Thanks for coming to this talk today, Behind the CHESS Transfusion and the ICU Guideline. My name's John Iaccarino, I'm the Director of Guidelines for CHESS, and I'm excited to start us off with this session, which should be pretty fun and interactive, a little bit of a different style than we're used to, where we're actually gonna ask you to be a guideline panel throughout the talk. So my goal, as I mentioned, this is me. I work at CHESS, and I have nothing to disclose aside from my relationship at CHESS. Except the only disclosure I probably have is this is not really my talk. This is a talk of Isra Sogier, who works for us at CHESS. She's a guideline methodologist. She couldn't be here in Hawaii, and so I'm really just being her voice and speaking on her behalf. So she deserves all the credit for this one. So I'm gonna start us off, really, for the first 10 minutes or so, just to kind of provide you with a framework to give you the ability to act as a guideline panel during this session. And so we'll quickly review a guideline methodology known as GRADE. We'll explore a framework where we use evidence to drive a decision to make a recommendation. And then we'll kind of just practice drafting and grading a guideline recommendation before I hand you over to the experts who will walk you through the evidence. So GRADE methodology is really just a cookbook for us on how we do guidelines. It stands for Grading of Recommendations, Assessment, Development, and Evaluation. Really, it gives us a protocol. It gives us a step-by-step formula for how we do guidelines to ensure that we're doing them consistently, transparently, and with rigor. And this is really a broad overview of what the GRADE methodology process looks like. First, once we have a topic, we build our panel. We ensure that our panel is free from conflicts of interest, according to the Institutes of Medicine. We develop questions related to our topic using the PICO format, Population, Intervention, Comparison, and Outcome. Then that's really when the real brunt of the work begins, where we perform a full systematic review, starting with a literature review and multiple databases. We gather all of the relevant studies for all of our questions and synthesize the data from those studies, performing meta-analysis frequently. We build what's called an evidence profile. And really what this is, it's in table form, explaining all of the evidence that we have for each individual question that we have for our guideline. That provides the framework for what is really the bulk of making recommendations, known as evidence to decision. And the evidence to decision framework, we're looking at the evidence, but also considering other factors, which I'll walk you through when deciding whether or not to recommend Intervention A or B or Intervention A over nothing. This is discussed by the panel at large, and ultimately recommendations are formed, which leads to the publication of a guideline. So this is just a very, very broad overview. And today we're pretending that all the steps have been completed really through the evidence profile. So what you'll see are elements of an evidence profile, and then my colleagues will kind of walk you through evidence to decision. So those will be the main steps we focus on during the talk today. So the evidence to decision framework is a essential part of GRADE that basically not only guides us through the evidence for each question, but it asks us to incorporate important components that we may not think about outside of the evidence related to a question. So first, we focus on the evidence, right? So how substantial are the desirable effects and the undesirable effects, right? So we're talking about adverse effects, we're talking about mortality benefit, we're talking about days missed from school or work, really patient-centric outcomes that might be desirable or undesirable. How substantial are those for that question? What's your certainty of the evidence, right? So are we talking about high-quality RCTs? Are we talking about observational studies, single-arm observational studies? Are we talking about experiential data? Maybe there is no published data, and so we're relying on the experience of our experts alone, probably the lowest quality of data we have. And then we're talking about the balance, right? So how do we weigh these? Is there a mortality benefit, but at the same time, a higher risk of decreasing quality of life, right? How are we weighing these against each other? So these are the things we're gonna ask you to think about as we go through the evidence. Beyond that, though, there's other questions that may influence whether or not we decide A or B. So is there variability in how patients value outcomes? The language here can be a little tricky, so an example probably best shows what we mean here. Let's say we're talking about a blood-thinning medication and maybe one outcome we're looking at is brain hemorrhage. Another outcome we're looking at is asymptomatic DVT. We both think they're critical outcomes for different reasons, but a patient probably values one of those more than the other, right? And so how do we weigh that? How do we weigh the variability when we're thinking about the outcomes? That's probably an extreme example. It gets trickier if we're comparing a pulmonary embolism to a brain bleed, right? So would patients prefer to have a PE or have a bleed in their brain, right? And that's a hard thing to think about in terms of how patients might value those different outcomes. We also want to think about the impact on health equity. If we recommend an intervention, does it have the potential to increase disparities, maybe due to availability of resources or inavailability of resources? So we always want to consider that whenever we're thinking about a recommendation. How will this impact all the different populations who should or shouldn't receive the intervention? We also think about implementation outcomes such as acceptability. So key stakeholders, the patient being one, the providers being another, insurers, healthcare systems. How would they perceive what we're going to recommend? Is it possible that, are we recommending something that has such high pill burden that patients aren't gonna be acceptable to take it? They're not gonna want to take 10 pills a day. So we have to incorporate that when we're thinking about whether or not we're gonna recommend an intervention. In addition to that, we think about cost and resources. So I kind of lumped these next two together a little bit. So resources might mean people, right? Do you need a specific staff to achieve an intervention? Do you need nursing staff? Do you need a patient navigator? Do you need, what kind of resources are involved in enacting the intervention? It might depend on the intervention, of course, right? If it's a medication, this might be very little. If it's incorporating pheno into pulmonary function testings, that might require a lot more expertise, training, and a lot more resources. And then similar to that, is there cost effectiveness we need to think about? So we already talked about undesirables and desirables. Now how does cost factor in? Maybe there is a small benefit to intervention A over intervention B, but intervention A costs a million dollars more per patient. We might not recommend intervention A, even though we know there's benefit, but the cost makes it so not worthwhile that we may actually say that intervention B is okay. So we do have to always incorporate cost into that. Feasibility is a related outcome, right? But if we recommend intervention A, is it possible that people actually can do this? I'll take lung cancer screening as an example. When we were going through our lung cancer screening guidelines, some of the recommendations we had in terms of what a lung cancer screening center should have, we had to think about, well, is it possible that patients can get lung cancer screening if they're in a remote area? Are we asking them to be referred to a main center? Can they be screened at their local center? How does that feasibility work? If we're recommending something that requires so much expertise, it can only be done in three centers in the country, that's not very feasible, right? So how does that impact the recommendation? So once we have decided all these things, we kind of need to summarize as a panel. How do the desirables outweigh the undesirables, or do they, and which way of the coin do we find? And then we start talking a little bit about strength. So how clearly, is it certain that one outweighs the other? Is it probable that one outweighs the other? Or is it truly uncertain that we cannot find a balance? And so that's sort of the next step in deciding which way on a recommendation we're going to decide. When we think of recommendations, and if you've looked at CHESS guidelines, you'll see there's a common nomenclature. And there's two pieces to these. There's a strength of a recommendation, and there's a quality of evidence. So you'll see a CHESS recommendation. We recommend A over B, strong recommendation, moderate quality evidence, or moderate certainty in the evidence. So the strength is how certain you are that the desirable or undesirable consequences outweigh the other. So it's how strong do you feel that there is a benefit to A over B. The quality is a little bit more subjective. And that has to do more with what is our confidence in the estimates from the data that we pulled. Again, if our data is from observational studies, I'm less confident that the true point estimate we've calculated from those observational studies is the true point estimate. If we have a lot of high-quality randomized trials, my confidence that that is the true point estimate is much higher. And so that is a very simplified way of how we determine the quality of evidence. There's much more objective ways we do this. We're not gonna get into that. It's a little bit beyond the scope. If you're interested, you can email me and become a methodologist, and I can walk you through it in very, very good detail. And so we're really gonna focus today a lot on the strength of the recommendation. And so really here we have two choices. We have a strong recommendation and a conditional recommendation. And when we say strong recommendation, again, this is how certain we are that the benefits outweigh the undesirable, sorry, the desirables outweigh the undesirables. Really what we're saying with a strong recommendation is that pretty much every patient should get this. Almost all the time, this is the answer. Just do it, right? This is something we might consider a performance measure. And if you go the other way, if you don't do this, you're wrong, right? You're not practicing good medicine if you don't do this. That's really the meaning of a strong recommendation. When we think of a conditional recommendation, we're saying most of the time, this is probably what you wanna do. But there's some instances where you may need to talk to the patient. We may need to think about values and preferences a little bit more. This is where it's sort of slow down, think about it. Does this apply to this particular patient? I know this is what I should do most of the time, but is this an exception? These are not appropriate performance measures. Often it's due to not having certainty of evidence that's strong enough that we know the desirables outweigh the undesirables. These are more likely to change in the future, right? So we may issue a conditional recommendation for A. More evidence comes out, and 10 years from now, we may actually say, you know what? Intervention B is actually better. Less true with a strong recommendation. A strong recommendation is pretty much us saying, we have all the evidence we need. We know A is better than B. It's possible, but very unlikely we'll ever go the other way. And so here's kind of how we think about it. So strong recommendation for an intervention. Conditional for an intervention. Either or, meaning there's equivalence. Conditional recommendation against, or strong recommendation against the intervention. And so this is what we're gonna ask you today to come up with as my colleagues present some data on our blood transfusion guideline. So here you become the panel. I want you to think through the evidence to decision framework that we went through. And then we're gonna ask you to vote on the recommendations. This is a real guideline. We're actively working on this. We have not finished the guideline. And so it's something to look for in the future. We will not be presenting the final recommendations today because we don't have them final. But we want to kind of just show you what the process is like. All the data is real data from a guideline that'll be published in the future. And so I want to thank you for your time. I'll hand it over to Kevin O'Neill, who's. Strapping on his leg. Yeah. So my topic is when to transfuse platelets. And in putting these two scenarios for platelets and red blood cells together, we decided to put it in the context of patients that we might see in the ICU. So we have a case, I have a case, and Angel also has a case that kind of outlined the clinical, the PICO question and why it's important. So I'm Kevin O'Neill. I'm the past chair of the Guidelines Oversight Committee and the medical director for respiratory care in the MICU at Novant, New Hanover Regional Medical Center in Wilmington, North Carolina. I have no conflicts of interest and no financial conflicts of interest to disclose. This is the QR code for voting so that you can be an active panel member. So please go ahead and lock in. I'll give everybody a minute or two to get that done. Okay. Looks like everybody's got it. So the patient that I chose was a 62-year-old gentleman who was admitted via the emergency department to the intensive care unit. He was found in a homeless camp in the woods, has a prior history of cirrhosis and polysubstance abuse. When he EMS arrived, he was unresponsive, hypotensive, and febrile to 103. On arrival to the emergency department, he vomited and aspirated a significant content of gastric material and was subsequently intubated to protect his airway. Because of his fever and hypotension, a sepsis pathway was initiated. He was cultured at a chest X-ray that showed a right lower lobe infiltrate. Empiric broad-spectrum antibiotics and a fluid bolus were initiated per sepsis protocol. But despite the fluids, he remains hypotensive and has poor IV access. So norepinephrine is initiated via 22-gauge IV in his right hand with a rapidly escalating dose. On arrival to the ICU, the nurse, who's assuming care of this patient, asks you for vasopressin and better IV access. I'm sure something that most of us encounter fairly frequently. His labs are remarkable for mild hypernatremia. He has a pretty significant anti-gap metabolic acidosis. The lactate is seven, but there's also probably a component of either renal failure or some unmeasured anion. He's mildly hypoglycemic. He's thrombocytopenic with a platelet count of 19,000. He has a significant white count with a left shift and mild anemia. COAGs are relatively, are okay, and his liver function appears reasonable as well. So the question that was asked was, should this patient be transfused before a central line is placed? So current guidelines for thresholds for transfusing platelets in a non-bleeding patient in any situation currently, both in the U.S., from the American Association for Blood Bank, and from European guidelines, recommend a threshold of 10,000 before transfusion. That's primarily based on observational data in series of patients with hematologic malignancies, and a lot of that data actually is pediatric. So the adult data is pretty thin, and when you look at these recommendations, they're considered low-quality evidence, and would be, as John described, conditional recommendations, although there does appear to be a threshold at about 10,000 for bleeding based on this observational data. For central line placement, there actually are two recommendations in the literature, one from the American Association of Blood Bank, which recommends prophylactic transfusion for patients having elective central line catheter placement with a platelet count of less than 20,000. Again, low-quality evidence and a weak recommendation. The British Society for Hematology recommends considering central line placement by experienced staff using ultrasound guidance when you have a platelet count less than 20,000. Apologize for the typo there. So the PICO question that we formulated for this guideline was in adult patients with critical illness in thrombocytopenia who are undergoing a bedside vascular procedure in the ICU, should they receive platelets? So that would be the intervention versus no platelets. And the outcomes that we were looking at that were deemed critical were ICU mortality, major bleeding, 30-day mortality, need for advanced interventions related to bleeding, procedural complications, and then adverse transfusion effects. When the literature search was performed by the methodologist, there were two case control, not case control, but basically cohort studies looking at patients in the ICU. One of them had 26 procedures. The other one had about 285 central lines. Neither controlled for, they were not randomized, but they did not suggest that there was any significant difference in terms of bleeding risk. We were getting ready to sit down and talk this out when this paper came out, and this is another thing that happens a lot in guidelines is evidence kind of changes as you're going along. So this was published in the New England Journal earlier this year. Many of you may have seen it, but it was a randomized control trial, unblinded obviously. It would be very difficult to blind this sort of a study, but randomized control trial of prophylactic platelet transfusions versus no transfusions in patients undergoing central line placement, and they had 377, I believe, central lines placed. They had a bleeding scale that they used, which is relatively standard and similar to most of the other studies in this literature, and this is their patient population that they looked at, and it's very evenly matched, but I'll point out this, which is that the population that our PICO question identified, which was ICU patients, represented only about 40% of their patients. So the rest of them were patients that were receiving their central lines at other places in the hospital, typically on a hematology ward. They were matched, and their conclusions were that withholding prophylactic platelet transfusion before central line placement in patients with a platelet count under 50,000 was not, didn't meet the margin or definition that they had established before the study for inferiority and resulted in more bleeding than prophylactic platelet transfusion, hence probably out of transfuse. So this is the subgroup analysis that they provided in the paper, and it turns out that for central line placement, like real estate, location matters. So this is, if you look at where the lines were placed on the hematology floor, it actually made a huge difference in terms of bleeding risk. If you looked at the ICU, there really was no difference in outcomes, and really the difference in the slide is one less episode of bleeding in the group that was transfused. It also matters, the location also matters in where you put the catheter. And so if you looked at subclavian lines, placing a subclavian line into a thrombocytopenia patient was associated with a higher bleeding risk if you didn't transfuse them. I think that makes sense. But subclavian lines and femorals, although to a lesser degree, really did not appear to have a significant bleeding risk associated with no platelet transfusion in this study. So this is really the only randomized control trial looking at this issue, and you can see that you can tease out some of the ICU data, but there's part of the challenge of dealing sometimes with making guideline recommendations is that you have to pull evidence that's maybe more indirect than you would like, or you have to find a way to tease out the information that's relevant to your population. So this is the evidence table that John was mentioning. The first piece of it is looking at mortality, really no difference in mortality. Second, major bleeding complications and how platelet transfusions affected that. And again, really no difference. Allergic reactions, there was one more in the platelet group, as you might expect. If you don't give blood products, you're less likely to have a reaction to them. Makes sense. And ICU length of stay was actually lower, although not statistically so, in the group that was transfused platelets. All of this evidence was downgraded for both potential bias and then also imprecision. There's limited numbers of patients involved and very low numbers of events, so there's a lot of uncertainty about the evidence. So low to very low quality evidence for making a decision. When we summarize this, this is what John was going through. This is the table that the group put together, so yes, this is a clinically significant problem. The undesirable and desirable effects were felt to be small in terms of whether or not, you know, for giving platelets. There was significant uncertainty as to the quality of the evidence, so it was rated as very low. We felt that patients would, there wouldn't be much variability in terms of how patients looked at the outcomes, that mortality and length of stay in the ICU would be things that would be something, would be the patients would not weigh differently than maybe the team did. We weren't sure about resources. We weren't sure about cost effectiveness or equity, but we thought that certainly platelet transfusions would be acceptable and feasible in most of the hospitals that we dealt with. So the question that was then asked of the panel is, should thrombocytopenic patients, that's your P, in the ICU receive a platelet transfusion versus no transfusion before placement of a central venous catheter? So I'm going to ask you to vote on my slide. It's missing my question. I pulled the question out. There it is. Go ahead. So strong recommendation against the intervention? Conditional against platelet transfusion? We're not sure and either one might be okay. Conditional recommendation for or strong recommendation for it? So, okay. I had a nice slide with the bars that John had showed you on the, in there, so they must have pulled that out when they did this a little bit earlier today. Okay. I think that's the vote. So, you know, there's some, this would be one that would have to go back and be, and have further discussion by the group because typically, typically we require, you know, 75 or 80 percent agreement, at least in the direction of the recommendation, and 75 percent of people voting. So this would be one that would have to be rehashed. But in any event, you can see that this process could take a little bit of time and require a lot of effort on the part of the person coordinating all of this. So continuing with my patient, the next day he has a new Rite-IJ triple lumen catheter without any bleeding, but I won't tell you whether or not he got platelets. He's stabilized hemodynamically on norepinephrine and vasopressin. His renal function's improved, but he's otherwise, his labs are the same and he remains thermocytopenic. However, the nurse notes a new erythematous macular papular rash on his trunk and legs, and when they do the wake-up assessment, he's less responsive and really not moving his left side to noxious stimuli. So his focal neurologic deficits in hospital, the nurses would automatically activate a code stroke. So that's done. He has a head CT that's normal. Neurologists also would come automatically as part of that, and they evaluated the patient and were concerned about the possibility of West Nile virus, which is where I am, has been fairly active this summer. We've had multiple cases in our ICU. And so the neurologist recommends that we proceed with an LP. So the question for this PICO, which is also in the guideline, was should this patient, who's thrombocytopenic, have platelet transfusion before proceeding with an LP? It should not be placed. There are two recommendations and other guidelines, one from the British Society of Hematology, which recommends a platelet count of 40,000 before proceeding with a lumbar puncture, and then from the American Association of Blood Banking, prophylactic platelet transfusion for elective diagnostic lumbar punctures below a platelet count of 50,000. Again, weak evidence, low certainty. Again, all of this data, primarily from observational studies dominated by oncology, and in this case, overwhelmingly by pediatric studies. So this is how the PICO question was defined. So adult patients with critical illness and thrombocytopenia undergoing a lumbar puncture, intervention would be transfusion of platelets. The comparator is no transfusion, and the outcomes that were deemed critical are pretty similar to the ones in the previous question, mostly mortality, bleeding, and need for surgical interventions or complications. So in doing the literature search, as I said, there were some old, basically cohort studies looking at platelet thrombocytopenia and risk of bleeding. Really, not a lot of data. This study actually was probably the best data, but again, not related to the intensive care unit and indirect as far as our patient was concerned. But this was published about three years ago in JAMA. It's a large study looking at marrying several databases together in the Danish health system where they looked at the association of spinal hematomas over a 30-day period after a lumbar puncture in patients who had a coagulopathy, and thrombocytopenia was also included in that. This is data from that study. You can see that they looked at 83,000 lumbar punctures in 65,000 patients, of which about 1,800 roughly were in the 50,000 or less group. The patients that were thrombocytopenic really did not appear to have a significant increased risk of bleeding compared to the non-thrombocytopenic patients, as you can see on the graph. Right here. But they estimated that about 30% of patients actually did get blood transfusions and they had no way of actually measuring that. But the overall incidence of a spinal hematoma was about 0.2% in the study. So low, but if it happens, it's a devastating complication for the individual patient. So this is the summary of judgments. There really was no direct evidence for this, so we really could not build the evidence profile that you saw in the last question. So is this a problem? Yes. Desirable effects of a platelet transfusion appear to be small because the bleeding risk is small. Undesirable effects in terms of platelet reactions from other studies also appear to be small. Certainty of evidence is very low. People probably are not going to have differences or different values about whether or not they would not want a spinal hematoma and there's really nothing to weigh that against. So the balance of effects was felt to not favor either of the platelet transfusion or none. We didn't really have a lot of data for any of the other pieces of the question that John showed you when he went through the summary of judgment table. So this question, should thrombocytic pain patients in the ICU receive platelet transfusions before lumbar puncture? So with that, I'll ask you to vote as the guideline panel and we'll see what people think. So low risk, probably very poor outcome if it happens. Okay. Coming down the home stretch here. Okay, so again, this one's probably going back for a revote. But you can see some of the difficulty there is in trying to put some of these guideline recommendations together. So you're trying to balance what's the potential benefit versus the risk and sometimes you don't have the data that you would like to have to make that decision. So based on that, I want to thank you. I'm going to turn it over to Dr. Cruz who's going to take us through red cell transfusions in the ICU. Now we'll talk about RVCs and I know I'm probably the last thing between you and going to the beach. So let's try to get this done soon. But I'll walk you through some of the evidence and try to kind of go through the process of how we develop a guideline. My name is Angel Coz. I'm a pulmonary and critical care. I work at the Cleveland Clinic. I have no relevant disclosures. And if you can do the same exercise again. I see most of the phones down now. Perfect. So we will present this with a case as well. We have a 73-year-old patient with history of severe COPD and coronary disease and cirrhosis presents to the ICU with acute respiratory failure secondary to pneumonia, failing on invasive ventilation. Patient is studying antibiotics and vasopressors are required due to heavy sedation to maintain synchrony with the ventilator. There are no EKG changes and the dependence are negative. And these are the initial labs showing hemoglobin of 8.6 and you can see the rest. Platelets are fine. White blood cell is mildly elevated. So what would you do now? Would you transfuse RBCs because the patient is anemic or would you stay the course? And those would be the thresholds. You would transfuse to a level of 10 or you would keep it at 7. Okay, so it seems like we're not going to have that much discourse in this one, Kevin. Most of you would transfuse to seven. So let's walk a little bit towards the evidence that informs the recommendation. So the question that we framed, the PICO question is, in a patient, a critically ill patient, should we use a restrictive or a liberal strategy? And looking at the literature, most of the restrictive strategies focus on a hemoglobin seven to eight, and then the liberal is 10 or above. So some of the trials that were available for review was this one. This one included around 150 patients on each arm that had, this was oncologic patients, but as you can see, the difference in hemoglobins were, you can see that they're present there in the group that was transfused versus in the group that had the restrictive approach. But when it came to mortality difference, as you can see there, there was potentially a trend over survival in the liberal group, which, as you will see, may be something that we will continue to see in the other groups or in the other studies or not. But here we have a trend towards liberal, keeping in mind that this is a study of oncologic patients. So that will add a little bit of variability when we talk about the final recommendation, because this is a specific population. It's not all commerce critical care patients. And then when we look at, this is probably the biggest trial that we have had. This is a TRIG trial that was published in the 1990s. And look at the same question, all critically ill patients. And as you can see there, in this case, there was no difference between patients who had a liberal or restrictive approach. The P value was .1, so there was no difference. And then when it was divided between patients who had a low APACHE score, less than 20, or APACHE score, or even patients who were younger than 55, you can see there that the restrictive approach actually had better outcomes, better survival. So, and this is another study that look at patients on mechanical ventilation. It was a smaller study, only included about 100 patients. But first, you can see that there's separation in the levels of hemoglobin. Patients who got transfused to a liberal study had close to 10, as opposed to the restrictive strategy that had close to 7. And when we look at the survival, what you can see is that there's definitely higher survival in the restrictive approach. So, so far we have two studies that show benefit of the restrictive approach, and one that in the oncology population that had some benefit in the liberal approach. So, this is when we, with the help of our methodologist, we put all the data together, and we analyze it. And this is a forest plot that shows the different studies. I did not include all of them in the previous slides, but these were the main studies that inform for ICU mortality. And what you can see is that it is a neutral effect. It is one, exactly on the dot. And then when we look at the quality of evidence, so something important to consider is when we're talking about guidelines, when we assess randomized control trials, typically the evidence starts as high quality of evidence. When we're looking at observational studies, it starts at low. But then we analyze several different variables that include the risk of bias, and these are the elements that are evaluated in the risk of bias that include the random sequence generation. And you can see all of them there. Then if there's inconsistency, then if there's indirectness, imprecision. And this all seems a little daunting at the beginning, but as you become more familiar with the guideline development, it actually starts to make sense. But what we see here is that basically there's no benefit of transfusing with a liberal strategy, meaning that by keeping the hemoglobin or transfusing to a hemoglobin of seven is just as good when it comes to mortality. When we go to a one-year mortality, the history repeats itself, 0.99. Again, almost as close to one as you can think. And there were only two studies evaluating that, and all the other parameters that I mentioned earlier were also evaluated. So now we're going to look at some of the adverse events, and this is data from the TRIC study that showed that there was no differencing. There was a difference maybe in patients with cardiac complications, meaning that those patients that had the liberal strategy actually had more complications, and also infectious complications were, in this case, were similar, but the cardiac complications were more frequently in the liberal strategy. And this is another study. This is a study that looked at patients in septic shock. So it's a little bit different population. Again, this is something that adds a little of complexity to the decision-making. But if you see, there was no really major difference in the degree of, in the number of complications. You can see that's very small. And one can argue that this study maybe was not powered to detect the adverse reactions, although it had 1,000 patients, but still they were so infrequent that the study did not capture them. So when we look at overall, this is, again, the study after cardiac surgery. So again, different type of population. As you can see, restrictive and liberal when it comes to infectious events were very similar. Overall complications were also very similar. And ischemic events, which is also something that we are concerned about after cardiac surgery, was also similar. So when we put all of this again in the blender and tried to come with a uniform assessment of all of this, what the data showed is that the adverse events, there's actually a trend towards favoring the restrictive approach, meaning that there's less complications if we use a restrictive approach. So, so far, we know that giving blood does not improve mortality, does not improve ICU length of stay and other parameters that were looked at. And there may be a potential for complications, especially when it comes to the infectious events. And then here, when we look at pulmonary adverse events, and again, this was something difficult to put all together because different studies define things different. But when the analysis was done, you can see it was one, or as close to one as you can think, and it crossed the line of neutrality. So, so far, summarizing, we know that giving blood does not necessarily improve outcomes. It potentially has some deleterious effects, and it's using resources, right? Giving blood does not only involve getting the back red blood cells, it involves nursing time, all the supplies that have different use of resources in different parts of the world. So if we look at what it may cost here, maybe what it costs in a different amount of resources in other countries. So when we put all of this similar, is this a problem? Yes. The desirable effects, there was basically no significant benefit when it came to transfusion of right blood cells. The undecidable effects, we see that there's a small effect favoring the, I mean, favoring not giving blood. The certainty of the evidence, it was moderate. Most of this came from randomized control trials that were downgraded for specific reasons. Values, there were probably no uncertainty or variability in how we practiced, and I think that was shown in the initial poll that most of you would not transfuse unless a hemoglobin is seven. The balance of effects, it probably favors the intervention, right? Probably favors not giving blood because there's no benefit and there's potential harm. So when we put both in the scale, it probably favors to be more conservative with blood. Resources required, it varies because it's different in different parts of the world. Certainty of the evidence of required resources, there was really not much data to inform that. Cost effectiveness, again, there's no data on cost or the data on cost is very variable, but we know that the benefit is nonexisting or it's not there. So basically that, and then the effects on equity, it was not really data to evaluate that, and acceptability and feasibility. So with all of this information, kind of summarizing that there's no benefit in decidable outcomes, but there's potential for harm. How would you grade this recommendation now? Oh, I guess they sub-attached my ARS. I don't see ARS here. Yep. Okay. Old-fashioned. Show of hands. Who would go for a strong recommendation against transfusing packed red blood cells? Two, three, four, five. Okay. Who would go for a conditional recommendation against? Okay. More. Who would go for a neutral recommendation? Not many. Conditional recommendation to give blood to a more liberal approach. And then strong recommendation for the same, to give blood? No. Good. So most of the hands that I saw were on the, let's not give blood so liberally, which is good, and we'll have to see what the guidelines show once all of this data is put together. So continue with this patient. We're at day five now. Patient got better. We don't know if they got transfused or not, but patient was extubated, no longer on vasopressors. And on day six, patient is doing well, waiting to go to the floor. But as you know, floor beds are very, very, very, sometimes very hard to find. So the patient was just sitting and waiting, and then now the patient has melanotic stools. You get called to the bedside. The vitals are stable. Blood pressure, heart rate are all stable. The patient looks fine, but is having several melanotic stools. And when you order a new CBC, this is what happened. You see that the initial CBC, when the patient came to us, was 8.6, and now it's 6.7. What will you do now? Transfuse to 10, transfuse to 7. It's a patient who is actively bleeding, hemodynamically stable. Excellent. You guys should be in this panel. This question is, I think, not as complicated as the one that Kevin got, so excellent. And next question. What approach would you follow? Would you give one unit at a time and recheck? Would you give two units before rechecking? Or would you, because you're an ICU doc, let's get the massive transfusion protocol? Okay. Excellent. Do we know how frequently that occurs in real practice, though, that we only give one unit? That's something that, I mean, I think it's very, very nice that we see that that's what we want to do, but in reality sometimes it's like, I mean, the resident sometimes orders two or three units right off the bat, and that's something that I think we probably should be a little bit more cognizant about. So the data that informed this recommendation mostly came from this randomized control trial. That's one of the ones that we looked at 2013 as a study that came from Barcelona in Spain. And as you can see there, those were patients who were hemodynamically stable, and this is the key, right? Because if we have a patient who is hemodynamically stable, the story is totally different. But as you can see here, there was really no difference when they received restrictive varicose transfusion strategy when it came to survival. And this was at the beginning, the one on top is at 45 days, and then as you can see here on the right, patients who had cirrhosis with a child puke A or B potentially benefited from that restrictive strategy. And this one is one of those that they hypothesized that maybe it is because of the acute transfusion of blood can have an effect in portal pressures that actually can make the bleeding worse. But I don't know that we have other than just being a hypothesis that that's necessarily the reason. So when we look at all the studies that informed this question, and again, here is the same question, patient with acute GI bleed, hemodynamically stable, should we transfuse to a liberal or a restrictive approach? All the studies showed here is .68, and the confidence interval is .48 to .97. In this case, we have high-quality events, three randomized controlled trials that show that restrictive approach is probably better. When we look at adverse events, this is another study that as you can see, the mortality was fairly similar, but in the serious adverse events, the liberal strategy had more, but it did not reach statistically significant. The treatment effect went from the negative to the positive. So there's potentially not necessarily adverse events associated with using a more liberal strategy. And when this was put, all the studies were put in the forest plot, also there was a signal of benefit towards using a restrictive approach when it came to adverse events. So, so far summarizing, using a restrictive approach improves potentially the beneficial effects and also produces less adverse effects. So when we look at this, is this a problem? Yes. Desirable effects, meaning is there a decrease in mortality with using a conservative approach? I think what we saw in the forest plot is that yes, potentially a moderate effect. The undecidable effect were not as big, but they were trivial, but they were still there, potentially favoring a restrictive approach. Then the certainty of the evidence was moderate. Most of them came from randomized controlled trials. There is no important uncertainty or variability. The balance of effects favors the intervention, using a restrictive approach. The resources required by not giving blood, there would be a savings in resources. We cannot quantify what that is because, again, it varies in different parts of the world. And then the certainty of the evidence of required resources, there's not much other than we know that if we don't use or give something, we will be saving resources. Cost effectiveness, there were no studies, and then the acceptability and feasibility, we believe that they are, there is, it's a yes. So what would you recommend if you were on this panel? Would it be a strong for being restrictive, a conditional for being restrictive, neither or, strong or conditional for being more liberal and transfusing to 10? I guess I'm going to have to do a raise of hands again. Who would favor the first one, a strong against recommendation? Okay, good. What about conditional? Couple hands. Neutral? Okay. Conditional for the intervention? No. Strong for the intervention? Okay. Sounds like in this question we have more agreement than with yours, Kevin. But not 75% of the audience. Yeah. But there were a few, several abstentions. So stay tuned. This is a guideline that should be coming in the next few months so I think it will be really, really helpful to inform our day-to-day practice. So thank you so much. Thank you.

John Iaccarino

Director of Guidelines for CHEST

GRADE methodology

platelet transfusion

ICU

prophylactic platelet transfusion

bleeding risk

red blood cell transfusions

restrictive approach