We also measured other components of the SOFA score on Day 3 and Day 7. Incidence of elevation of [inaudible] of SOFA score higher than 2 was similar between groups for most components of SOFA score, except maybe for in the -- on Day 3. That SOFA score higher than 2 was likely more frequent in the balanced solution group. This effect disappeared on Day 7. However on Day 7, we can see here that abnormal neurological SOFA score higher than 2 was likely more frequent in the balanced solution group. No other differences for the other components of score, and of course, we have the outcomes and we are not adjusting the [inaudible] for this [inaudible]. Regarding the subgroup analysis, we can see that the effect of balanced solution versus saline on mortality was similar across the groups, except for the subgroup of patients with traumatic brain injury. As you can see, we have a little bit more there for 100 patients with traumatic brain injury, and among these patients, mortality was 31% for those who received balanced solution versus 21% for those who received saline. Hazard rate was 1.48. And the p-value for interaction was significant as 0.2 suggesting [inaudible] infusing balanced solution for those patients. Now for the results of this lower versus faster IV boluses. The flow of patients was basically similar. And we had 5,276 patients included in the [inaudible] for this [inaudible] future group, and 5,244 patients included in the analysis for the faster infusion group, the control infusion group.
>> Fernando Zampieri: So I'll take over from where Alexandre left. So we got an infusion speed. We had the same amount of patients randomized. We had a very good distribution among the groups. We lost around 14 patients fall out for the slow infusion speed, and 11 patients for the fast or the control infusion speed. And we had some patients that were [inaudible] that were also discharged, but we missed and the primary point was missing in 10 patients that were [inaudible] for the primary endpoint. So there are no big surprises in Table 1 here too. So both groups were similar regarding major features, regarding SOFA, KDIGO, [inaudible] sepsis, TBI, hypotension, et cetera, so there is nothing that is much -- that brings you some [inaudible] impressed by this. Regarding the fluid use, on Day 1, around 75% of all fluids that were infusion were infused as a bolus infusion, right? So then over time, the proportion of fluids that was infused as a bolus decrease. But in the first day, we had around 75%, then followed around 50%, then 38%. And also we had around 20% of fluids that were issued as maintenance fluids in the first day. then 34 or 35 on the second day. And then dispersion [inaudible] increased on Day 3 and Day 7. And other fluids that were infused around 6% on other infusion speeds. So most of [inaudible] was pretty good in the trials. So around 97% of all patients required at least fluid bolus, which was one of the inclusion criteria. And on Day 1, 2, and 3, the vast majority of patients had received, that had required a fluid bolus received a fluid bolus at the allocated rate. So the [inaudible] was pretty good in all the arms. And for the primary result, we also reached a neutral result of 90 days mortality was 26.6% for slower infusion, and 27% for fast infusion rates, and that resulted in a hazard ratio that is very well distributed around the no effect, with a p-value of .46. Regarding -- there was no difference across all the several subgroup [inaudible]. Estimated delivery date: There was also no or far more benefit for most secondary endpoints, so regarding renal replacement therapy, was also neutral, so no different in the [inaudible] ratio for AKI needing renal replacement therapy. KDIGO of above 2 at Days 3 and 7, et cetera. And we had, of course, in the similar to the fluid type, we had some secondary endpoints that were slightly different between groups. Of course, we have technology [inaudible] for comparison as Alexandre said. But we had lower number of patients requiring vasopressor at Day 3 in the slower infusion rate. That was [inaudible] odds of .09 -- .89. Sorry. And we had more patients -- less patients that had respiratory SOFA both treated [inaudible] mechanically ventilated patients with a low PF rate too, at this lower infusion speed. On the other hand, the normal population SOFA at Day 3 was slightly higher in the slower infusion group than in the control group. But all these differences, they vanished in Day 7. And of course, they are not patient-centered endpoints for what it's worth. But they bring up some interesting discussion I hope for the panel. Of course, this trial has several limitations. So we did not -- use of fluids was limited to the ICU and while we collected data on using fluids in the emergency room or operating room, that was not considered when some patients arrived in the ICU after receiving a significant amount of fluids. We also allowed dilutions that were lower than 100 mLs to be on open fluid. So that result in some degree of contamination. We had some post randomization exclusion, mostly due to lack of consent. We have no physiological data that what happens et cetera. One very interesting subsidy [inaudible] on what triggered fluid bolus and what happened with fluid bolus. And of course, patients consider what might be considered as low -- a low volume of fluids during their ICU stay, which is similar to the clinical practice, just to remember the SMART trial. We had around 2 liters in the first three days. And that's pretty similar to what we had here. So in conclusion, among patients in the ICU that required fluid challenges, the use of balanced solutions or 0.9% saline did not reduce 90-day mortality. We have a very -- of course, all the secondary endpoints and all the subgroup analysis, one thing that we should keep in mind is that there is a signal for harm with TBI patients. And for the infusion rate, we could not find a difference in 90-days mortality for slower infusion rate versus a fast infusion rate. Well, with that, so I believe that I end my presentation now. Alexandre, if you want to take over. I gather Alexandre is not here?
>> Rob Mac Sweeney: He is just coming back into the meeting I think. Here he is. He'll be with us in just a moment. But Fernando and Alexandre and the entire team at BaSICS, congratulations. What a huge achievement to design and execute that trial, and especially finish it under such difficult circumstances, during the pandemic. Alexandre, I hope you were able to hear the end of Fernando's talk there. He was just handing back over to you. I think maybe you have a concluding remark?
>> Alexandre Biasi Cavalcanti: So in conclusion, I think Fernando has commented on the limitations of the trial, and we can conclude that among these ICU patients requiring fluid challenges, use of a balanced solution compared to .9% saline solution to their mortality, and there was a sign of possible harm for traumatic brain injury patients with the use of balanced solution. Infusion of crystalloids at this lower rate also did not reduce 90-day mortality. Therefore, these findings do not support either the use of a balanced solution as compared to saline, or as lower as compared to faster infusion rate. I'd like to tell you, and we are very happy that the results of the trial have been published today [inaudible] in JAMA, both of the comparison of balanced solution with saline, and also the paper comparing the slower versus the fast infusion. Available for you to read. And with that I finish and thank you. Thank you everyone. And thanks Fernando also for your support during this end of the presentation.
>> Fernando Zampieri: My pleasure. My pleasure. My absolute pleasure.
>> Alexandre Biasi Cavalcanti: And [inaudible] thank you.
>> Rob Mac Sweeney: Thank you very much, and indeed as Alexandre said, the papers are just out on the JAMA website. And my thanks to JAMA for coordinating the publication of these trials with this livestream. It's great to be able to do this and get both [inaudible] together. So all the information is there, supplemental [inaudible] I'm sure our viewers will want to get into and read. So it's with great pleasure now that I'll hand it over to Sydney and Simon Finfer. Simon is the lead investigator for the PLUS trial, which is similar trial to this. So it will be fascinating to hear his thoughts, and hopefully we might be able to bring you the PLUS trial results as a livestream later on in the year. So Simon, early morning over in Sydney, thanks again for gettin up. Over to you.
>> Simon Finfer: OK, well, before I summarize my entire feelings about this trial with one word, I'd just like to obviously congratulate [inaudible] Alexandre, Fernando, [inaudible] and all those wonderful people who've done this extraordinary trial. Also Rob for continuing to run this extraordinary resource for the critical care community, which is pretty unique, and incredibly valuable for our specialty. Particularly as we're unable to get together physically in these dreadful times that we're living in. So really that is an extraordinary trial. It's great to see it's published today. It's freely available online, an dit's my pleasure to make a few comments about it. I'll start just by disclosures in relation -- obviously I've been involved in fluid research for many years now, and I have received -- I haven't taken any money personally from any of these companies, but the CSL for Sydney [inaudible] have funded travel for academic reasons for me, and research funding, all of which has been given to my employer, either The George Institute or University of Sydney. As Rob has said, I'm also leading a large trial which is very similar to the fluid type component of BaSICS. And we have actually coordinated our data collection and endpoints, et cetera, so that we'll be able to put those trials together at the end of the day, and provide some pretty compelling data I hope. So the background I'm not going to go into in great detail, but obviously I think we all know why we're here, and there's a group of people who don't like saline, who like to refer to it as abnormal saline, and have told us all its ill effects over time. And probably the strongest evidence, trial evidence in support of those contentions comes from the SMART study, which again was an extraordinarily good and well-run trial. And you've already heard about the outcome, primary outcome measure from that, which was a [inaudible] renal replacement therapy and doubling of serum creatinine concentration which produced a just statistically significant result in favor of using a balanced crystalloid. And Naomi Hammond led a observational study, international observational study, across several hundred intensive care units across the world, demonstrating or reporting that the use of balanced solutions have increased significantly over a period of about seven years, from when we had done a study, a similar observational study after the SAFE study. So there's been a practice change, moving away from the use of saline to use of more balanced sorts solutions. So it's great that we're now got, from [inaudible] better data to tell us whether this is a good thing or not. I do feel my initial reaction of "wow" kind of summarizes my feeling about this trial, but of course we have to try and be a little bit more systematic about how we look at data. And I usually go back to this. What might now be viewed as a very, very old paper that I think is still a very good paper about -- look at what features should we look at when we read a clinical trial to decide the most important thing is, will the results help me in caring for my patients. Because clearly not all randomized controlled trials do produce the least biased evidence for us. But not all randomized controls are created equal. And you can have some serious methodological issues, and you can produce results that are misleading. And that can result in patient harm. So to look at some of these issues quickly, The eligibility criteria I don't think one has any particular issues with that. And the selection of patients who are going to be given the ICU the day after tomorrow effectively I think is very sensible. And clearly, we would like to see a decent exposure to this of the interventions. If patients are only getting -- are staying in the ICU for a very short period of time and getting very little fluid, then despite the numbers, we might think that there are some issues about whether the exposure to the interventions have been significant enough to give us a true result. One of the most important, the randomization in a big trial like this, one would expect to see balance between the two groups. But really the feature of randomization that's most important is allocation concealment. That investigators don't know which group the patients are going to go into until they have actually been included into the trial, and they're going to be tracked. And that certainly was done in BaSICS. It's very important. Prevent people from -- we can all look at two patients who look reasonably similar on paper, if you're an experienced intensive care clinician and know that one of them's got a much better chance of survival than the other. And if you then know which group the patients go into, you can game the recruitment a bit. So the allocation process I think was very good. You can argue a teeny bit about whether you have fixed block sizes, but I think it's stated in the papers that the sites didn't know the block sizes. So I'm not concerned about that. Blinding is really important and especially important I think in fluid trials, because we all have our own biases. And so knowing or not knowing, which fluids a patient is getting is important. It's important, as I say, there are subtle things that happen when we allow our biases to interfere with how we're treating patients. It sounds like it shouldn't happen. But there's a long history in critical care research of single-center unblinded trials producing results that can't be replicated in multi-center blinded trials. So I think that's a really important thing. It's important the patients are treated equally in all other fashions. And with a sample size of 11,000 and everything else left to the treating clinicians, it's extremely unlikely that there would be imbalances in other aspects of treatment that could explain in this case, why something that should have been a positive result turned out to be a neutral result, for whichever fluid you consider. It's important that this statistical analysis plan is published beforehand, as was done. Because otherwise, obviously if you can [inaudible] crawling through the data until you find something you like and publishing that used to be a common occurrence. And hopefully is something of the past. And large sample size with a, you know, reasonable reduction -- reasonable hazard ratio, rather than looking for a very unrealistic treatment effect. The numbers [inaudible] to follow up a very small -- I guess one thing we might like to know is how many of the patients follow-up came from the National Database rather than from individually knowing. I mean, I don't -- [laughs] maybe the Brazilian database is much better than the Australian one, but you can't -- nobody gets into our death data for months and sometimes years after they die. So that is a potential source of error. And the loss to follow rate is quite extraordinarily low. That's really quite remarkable. But again, might like to know how many came from the National Database. So did the patients sound similar to the patients that I treat, which would be something I would look at if I'm deciding whether to allow this to change my treatment? The traditional ICU trial has patients who are aged round about 60 and 60% male, which BaSICS has. Half the patients had elective surgery, and again, this I think is a feature of that varies nation to nation. Australia is very well-supplied with intensive care beds. You can get into one of our ICUs with relatively minor comorbidities, and most of our elective surgery patients we would discharge the next day. So I guess I was a little surprised that half of them were elective surgical patients that you're not expecting to discharge the next day. But again, that may reflect differences in national practices. I think the median APACHE 2 score was relatively low, again for a population that you're expecting to stay for a couple of days. And obviously only half were ventilated. So in terms of severity of illness, possibly a little less than I thought it might be. What about exposure and contamination, which Alexandre and Fernando have alluded to. I think that the exposure to study fluid is good. It's better than was seen in, and certainly in terms of medians, it's better than was seen in SMART or more exposure. There is always going to be some contamination, and you can see from this graph, I think the amounts, I think that it could completely obscure any treatment effect of one fluid versus another. It seems unlikely to me. The chlorine which is supposed to be the mechanism by which normal saline is hung for, again this is quite a similar -- if you look at the left-hand panel, which you've already seen, it's quite similar to -- we've seen in SMART. In fact, I think a little bit, the maximum difference in serum chloride concentration's slightly more in BaSICS. And the box of whiskers plot next door, which has the median to quartile range, and the range is interesting in that it shows you the degree of overlap, even though the means are quite different. And I think that that -- you would see exactly the same thing from the SMART data, which does make us wonder whether that mean difference, which looks quite impressive, is actually with a truncated Y-axis, is actually reflecting something that would -- one would expect to change significant outcomes for patients. Again the -- in terms of the exposure and the -- I was very interested here when I looked at the maintenance fluids. It's basically on Day 1, it's like 10 mills an hour. We've looked at that in our previous large fluid trials, SAFE and CHESS, and looked at -- we found that far more, in those studies and also in some point prevalence work we've done, maintenance fluids work much -- by far the biggest volumes of fluid, and the volumes of fluid that our patients get were more through maintenance fluid and drug delivery than they were through boluses. So that's -- that may be something that Alexandre or Fernando would comment on when the Brazilian practice is to be quite restrictive about maintenance fluids. Because certainly that would be a lot less than I would expect to see. Before seeing the results, and I think what's the -- how precise is this estimate? So the 95% confidence intervals from .9 to 1.05, obviously there's a reasonable, good sample size [inaudible] event rate. So that's pretty impressive. And the same for the speed. I admit it's pretty hard to argue with that. There is this one subgroup effect, where there's a significant treatment effect, patients with traumatic brain injury is significantly different from the treatment effect with patients without traumatic brain injury. In PLUS -- I mean, we published our protocol's statistical analysis [inaudible]. I'm not -- this is freely available information. We've excluded patients with traumatic brain injury from PLUS, because we thought that possible Plasma-Lyte, which is the balanced solution used in both these trials, could be harmful in patients with traumatic brain injury. This is based on one very small trial published in Critical Care in 2013 by Roquilly et al, from France. Was only a 40-patient trial, and they reported that using a balanced salt solution, which is a lower tonicity than saline, didn't affect ICP, but because their trial was only 40 patients, the [inaudible] wasn't different, but if you actually look at the graph of the ICP in that paper, immediately following a bolus of the balanced salt solution, the ICP increased 10 millimeters of mercury on average above that seen in patients who got saline. And [inaudible] has done some very nice [inaudible] with albumin. If you put albumin in saline, you don't see any effect on ICP in animals. If you infuse it in its usual slightly hypotonic fluid, you do see an increase in ICP. So I think there's a -- to me that is -- that has face validity, that actually saline is better in those patients. There are some sensitivity analyses which are in the supplements that are available on line with the paper, looking at things like whether you -- if you look at just the people who didn't get fluid before enrollment, does it affect the difference? Does the baseline serum chloride have a different -- make a difference to the treatment effect? What about people who did or didn't have kidney injury? And I think the interesting thing here is this one, where I think, you know, there are many comparisons and we haven't really had any comment about multiple comparisons that you expect something to come out as significant when you do so many comparisons. And really there isn't anything there. Is there anything I don't like? Not really. I sort of thought very hard and I -- these are very trivial, minor trivial points. I mean, my preference is not -- in a trial of this size, would be to have a primary outcome that was not adjusted. That was landmark mortality, without implication of missing values. But these are really, really trivial points. I guess the conclusion that's in the published paper's that the balance salt solution isn't indicated. I'll probably get the wording wrong. We can discuss. I mean, I think the answer to me is, you know, you could probably use both, apart from -- both or either. And also the conclusion in the paper that the slower infusion rate isn't indicated. Well, it's not really evidence that either -- that using a faster infusion rate is better. So again, I think that's -- and that's a unique feature of this trial, I think. That's what's really important and interesting, that people think about other aspects of how we use fluids rather than just which particular type of bag we're hanging up. Will it help me care for my patients? Well, the population I've mentioned, apart from the maintenance fluid, but I think -- and most has been given as boluses. I recognize the population may seem to be very like the patients that I treat. I did not expect the overall result. You know, I have uncertainty about the choice of fluid, which is why we're doing the PLUS study. But I'm very happy to accept it. I think that there's no signal for renal injury is a little surprising. It confirms my bias about patients with traumatic brain injury, so of course I'm very happy about that. Because we love having our biases confirmed. Maybe I think what I see a lot in practice, where people are picking fluids based on blood tests and by chemical parameters, it may not help your patients, but it probably won't harm one. And obviously the rate of infusion doesn't seem to have a strong effect in this study. So in summary, I think it's an extraordinary achievement. We have to congratulate everyone. As we know, Brazil has been hit very, very hard by the pandemic. I'm not sure that I would want to import their president to my country. But to achieve this in normal times is extraordinary, and to do it during the pandemic is really unbelievable. So I think it's brilliant and fantastic. It's another piece of really solid evidence to help us decide how to treat our patients and hopefully improve our patients' outcomes. But thank you very much for giving me the opportunity to comment.
>> Rob Mac Sweeney: Simon, thanks very much. That was a wonderful and erudite editorial. So we'll go back to São Paolo now, and Fernando and Alexandre have the opportunity to reply to Simon's editorial, and answer some of the comments and questions that he posted as well. Over to you guys.
>> Fernando Zampieri: Thank you. Thank you, Rob. Thank you, Simon, for the lovely editorial. Thank you very much. So regarding some points in the editorial, so regarding the use of the Disease Database for patients with missing data. So that happened for something -- I don't remember, there's like fewer -- something like 40 to 50 patients that were enrolled early in the trial. The National Disease Database has a delay of 18 months. So we could only use data for that for the first few years of the trial. So it was around 40 patients [inaudible] something like that. So that was not common. Regarding the other things, you brought up some very interesting discussions on this. So I believe that I had the same personal bias for TBI too. And now that seems to be, you know, justified now. So it seems like saline should be the go-to fluid for TBI. Regarding some of the minor points you mentioned, so we decided to adjust the primary endpoint for several variables because that perhaps can increase the precision of the effects size I estimate. So when this is done a priori, as we did, adjusting the effects size in the clinical trial, so Angelo is here, and I hope we can discuss this further. Usually increased precision in the effect size estimates for that. And that's why we planned to run an adjusted analysis for the endpoints, and that includes the primary and secondary endpoints for most of the trials. Regarding the missing value [inaudible] so the concept here was mostly similar, right? So we could not have included an only [inaudible] complete case analysis or a worst and best case scenario for example. But my understanding, and please correct [inaudible] if I'm wrong. If you have a large mass of data, it's better to just do the data and not, you know, [inaudible] things up. So you end up losing relevant information because some patients you have information on, and how much fluids they receive and [inaudible] the primary endpoint, et cetera. So when you [inaudible] data, you perhaps waste data than if you just remove patients with missing relevant information. And of course, that was a very small percentage of patients [inaudible] but we in sensitivity analysis did not change the results of the trial in any way. Regarding the conclusions of the paper on why should we use saline if -- outside TBI. So the main point here, and this is something we learned from BaSICS is that when you have a very large country -- so it's not just a matter of cost. So people say, well, Plasma-Lyte is not much expensiver [phonetic] than saline, and that's true in many scenarios. Not precisely the case in Brazil. But if you have to move fluids from large distances of land, and controlling stock, and how to stockpile two different types of fluid, it's a little bit more complicated, and especially, you know, because Plasma-Lyte is not as compatible for dilutions as saline is. So [inaudible] population now affect, or for a population level, it's easier for a site that is outside, you know, São Paulo or Rio or the center of southeast and south of Brazil, it's easier for a site to only stock one type of fluid than to stock several different types of fluid and have to consider all the logistics involved. So if one fluid is not harmful, and if saline is not harmful, why not only using saline and avoiding all the issues of stocking two types of fluid? So that's one thing that we thought about, and I would like to hear what Alexandre thinks about this.
>> Alexandre Biasi Cavalcanti: Yes. I agree and there is really this fact that although for many countries, cost for balanced solutions and saline are similar. This is not the case for many other countries. In Brazil, costs are completely different. Some balanced solutions, for example Plasma-Lyte can cost many, many, many times the price saline costs. So this makes a difference for many. And one other question that I think Simon had is regarding intuitive values in Table 2. I think Simon, the last, last version of the paper is us changed. We have published the paper that was actually published in the JAMA. Numbers in Table 1 are for the complete case that the BaSICS without [inaudible] numbers. Again, the number of patients without -- missing variables was small. And one last comment is regarding the big question of maintenance fluid. I think in the first day that actually the volume of fluid used [inaudible] as fluid challenge was much higher than the maintenance dilution. But then as the days go by, you can see that the number of maintenance and dilution in increasing proportion to the number of fluid challenges. And I think that part of it has to do with inclusion criteria. To be enrolled, patient had to be eligible for a fluid challenge. So in the first day, all the patients were supposed to receive a fluid challenge. And finally there is this thing that if you -- even if the use of maintenance fluid was maybe smaller than expected, I think it really reflects [inaudible] here use of maintenance fluid is limited in most ICUs here in Brazil. And nevertheless the total amount of fluid received in the first three days is similar to other trials in for example, higher than they used in [inaudible] trial. So yes. In the end, we have a lot of fluid challenge in the first days, but then is it crazy to -- so that you could see more maintenance and dilution fluids composing the overall amount of fluids you use?
>> Rob Mac Sweeney: Thanks, Alexandre. What we'll do now is, we will move on to our questions from the audience. We're well behind time, so Bronwen Connolly, my colleague from Queens University in Belfast is going to join us. Bronwen, maybe just have to limit this to three questions, please. And we'll pick up the rest in the podcast that I'll do with Alexandre and Fernando in the next week or two instead.
>> Bronwen Connolly Thanks, Rob. And another congratulations to the trial team around the delivery of this study from the discussions. I've got one general question, and then one of the fluid type and the fluid rate components of the study. So in general, we touched upon the notion about whether or not this is a big enough group of patients, given the APACHE score and mortality rates. Wondering whether there are plans to look at the subgroup analyses around the emergency cohort, and excluding the roughly 50% of patients who are elective surgery?
>> Fernando Zampieri: Yes, so those are very good questions, right? So we actually needed up with slightly lower [inaudible] than we actually planned before. So we have some secondary analysis coming on, including -- excluding elective surgical patients for other subgroups, et cetera. So that's coming. There was no interaction a priori, and the trial was slightly well-powered to check for interactions between fluid type and infusion rate, so this is on the [inaudible] of the paper, so there was no interaction. But we have some secondary end analysis coming up for -- excluding let's say [inaudible] patient et cetera. But I'd like to remind when you look at Table 1, so -- and at the figures -- for the subgroups, the signal or the absence of a signal there remains absolutely constant, or mostly constant in all subgroups except for TBI. So I would be surprised if any subgroup will have a very different result than from the major trial. Alexandre, if you want to comment on that, or -- .
>> Alexandre Biasi Cavalcanti: [inaudible] Fernando, we have found no [inaudible] up to this moment [inaudible] in the analysis and subgroup analysis [inaudible] affecting in [inaudible] patients. Another interesting question, among patients who received more fluids, of course it was randomization variable, but even considering patients who are receiving more fluids, there was no effect of choosing [inaudible] versus saline.
>> Bronwen Connolly Thank you. And then with regards to fluid type component, many of the patients enrolled in the study had suspected or proven sepsis. Do you think there were other factors such as time to administration, antibiotics, and so forth that could have had an impact on your findings? Especially around mortality.
>> Fernando Zampieri: Yes, it might be related. So we have absolutely no data on that. And we expected that will be distributed among the group, so most of the patients were remaining in ICU after initial resuscitation efforts for sepsis. So they also got a few boluses in the emergency department, or in the operating room before our randomization. So that's possible. We are currently working on a secondary analysis, checking whether fluid use before enrollment could modify treatment effect for the trial. So -- but for the other variables, we actually we have no data on that to support either statement. Yes.
>> Bronwen Connolly And regardless of fluid type, was vasopressor therapy protocolized at all?
>> Fernando Zampieri: No. So we had no protocol for vasopressors use or vasopressor preferences. So that was left at the discretion of attending physician. So some -- I can say that the usual practice in Brazil is something. But we don't have data on that. So I would just be providing my opinion and not anything based on data, so we don't know [inaudible].
>> Bronwen Connolly And then final question around the fluid rate. What were the clinical or pathophysiological triggers which led to treating clinicians administering fluid boluses? Was there a difference in the propensity of clinicians to use higher [inaudible] amounts of fluids according to clinical pathological status?
>> Fernando Zampieri: So that's my greatest regret in BaSICS is not collecting enough data on why fluid boluses were indicated, right? So we provided some guide on that. So we just mentioned that the physician should consider a fluid bolus in that scenario. But we found that it wasn't feasible to do a very protocolized trial with such a large sample size. And we -- I kind of regret it a little bit, to not have enough data on that. I don't know if any of you guys want to comment on this, but I truly don't believe that the decision to start vasopressors or not to be much difference, let's say. So when we [inaudible] initially some sites were afraid to use lower infusion rate. So when we work a lot on engaging sites to adhere to this lower infusion rate, and as time went by in the trial, eventually we saw that most of sites were absolutely OK with lower infusion rates. So this was very interesting to see at a site level.
>> Bronwen Connolly Rob, I'll just check in with you before I go to any other questions.
>> Rob Mac Sweeney: Thanks, Bon. I think we better move on, I'm afraid. And like I mentioned, we are well behind time. So we'll move on to the panel discussion now. We've already introduced everybody, so we'll just get straight into it. Maybe I'll turn to you, Kath, and hear your initial thoughts about the trial, and I'm sure you probably have some questions for the investigators.
>> Kathryn Maitland: Thank you. I mean, I'm used to running some difficult trials, but I'm really in awe of this trial. Seventy-five ICUs, all following a protocol and the sort of the conduct of the trial ticked all the boxes in terms of excellence of conduct. And when you actually looked at the thought that went behind, so designing a trial and getting the design right, and the whole protocol is really -- you have to really, really think about absolutely every aspect. And the -- obviously the PIs have done that. Because that meant that people could implement it and follow the protocol. So I am very, very as I say, amazed that you were able to do this, and especially when the latter part was in times of COVID. So yes, absolutely congratulations. In terms of the two questions, in terms of the type of fluid, I don't really have any skin in the game there. And I'm going to let other people comment on that. But also in terms of the rates of fluid, I had a couple of questions around that. Obviously, I would be interested in this secondary analysis of why clinicians decided to give fluid boluses. I looked at the entry criteria, and you could actually be enrolled into the trial purely on an age criteria over 65 alone. But up here, I saw that there's quite a lot of patients who were admitted with elective surgery. Is it usual that you need fluid boluses after elective surgery, particularly on the first day? I mean I was quite curious about that. But overall, if you actually look at the volumes that were given on the first day, in respect to the rates, they were roughly even. So you know, there were still quite a lot of volume being given overall. So although the rate was slower in the slower arms, and they followed that, there was almost the same amount that was given between the two arms. What else did I -- had I picked up? So I did -- I was privileged to be able to have this shared with me earlier this week, and so I looked at the additional, the supplementary data on -- so the group that I'm very interested in, the sepsis group, and I saw that obviously there was no difference in mortality between the two arms and overall, it was between 47 and 48%. Is that what would normally be expected? I'll leave it there, because I know that you're pushed for time. Thank you very much.
>> Alexandre Biasi Cavalcanti: Yes. First, thanks so much, Kathryn, for -- Professor Kathryn for comments and [inaudible] thanks. Regarding the amount of fluid, really where the very similar [inaudible] is lower and faster infusion arms, with -- what hypothesis we had before the results [inaudible] may be for the patients who would receive fluids in a lower rate, maybe the final amount of fluid would be increased if the doctors would stop giving you fluid boluses as the patients would stabilize. That was not the case. The total amount was really, really similar between groups. And we have a lot of those patients as mentioned coming from the operating room, from elective surgery. Most of those were big, complex surgeries and this is why patients need ICU. Many surgeries were cardiac surgeries or other large surgeries, so this kind of patient, it's very common that patients have hemodynamic abnormalities, such as hypertension and other signs of [inaudible] cardiac output after the surgery. And the physician understands that, judges that those patients need fluids, and gives the challenges for those patients. So it's very common. And regarding the mortality of patients with sepsis, we have as mentioned a 47% mortality with this. This is somewhat higher mortality than many countries. We have had historically differences among countries in terms of mortality in sepsis. However, I think part of it has also to do with the eligibility criteria of the study, when we consider the set of risk factors that patients had. We knew that those patients that are eligible had those risk factors, would have a higher mortality than those who do not have. So I think are a special subset of patients with sepsis than other [inaudible] patients. So maybe it helps explaining the high mortality in the subgroup.
>> Rob Mac Sweeney: Thanks, Alexandre. We'll cross to Copenhagen, and Tine, your thoughts on the trial, please.
>> Tine Sylvest Meyhoff: Yes, Rob. I'm very humbled to be part of this panel, and Fernando, Alexandre, just many, many congratulations on this trial. It's extremely large and also individually randomized, robust trial [inaudible] but relevant clinical question. So I'm very impressed. And overall, general first thought is also that I'm actually excited to see all of these aspects of fluid therapy on the research agenda. They're being taken very seriously and also discussed, and also prioritized into high-quality trials like BaSICS, so I think that certainly helps us improve therapy. I'll also -- I'll try not to repeat any of the other comments, and I'll also go to the speed part of the trial. So I'll get in line with Simon, and your conclusion was that these data do not support slower infusion rate. And I get that this was also your intervention, so obviously it was phrased like this, but I'll get in line with Simon's comment that equally to me they do not support a faster infusion rate. So first comment or question could be if you're in a ICU setting and not routinely using infusion pumps, which was used in BaSICS, could this also support that you do not need to go on and do that for your -- invest in those for your ICU patients. You can go ahead with slower infusion rates. And a second comment could be regarding the different SOFA scores on Day 3, the patients in the lower infusion rates had lower at least cardiovascular and respiratory SOFA scores, could this -- do you think -- you mentioned several secondary outcomes, so obviously we have to be cautious, but could it be in line with some of the potential paradoxical effects of a fluid bolus that we've seen in FEAST and also in Andrew's sepsis studies from Zambia? Or do you think we should not pay much attention to them if the downline patient importance outcomes do not differ?
>> Fernando Zampieri: Those are great comments. So of course, I'm very fond of the infusion rate part of the trial, because that was something that we -- I'm very fond of it, and of course that was inspired by Dr. Maitland's FEAST study. So that was where it all started. So of course if you have a look at the secondary endpoint, and I know they are secondary endpoints and very exploratory. They all follow the logic or reasoning we had behind, right? So if you give a last fluid bolus perhaps you'll have -- if you did not have the harmful effect of lowering arterial [inaudible], perhaps reduce lung edema, et cetera. So that's in line with physiological reasoning, right? But of course we have to be very careful not to extrapolate physiological reasoning from patient-centered endpoints, and we have to be very modest on that. But of course, I truly believe that that's something we should discuss. And especially for the first part of what you said, of many sites before BaSICS did not use infusion pumps for fluid challenges, right? So it was common practice in Brazil to run 500 mLs at any speed you can, just wide open. And wide open can be anything from 20 minutes to three hours, right? Because it depends on venous access, et cetera. So when we protocolized that, that was a very -- it was a quality improvement for sites, and sites were very happy to do that. but I believe that what BaSICS can say is that for patient-centered endpoints, infusion speed [inaudible] infusion speeds we tested for patients that were already resuscitated in the emergency department doesn't seem to make much of an effect. If it does, we have to explore that a little bit better, and perhaps we should still work on that and do some secondary analysis, perhaps run other trials on that to understand further whether that translates to other relevant endpoints. But that's my favorite question of the trial, because -- and that's very interesting to see the secondary results. And so literally we had to keep a lower tone on that. Because a lot of secondary et cetera. Yes.
>> Rob Mac Sweeney: Great. Thanks. We'll cross to Pittsburgh, and we'll go to Chris first. Chris, as an associate editor at JAMA, you no doubt have seen this paper for a period of time. Perhaps you could give us your opinion, and also maybe some of the comments that were made as this was reviewed as it went through JAMA.
>> Christopher Seymour: Great. So hopefully everyone can hear me, and thank you for the opportunity to join the panel. There's not enough superlatives, right, for these papers. And I would be repeating those that others have shared, but of course, congratulations. This is quite the duet. And that, wearing the JAMA hat for the first few comments here, you know, there wasn't necessarily a question, right, of whether this data and the quality of this trial was going to land in our pages, but rather how to present that for the reader. Was this one manuscript? Was this two? were these questions truly scientifically separate? Certainly we had the absence of the interaction, and so that gave us a little bit of a foundation there to make these two separate pieces. In the end, you know, we had Professor Coopersmith run a fantastic editorial that should also be released today. And I think we were excited to get this data. The [inaudible] been producing fantastic and rigorous trials that are very large now for a number of years. As a clinician, I'm also struck by this fantastic work, and can only imagine the critique from others might be, well, you didn't answer the question I wanted you to ask, right? Because the clinical moment of deciding how to administer fluids is so different, depending on where you are in the hospital, what phase you are in your intensive care, and even of course what country you're in. So that's what I'm struck by. As I listen to this amazing presentation with important data that we'll be analyzing for years is there are so many other clinical moments we could test. Not only do we have an average treatment effect across the patients, but the trial design averages the clinical moments across their stay, and says we're just going to implement one strategy for all moments. It makes us think about the AI clinician and what Dr. Komorowski's group is doing as ways to sort of find these individual treatment effects. So all told, this is fantastic work. We were so excited to get this and to partner with Critical Care Reviews and everyone here to release this data today. So thanks.
>> Rob Mac Sweeney: Thanks, Chris. We're going to cross town to the other side of Pittsburgh, and to Andrew Althouse, our expert statistician and methodologist. Andrew, you've had a chance to look at this. How does this statistical plan and the execution of it look to you? It looks to my amateur eye very robust. We've had some comments made once already by Simon in his editorial. Perhaps you could touch on some of those, and give us your overall feeling for this.
>> Andrew Althouse: Yes, thanks a lot, Rob. And again, congratulations to the team. You know, I'm glad to have had the opportunity to contribute here. I mean, I would agree with your comments, Rob. This is very -- a robust-sized trial. I think it's rare that we get trials of this magnitude in the ICU, although we're starting to see more and more of them, thanks to some really great efforts around the world. I think it had very sound and very transparent statistical anwyslis plan. You know, I certainly applaud the authors for publishing their statistical analysis plan, being very, very clear about what they intended to do. I guess there's been two comments about this statistical analyses thus far that are worth addressing. The first about the imputation. So practically for the reader here, the key takeaway is that there was actually so little missing data, I can't imagine this made much of an impact once way or the other. And I mean that's thanks to the incredible system the trial team put in place to make sure they had fairly complete outcome data. It think it was, you know, on the order of, you know, no more on the order of a few dozen that would have been missing primary outcomes data that had to be imputed. So that does not trouble me very much, because I just think that there was so little missing outcome data that it would have made very little effect. Although I do agree that imputation is generally preferred to a complete case analysis because of the concern that some sort of bias could be introduced, depending on the mechanism of missingness. But there's just so little missingness here, I just doubt that it would make much difference. I think the more interesting comment that came up that I can bring a little punch is the question about using an adjusted survival analysis as opposed to an unadjusted analysis. And the truth is, these results, because the trial was so large and so robust, and the results are sort of what they are, I doubt this would have made much difference either. But I will say that it's probably somewhat of a little-appreciated fact that you do gain something in your analysis by adjusting for a few baseline variables. And that's actually true, even if they are balanced between the groups. I think often there's this belief, oh, it's a randomized trial so you're guaranteed -- you're "guaranteed" balanced groups, why would you need to adjust for any variables? But when you adjust for something that is associated with the study outcome by explaining a little bit of that outcome variation, you actually end up getting a little bit more precise estimate of the treatment effect, which can increase your power to detect a difference if there is one present. So you know, I actually applaud the authors for doing that. Something I'd like to see used a bit more in randomized trials, although I will concede the point again here, it probably, because the trial's fairly large and robust, would not have made a great deal of difference one way or the other. I guess my last point that I'd bring here, you know, I think we always have to be careful ever saying that we've proven that there's no difference, or anything like that on a negative -- so-called negative trial, because we know or we should know that the absence of evidence is not always necessarily the same thing as proving there is no effect. But with the large sample size we had here, the resulting precise, you know, fairly tight confidence intervals on the treatment effect and in the two trials, it certainly suggests that sort of on that broad scale, whatever effect may exist of the -- within the type of fluid, or between a faster versus slower infusion, at least in this type of patient group, and in this setting, seems likely to be small. And I love that point that was made by a couple of our panelists [inaudible] which is being mindful of the local context. Thinking about things like is there a difference in cost of these fluids in a location, or is it easier to deliver a faster or slower infusion in a certain location. So I think this data is great because it suggests that at least in some settings, the difference is likely to be fairly smaller, perhaps trivial, and in some ways maybe that's reassuring that whichever you do, it's probably OK.
>> Simon Finfer: Can I jump in, Rob?
>> Rob Mac Sweeney: Yes, of course, Simon.
>> Simon Finfer: I mean, I think Chris raises, you know, some interesting point, because you know, around this whole issue about how we have these very different time points and what we're studying, and what basic study that we want [inaudible] is does the election of a type of fluid in the ICU influence patients' outmode? Because obviously we're working in ICUs and we're deciding what fluids to give patients, and that's kind of ignoring what's gone before, and what's going after, and where it [inaudible] different time. So I think -- one of the things I looked at in SMART because they did their electronic health record for all their data collection, they actually had a record, and I think this is a particularly U.S. kind of centric thing, that they actually knew all the fluid that someone was given throughout their entire hospital stay without having to go away and collect it on bits of paper. And basically in that trial, patients who had gone through the ICU, 2/3 of the fluid they had received during their entire hospital stay was in the ICU. So you'd expect that that would have some impact. But I think, you know, we can't get away from COVID. And I think to me, COVID, you know, with recovery and [inaudible] and all these adaptive trials with stopping rules that we've all kind of shied away from in the past, and adapting, you know, that that's maybe a more efficient way for us to do research. I mean, I've been very much in the Gordon Guyer [assumed spelling] camp of thinking we shouldn't stop trials ever. But in reality, the resources required to do these trials, and there are so many questions, I mean I'm wondering if, you know, maybe the landscape's changed, and maybe we're going to have to think a bit more about efficiency. I mean, our government has spent so much money on COVID. I mean, they're trying to get everybody tested every other day, goodness knows how, running up, you know, billions, trillions of dollars like most governments around the world, in debt. And that's impacting their funding of research. They're cutting back on -- you know, there's not enough money. They're spending it all, we're giving it all to the vaccine makers and to the people who run tests. And it is actually impacting on research. So I'm wondering whether we need to change our mindset a bit about that. And I'd be very interested [inaudible] maybe people can tell Bronwen or through Twitter or whatever means, what are the questions that people actually want more questions we want answered about fluids? I mean, there's some good -- clearly there's a lot of interest in whether we should be using vasopressors earlier. We do that anyway. But I think, should -- my thought's around maybe we need to think a bit. You know, maybe the 11,000 patient trial is phenomenal, fantastic. But maybe if you've been -- using futility boundaries, you'd have stopped earlier, and then taken those resources somewhere else. Just some thoughts.
>> Rob Mac Sweeney: Fantastic. Thanks, Simon. We'll try and magnify that comment and send it out, certainly across Twitter. I'm going to turn to both Kath and Tine. Now Kath's FEAST work showed previously that fluid therapy, certainly in febrile African children who were shocked, worsened outcome. Tine's classic program is looking at a restrictive fluid regime. I would have through that the slower infusion rate would have chimed with both programs. Perhaps I could get your thoughts on that.
>> Kathryn Maitland: So the trial that we did was obviously in children. I probably don't need to expand. I think most of the audience may be familiar with it. And we literally gave a homeopathic dose of fluid, which caused harm, at 20 mills per kilo, whereas many of the guidelines were recommending far, far more fluid. And yet, that caused harm. And the harm we started to see within hours of them receiving fluid. The harm was early, but it was also -- it hadn't, because we followed the children up for up to about 100 hours after they received that single, largely single bolus, that the curves did not come back together. So they weren't able to get better faster. So that was very surprising. And so I think it was -- it's not just the rate, but it's also the volume. And overall the rest -- you know, all the sort of fluids that they were given were relatively modest compared to that. So there was an ovine model of this FEAST-in-Sheep, by John Fraser's group in Brisbane, and they also showed that actually troponins went through the roof once they gave fluid boluses. I know that's in sheep and not humans, and it's an anti-toxin model. But again, that was early, around about eight hours after they started the saline bolus. So I think there's still more questions that we need to ask around volume of fluid. Thank you.
>> Tine Sylvest Meyhoff: I'll just join on that. And I was actually thinking of that ovine model as well. We also saw in the classic feasibility trial at that time it was conducted, it was expected that the patients randomized to the restrictive strategy would have a way higher noradrenaline requirement. And in a post hoc analysis, they had the same noradrenaline requirement as the [inaudible] group. So I still wonder if this could be in line with some paradoxical human [inaudible] response, maybe also seen in BaSIC. I'm not sure. I think we know way too little about this initial human [inaudible] risk response of a fluid bolus. In terms of the population, in the CLASSIC trial [inaudible] et al with septic shock, so they differ from the population basics. We'll expect them to have received both prior to randomization, quite a lot more fluids going into the trial, but also during the first days of randomization, receive quite a lot more boluses and also I guess maintenance fluids in the setting that we're conducting them.
>> Rob Mac Sweeney: Fernando and Alexandre, can I ask, have you had the opportunity to take any mechanistic studies to look at how fast or slow infusion rates could either cause benefit or harm relative to the other group? 3
>> Fernando Zampieri: Yes. So we had one such study that was led by Flavia Machado that was a subset of 120 patients or something in the BaSICS trial where we have somewhat [inaudible] negative values before and after fluid expansion. We are actually analyze this data, so we do not have data on that to present. But overall, I believe that the questions that were posed and the question is why are we still doing fluid bolus, right? So -- and somehow we know for a long time that fluid bolus have a very short half-life. There is a paper by Flavia too where I know the duration of the -- and [inaudible] benefits of fluid bolus is limited to half an hour or so. And perhaps it comes at the cost of increasing vasopressors or [inaudible] tissue edema and organ failure. So I totally agree with what Professor Maitland and Tine said regarding fluid boluses. It's time to reconsider why we are doing fluid bolus. And not only speed, but the volume, the time of using fluid, et cetera. So it's time to reassess everything again, and stop doing the same mistakes. So I would love to have done something similar to FEAST [inaudible] but of course it will be very hard to get people to randomize a very large number of patients. Even though the slower infusion speed was not very well received, you can imagine, because people were afraid that we were not going to correct the cardiac output and the blood pressure as fast as we could, and that's harmful for patients, et cetera. So if I had to give a step forward, that would be do not give fluid bolus, or something like that. That would not be well-received. So I believe that BaSICS was somehow an intermediate step between that was just -- now we can say that perhaps it's reasonable to think that giving fluids lower is actually safe so there is no rush to use a very fast infusion rate or something. And now we can cool down, perhaps reconsider whether we should or not give fluid bolus, and not say that we have a very narrow window to give that amount of fluid, et cetera. So if that's the legacy of BaSICS, I would be truly happy, and that's enough for a trial, you know.
>> Rob Mac Sweeney: Thanks, Fernando. Chris, can I ask, where do you think this question goes next? The most obviously example I can think of is the CLOVERS trial, where we're looking at less fluid and early vasopressor. Is there anything else that you can think of, or where do you think the next step in this field is?
>> Christopher Seymour: Yes, that's a broad question. I think you could ask all of us on the panel, we might give you a different answer, to be honest. Which probably speaks to the number of unknowns. So yes, I'm excited to see results of CLOVERS as well. So as some know, CLOVERS is a trial by NHLBI, as part of the PETAL network, testing different fluid and vasopressor combinations, looking to give a smaller fluid bolus size with an earlier initiation of vasopressors, similar although not exactly the same as the classic program. And so, yes, I think that's an exciting future direction. I mean, when the trial controls the type of fluid and the pace, what's not controlled again, is the clinical moment at which -- I mean, what's left open are the other times. Is this in the ED? Is this later? Is this the dose altogether? And then, of course the target. What is the clinicians administering their fluid to? And is that also an open question? I think Simon brings up a fantastic idea. Why not have a remap for all of it, where we have domains and arms that are incorporating all of these together? The complexity of a trial like that is daunting, but the alternative is perhaps doing one trial after another after another for many, many, many years.
>> Rob Mac Sweeney: Thanks, Chris.
>> Simon Finfer: Have you got time for another quick comment from me, Rob?
>> Rob Mac Sweeney: Of course.
>> Simon Finfer: If we look at -- I mean, one of the things, and again, Chris touched on it there, what is our target? And I think that's really where our challenge is, is that we have -- I mean, I'm so old that I remember, and none of my junior colleagues see this now, people on ventilators with three chest tubes in each side, because they'd had their lungs blown apart by us, you know, people trying to normalize CO2. Because that is what we were taught we ought to do. And if you look at, you know, odds that low tidal volume basically said that it's better to -- the mini-ventilation was the same, so better to have lots of small breaths than to have a few large breaths. If you want to really simplify it. But nursing staff don't like that. They don't like patients who are breathing very fast, and looking upset. And they don't like tachycardic patients, and they don't like patients who look like they're a little bit hypovolemic. And that I think drives a lot of practice. And we [inaudible] even in the hospital setting, back to the study, you know, they didn't practice low tidal volume ventilation when we looked at afterwards. So I think one of the -- that question, what are our targets? And if we know our targets, which I think we do pretty well in, you know, the ARBS and the low tidal volume stuff is pretty convincing and compelling that people don't do it. How do we then translate if for instance CLASSIC and CLOVERS, and there's a RISE fluids program in Australia, which is going to start with no -- less fluid, more vasopressor in the emergency department for sepsis. So right at the front door. If those things do suggest that we should be practicing differently, a lot of trying to change that into improve patients' outcome is going to involve reeducating certainly my country entire critical care nursing population. That what they currently view as a patient who looks untidy and unwell, and they haven't got nice straight lines on the monitor, is actually -- fixing those things is actually harming the patients. So I think that question of what is our targets is a really important one.
>> Rob Mac Sweeney: Thanks, Simon. Andrew, want to bring you in on kind of a methods point. Simon had added in his editorial that balanced solutions led to a worse outcome in those that were associated with a worse outcome in those who receive balanced solutions. There were 18 or 19 secondary outcomes which were considered exploratory. This was a tertiary or subgroup outcome. As a methodologist, how much value do we put on this result? Clearly Simon's SAFE trial from 18 years ago maybe demonstrated harm with albumin in traumatic brain injury as opposed to saline in a subgroup. So he has that in his head to think about when looking at these results. But as a statistician, perhaps somewhat separate from a clinical ICU background, how would you view that?
>> Andrew Althouse: Yes. Boy, this is always one of the hardest things I think to look at with trial results. Because on one hand, you say we've got all this data, and we want to learn as much as possible from the trial. You know, it seems silly almost to say, well, we're just going to look at one outcome, or we're just going to look at two outcomes. We should look at a lot of things to try to understand what happened to these patients, and whether it be subgroup analyses or just looking at secondary and tertiary outcomes to understand if there's other things that maybe we didn't expect, or that have clinical relevance, you know. So on one hand, we say great. We want to learn as much as we can from the trial. On the other hand I think, as most of our folks probably know, the number of statistical tests that you do without any sort of correction for the fact that you're doing more statistical tests does increase the possibility that one or two of them will, I hate to say look positive by chance, but I think that's the term that resonates with most people. You know, I always joke, if you stand there flipping a coin all day, at some point, you're going to get four or five heads in a row. And that doesn't mean that the coin was suddenly unfair at that time, it just means you stood there flipping a coin all day. The same thing happens when you do a lot of statistical tests. So I think with secondary outcomes, they're generally best viewed as some combination of confirmatory and exploratory. You know, maybe they bolster your confidence in the primary outcome if they're kind of going the same direction as the primary outcome. In some cases maybe they go in the other direction, but you have a sense of why that would happen. Like you know that the treatment might get you a benefit in one regard, but potentially have a possible harm or side effect in another regard. Statistically, there's no great answer to this for me. Because in one respect, at a very rigid side, you say you should just correct for all of the multiple comparisons you do. You know, you take some sort of penalty, statistical penalty, and you adjust all of your comparisons to make sure that you don't increase the overall risk of any sort of false positive result. But on the other hand, that's actually very, very harsh. That makes it very hard for almost any finding [laughs] to become significant, particularly of secondary outcomes or subgroup analyses that you're looking at. So I think there's no easy answer to this. I think it has to be this combination of plausibility of understanding is there other prior evidence? Or evidence from this trial that makes -- that says why we might see a particular out there effect in a certain subgroup or in a secondary outcome? But it's really, really tough, Rob. There's no great answer to that. I wish I could be giving a more satisfactory answer, but the truth is, I don't have one.
>> Rob Mac Sweeney: Fantastic, Andrew. Yes, it's a hard question. We're over time. What I'm going to do is quickly go around our panelists, and just ask each person the same question: what is this going to do to your practice tomorrow morning? Kath, can I maybe start with you in Kenya?
>> Kathryn Maitland: Well, not much, actually [laughs]. But in terms of, as I said, there's no skin in the game in terms of types of fluid, because the Plasma-Lyte is not available here. But in terms of rates of fluid and overall volume, I think we've already hopefully addressed that question. Sadly, a lot of people are still wedded to fluid boluses in Africa, pediatricians are. I just keep on trying to get the message out there. But yes, just congratulations for a fantastic trial. And I think it's a step forward in terms of moving the goal towards saying well yes, we can give slow infusion rates safely to the next question. And I think that allows you to ask the next question.
>> Rob Mac Sweeney: Fantastic. Thanks, Kath. So Tine, just very quickly.
>> Tine Sylvest Meyhoff: So Plasma-Lyte's also not routinely used in Denmark. So that part wouldn't change practice a lot. And I would be safer using the slow infusion rate, but wouldn't worry too much about either or from these data.
>> Rob Mac Sweeney: And Chris?
>> Christopher Seymour: Yes, so I think in the U.S. and frankly even myself have been enamored with the trials from the Vanderbilt Group. I'll probably be using more saline than I had six months ago, with a little bit more of an open mind to equipoise.
>> Rob Mac Sweeney: Yes, and I think as a clinician I might maybe use a little bit more saline as well, which I wouldn't have thought of saying before these trial results. Andrew, I'll maybe not ask you that question. Might be a touch unfair. So we'll finish the panel discussion just with a quick comment from Alexandre and Fernando, and then we'll move into Fernando's last talk. So just a quick comment, gents, please.
>> Alexandre Biasi Cavalcanti: I'd like to thank everyone for the excellent discussion, and we're happy that the results may be contributing somehow to the decision regarding the best choice of fluids. And I think this was a large study, and Simon comment before that maybe we [inaudible] stop it earlier because of futility. On the other hand, as this is the most commonly used drug for [inaudible] patients, I think even a small difference would count. So I think even if the data is not [inaudible] for this study, the possibility that we have to have other studies and [inaudible] other databases together with [inaudible] and SMART and all the trials, will be super helpful. And yes. I think what we have -- we are very pleased to have had the opportunity to be with this data for the [inaudible] patients. So thanks.
>> Rob Mac Sweeney: OK. From me, thanks very much to our panel. We're way over time. If I had a boss, I'd be fired. But I don't, so I'm not. We'll move on now to Fernando's final talk. It's a brief talk, the learnings from BaSICS. Clearly this was a huge, huge trial and there will be lots of learnings for other trials out there to glean from this talk. So Fernando, back to you.
>> Fernando Zampieri: Oh this is just -- mostly comments here were already addressed. So just to make some formal final comments. So I'd just like to reinforce that the benefits of the clinical trial far exceed the research question, right? So sites we improved their registry in fluid balance. BaSICS created jobs, right? So some research staff and data collectors were employed, because we were able to refund the sites for -- reimburse the sites for their work. This trial happened to consolidate BricNet. It's a very large research network. And of course, the money that we gave to the site, the reimbursement rate was used to several questions from sponsoring young physicians and young nurses to go to conferences, to buy new ultrasound machines, et cetera. And we are very proud that we do that. I don't regret a lot of things. So I don't regret doing a very large trial. I don't regret doing a pragmatic but not sloppy design. I don't regret using 90-day mortality as a primary endpoint, because if I did a composite endpoint, people would say, well it's a composite endpoint. If we did renal replacement therapy, but it was positive for renal replacement therapy, but negative for mortality, people would say well, but it does not change mortality. So what's the point? So mortality is pretty much straightforward and [inaudible] very straightforward. And I don't regret pre-planning a IPDMA with other trials, and I don't regret all the work that sites had for using fluids for challenges [inaudible] and dilutions. There are some things that I could -- perhaps we could have done differently, right? So looking retrospect, we had different rationales and different concepts for interventions. And it might seem that perhaps we could have considered different endpoints for some secondary analysis. I also regret, partially regret, using some secondary endpoints, including SOFA score and using [inaudible]. We have data [inaudible] two that probably didn't add much. And we failed at using the main analysis to consider all the time dependencies and the competing risks for these endpoints, right? So we amended that with some secondary analysis that are in the paper. But perhaps we should have thought a little bit more about this. Perhaps we should also have collected a little bit more data on some aspects like, you know, type of surgery, the ICP values in TBI for example. I regret not collecting data on why the fluid challenges were formed. So overall, of course it's easy to say everything afterwards when the trial is completed. But most of all, I believe that we could have worked a little bit more in the infusion rate study, or considering other different analysis. So if I may, so I don't think we should abandon mortality as an endpoint, when this data is actually needed to change practice, right? So if this trial was positive for Plasma-Lyte or whatever, results would be, wow, it's very easy to make clinicians understand effect size when the endpoint is mortality, right? On the other hand, as was said by many, we did not consider the patient journey inside the hospital, right? So this was a point that was very well made by all the panelists. So perhaps we should have considered the patients from admission to discharge. And of course, we are just working on a small timeframe. And finally, the only thing that perhaps we can discuss, when BaSCIS was designed, and so this was the standard, right? So BaSICS is a very traditional trial in the sense that there are no adaptive rules. So it's not a vision trial. We did not have different endpoints that require some more advanced modeling. But of course perhaps for future trials, some different endpoints like, you know, "days alive and free of" which can be assessed in several ways, either [inaudible] could be very interesting endpoint, because it can bind, you know, mortality with need for other support. So that's going to be interesting. And perhaps in ratio with -- comes from the cardiology field is something that perhaps we should work a little bit more in critical care. We are going to run a secondary analysis of patients using the ring ratio. And perhaps other studies should consider one of those approaches to maximize our -- or to maximize finding a difference and provide clinically meaningful endpoints. So that was it. It was just just very quick comments here. Thank you.