Please welcome Dr. Greg Martin. Good morning. It's my pleasure to introduce this morning's thought leader speaker, Dr. Maurizio Ciccone, who will present our Max Harry Weil lecture. Dr. Ciccone is an anesthesiologist, intensive care specialist, and chair of the Department of Anesthesia and Critical Care units at the Humanitas Research Hospital in Milan, Italy, where he's also the vice president of the MedTech School, the combined medical and engineering school at Humanitas University in Milan. Maurizio completed his studies at the University of Udine, Italy, as well as time at the Universidad Autonoma in Madrid, Spain, and received his doctoral degree from St. George's Hospital of London, and then participated in the GCSRT research program at Harvard Medical School. Maurizio worked for 14 years as an NHS consultant at St. George's Hospital in London, where he chaired one of the largest critical care departments in Europe, and he is the immediate past president of the European Society of Intensive Care Medicine, which he led quite ably during the COVID-19 pandemic. Under Dr. Ciccone's leadership, ESICM established its reputation within European institutions. In 2018, Dr. Ciccone was awarded the Silver Griffin by his alma mater, the University of Udine, and he designed and directed the EU-funded C19 space program to over 20,000 healthcare workers, where he also worked with WHO on the COVID-19 clinical guidelines, and not surprisingly, he was mentioned by JAMA in 2020 as one of the most influential doctors of the pandemic, after which he was nominated Knight of the Order of Merit of the Italian Republic. In 2021, he was awarded the Gastine Health Forum European Health Leadership Award. Dr. Ciccone's research focuses on improving outcomes in perioperative care, in data science, and the physiology of shock, as well as acute respiratory failure and sepsis. And today Maurizio will present Fluids from Theory to Bedside Practice in Hemodynamic Management of the Critically Ill. Please join me in welcoming Dr. Maurizio Ciccone. Good morning, everyone. And thank you, Greg. Thank you, Society of Critical Care Medicine, for this award and this recognition. I'm very, very honored to be here and to be recognized for my research and the research that I've done with my groups, first in London and now in Italy. And I will take this opportunity really to show you what work we have been doing for the last 20 years. First of all, this is my disclosure of conflicts of interest. I've done research and collaborations with a series of hemodynamic monitoring companies. Probably my biggest conflict of interest is that I do believe that when patients are very complicated, they don't improve quickly. Probably having extra information about their physiological status can inform us to make better decisions and hopefully to make them better. And it's a real honor to have a lecture named after the father of critical care medicine, the founder of the Society of Critical Care Medicine. And so going into the topic now, we have to start to talk about shock. And the first definition of shock by Max Harry Weil was this one, a reduction of effective blood flow and inadequate tissue perfusion with decreased oxygen, delivery of oxygen to the capillary exchange bed. We've done lots of research and we have a better understanding of this now. So the most recent definition of shock, which we did in 2014, is very similar to the previous one. But we introduced the concept also of the inadequate oxygen utilizations by the cells. Because what we know is that we have to transport oxygen to the tissue level and we have to try to preserve the efficiency and the energy production of our cells. When in some districts of our body during a state of shock, this does not occur, well in those districts, cells are fighting them for survival. So normally when I have an aerobic metabolism, I can produce 36 moles of ATP. If I'm switching that particular district of the body to an anaerobic metabolism, that the cell is still fighting for surviving, it needs to get energy. And actually the energy that it uses is the production of lactate. Indeed, even if we measure lactate in our blood gas to understand what's going on, lactate is not the problem. Lactate in this moment is the solution for survival of the cell. What happens if I'm producing more lactate? Well, I will also produce more carbon dioxide and there will be also more hydrogen ions. So what's going to happen to my pH? My pH is going to decrease. And the way my body compensates to that is going to be an increase in the respiratory rate. So this is why the respiratory rate is pretty much in any score that you will find in any paper of looking at clinical deteriorations of patients in our hospitals, it's probably the most sensitive and one of the earliest signs of deterioration. So an increased respiratory rate is very important. And this brings me to probably the most very important concept of today, which I want to share with you. To diagnose shock and to start to understand how well or unwell my patient is and how the clinical trajectory is going. I don't need fancy machinery, I just need my clinical examination. And within our body, our system, they always work as a several mechanisms, mainly the cardiovascular system, the respiratory system, and the renal system. And if I understand what's going on, I can look at clinical windows as an opportunity to try to understand if I spot that something is occurring in terms of oxygen transport at the cellular level. So if I was sitting in the back of the stage, immediately my baroreceptors, they shoot so that I could keep my blood pressure up. I'm now probably using a bit more oxygen than you that are still waking up maybe from the night. And so I will probably breathe a little bit faster than you. So these signs are always in front of our eyes. And of course, we can elaborate this even more. I didn't mention all the clinical windows here, but very importantly, to touch the patient, feel the temperature, measure a capillary fill time, see if the patient is getting confused. Too often with my trainees, I hear sometimes, this is a difficult patient, he's confused. Well, this patient probably was not difficult at all and was not confused. So that could be a very early sign of a problem that is occurring there. I'm often asked, what is the most important variable to focus on when we want to revert the status or we want to prevent shock occurring or we actually want to treat shock and to go back to a normal state? And I don't like to say which one is the right variables, especially when it comes to blood pressure and blood flow, I think they're equally important. I come from the generation where we were always told, and I will come back to this later, give a bit of fluids to see if the blood pressure goes up, then maybe we can see what's happening. And then we had moments where we were thinking, but the most important thing is cardiac output, so blood flow. So as long as the blood flow is good, maybe the blood pressure is not so important. I don't think we should reason that way. I think we have to try to have a holistic approach to all the clinical variables that we see at the bedside, and particularly when it comes to hemodynamic variables, I think cardiac output, blood flow, and blood pressure, they should be integrated. I always like to show this study, modify slightly the graphic from this paper, but I really like it. It's a paper from the group of Donino and Shapiro in 2007, so very old. But it was the first one to introduce a concept of synergy between variables. As you see here, the data is separated in strata for systolic blood pressure. So systolic blood pressure above the 90, between 70 and 90, or less than 70. And lactate, lactate levels that, at least in the initial phase of septic shock, we can consider as a proxy of inadequacy of perfusion, so blood flow. And as you see, we have patients with a lactate less than 2.5 and so on up to more than 4. I would like you to focus on the right side, where we have a systolic blood pressure of more than 90. Very often, we don't get cold to see these patients in the ward, because the blood pressure is still holding, and if there are no other signs, maybe we think that the patient is doing well. Well, this patient may still have hypoperfusion if he's developing sepsis, and so if I have a so-called normal blood pressure, but a lactate of more than 4, one in ten of these patients will die within hospital admission. That's why it's important to recognize these patients. You can look at the things in the other way. So even if the lactate is not so high, but if there is vasoplegia and my patient stays on a systolic blood pressure less than 70, then again there is an association with mortality of more than 10%. If I look at the highest column, the one in green in the back, that's basically the combination of the two variables they arranged together, high lactate, low blood pressure. This is not just a sum of the two, there is a synergy between these, and I think we need to learn really to get more science together to understand what goes on at the bedside. So ultimately, what I want to do is to improve perfusion pressure, to improve blood flow with my treatments, whether these are fluids, vasopressors, or inotropes, and what I think in the back of my mind is that I want to improve the microcirculation of my patients. These are some videos that I've taken with some camera. When I was in London with my research group, you can see that when the microcirculation is good, the capillary bed is quite dense, you can see the red blood cells flowing. You can imagine that the distance between the capillaries in the cell is very small, so I would have oxygen and nutrients going there, catabolites come back. In those areas that develop an heterogeneous blood flow, the microcirculation can start to become poor, and so I will have vasoconstriction. The density of the capillary bed becomes smaller, and I can really imagine there that it is more difficult to transport oxygen and to get rid of carbon dioxide metabolites. So if this is the problem, and by the way, I don't think you need to use these fancy devices to diagnose shock, and I was saying to you just a few minutes ago, this is a clinical skill. So if I have to add one exam, I would say a blood gas, venous or arterial, or maybe both, we can come back to that later, and look at the lactate, look at the pH, but really it's a clinical examination, it's the clinical understanding that helps me to make the diagnosis. When I've identified this, what can I try to do to improve the situation? I will focus on fluids in my talk today. I'm going to apologize, but for the time given me for the talk, I cannot go into all aspects of fluid management. I'm happy to have a conversation with you after my talk later on outside, or even to show you some references. So I will not go into the composition of fluids, crystallites, or colloids. I will focus more on the physiology of fluid administration and their effect on the cardiovascular system. And to do this, I am going to start from Gaiton. Everyone is familiar with the Frank Starling mechanism. We will come back to that, but if we really want to understand how the oxygen demand of our body is driving venous return and cardiac output, we have to go to the physiology described by Gaiton. In this talk, I try to summarize how we can describe this. If I imagine all my vessels, veins, and arteries together into a single elastic band compartment, then I can think of some blood into this compartment that doesn't generate any stress on the wall of the system. But then one part of the blood is generating stress on the system. This depends on my bulimia, but also on how vasoconstricted or vasodilated my patient is. And we will come back to that later. Gaiton described the pressure generated by the stress volume as the mean systemic filling pressure. And it is a pressure that is there in the system in a moment of no flow. So if you think about it, it's very difficult to really measure this pressure because you should have a patient with no blood flow. We don't want that. But over the years, we have developed proxies, either through ventilation techniques with fast tourniquets and so on, to estimate the changes in mean systemic filling pressure. These are some of the things that we did also with my research group. So everyone understands cardiac output is liter per minute. Everyone understands that is the blood that leaves my heart and goes into the systemic circulation, right? But in a closed system, everything that goes into one direction and comes back needs to be exactly the same quantity. So when I think about venous return, I'm just describing the cardiac output from the venous side of the circulation. If my venous return is four liters per minute, my cardiac output will be four liters per minute. But what happened is that it is the oxygen demand of the tissues that is driving this. It's not vice versa. I like to jog sometimes. When I go for a jog, it's not that my heart starts to pump before I start to do exercise. I start to do a bit of exercise, and my legs, my muscles are consuming oxygen. There is a release of carbon dioxide, there is a vasodilation, there is an increase in the venous return, and a good heart pumps all this venous return forward. So if we understand this, we understand the very important physiological concept about rate atrial pressure, for instance. If this is my mean systemic filling pressure, there is another pressure here. In the small quadrant goal, I have the heart and I have the backward pressure, which is the right atrial pressure or the CVP. Now if we understand this, there needs to be a gradient, right? For any flow in any hydraulic system at home and in the human body, wherever you want, in order to have flow of fluids, I need to have a gradient. So from this, I understand that for any given value of mean systemic filling pressure, the CVP can only be lower than that. Do we all agree? So if it is lower than that, this is a very important physiological concept. A good heart pumps the venous return forward, keeping the CVP as low as possible. My goal is not to increase the CVP. My goal is to increase the mean systemic filling pressure. When the CVP rises, I'm starting to pay the price of my interventions, and it should never be my goal to increase the CVP. For two equal cardiovascular states where the only difference is the number that you read on the CVP, it's a better situation when there is a low CVP compared to a high CVP. So why for many years we had protocols saying give fluids until the CVP goes to 8 or 10? It's not that we preceded us, didn't think about this thing. We didn't have technology to look at what really matters, which is the change in blood flow. We didn't have tools to estimate how the mean systemic filling pressure was changing. So what really matters is that when I give a bolus of fluid, I give enough fluids to stress the system, to increase the stress volume, to increase the mean systemic filling pressure, and hopefully to see if there is an increase in grid. Does this work? We demonstrated in this paper 10 years ago. When I give a bolus of fluids, patients that increase their cardiac output are patients in which the gradient between the mean systemic filling pressure and the CVP has increased. If I give enough fluids to stress the system, but there is no change in cardiac output, it means that the venous return has not changed, and so the mean systemic filling pressure is increasing, but the CVP increases by the same amount, and therefore there is no increase in gradient. Let's bring back these two theories, Guy Tong and Frank Starling together. So we just said that the venous return and cardiac output at any given time are exactly the same thing. So the two lines that can only cross into one point in the specific moment. If I give fluids in this situation, I am working on the ascending part of the Frank Starling curve, right? So if I give fluids enough to stress, to increase the stress volume, I will shift my venous return curves to the right and to the top, and now it will intersect into a different point. This is the principle of fluid responsiveness. This patient is responding to fluids with an increase in stroke volume and cardiac output, and if you see what happens to the CVP, to the radiatory pressure, that there is either no change or is a minimal change. This is the optimal situation. If I've decided that there is a problem with perfusion, which I want to try and correct with fluids, this is a good response. Then I have to go back and examine my patient and decide whether I would still give fluids or maybe stop there. What if I'm working on the plateau part of the Frank Starling curve? Well, in this case, I start from the same venous return curve, so I could have the same mean systemic filling pressure there, but if I am giving enough fluids to stress the system, I still move my venous return curve to the right, but this time the change in stroke volume, the change in cardiac output is minimal, or maybe there will be no change. But if I look at what happens to the CVP there, a rise in the CVP with no changes in stroke volume and cardiac output suggests that that's a situation where fluids are not indicated I should stop. I'm not saying that you need to use a cardiac output monitor necessary for this or a CVP. If I have a central line, I do like to measure the CVP and use it to inform me about these changes when I do that, but if you're familiar with ACOE and you look at markers of congestion and you see how these things are changing, the principle applies in the same way. When I started to approach this physiological research question, we also wanted to see what was the variability of one of the most common practices in intensive care. This is the story behind FENICE. FENICE was the Fluid Challenges in Intensive Care Study, which I conducted with Daniel DeBacker and several investigators and also many investigators around the world, and we were asking, how do you give a fluid challenge in these patients, and can you tell us which variables you measured and what are you using to guide the treatment? Surprisingly, in 2013, still the majority of clinicians were using CVP, despite all the limitations that we say, to guide fluid management. But the most interesting thing, and that's what prompted me and my research group to start to do research in this sense, was actually the huge variability in what we call a fluid challenge. We realized that what I call a fluid challenge in my unit, my hospital, is very different maybe from the unit in the same hospital or in other areas. We found an average of 500 mLs, but you see with huge variability, huge variability on the rate and the duration and so on. So how can we then compare studies or make recommendations if for the same term, we actually mean very different things? So in order to study this, we thought, why don't we go back to basics, and we have to remember that fluids are drugs, and when we are interested in increasing cardiac output, why don't we look at that effect of this drug, of fluids, on the system, and then we see how the body is making an effect of the distribution of this fluid. So we started to do pharmacokinetics and pharmacodynamic studies of a fluid challenge. This was the first paper that we published. We gave about 250 mLs of fluids in a very controlled environment. These were post-cardiac ICU patients that were coming to the units of very standardized, and this was important to control pretty much all variables. And we looked at the area under the curve for the effect of cardiac output over time. So we looked at the total area under the curve with the separated patients between responders and non-responders. We looked at the maximum change after the end of the fluids. We looked at when it occurred, this maximum change. So was it one minute, two minutes, five minutes after I stopped giving fluids? And then after the maximum change, we followed these variables for 10, 20 minutes, and we saw what happened. So of course, responders and non-responders are the different area under the curve. That's how we were separating the two groups. Look at this. The maximum, the time of the maximum change occurs one minute after the end of the fluid challenge. This is something I keep repeating to my trainees and to my students when we do fluid challenges at the bedside for the first times. We are doing a physiological test here. We're using a very small bowl of fluid, like a little glass of water, to test our hypothesis. I need to be in an almost quasi-experimental condition. Nothing I should change. I should not move the position of the bed. I should not aspirate my patient. Really, I should keep everything as it is, and I should not go away. It's a short period of time. I want to follow these changes. Because if I don't follow these changes that occur almost immediately, I may miss these changes. Because indeed, what we found after 10 minutes, both in responders and non-responders, the effect got dissipated. And that prompted a lot of more research questions, because why is the effect dissipated after 10 minutes? Well, we have different questions that we try to answer. There are different answers. One possible answer is that these patients were well-resuscitated in the operating room. When they come, we are just testing the Frank Starling mechanism. But if the body does not need an increase in oxygen delivery, maybe it redistributes very quickly this volume into the unstressed volume. Another hypothesis is capillary leak, even though I don't think it can apply to this group of patients. But certainly, it can apply to septic patients, and those patients which I'm sure you all relate to, they are so difficult to treat sometimes, very hypotensive. They seem to respond. After five minutes, they go back to normal. And then we start to repeat this. And when do you stop? When do you start to think that maybe you will pay a price for a too positive fluid balance? So lots of questions there. But one very important question was, if there are so many unknowns when we give fluids, when I give a fluid challenge and I'm doing a test, I really want to try to find ways to standardize this so that I can trust both a positive response but also a negative response. Because if I don't do this, I run the risk to feed my internal bias. If I think that this patient is to stay dry, I will not give fluids. If I think that this patient is fluids, I will say, maybe I didn't give enough fluids with this challenge, let's give another one. And we keep giving and giving and giving and then patients become too positive. So what did we do? We randomized patients to receive in a fixed period of time, so we fixed the timing of fluid administration to 10 minutes, but we randomized them to receive other 1, 2, 3, or 4 milliliters per kilogram. And we wanted to see which dose per kilo was the one that would always increase the mean systemic filling pressure. And we found that basically moving from 3 to 4, probably at 3 you can already argue there was a signal there, but it is at 4 milliliters per kilogram the 95% of the times the mean systemic filling pressure increases. And so I can trust that I've given enough fluids, and it's not a lot, 4 milliliters per kilogram is about 250-300 mL for most patients. How was then this reflected to the percentage of responders? I was expecting, of course, in the lower volume, in the lower dose, to have a few less responders, but what I was surprised to find, the way we analyzed the data, there was almost a linear effect. So there is a dose response effect on the fluids that we're giving. And so it means that if I'm giving only 1 milliliter per kilogram in 10 minutes, probably I will miss a lot of patients that would benefit from fluids according to my initial strategy. We stopped at 4. I don't know how it would have been if we randomized patients to 6 or to 7. I think this is important. I have the feeling that we, for many years, oversimplified the concept of fluid responsiveness as a binary yes or no. I think this is how I interpret my dilemmas with my team at the bedside. If I have a patient that comes to the emergency department profoundly hypertensive, high-lactate, I have to salvage this patient now, probably I'm happy to use a bit more fluids to get all of that physiological reserve very quickly. Maybe day 2, day 3, when we're giving enough fluids, we know that there will be some redistribution there, which is normal because any drug redistributes. Then probably I want to be a bit more conservative with my strategy and I will decide to give fluids only if really strictly necessary. So we talked about a different response. We talked about which dose to give to give a fluid challenge. What about the rate of administration? Does it have an impact on my definition of fluid responsiveness? We tested this in the operating room and we randomized patients to receive this time. We thought, okay, we have an idea of the 4 milliliters per kilogram. It may be a good way to standardize the dose. What about the timing? Some people argue you can give it as fast as possible. Why don't you just give it in a minute? I don't think we know enough to say that as fast as possible is good. There are some theories, for instance, about glycocalyx and so on when you give too fast the bowls of fluid. Maybe that can impair and maybe there will be a price to pay later. So we thought, why don't you just go with some times that are normally found on paper. So we randomized patients between 10 to 20 minutes to receive a fluid challenge. And what you find is that in the 20 minutes group, basically we labeled as responders 30% of patients, 29 to be precise, while in the 10 minutes we label 51% of the patients. So you see, same amount of drugs, same amount of fluids you're giving, just by changing the time we would correctly or wrongly label these patients between responders and non-responders. I think overall clinical practice is moving towards a faster time for fluid administration. We did this meta-analysis recently, which we published in Critical Care, in which we looked at what is the time for a fluid challenge in a decade between 2000 to 2010 and after 2010 to 2020. And as you see here, there is basically a decrease in time over time. So we seem to be giving fluids for with smaller boluses and a shorter duration. I think it's a good thing to move in that direction. But we don't have overall consensus for everything. As I said, I will not speak about balanced crystalloids today, but in one of the arms of this crossover trial of the Brazilians of the BASICS randomized control trial, they also tested the hypothesis of different fluid challenges that they called. So in between 333 mLs per hour to 999 mLs per hour. To me, that's not a fluid challenge. I don't see the rationale why this would work physiologically. It's informing as probably you can use infusion when you want to give large volume of fluids as you wish, but that to me is not a fluid challenge. A fluid challenge, as I said, is a way to both administer but also to do a small bolus, to use a small bolus to actually to do a test to inform then my, to check my hypothesis and then to inform about further fluid administration. Of course, I'm not showing you today all the literature about positive fluid balance and outcomes. I think everyone remembers that and I do believe that it's wrong to give fluids when they're not necessarily indicated. So we also know that there have been a lot of research in trying to predict which patient would respond to fluids before we give them fluids. So all this chapter about heart-lung interaction and pulse pressure variation and stroke volume variation and so on. We also did some studies. If you're in a very controlled environment like the operating room or a patient you've just intubated in intensive care, 8 mLs per kilogram, no arrhythmias, fully adapted to the ventilator, not to tachycardic, not to bradycardic, they perform very well. They perform with areas under the curve approaching 1. So you can be pretty sure whether the patient will respond or not to fluids even before giving fluids. If you cannot use these due to some limitations because you don't have all these criteria that are matched, maybe we can do passively grazing and test the system by changing the position of the bed. If I have a cardiac output monitor connected in real time, an increase in stroke volume during that maneuver will predict that when I give fluids for real, that patient will increase cardiac output. All very interesting, but I work in very complex environment, both in anesthesia and intensive care. I will show you that, for instance, even for pulse pressure variation and stroke volume variation to have the ideal conditions is not always the case. I work with trauma patients a lot. I work with patients that have an open abdomen. I don't like to keep my patients sedated anymore as 20 years ago. If you do a passive leg raising to a patient that is still in a bit of pain, they may look at you and say, I don't like this test. So, of course, of course we have this test that when they are applicable, we should apply them. But I think also we have to be pragmatic and realize that it's not always the case. And in a very recent meta-analysis, we sparked a lot of debate over the internet, even with some letters that we received. We meta-analyzed all the papers reporting the positive predictive value and the error under the curve for these pulse pressure variation and stroke volume variation in the operating room. And if we plot all data together, the error under the curve was around 0.79. So not 0.9, not 1, but actually a bit lower. This to me is not a problem. This is how I practice. I try to bring more information. I don't think I can rely just on one variable. 0.8 is much better than tossing a coin. Tossing a coin would be 0.5. But I don't think you can trust this all the time. We received some criticisms for this from some groups. They were telling us we should include only studies that have the perfect clinical situation. But I'm not interested in that. I am interested to find tests that matter for patients and not patients that matter to validate tests. So I use this in my practice, but it's not the only thing I'm using. I'm trying to put this together with the clinical judgment and we're trying to have this holistic approach. And I'm very confident and hopeful that actually artificial intelligence in the future will help us not to make everything for us. I would find the job very boring. But to visualize this association and to identify physiological phenotypes in the response and maybe before even we test our hemodynamic therapies. Artificial intelligence is becoming more and more a reality. I have to say I'm showing you a paper which is quite old now, 2018, but it was the first paper that tested in two independent cohorts the idea of an artificial intelligence clinician that would suggest which dose of fluids to give and which dose of pressures to give in patients with sepsis. You may have seen this paper. Confirmation on the fact that if I give too much fluid there was worse outcome there. Mind you that every paper that we find on artificial intelligence is by definition observational data. We have not been able to randomize yet into studies with artificial intelligence, so I think it's important. I will come back to that at the end of my talk. Now maybe we can use big data to inform about big trials. A very interesting signal I found, maybe the most interesting signal for me of this paper, this U-shaped relationship for what concerns vasopressors. And this to me is in line with the practice that has been changing. So giving too little vasopressors and maybe too late is not good. Giving too much probably is also not good. And why is that? Well, if we go back to physiology, when sepsis is in the phase already with vasodilation, what's going to happen to my stress and unstressed volume? Blood is going to redistribute, so I'm going to lose some of the stress volume into unstressed volume. Of course I can give fluids. And I remember when I started this job, I was always told, so give fluids first. We can argue about the 20-30 mers per kilogram. I don't care to go into the debate now, but give fluids first, and then if the patient does not respond, then give a vasopressor. But the clock and the time is ticking, and sometimes you may lose minutes, sometimes you lose half an hour, sometimes you lose a couple of hours for that. And physiologically, there are some situations in which it's just impossible that the bolus of fluid will increase my blood pressure. If I have a patient that has a heart rate of 140, with a diastolic blood pressure of 35, and the vascular tone does not change, I would have to multiply my cardiac output by so many times with the bolus of fluid to increase the blood pressure. So it's just impossible. So that's why I think it makes a lot of sense now, and I'm very pleased to see Laura Evans here. We also recommend with the surviving sepsis campaign, not to delay the start of a vasopressor. This to me is one of the most important pragmatic recommendations we did in recent years. It makes sense also physiologically. We are recruiting some stress volume from the unstressed volume, and I think this is the reason why when you give norepinephrine, cardiac output goes up. Some people believe that it's the beta effect. There is a bit of a beta effect from the norepinephrine, but actually the biggest effect, I think, at the beginning is the fact that you are in a patient that responds to a vasopressor of course. You are actually recruiting from the unstressed volume to the stressed volume, and that, for what we were saying before, will increase the mean systemic filling pressure, increase the gradient for the venous return. So a very powerful intervention, very useful in very unstable patients. Imagine when you're about to intubate a patient, the faster intervention to optimize and to recruit volume is actually a vasopressor. Fluids will never be as fast to give you a bit of safety for that situation. But again, when we've given fluids and we tested our hypotheses, what do I use to understand if what I'm doing is working? Well, I go back to my clinical examination. My clinical examination, and always, if I can, also with the blood gas. I prefer to use a venous blood gas rather than an arterial blood gas, but if you can do both together, even better. And if I have to use a device in patients that are not improving quickly, it's probably echocardiography. I remember when I started to, as a trainee, we had our first echo machines. We were very excited. I think it was about $80,000, $100,000. It was as big as half of an ICU bed. You have to find space to move it there, and how times are changing so fast now. With technology that is cheaper, maybe it's not as good as the most sophisticated one. Most of the time, we don't need the most sophisticated ones. I have no interest in any of these companies. I'm just showing you some examples here. We also know that echocardiography is a skill that you need to learn. You need to get used to the images, but you also need to get used to take good images, and so we always say that it is operator independent. But interesting, there are now devices that are coming out that are actually identifying how you're putting the probe on the patient, and they can help you to change and tilt in the probe and actually get the right image. So I think we will see more and more of echocardiography in the future. But what do I do with all this information? We will soon have devices that maybe we can stick to our patients. Think about the pre-op clinic and see how the heart of the patient behaves when he does some exercise a few weeks before surgery, or how we can follow up patients after discharging them from the intensive care. This almost looked like science fiction 10 years ago. It will be years before these things become mainstream, but the technology is there and will come, and I think we have to embrace these changes. Not without the understanding of physiology. I don't care if my patient with an echo image has ventricular dysfunction, if all the other clinical parameters and my clinical examination is now better. But if the patient is getting worse clinically, then to have that information, of course, is useful. And as I said, if I have to complete my clinical examination with something, I would say, take a blood gas. I know there is a lot of debate about central venous saturation. I'll come back to this in a second. But this is the complex physiology around that number. It's not, it's a number that can depend on so many variables. And I think remembering this physiology is important when we have a patient at the bedside. If I find a low central venous saturation in a sick patient, I'm actually happy. It means that I have not yet found the balance between my transport and my demands, and maybe I have some room to improvement there. And indeed, if you look even at the baseline characteristics, for instance, of the recent multicentric randomized control trials, they found no differences whatsoever using SVO2 against just the clinical management. Well, I think there was a difference with reverse. If you look at the mean SVO2 in their eyes was 73 percent, in process 71 percent, in promise 70 percent. So physiologically, this patient, the average of the population, had achieved already the goal even before I was starting my protocol. Reverse at the mean SVO2, 49 percent. I don't think the reverse patients are the ones that we see nowadays, but this is also thanks to reverse. I think reverse taught us about being very fast and aggressive with the first hours of resuscitation. So this is almost like reverse protocol against himself for the success in understanding the concept of early resuscitation. So it is true, the majority of patients will receive fluids, and I don't think you need a monitor there. If a patient is in septic shock, just give the fluids. We're in the salvage phase. But in those patients that don't improve quickly, I do like to understand, and I'd like to go back to a bit more physiology. And I participate also in randomized control trials. Even if I love physiology, it's almost like gangs sometimes. Who likes trials? Who likes physiology? I think it makes no sense. I think we have to learn from both types of studies. When you do a study like this, like the one that we published with the group of Anders Berner, for instance, yes, we found that there was no difference. But sometimes you do find in synchrotron analysis some signals there. Does it mean that there is no difference in a large trial? I just go back to my team and say, do whatever you want with fluids because it makes no difference. No, of course not. This is just something that is informing me more. And probably why is it that when we do a small study, we find a signal. And when we do large trials, nine out of ten, pretty much everywhere in medicine, we lose this signal. When we lose this signal, because when we have to power our studies for very large populations, we go from an initial idea, but then we're going to compromise it for the sample size. And we enter into a very heterogeneous populations. I think we need to learn from other fields of medicine. Of course, they had it a bit easier. Patients with cancer, they may come, they have a bit of time maybe to do a genetic testing to understand the phenotype or genotype, and then you can do trials on a new drug just on 15 of 20 patients. But imagine now, and this again, this is possible, that we have all these large databases, our electronic records. If we federate this together in the same network that we use for doing trials during the pandemic, we can have massive registries in real time, identify subphenotypes where maybe that intervention I can really randomize. And I think it's important to randomize in that homogeneous group of patients, because it's still important to look if my association really holds when I test it against causality. And randomization is still the only way that allows you to compensate even for factors that you don't know yet. There are some approaches in this. I love this work by the group of Derek Angus and Chris Seymour. You may have seen it already. They simulated changing the phenotypes of populations on some studies, and by changing the proportion of different phenotypes, they found that maybe when one proportion goes higher, they would have, you would have seen benefit in the trial. When another proportion of patients becomes prevalent, maybe you would see harm. And I suspect this is really how we work more and more. With my group, we're also looking at phenotypes. There are phenotypes also in the response to fluids. So I have an interest in improving perfusion, but how do this patient respond? Can I predict the type of response? There are patients in which, as we were saying before, cardiac output will increase and blood pressure will increase. There are patients, and these are the easy ones. In those ones, you don't even need a cardiac output monitor. If the blood pressure goes up, obviously the cardiac output has gone up. But there are patients in which nothing happens, but if I don't measure, I don't know, because there is a third group in which the blood pressure does not go up, but the blood flow, the cardiac output goes up. And if I just look at blood pressure, I may be blind to these changes. My last few slides, whatever we decide to use to monitor these patients, our patients, it's important always to think about the initial clinical problem. When we've done the debunker, we looked at the first analysis of Fenice, we looked at one part of the research question, which was, in patients that received the first fluid challenge, did you give a second fluid challenge? And we found that in about 50% of patients, this was the true. And we find that in 48% of patients with an initial positive response, clinicians didn't give a second fluid challenge. And we started to look at the data and say, well, this is great. Actually, clinicians are not just giving fluids only because the patient is responsive. Maybe they stop because the clinical situation got better. But then we ask about an uncertain response, and it was still about 50%. And what is striking is that with a negative response, that clinicians defined as negative, they still gave fluids. And this, to me, goes back to what we were saying before, to our internal bias. And we try to find confirmation in what we already have. That's why it's so important, I think, that you try to find a way to standardize your hemodynamic tests at the bedside with your teams, to really try to find a way to trust the test when it's positive and when it's negative. In a summary, how did we recommend about monitor? You don't need monitor for all patients, I think. You do need, we do need the monitors, echocardiography, continuous cardiac output monitor, if you can, in patients that do not improve quickly to the initial therapy. Just to let you know, we've been working for the last two years on an update on recommendation for shock and monitoring. I'm sure you know something that was already presented in Milan this year, but I don't have the final version. This group is led by Xavier Monet, Antonio Messina, Michel Chou, Ian Bakker, and many others, and will be released probably very soon in the next few months. And in my last slide, just a reflection. We live in a very privileged part of the world, and we have access to technology. But actually, we also have problems here, and we need to think how we bring these advances, maybe also to less privileged parts of the world. I am confident that technology, if we embrace the revolution and we guide it, actually can help in this. But we have to realize that for any technology that we bring in our intensive care unit, we really need to find a way that our jobs become smarter, and our job is allowing us maybe to spend more time with patients and families, and to do smart diagnosis rather than focusing too much on the tech. We are living in a time of massive crisis. Even before COVID, we knew that the age parameter was changing massively in our populations, which means that we are seeing older and sicker patients, and the proportion of healthcare workers will be lower and lower. So I don't know how the future is going to look, but I am sure that in 10-20 years, our job will be very different, and I am sure that technology is the only way for us to work, probably in a smarter way. But again, we have to embrace this and to guide this revolution. This is my last slide. I didn't have the time to go into the nuances of fluid administration today, but if you want to read about this concept that I described and find even more about clinical examination type of fluids, this is a review which is free to download from Intensive Care Medicine Experimental. And if I had to give some very practical take-home messages, remember the fluids are drugs. Give only if clinically indicated, so always a clinical question first. If you can predict the response better when you decide to give fluids, use a fluid challenge. Check the initial and the late response. Consider vasopressors and fluids together, and always remember that the clinical judgment leads the way. Thank you very much.

hemodynamic management

critically ill patients

shock

fluid management

fluid responsiveness

FENICE study

artificial intelligence

patient-centric healthcare