TIRBO #28: How I set PEEP

Hey everyone, I'm Brandon Odo and I'm Brian Bulling and this is Critical Care Scenarios, the podcast where we use clinical cases, narrative storytelling and expert guests to impact how critical care is practiced in the real world. Hey everyone, welcome back. It is Brandon Odo with another Turbo. I want to spend today talking about peep. We recently did an episode on what I called radio peep discordance when the peep that you're giving a patient seems to exceed what makes sense given their imaging findings. And I thought this would be a good time to just get into a look at how in general I do set peep. You know, what is my method for figuring out the patient's best, most appropriate peep? And this is an area that I find interesting in critical care because there are a lot of ways to do it. It's certainly not clear the best way if there is a best way. You know, I don't think there probably is a best way to do most things we do. There's a variety of ways that may work for different people or for a particular patient. But my kind of core philosophy here is that for any particular patient at a given moment, there is an ideal peep. There is an amount of peep which will hold open the majority of their LVL without over distending the remainder. And best here is admittedly always a compromise. And I say that because even though we often will draw this out as a model where there, let's say just two LVL, one that's shunted and one that's open, that's not how it is, of course, right? You have many thousands of these and they're on a spectrum of their degree of adelectasis and shunt. And so when you apply a single pressure to the airway, you can't target any specific LVLAs. For a particular LVLAs there probably is a correct pressure, which is just enough to keep it recruited without over distending that one or any other. But because you're in fact averaging things, you're always striking a bit of a compromise. That being said, there is some compromise. There is some average pressure which most approximates what's correct for the patient. Now that will change with time, potentially change rapidly, but certainly change over let's say days as their disease gets better or worse or they plug or who knows what else. But there is a pressure. And I emphasize that core belief because that is not how everyone approaches this, whether or not they believe that's true. Sometimes they treat peep as some kind of a marker that's more tied to degree of illness or to the time course and the disease. And that just changes with that kind of like the hands on a clock. As opposed to being a physiologic variable. So let me describe some of the ways you can determine peep that I don't use. I'll kind of spell them out to give you some idea for what they're about. And then I'll explain what I actually do. So the simplest way you can do this, and I think you can make a good argument for this and it probably is the most appropriate way as a starting point. If you're writing a protocol for non-expert users, just kind of a way to get into the right ballpark is to use a peep table. So the most well-known one was from the original ARDSNET trials and they tried some different ones. There's a high and low peep version, but essentially correlating peep and FiO2, meaning as you turn up the FiO2, you also should turn up the peep until you get to a point where you kind of quote maxed out on both. The idea here is just that the worse their hypoxic disease, you know, the more of both they'll need, which is more or less true. We talked about radio peep discordant circuit also probably be a peep FiO2 discordant, which should at least give you some pause. If you walk in a room and find a patient on, let's say, 100% FiO2 on a very low peep of like five or vice versa, people like 20 and a low FiO2. Now those might be the right thing for the patient, but they usually at least wonder because they do tend to go together. Now the problem here using this as your approach is that, again, while these are roughly correlate, they don't correlate very well. This includes the basic fact that in many ways the purpose of peep is to correct the underlying shunt physiology, so they don't need so much FiO2. FiO2 is not a good treatment for shunt physiology because as we all learn, when we learn about this physiology, first of all, the worst is shunt, the less responsive they are to it. You can't overcome shunting of one lung unit by giving more oxygen to the other because you're always going to end up mixing with that deoxygenated blood. The only reason it works at all is because it's not just too long units, it is again a spectrum and you may be able to overcome some kind of middle ground, the QMIS matching with more FiO2, but not a good tool. And of course, we know that it can have toxicities. So in a way, you could argue that having a relatively high peep and a low FiO2 is actually the goal. In any case, this is a blunt tool and I think ideally we would move past it. A somewhat more physiologic tool might be something like the stress index. The stress index is a concept that if you look at the pressure curve, the scalar on the ventilator waveforms that traces that pressure over time, it should have a certain shape to it during inspiration as you flow pressure into the lungs, pressure increases as you inflate the lung. It should have a certain shape, particularly when you are in the right modes. This is usually looked at in a continuous flow situation, like volume control without a decelerating ramp, like a square ramp, not so much in something like pressure control. But the shape that you should have there should be a kind of shark fin or a humped curve, which is either relatively straight on the upslope, like a straight rooftop. Or you may find that it is either domed, kind of convex, or scooped out in concave. And the implication here is that as you put volume into the lung, the compliance of it is changing during the breath. And if you find that it is swept up, it is kind of scooped out, that implies that towards the end of the inspiration, the compliance is getting worse. The lung is getting tighter, which suggests that you are over distending it. Whereas if you find that it is domed, kind of has a hump back to it, it's the opposite. You are actually recruiting lung during the breath and the compliance is improving towards the latter part. Now stress index people would say that that first one, where it's scooped out, implies that the tidal volume is too large, and therefore you're maybe causing injury. Whereas the kind of domed version might mean that you should increase the peak, because we don't really want to recruit lung during the breath. I mean, to some extent, you always will, but we want to have optimal recruitment at the start of the breath, even at end expiration. And then when we actually take the breath, we take it on the relatively flat part of the compliance curve, where we're putting air into an already inflated lung, but neither recruiting it nor over distending it at the end. As an interesting concept, I find it hard to apply in reality for a variety of reasons ranging from vent modes that have varying flow, which are common. And I think often good for patients, because they're more comfortable, to having distinct journeys that make it hard to interpret these things. But potentially something of interest, and certainly a good illustration of physiology. A related but different concept is PV loops. A pressure volume loop. So if instead of the scalar over time, you graph out pressure, not against time, but against volume, and put it on a loop. Loops are more often used by pulmonologists in PFTs and things, but pretty much all modern ventilators can give you loops if you just elect to view that on the monitor. And principle, a pressure volume loop would trace out a sort of sigmoid curve during inspiration, and then during expiration, it would kind of come back and recapitulate it. And sigmoid meaning S-shaped, and the two inflections of that S are the first one, as it steepens, is where you've finished recruiting the lung. The first part is very flat, because the more you get to put in a lot more pressure to get a little bit of volume out, once you've recruited a lot along, you get a lot more volume out of it. And then the second inflection, as it flattens back out at the top, is where you've maximally inflated, and now you're over distending, kind of like we talked about with stress indexes. So in theory, you can inspect one of these loops during the breath and say, I want to set my peep so that I avoid the lower inflection point, because I don't want to de-recruit at the end of my breath. I want to have my entire breath, again, kind of in that flat part. Again, avoiding the top part is more a matter of limiting your tidal volumes. But I want this whole thing to be pretty straight. Again, easier said than done. One of the challenges here is that the appliances you're interested in are not the resistance to flow. And a lot of the pressure you're seeing is just due to the kind of conducting airways, the vent circuit, the tube, and all your non-exchanging parts of your lung. You're interested in the alveolar pressures. So really the pressure you want is the plateau pressure, not the peak pressure, which is what is being dynamically measured over time. So how do you get that? Well, instead of just watching this curve during the breath, you could measure a series of plateau pressures, so static pressures during a breath hold like we typically do, and then plot those points out over time. That'll give you a series of plateau pressures. And then if you correlate them with a volume at that time, then you would get a curve. But then you have to take a whole bunch of breaths and graph it, which is a little bit infeasible. You could give a very slow breath, and it should be a breath again with constant flow. If the flow is changing over the breath, this is all out. It won't have any accuracy. But you can give a constant flow breath, like a square waveform, which is slow enough that the amount of pressure created from the resistance to air flow is just very, very low. So just a very slow, steady breath. And that may allow you to trace out this with a limited confounding from resistive pressures. Now, can the patient tolerate you giving them a breath over like, who knows how many seconds? I don't know, maybe. A lot of the research on this was done with some of these techniques, or more realistically, even more esoteric ones using super syringes and things like that. So again, something that I find is a wonderful demonstration of physiology, but a little hard to apply at the bedside for a lot of patients. And again, never mind patient interactions. You might need to paralyze them and so on. The last technique you might consider would be using something like an esophageal balloon to measure esophageal pressures, which these allow you to get a sense for what the intra-plural pressure is. As opposed to just the intra-alveolar pressure you're getting from your vent and your plateaus and things. And you could use this in a couple of ways. One would be to better approximate their transpulmonary pressure as in a patient who had a reason for them to be atypical, such as a very heavy obese patient, the very heavy chest wall. You know, in these patients, we may say that our typical goals for like a plateau pressure to limit lung injury are not correct because really, I don't want to dwell on this because it's not really the topic. But let's say you want to keep their plateau under 30, typical goal. What you're really saying is you want to limit their trans-alveolar or transpulmonary pressure to, you know, let's say something like 30. The reason being that's the actual distending kind of injurious pressure being applied across the wall of the alveolus. That's the kind of stretching, perforating, distending pressure. It's not actually just the pressure inside the alveolus, and you can see this if you imagine matching the pressure inside with pressure outside. So let's say the pressure in the alveolus was 40, but the pressure outside of it on the other side of that wall was also 40. You can see how there is kind of no gradient there. I mean, there's pressure kind of squishing or sandwiching the wall, but that's not really injurious. And in fact, you see this situation where there's a very high pressure in the lungs, but also a very high pressure in the plural space outside of the lung. Anyone who is forcibly bearing down, a very strong valsalva, let's say, you know, someone playing the trumpet or a weight lifter doing something like a squat, or they often give a very vigorous valsalva to tighten their core and strengthen to bear the weight. I mean, the pressure in there may be extraordinary, but none of these people are getting pneumothoruses or, you know, pneuma-media stynomes because there's no increase in their trans pulmonary pressure. So in a patient who has a very heavy chest, you may see high airway pressures, but maybe in reality, there's not an increase in their trans pulmonary pressure. And therefore, not an increase in lung injury. So this would be an argument to say we can give this patient a higher tidal volume or driving pressure, and it would still be safer their lungs. But it's hard to do that accurately without knowing what their plural pressures are, and therefore maybe placing a esophageal balloon would let you titrate that. This is, again, more a matter of setting things like tidal volumes. What about peep, though? You could use a balloon to set that as well, targeting something like a trans pulmonary pressure of around zero, because your goal is not to have any pressure in the lung or any, you know, relative trans pulmonary pressure. Your goal is to have enough peep in there that at the end of the breath, the pressure is not negative. You don't have a collapsing pressure from the outside. Again, especially in maybe a patient who was obese and there was a great deal of interpolar pressure. You just want to match it. You want to have enough pressure in the LVLI that they stay recruited, but you don't want to be extremely high as well as not being low. So that would be one approach to take here, measuring those pressures and trying to kind of match them with your peep. A lot of places don't have esophageal menometry. I think it's a cool idea. I've never worked anywhere that really had it readily available, but I think it could have a role, especially in, again, certain patient populations. What do I actually do? Okay. So I'll start by considering the patient to kind of form a pretest probability for what sort of peep I think they're going to end up needing. And this could be based on their imaging. So what is their lung disease look like? How much of it is there? And sort of how dense does it seem to be, meaning is this some relatively mild, maybe widespread, but mostly interstitial disease that seems pretty readily recruitable? Like I can open it up without too much pressure, or is it very, very densely consolidated? There's large areas of the lung that are completely without any air. And you never really know this, but based on their disease, they have pneumonia or air or yes or something. Do I have a sense for how sort of thick this stuff is? Or is some of their disease look completely unrecruitable? Maybe it's due to a mass effect, tumor obstructing an airway, which I don't expect to be able to open with any amount of pressure. So get some sense for that. And also how heterogeneous is it? Because if you have some dense disease and some completely normal lung, that's a situation where I might be more sparing because maybe I should kind of give up on those areas of consolidation, understanding that any pressure I apply is probably just going to go into the good lung. The more difference between the diseased lung and the healthy lung, the harder it's going to be to selectively recruit the bad lung because of course it's not selective. I'm applying just one pressure to the whole lung. And also perhaps considering their body habitus. You know, again, a bigger patient, maybe someone who is going to need more peep, even with a relatively healthy lung, just to fight against the weight of their chest and abdomen. So if this is a patient we're freshly intubating, we get them intubated, they're going to start on 100% FIO2, sure. And then we're going to set the peep kind of to whatever. You know, I'll base this somewhat on those factors. But I don't feel that strongly about the initial settings because it's just about getting them stable and situated on the vent. Once we've reached a point where we can start kind of experimenting, and this may be once they've kind of otherwise settled out, maybe they're a little sedated, maybe they have some lines or something, maybe they've had a blood gas, although I don't feel too strongly about that because mostly what we're interested in here is in oxygenation. And I'm typically happy to follow oxygenation using their non-invasive pulse ox symmetry. I am not someone who turns to like PAO2s on a blood gas for this. I think that is more specificity than you need here. And enough that it could even mislead you. You know, you go up on your PAO2 by four points and you think you did something good. I mean, that could easily be within the variation over several minutes of the test itself. Plus it's slower. I'll usually use SATs. So wherever we set the peak, let's say we started at a peep of five just to start out. It's still on 100%. I will only, I just, I want the FiO2 to be somewhere where we have a SAT that is safe, so probably over 90% or so, but less than 100 because I want to be able to see if it goes up. And if we're already at 100, then we can't tell and then you do need a blood gas. So turn it down until they're somewhere in the 90s. And then I will start to up-titrate the peep. And what we're doing here is a sort of peep trial where we're going to assess the driving pressure to use it as a marker for compliance. And if you're driving pressures are relatively new to you, here's the concept. They are best known out of some work that was done, reanalyzing the ARDSNET trials to look not at their plateau pressures, but at their driving pressures. The driving pressure is the plateau pressure minus the peep. So it's not the total pressure in the airway. It's not even the peak pressure. It is the plateau minus the peep. So it is essentially the pressure gradient that is actually changing in the LVOI between inspiration and expiration. You subtracting the peep because that pressure never goes anywhere. There's no delta it stays in. And of course you're subtracting anything on top of the plateau because that doesn't involve the LVOI. Now the way a lot of people use this is to say that a lower driving pressure is associated with less lung injury. And that was one of the things they got from the reanalysis. Driving pressures under maybe 15 or so seemed like they're actually a better correlate with the risk of mortality than the plateaus and tidal volumes as many people use. I don't know. I think there's some sense to a lot of this. I'd still like to see some better data on it. But that's actually not how we're using it here. Here we're using it as a measure of the peep you need because driving pressure is a good way to assess the compliance of the lung. And that is a good marker for how much lung you have recruited. So if you give somebody a certain peep and their driving pressure is X and then you give them more peep and their driving pressure is less, that must mean you recruited lung with the higher pressure because how else could you have improved the compliance of the lung? You're putting more pressure into their lung because it's the higher peep. So you would think their plateau pressure would have gone up if you give them the same size breath or the same driving pressure. And that could happen. It could be the same, but if it was less, it must be because you have a bigger lung to put breath into. You have opened up areas of lung that were not ventilated before. Conversely, if the driving pressure gets worse, you have recruited nothing, you are just over distending lung. And this is giving you a way to start understanding what is kind of good peep that is recruited lung, what is bad peep that is just over distending or at least has not helped you. So let me show you how this works. We're going to start up titrating. Now you could do this trial upwards. The problem is that it takes some time for the lung to recruit with a higher peep and how long do you want to sit here. So I prefer to do it as a downward titration, which means you start high and then you turn down. Now if you start out low, you still need to get up to that high number, but I'll do it relatively quickly. The only reason to titrate at all is just to make sure they tolerate it, which mostly means human dynamically. Some patients with high peeps may have trouble with preload and stuff. But I'll dial up by two or three points, you know, every minute or so, just watching their blood pressure and things. And I'll get them to around 20, usually around there, maybe a little more, maybe a little less, and then I'll start our process. And the process is leave them here for a few minutes, two to five minutes. Again, if you're going upwards, I would say at least five minutes. You should give them a few minutes though to kind of equilibrate. And then check a plateau pressure. And you can check two or three if you want to make sure you're getting a good reliable number and average them. Take the plateau, subtract the peep, and get the driving pressure. And you're going to forget these. So I write them down. If there's like a white board in the room, it's often a good place for this. Write down the plateau over the peep, subtract them to get the driving pressure at that peep. You should also look at how they're sat is, how they're oxygenating. And you can drop that down as well. Great. Now, go down. And I usually go by two. So if you started at 20, go down to 18. Give them a few minutes. Check again. Check their plateau, calculate the driving pressure. And what you want to see is is it more or less or the same as it was? So let's say at a peep at 20, their plateau pressure was 35. So their driving pressure is 15. OK, not terrible. If you go down to 18, now their plateau pressure is 33. So that's the same, right? That is still a driving pressure of 15. OK, so we haven't really done much. But we have identified that perhaps 18 is at least as good of a peep as 20. We didn't get anything more at 20, and therefore less is probably better. OK. And check your stats as well. Now we go down to 16. Maybe at 16, our plateau pressure is 30. So that's a driving pressure of 14. Now that is less than the 15 we had. Now, it's a small difference. Again, you can recheck a few. But that implies that this is a better peep. 16 is better than 18 was. At 18 and at 20, we were slightly over-distending. We were getting a needless amount of peep. And now we're getting more into the realm of goodness. We keep going, though. Now we go down to a peep of 14. Now, let's say our plateau is 27. That is a driving pressure of only 13. So it's even less. So now we're doing even better. We were way too high before. Yeah, that doesn't mean we were wrong. That's the purpose of this trial. You want to start high enough that you're pretty sure you're above what you need. I mean, some people might need a higher peep than 20, but typically for a patient. So we're going to keep going. Let's say we go down to a peep of 12. Now when we check a plateau pressure, it's 28. Our driving pressure is 16. It actually went up. So what does this tell us? Well, perhaps what it tells us is that when we turn down the peep now, we dropped below the collapsing pressure of a significant portion of our lung. And we de-recruited. And now when we give the breath, we're giving it into essentially a smaller lung. So we've gone too far. So if 14 was good and 12 was less good, we should go back up. We go back up to about 14, back up to where you were, and maybe that's your best peep right now. Maybe you want to go a little higher for a minute to kind of re-recruit them. Maybe as high as at 20 for a few minutes, they'll probably recruit at their optimal peep eventually, but maybe if you want to speed it up, that's where you're going to leave them. So that's the process. You go down slowly, you check your driving pressures at each number, and then you find where you have the lowest driving pressure, and that's your best peep. Now you should again watch your oxygenation, and you won't necessarily find that their oxygenation is best at the best peep, but it's usually in that ballpark. If you are way too high, you're going to be over to standing some of their good lung, and if it's way too low, of course, then you're not recruiting enough lung. But the driving pressure is a little more specific, I think. You don't usually have to choose between optimizing one or the other. Although that does depend on the lung physiology, and for instance, if you have one of those very heterogeneous lungs. So that's the general approach. A few considerations and caveats. You should do this relatively frequently because their peep is not set in stone, and it will change over time. Minimum, let's say daily if you're managing these patients, probably a couple of times a day is better, and certainly if anything changes. So the nurse comes and says, hey, they were satting at 94, and we turned them, and now they're 85. They're being changed. Sure, some mucus moved around, they plugged a little, they changed. Maybe you should reassess now. Their physiology has changed, and what was best peep before may not be now. Or you set their peep, and then three hours later, they're satting 100 at the same FIO2. Well, they probably recruited some, which can happen with time, with appropriate peep. Now, maybe you can come down in your FIO2, certainly. But at some point, you may want to reassess your peep. Maybe they don't need so much to maintain recruitment as they needed to achieve it, which is normal. What do you do in general with the FIO2 in the background of all this? Well, if you're using these physiologic methods to set your peep, the FIO2 you can just titrate to their oxygenation. This makes it very straightforward for the nurses and respiratory therapists. Just target a sat of whatever. I usually like it above 90%, but certainly less than 100%. But if you have the optimal peep to recruit as much along as you can without causing injury or over-distention, then FIO2 is just something you can titrate within that to keep their oxygenation as normal as possible. So if you end up titrating down to 40% with this peep, great. Again, reassess to some point. Maybe they won't need as much peep, but if that's what they need right now, then perfect. Now, if you get to a point where you're using more esoteric tools to recruit them, this may fall by the wayside. If you are turning to something like inverse IDE ratios to try to get more recruitment and mean airway pressure during inspiration, supposed to end expiration with your peep, or something like APRV, which is kind of the all-in daddy version of that same strategy, not all the supplies. Although some of the physiology may apply. But this is a way to set your peep kind of in ordinary situations. One of the last caveats is that many ventilators now will calculate an actual compliance for you. And you may think that when you don't need to look at driving pressures because the compliance itself tells you the same thing. Maybe. You can certainly watch compliance and even jot that down and follow it as well. Two caveats are that, first of all, like any auto-calculated number, they can be a little bit confusing and black boxy. It's not always clear what goes into it. And the other is that if you're not actually performing a hold, an history hold to get a plateau pressure, and you're not in a mode where you're going to end up with something equivalent, you're typically going to get a compliance to reflect their peak pressure, meaning consideration of their resistive pressures of the airway, not just the LV-LI. And therefore, it's not really what you're looking for. That's why you have to go to all this trouble of doing all these holds. The last point here is about recruitment maneuvers. A recruitment maneuver is a very high kind of superphysiologic pressure in the airway, intended to open up lung, which, again, you can then drop them down to a lower peak to maintain that recruitment. So you could do this as simply as turning their peak way up to like 20, 30, 40. There used to be a traditional thing of putting them in pressure control and then dialing them up for a sort of ladder. It makes physiologic sense until the arch trial, A-R-T, which was of recruitment maneuvers and actually showed worse outcomes when they were routinely used. Now you can argue about whether they are accruements were too aggressive or a variety of other things, but it's given a lot of us pause about using these routinely. So while it might make sense as you do a peep titration to, especially if you're going upwards, you may be recruit them with each step. So you don't have to wait for them to slowly recruit over time. You know, maybe it's not good for you, especially if you're aggressive. And that's part of the reason why I like to do a downward trial instead. In a way, my starting peep and whatever airway pressure they get there is the recruitment, but it's all reasonable. I mean, I'm not going to do insane airway pressures. Therefore, I think it's pretty defensible and pretty safe. And again, there's that time elements too because you're coming down slowly. All right, guys, that is a general look at my approach these days, typical patient for determining peep. Again, I think the kind of starting point is just to believe in your heart that there is a right number here and not just to let this become one more thing to kind of blindly titrate to some output at understanding that there is no perfect number, understand that the lungs are complex, but nevertheless kind of aiming for that. And I think you will find that sometimes you end up at a very different place than you would have predicted based on something like the FiO2 or a protocol. Give it a shot. Let me know what you think. Talk to you next time. Thank you. the next time.