Well, good afternoon, everybody. Thank you for attending on a Monday afternoon when the sun is out. I appreciate your attendance, and hopefully you'll enjoy the session. My name is Jamie Krensley, and I'm going to moderate the session this afternoon. I have a few words that I'd like to say as a small introductory set of statements about glucose control. I think we're at an inflection point in the history of glucose control, which started quite a long time ago. And so my little talk is called Moving Towards Increased Clarity, Glucose Control 2020. I do have a few disclosures. I've worked with a couple of companies trying to design continuous glucose monitors. So as we all probably remember, 2001 was a landmark year. That was the year that Dr. Vandenberg published her study from a single center, surgical ICU, ventilated patient cohort showing a marked reduction in mortality and morbidity with tight glycemic control. This put tight glycemic control on the ICU landscape and really was a wake-up call for many of us. The history after that is that in 2004, from Stanford Hospital was the first confirmation of these findings. It was a non-randomized before and after study, but it was strongly positive. In 2006, Dr. Vandenberg published her medical ICU study that was sort of weakly positive for glucose control. And then in 2008, there were two studies that crashed and burned due to high rates of hypoglycemia and low protocol compliance. And that led us to the big granddaddy, 2009, the publication of the Nice Sugar study. This is the study that showed increased 90-day mortality among patients who had the tight interventional arm of 80 to 110 compared to the conventional arm that this time was 140 to 180. So the Nice Sugar trial led to a period of increasing apathy for glucose control pretty much worldwide. Enthusiasm started to wane. People said, ah, forget about it. Just do 140 to 180. That's what the guideline says. Easy peasy, no big deal. Well, we've sort of learned a lot in the last 10 years about the science and the outcomes and the metrics of glucose control. And some of this we'll be sharing with you today. First, why were the results of Leuven 1 not replicated? I think perhaps the most important reason is the single center effect. And this picture is of one of my true heroes, the Mickey Mantle and Babe Ruth of glucose control, Dr. Greed Vandenberg from Leuven, Belgium. She's passionate. She's brilliant. And she had her ICU taking care of these patients very carefully with low rates of hypoglycemia and pretty good time and targeted range. Second, of course, everybody knows about hypoglycemia. There's a whole raft of observational data, as well as data from the randomized trials, this one from Nice Sugar published in 2012, demonstrating that hypoglycemia was independently associated with increased risk of death. Third, monitoring technology and monitoring frequency contributed to low time and targeted range. In fact, if you look at the Nice Sugar data, you could extrapolate. And there's paper that shows this, that probably only a third of the values were in their targeted 80 to 110 range. So they weren't achieving what they were supposed to try to achieve. Finally, and what's most relevant, I think, for the current literature and where the field is going, is one size may not fit all. Perhaps 80 to 110 was an incorrect target for patients with diabetes, especially those with poor pre-admission glycemic control. One thing I like to know, one of the most important lessons I try to teach my residents and interns on rounds is that the risks and benefits of an intervention are not evenly distributed across a population. So you might have a study that shows a 5% improvement in mortality, but of the people in that study, perhaps 20% had significant benefit, 70% had really no benefit or no harm at all, and 10%, the benefit of the intervention was harmful. So there's a strong signal from observational as well as the interventional trials that patients with diabetes and patients without diabetes have a different relationship with glucose control and how it relates to outcomes in the ICU. So this is a 45,000 patient nine center study that shows the relationship between mean blood glucose during ICU stay and mortality when you compare patients without diabetes, which is the top slide, whereas mean glucose goes up, mortality goes up, and patients with diabetes where there's a flat relationship between mean glucose level during ICU stay and mortality. Subsequent data has tried to make this information more granular by not just looking at diabetes status as if it were one monolithic cohort, but by looking at hemoglobin A1C values and stratifying diabetes patients based even on A1C, because a diabetes patient who's A1C is 6.6 is actually rather different than one who's A1C is 9%. So putting these principles to the test, Stanford Hospital conducted a two-year before and after study that was published in 2017. This is the first study of two targets in the ICU. In the first year, about 1,000 patients in each year, we had everybody treated at 90 to 120. That was the standard of care at the time in my ICU. In the second year, we stratified patients based on A1C measured at admission. So A1C less than 7%, you were treated with a tighter target of 80 to 140, and A1C of greater than 7%, you had a looser target of 110 to 160. What we found is that for patients without diabetes and with the lower A1C, there was no change in severity-adjusted mortality, not so surprising over a two-year period. But with patients with diabetes, there was reduced severity-adjusted mortality during the second era, and that was driven by a reduction in mortality in the patients who were treated with a loose target. So looser was better for those patients. Now we have data that are under review right now. I can't show you a slide of it, but I will tell you that we have stratified patients by A1C, and for patients, this is a 5,500-patient cohort, patients with A1C less than 6.5% as mean glucose in ICU increases, mortality increases like a rocket, and for patients with A1Cs of 8% and higher, mortality plummets as mean blood glucose increases. In fact, the mortality of patients with mean blood glucose of over 180 in this cohort was about one-third of the patients who had a mean blood glucose of 80 to 140. So mortality goes down as mean blood glucose goes up. So as I said at the beginning of my talk, I think we're at an inflection point in glucose control in the critically ill, and I believe we're entering the era, and this is a theme you're hearing throughout the meeting. Every organ system, we're now entering the era of personalized glycemic control. So this is a review article that was published in 2017, and I guess somebody was listening, because two months ago, this review article was published in another journal, not by me. So the long and winding road towards personalized glycemic control. So today we have, I think, a rather interesting multidisciplinary group of presenters. It's a powerhouse group, all four of the presenters are on this year's guideline committee for glycemic control that is meeting and hopefully within a year or two we'll produce more guidelines for control. Beth Taylor's gonna lead off with the role of nutrition and nutritionist. Judy Jacoby is going to discuss pharmacologic aspects of treatment. Dr. Jacoby is a former past president of the SCCM and also for the second time. It's amazing that you're doing this again, taking this on. And she's the head of the guidelines committee again. Then Karen Sands is going to talk about nursing implications and monitoring considerations. And then Dr. Ellie Hirshberg will talk about glucose targets. And at the end, if there's time, we'll have a few case studies to discuss as a group and with you. And the structure of this will be each speaker will have about 20 minutes, and I'd like to leave time after each speaker for the audience to ask some questions. So you don't have to wait for an hour and a half, since we have two hours with you this afternoon. So thank you again for your attendance and sit back and enjoy the show. So. This is where I bring out my song and dance. No, you do not want to see that. Is it on the desktop, would I go to the desktop? All right, so I've been tasked about talking about the role of nutrition and blood glucose control in the ICU. The only disclosure I have is not related to the current presentation. Okay, and these are my objectives. We're going to talk about, obviously I'm going to talk about what the dietitian, what their impact would be in the control of blood glucose levels, and I'm going to talk to you about that in terms of permissive underfeeding versus full feeding and using some of that data to illustrate my point. So I think first and foremost, I think we can all agree that it takes a team, and that's why you have someone, all those different members from the team here today. And so what I'm going to focus on is the nutrition prescription and a little bit about the carb to insulin ratio. But the transition from insulin drip to scheduled insulin will be covered later. And so kind of like what Dr. Crinsley was saying, I think we are a little bit at a crossroads because we, I don't know about you, but on our dashboard right now is both hyper and hypoglycemia that we have to report, and we'll get nicked if either one of those are high. So we want there not to be hyperglycemia, but we also don't want there to be hypoglycemia. So we're trying to figure out how to effectively control glycemia for each individual patient. It makes it a little difficult. So when I think about nutrition and what's the dilemma for me as a dietician, I want on the one hand to avoid the highs and lows of blood glucose levels. I want to keep in mind that changing patient status in the ICU, that may affect how I can feed them and their blood glucose levels. Some may say I want to limit the insulin. Not everyone will agree with that. On the other hand, I want to give them the nutrition they need, right, to heal. And I want to be able to in real time adjust their insulin just in real time as I may be adjusting their carbohydrate. So I think that it's key that if you're coming up with a protocol for your ICU that you have to keep in mind it shouldn't be carbohydrate blind. You have to keep in mind what will happen when I'm feeding the patient, and if I'm not feeding the patient, how will this protocol work? So I'm not going to go into this in great detail because I think as ICU clinicians we're all aware that we know there's a changing metabolism of how we metabolize carbohydrates during the stress response, that we are overproducing glucose in our body. It's being, there's an uptake peripherally, and then we also have an increased resistance to insulin. And a lot of the times we call, you'll hear this called acute or stress hyperglycemia, and just as a quick review that even though this stress hyperglycemia can exist whether a patient is diabetic or not. We've seen it all the time where we have a non-diabetic that has high blood glucose levels because often during the stress response we have decreased gastric emptying, more gastroparesis. The carbohydrates that I would feed innerly are going to hang around a little bit longer and may have more of an effect on their glucose level. Right now we treat stress hyperglycemia with insulin and we think about when we're giving insulin not only are we trying to treat hyper and hypo but we also want to think about the glucose variability. So when we think about ICU patients I think more and more we're starting to categorize them acute phase, late acute phase, recovery phase, but unfortunately we don't have a single marker that tells us when a patient goes from one phase to the other and I'll have some physicians say I'm going to wait to the late acute phase or the recovery phase before I feed the patient, but think about the patients we save now versus what we saved 30 years ago. We have some really sick patients and I explain to them you can't just starve them for weeks until you feel like they're in that late phase. So we have to think about what's the best way to feed and control their blood glucose levels in that acute phase. And so we can get some of that information by looking at the early internal feeding studies that are done in the ICU and from those studies and there's four listed here and I'll look at each one of them a little bit more individually. We know that if I give full internal nutrition what it does not do is it does not lead to worse clinical outcomes or improved clinical outcomes. So in other words, early feeding with not goal internal versus full feeding is really the same in terms of clinical outcomes. But what I do know is if I give full internal nutrition early in that ICU stay they will have a higher blood glucose level, they'll use more insulin and they'll have more GI complications. On the other side, if I were to right off the bat give someone full parenteral nutrition I will not improve their outcomes but I will probably cause some refeeding and again maybe a higher risk of hyperglycemia. So these are those four trials and I just put up the survival curves to show you there's no difference. And now I'm going to look at those same trials but I'm going to look at what the authors told us about their glycemic control and insulin usage. This one I put up, this is a trial by Zussman where they did indirect clemetry or measured needs. And the only reason I put this one up is for you to really see that when they measured their needs on those first few days, that first 48 hours they actually measured lower than the rest of that time in the ICU. So that gives you an idea that perhaps we don't need to feed full feeds, right, in those first few days so their needs aren't quite as high. The only comment they made about daily insulin therapy was total amounts, so not too much on this trial. But if you look at this one in the EDEN trial, if you looked at their patient populations, if I looked at how many of them had a diabetes diagnosis and they were about the same whether they got trophic feedings or full feedings, they gave no specific formula as far as the different groups. But if you look here and I look at the mean blood glucose levels and on the axis is here, it's milligrams per deciliter and I've also translated to millimoles per liter. The full feeding group obviously had higher plasma glucose levels compared to the trophic feed. Full feeding group used more insulin but not enough insulin to get within the range of the trophic feed. The ARABI permissive underfeeding, you can see about 35% of the patients were diabetic no matter what group they were in. And then I look at what happened with their glycemic control. The full feeders and the underfeeders actually were about the same, not a whole lot of difference. But they were, look at them, they were about 165 to 170 plus or minus 95. The full feeders took twice as, almost twice as much insulin but again, some of these patients were probably over that 180 if we're trying to shoot to keep them under 180. And as far as formulas, they did use some that were lower carbohydrate but I couldn't tease out which ones got the lower carbohydrate. And then in the target trial, they had a much lower percentage of patients that were diabetes, they had diabetes. They had similar carbohydrates for both groups. They had a very small use of insulin but they ran high blood glucose levels, over 200. So what do I think this tells us? Why even bring this up? I think that from these trials, I can learn that probably 30% of ICU patients have diabetes. Maybe about 7% of them required insulin on admission. And that by giving more nutrition in that acute phase, I'm not really going to make, again these studies tell me I'm not going to improve their outcomes but I'm going to make it a little bit more difficult to control their blood glucose levels. Now these trials didn't tell us anything about hypoglycemia but if I could come up as a dietician with something that would improve all these things, well then I'd be a hero, right? I'd make a lot more money but I probably won't come up with that. But this is what some of the diabetes formulas that have kind of been developed over the years and they've changed a little bit. And I think there may be a place for these. It's not for all patients but if you're having that individual patient that's really tough to control, you might consider something that would give you a lower percentage of carbohydrate and different fat to help stabilize. And some of them are also higher in formula. Let me show you a few of those trials. And this was a randomized control trial where group A and group C are the diabetes formulas and group B is the standard. And you can see the breakdown. So standard 43% carb, the others are 30 and 33 but pretty similar in protein. A has a little bit more. And then you look at their results. And what they found, right, is the, remember A and C are the diabetic groups. They used less insulin. They had a lower mean blood glucose in group A especially. The hypoglycemia, which remember we were maybe worried, am I going to cause more hypoglycemia? Did not come to fruition. They actually had less hypoglycemia. And in this trial they used bronchitis as a way to determine infection. And when they looked at the relationship between bronchitis and glucose variability which was statistically different between the groups, they found in fact there was a relationship that if in the diabetes formula they had less glucose variability and less infection using bronchitis as their infectious. This was another one. It was a little smaller, 41 ICU patients. Again, carbohydrate differences, 44 to 29%. Similar protein. Sick patients. These are sick patients, vented. Diabetes, almost half of the patients were diabetics. And they fed 44 hours to give us some information. Their results, same similar findings, significant difference in their glucose levels and the probability of a glucose level outside their targeted range. But no difference in clinical outcomes. The divine study that was just done more recently was an open label, randomized control trial of 150 patients. And it was only looking at these overweight patients. So BMI 26 to 45, interval for five days or longer. They excluded type 1s in DKA. Again you can take a look at the products they used. But the low carbohydrate also was much higher in protein to meet the protein needs of these obese patients. Now the results, what they found was there was no difference in the events outside of the target range between the two groups. But they did have a significant decrease in the values that were higher than 150. So the lower carbohydrate group did have less excursions on the high end of hyperglycemia. There was actually no difference in their hypoglycemia rates. Similar glycemic dispersion. A little decrease in frequency of insulin administration. And again no difference in mortality. And they actually stopped that trial early for futility. So let me give you a few summary points. I think that when you're feeding patients, it does make glycemic control kind of challenging. We want to maintain a consistency of carbohydrate infusion as much as possible. But we always have to be mindful when that's going to be interrupted. That nutrition support. And you want to have a planned approach. You may need to personalize your approach. Just like Dr. Crimsley said, if I have someone who I know is on insulin at home and I'm starting their feedings in the ICU, I take that into account. What was their insulin requirement at home? And I know that I might start them at a scheduled insulin dose, you know, half at least of what they were making at home to cover their carbohydrates. And it's because they're critically ill patients, okay to start slowly. I don't use routine diabetes formula routinely. But if I have that patient that's really kind of tough to control and it's time to feed them, then I will give that a try for a little more consistency. Typically if they have the stress hyperglycemia and it's tough, I usually don't put their feeds up to go though until we've got that blood glucose control under 200. Fiber has not been shown to make a difference in the ICU patient for glycemic control. My thoughts, and I'll leave you with my thoughts on insulin management, and we'll hear a little bit more about this from the next speakers. But I usually start with an insulin drip when I start to feed them. I use continuous internal nutrition or parental nutrition and I kind of see where they're going. And I have it written in my protocol. This is part of our online protocol that if we're holding their, or is this discontinued, that we're going to hold that infusion and I make sure dextrose is started to cover what they've already received. And as far as translating in non-diabetics, I just go with the formula that best meets their needs and then work to control their blood glucose levels unless they really give me some grief. If it's a diabetic, I may try with a diabetic formula if I'm having trouble. And then as far as whether I, when I go to basal bolus, I'm not going to get into that because Judy's going to talk about it. But I think it's, the key is always have plans in place or an alert so that you know if two feeds are changed or as you're increasing tube feeding, you, I'm right there, I'm working with my pharmacist is saying, okay, this is when I'm increasing, so this is the timing of when I'm going to increase that bolus insulin. So we work together, really work together. And if you're using the DRIPs, continuous feedings go with DRIPs, not bolus feedings. I left, we didn't really talk about this, but a few thoughts on prenatal nutrition, how I may calculate that. I think, I don't know if Judy will cover that, but I usually start with a DRIP, look at my total dose and would start to add some to the TPN. And then I also, the other thing I was going to say, and I bet they'll mention it, as they get better, right, as they get out of that acute phase, their need for insulin may decrease. So you kind of have to watch how that changes. Thank you. So, so we do have time for some questions if you'd like to step up to a microphone. Yes, and you can introduce yourself and speak up. Hello, my name's Kelly. I'm a pharmacist. So how, what's your take on the diabetic formulas on the mechanically ventilated patients as sometimes those contain higher fiber formulas and risk of boluschemia in that patient population. Because a lot of times we definitely steer away from them in my current practice on any of our ventilated patients. So really, I'd say 80% of our patients are ventilated. So first of all, so that is, I cover, it was a surgical trauma unit where I had mostly ventilated patients. And so you, what you have to watch is the type of fiber, soluble fiber is probably fine. In that acute phase, I wouldn't go straight there, right, if I'm trophically feeding, it doesn't really matter if I'm using a diabetic formula or not. As they get out of that acute phase, so they're going to be at less risk for the non-inclusive bowel necrosis. It's not the ventilator so much as the perfusion. So if they're a patient that's on multiple pressers, as you know, I would not go beyond trophic yet in that patient. If they're just on the norepi and the rate's coming down and they're seeming to stabilize, then I would go ahead and start pushing them to goal. And then if I was having trouble controlling their sugar, I could think about that. I'm also still using the drip at that point. So usually when I switch them over to diabetes formula, it's probably a little bit later, it's a little bit beyond the acute phase because you bring up a good point. You want to watch against non-inclusive bowel necrosis, happens very infrequently, carries high mortality as you know, yeah. Great question. May I ask a question? Absolutely. Oh no, from the expert. Okay, don't make it too hard. So can you make any comments about continuous versus bolus enteral feeding since we don't eat continuously, most diets. Right, I know. So from what standpoint did you want me to make that, from glycemic control or otherwise? You could start with glycemic control. So in the concept of glycemic control, continuous carbohydrate infusion is much easier to use with the insulin infusion, right, than bolus because if I'm going to use a bolus feed in the ICU, which you can do, you have to do it a little differently. I usually don't use an infusion at that point. If they're getting well enough to handle a large bolus, which would be later phase, then I'm switching them over to basal bolus type regimen. From just the standpoint of continuous versus intermittent feeding, the reason we think about that is, especially now when we're concerned about muscle wasting and frailty and weakness, that we are more likely to synthesize muscle after bolus feeding, especially a protein bolus. So that's the argument against maybe going back to bolus feedings more often in the ICU. And I think that's fine. I think in the beginning, so if you were doing 50Q, four hours bolus, like of a tube feeding product, maybe you could do an infusion. It might, but maybe you could cover it with a slide, you know, a nutrition correctional that way for glycemic control. So I think you can go either way. So in our acute phase, we tend to do continuous and get to bolus in late phase because it may be beneficial for protein synthesis. Thank you. Any other questions? Okay, so our second speaker is Judy Jacoby, and she'll be speaking about pharmacologic aspects of insulin management. Okay. So my particular focus was on the transition off of an insulin drip to the basal bolus or other forms of insulin management. And I always like to predicate it. Clearly, Beth talked about her population and fairly extensive use of an insulin infusion. In my practices, where we had patients who had difficult-to-control sugars or certain other clinical situations, which we'll talk about a little bit more, certainly we heavily used insulin infusions as well because we had a protocol that had demonstrated very low hypoglycemia rates. But one of the things we learned is that not everyone needs an insulin infusion. And certainly, Dr. Crinsley's group at Stanford really published on this first. And so he gets credit. We emulated his program extensively. And with at least a portion of your patients, an approach with a fairly intensive regimen of sub-Q insulin may be enough to achieve certainly a moderate level of glycemic control. Probably if you have a patient that you really want tight glycemic control, it's not going to be adequate. But it certainly is very important from a workload perspective. And so we would give every three to four-hour sub-Q insulin, pretty aggressive dosing. We used a DOS program calculator. Later versions became the CGS product. And basically, it uses the perspective of insulin sensitivity. And so we selected an insulin sensitivity factor, which if you think about it, if you gave one unit of insulin, it's a guesstimate as to how much you think that blood glucose would change. Your patient that's on 250 units of Lantus a day you know is pretty insulin resistant. They're going to have a very small number for their insulin sensitivity factor. And you can guesstimate it by maybe about 1,700 divided by their chronic total daily dose. Obviously, that also depends how they eat at home. But anyway, so generally, an insulin sensitivity factor we would set between 15 to 60. For an intensive insulin regimen, we would pick 15 to 20. And one of the other institutions using this approach in Indianapolis at Eskenazi, Serena Harris and her group looked at almost 1,400 patients retrospectively with various settings of the ISF, 15, 30, and 60. And what they were able to achieve is that for a population where more than half the time the blood glucose was above 150, they were able to get the patient to a goal of 100 to 150 in about 19 hours. And it was quicker in terms of titration with every three-hour dosing versus every four-hour dosing. And the lower ISF, which would give you higher doses of insulin, was faster to the target than the higher ISF values, overall achieved about a 50% time and range. These are patients that may or may not be eating. And found a pretty low hypoglycemia rate, at least if you look at it based on the number of blood glucoses. And so if you do a lot of blood glucoses, that's going to influence your percentages. It's better to do a hypoglycemia rate relative to a time on insulin or at least per patient. A low rate of blood glucoses reported as severely hypoglycemic, less than 40. And it was a bit more common for somebody with that more aggressive insulin regimen to hit a blood glucose less than 70. But it's not perfect. And a certain population of patients, despite this intensive sub-Q insulin, would still pop their blood glucoses above an acceptable range. And so some of these patients may indeed need insulin infusion therapy. Now, am I going to do this on a type 1 diabetic? No, I'm not. Because I want those patients to get basal insulin always. But if I have a type 2 patient who's got some hyperglycemia that's not excessive, it may be enough within every three hours sugar to get a reasonable management of that patient. Now, if you take that type 2 patient and start to feed them extensively, you're going to find this might not be enough. But hopefully by then, you can add some basal insulin. Now, all of this insulin management is very complex. And what we do in the ICU may or may not translate out to the general ward. And so your diabetes team really ought to be like an antimicrobial stewardship team. And certainly, there's a lot of support for that if you're looking in the diabetes literature to really have an overall program for glucose management for your hospital or health system. And because it's a pretty complex thing, when you think of the patients coming from home and doing whatever it is they do at home, hopefully some of it's correct. In the pre-op space, in the peri-op space, post-op, ICU. And so those are all transitions. We go from the ICU out to the wards. We may have patients who use an insulin pump in their home settings. And how are we going to interface with that in the hospital? A lot of pretty complex issues. And so in my practice, we had a multi-professional team extensively involving nurses and pharmacists for a lot of the day-to-day, but certainly with key input from our endocrinologists and critical care specialists. And other places use advanced practice providers as a big part of their management of a diabetes team. So there's a lot of different ways to do it. But it certainly warrants the kind of attention that most of your institutions have given to an antimicrobial stewardship team. So all right, let's think about where we all live in the ICU, or most of you live in the ICU. And we have a patient, and they've been on an insulin infusion. If it's for DKA, obviously the key there is a safe, well-monitored program of insulin therapy during DKA. And obviously, continuing that infusion until the gap is closed for the majority of patients. There are perhaps some new models being proposed using a greater component of sub-q insulin for DKA. We're not going to go there today. So once the gap is closed, you're going to be thinking about transition. I usually go back to the home regimen that the patient is supposed to be taking. Key word being supposed to. And yeah, that's usually a pretty good starting point. Because we're changing a lot of things in IV fluids. And so you have to be really cognizant what you've done with your dextrose intake as you're transitioning that patient to an oral diet as well. If I have just kind of my routine ICU patient on an insulin infusion, I'm probably not going to think about transition until things are pretty stable. I want to make sure that certainly the blood glucose is stable. I want a pretty stable insulin infusion rate. So if you're having to go up and down like crazy, and it's not related to the patient eating a meal, that's a patient that you probably want to wait to consider transition. We'll talk about some strategies for those really difficult to transition patients. I prefer the patients to be hemodynamically stable. We know that vasopressors will decrease the absorption of sub-Q drugs. That's well demonstrated with low molecular weight heparin, among other things. Some evidence for especially our very long-acting insulins being poorly absorbed when patients are on vasopressors or with significant edema. So that's my preference is for the patient to be hemodynamically stable. I also like them to be nutritionally stable. I like to only change a small number of parameters at once. So I don't want to be titrating my nutrition while I'm trying to transition my insulin, just adding more trouble. And certainly Beth has talked about that extensively. But I always like to mention with that difficult-to-control diabetic, make sure that they're on some kind of carb-controlled diet. And because that's the sort of thing that sometimes if it's not protocolized, they end up on a regular diet and they're happy as a clam because now they get to eat stuff they're never supposed to eat. And of course, you want to look ahead a little bit to make sure you're not going to have a planned interruption of your nutritional intake, which would impact the use of long-acting insulin. And gosh, you have to decide if you even think they need ongoing insulin. And again, your diabetic type 1 is certainly going to need a basal insulin of some sort. But if you have somebody who's getting a really low rate of an insulin infusion, it may be that just monitoring will be adequate with some call orders for the future. Now insulin isn't a thing. It's a whole family of things. And it's got ridiculously complex kinetics in healthy people and even more challenging in the ICU patient. Our long-acting insulins are these hexamers. And you may have groups of these hexamers bound to albumin. Big molecules are not quickly absorbed, thus it's a nice long-acting insulin. In order for that insulin to be absorbed, these hexamers have to break apart into dimers and ultimately the most easily absorbed component will be a monomer. But it's a very fluid thing and they go back and forth into different configurations. And that influences how quickly that insulin is going to be absorbed. And I really think that that's the reason that when you get to these long-acting insulins in somebody with a lot of sub-q edema, that the absorption rate can be really variable. And I will tell you that I have given patients long-acting, mostly glargine insulin, and you think you didn't give them anything. Now, it's never come back to haunt me like it didn't come back three days later. Some of it gets cleared through the lymphatics and some of it just degrades in the body. The site of injection is important in terms of speed of onset. I'm not sure we're ever going to appreciate that in the ICU. The patient position is going to be important. And so if you're mobilizing your patient, their insulin might behave a little bit more differently than if they're always in bed. And nobody studied to see if it changes from nighttime versus daytime. I'm sure there's a variety of other diurnal variations that are going to influence that. Skin temperature, blood flow, edema, as I've mentioned. And then in addition to the insulin type, the volume that you give per dose makes a difference because these lovely little monomers really like to get together with their friends and make dimers. And those aren't as quickly absorbed. And so a big dose of insulin in a single location is going to want to make more dimers and that will slow the absorption. So it's kind of a complex thing. And so to me, OK, you may or may not have to know all of this when it comes to prescribing insulin. But if you have a product that you're comfortable with or you've developed a routine, it's probably best to stay with that routine because you have a compound that you will find more predictable, hopefully, from patient to patient. So when I'm looking at a transition program, just like you would use for any of your other patients on insulin, I'm probably going to be thinking in terms of a basal, nutritional and correction components. And again, you can use insulin sensitivity factors to calculate those replacement doses. You want to consider a carb ratio that I usually use about a third of my ISF as something that usually works pretty well. And again, being attentive to the type of diet. But I'm going to say that, again, you want something that's protocolized, that's routine, that's consistent so that everybody knows the expectations. And most importantly, that the bedside nurse is involved in this system to make adjustments or raise the red flag when necessary. You've got to have some flexibility for that patient that isn't eating now and was before or has a sudden interruption of an enteral feed or other changes in medication or nutritional intake. And so having that flexibility in a bailout strategy is important because it's hard to transition these patients. They're sick. There's a lot of variables. None of my patients who are type 2 ever eat in the hospital like apparently a lot of them do at home. And that's going to be your overweight patient who that you know is not terribly compliant with their diet. We can't use most of their home or I'm not willing to use most of their home diabetic agents. So think about that. And as Jamie introduced, a diabetic's not a diabetic's not a diabetic. And so you have to really be cognizant, is this a type 1, a type 2, or what we would put into the category of critical illness, hyperglycemia, that it's a little bit situational and hopefully will get better. Now the insulin in the bag. We put 100 units in 100 mLs of fluid, but some of that insulin sticks to the plastic. Actually a fair amount. The most recent study suggested that only about 57% will come out of that bag. So if you put in 100 units, you get 57 units out. I don't have data on Baxter's new product. It's a different kind of plastic bag. And so I don't know their delivery. That's on my to-do list to find that out. As long as you always use the same methodology to mix, it should be consistent for you. Now the insulin sticks to the plastic in the tubing as well. You could use low-absorbing tubing. You've got to have that special tubing that you save perhaps for other special drugs. Ideally, you let the insulin soak in the tubing, and there are a lot of ways to maximize that delivery, none of which is very practical or efficient. We tended to just, in addition to filling the tubing with a new tubing, flush an additional 20 mLs of fluid out and waste it as demonstrated as a way to help saturate a lot of those binding sites on the plastic tubing. So that when you do hook it to the patient, you're going to get pretty consistent delivery. And so the same methodology with your IV delivery makes it more predictable as you transition. And there's some data on syringe pumps perhaps being a bit better in terms of delivery, depending upon what kind of syringe your plastic is. All right, again, reiterating your type 1 patients need the basal. I used basal for type 2 patients who had high insulin requirements or really difficult to control patients while they were in the ICU. And somebody who comes in with prior poor control, I anticipate that they are going to be a long-term insulin patient, at least while they're sick. And so I'm much more willing to use long-term insulin or basal insulin in those as well. But again, as you heard from Dr. Crinsley, how intensively you do that is going to depend on your population. We've heard about nutrition. One comment, if you do use bolus feeding, time your blood sugars before that bolus feeding. And it's really hard to maintain consistency between your bolus feed schedule and your blood glucose schedule. And otherwise, you're going to be kind of chasing around post-prandial bolus. Anyway, think about your IV fluids and the influence of any of those changes that are going to be made. All right, I like to overlap. Start my basal before I stop my infusion. Again, especially important in my type 1 patient or the patient who is more difficult to control. And in terms of which long-acting you pick, again, there's no absolutely right. I tend to use NPH every, I prefer every eight hours, for a difficult to control patient because that's more nimble. I can make a change every eight hours. If I commit myself to a once a day long-acting insulin, maybe I'm only making a change once a day with correction insulin. And this particular study from Dickerson in J-PEN looked at some of these really difficult to transition patients. And they said, all right, what if we sneak up on them with some long-acting insulin? I'm going to start some NPH, but a pretty low percentage, 30% to 50% of their total daily dose based on the IV. I'm going to go up every day on my NPH. I'm going to go down over time with my IV. If they're hanging out in your ICU and you've got lots of time to do that, that might work. And so what you can see is that for diabetics, they got more in the dark bar versus the non-diabetics. And the doses kind of crept up over time. Here's their insulin infusion doses. And those came down over time. Some of it's the NPH. Some of it's that they're probably not as sick. And their carbohydrate intake actually was pretty stable over time. So if you have a lot of time to do all that overlap, you certainly can do that. But that doesn't usually match most of my patients where it's like, let's get them out now. If you've got that patient on a really high dose of insulin, to me, it's much harder to match that. I like my insulin infusion rate to be stable at least six hours. And so to me, that's going to be very helpful to predict my transition dose looking forward. In terms of transition, I'm probably going to go, I've got a slide actually on how much I'm going to use. The other patients that I'm going to consider basal insulin beyond my type 1's are going to be my type 2's on lots of insulin, somebody on lots of corticosteroid, maybe my high BMI patients. And then as I mentioned, I prefer more frequent dosing to be a little bit more nimble with my basal dose. If you really want to use it once a day, such as a glargine, and you like to give those at night, but you want to transition at 10 in the morning, use a dose of NPH to fill that gap. So there are a lot of ways you can combine insulins to accomplish consistency. As I said, I like to overlap a couple hours. But if you can't, again, if you got that patient, you just have to get them transferred. A suggestion in the literature is you give a rapid insulin dose, a log insulin dose, about 10% of what your basal dose would be, plus your basal dose sub-Q, and stop the drip right then and send them on their way. They might rebound. In terms of how much basal, 50% to 80% is a typical reasonable approach in terms of your total daily dose. If you look at the literature and if you look at practices, underdosing is very typical. And so where I've got that patient and somebody's like, I just don't know, I'm afraid, and especially I'm not going to commit to a 50% or 80% once a day dose, again, that's when your NPH dosed more often can give you a little bit of security because, again, you're nimble to make changes. Looking at percentage for transition in this particular study of a mixed surgical medical burn ICU, they didn't really have an insulin protocol. But they found that, in general, their hypoglycemia rates were low, and they did pretty well in the 50% to about 70% range. So certainly a little bit of evidence for that. Here's a multidisciplinary team that, with their protocol in these lower bars versus no protocol for transition, you can see that a team with a plan does better than no plan. I'm not surprising anyone here. And so ideally, you've got a fairly consistent plan, and the players involved know how to carry it out. Ideally you're going to look at the impact of that. This was a group that looked at their impact of transition. They pretty much used half basal bolus, and they found that their more difficult patients were older, on more than 1.6 units per hour, and at baseline had wide glucose variability. So that more difficult-to-control patient, to no great surprise, is more difficult-to-control. There is some potential role of oral hypoglycemics. Again, it's going to depend on your patient population and the circumstance. Metformin you can use if the creatinine clearance is above 30. In the hospital, heck, we're monitoring that closer than they are getting monitored at home. If there's an obvious risk factor for acute kidney injury, all right, don't go there. Certainly not if they've had contrast, or you know they're going to plan to use contrast. But for a lot of your type 2 patients, metformin is not an unreasonable choice. There is some literature looking at sitagliptin and some of the other gliptins continuing in the hospital. And this was actually, they started it specifically in cabbage patients pre-op and for 10 days post-op. It didn't prevent stress hyperglycemia. It's a tough population, the cardiac surgery population. Hyperglycemia is pretty typical. But it did apparently reduce their insulin requirements in the first 48 hours. You do want to obviously monitor for an impact in heart failure, and I probably wouldn't use it for somebody with pre-existing heart failure. I would always avoid, however, an SGLT2 inhibitor, and that certainly gives us fits for patients who are on it at home in terms of that persistent glucose urea and the potential for euglycemic DKA. But that's a whole other topic as well. So that's how I do transition. Well, do we have time for a question? That was a great presentation. Thank you very much. There's time for a question or two? Anybody? Hi. Rachel Krueger from IU Health. Is this on? It's on. Yeah. Thank you for your presentation. I wondered if you could comment on using nutritional insulin in patients on continuous tube feeds and how you go about doing that, especially if the rate is barely consistent. Do you use a fixed regimen or try to just accomplish that with an ISF versus doing something else in between? Yeah. So for a fixed rate continuous enteral, or you could say per enteral, continuous dextrose intake essentially. For the most part, I haven't needed to use any kind of specific nutritional insulin. And so basically it would be a basal and then maybe correction at a reasonable interval of maybe three to six hours. I kind of hate to go longer than four hours in the ICU for anybody in terms of glucose monitoring. And so to have a specific nutritional component, not done that. But for a bolus tube feeding, then you would want to consider how many carbs in the bolus, do a correction, and you could have a basal on there as well. So those patients are a little bit more complicated potentially. I'll ask a question of the audience. So this is a pretty complicated car to drive. And it's especially hard to drive if you have your windshield all blocked up. So how many people in this audience have data monitoring systems that would allow you to see what your hypoglycemia rate is on a per week or per month basis or hyperglycemia rates per week or per month? So in this glucophilic crowd, we might have 25%. So the rest of you are trying to monitor glucose without knowing your glucose values on a regular basis, which is really a challenge. And I will add that CMS has a new monitoring criteria for reporting for hypoglycemia in hospitalized patients who are on anything that is an anti-diabetic agent. So orals as well as insulin. This will be adults only, excludes liver failure patients, that sort of thing. So they're not getting anti-diabetic agents. But anyway, so if you have an adult who's got an anti-diabetic agent and has an episode of hypoglycemia that everybody's pretty sure is not just a lab abnormality, that rate is going to be reportable to CMS. And so your quality departments are probably looking and or will be looking at your hypoglycemia rates a whole lot more closely. So there will be additional accountability. Hey, Dr. Giacobbe, this is Dennis Parker of Detroit Steven Hospital. I'm a pharmacist. I was wondering about, I saw you had metformin there. Our formulary doesn't even allow any oral hypoglycemic agent. I was wondering if anyone or you had experience with using the metformin recently in the ICU because that's something that I would like to do, but people are afraid of it. Yes, people are very afraid of metformin. And clearly, if somebody has acute kidney injury and you keep giving a lot of metformin, you can get them into all kinds of trouble with lactic acidosis. And I've over the years had the odd patient that that happened at home. They're in the hospital. We're monitoring renal function. And if I'm not doing something that specifically is going to provoke acute kidney injury, I will have that discussion with my team. You're right. And I will get different responses depending upon my team. I most recently practiced in CV surgery ICU and those patients who were very stable, certainly maybe even a TAVR patient, but they usually weren't in the hospital very long, but patients after a CABG who are real stable, we would discuss continuing or essentially restarting metformin if I felt they were at low risk for acute kidney injury. Hi. Judith, excellent presentation. I'm Juan Maldonado from Puerto Rico for Puerto Rico Trauma Center. I have a question regarding the possible use for continuous glucose monitoring in patients in the intensive care unit because for years and years, we hear about the same issue, hypoglycemia, hyperglycemia, but nothing specifically like an option or good results from this. Could you please comment about the news, the possible implementation of this with the new devices that are in the outside? Could you please comment on it? Absolutely. Unless, Karen, were you going to talk about that? All right. Karen's going to talk about that. There is one device that has an approvable status by the FDA in the US, a couple in Europe, one in Japan that I'm aware of. They're out there. It's just every time there's a big clinical trial that puts less emphasis on glycemic control in the ICU, all the funding for product development goes away. That's the reason we haven't seen it in the hospital extensively, at least for the intravascular products. I'm not going to take all of Karen's thunder on that one. We'll come back and discuss that some more. Great. Thank you very much. So our next speaker is Karen Sands, and she's going to speak about nursing implications of insulin monitoring and monitoring frequency. Good afternoon. Glad you guys are here. My name's Karen Sands, and I'm a nurse practitioner. I spent about 30 years in the med-surg ICU and have just transitioned to a hospitalist team doing admissions. It's been kind of a bit of a transition, but we're going to talk about the nurse's role here for the next few minutes. And as we all should know that it takes a team to manage these patients in the critical care arena, but I dare say that the nurse is probably the most important person on your team when it comes to glycemic management because they are the gatekeeper of that room and the gatekeeper of glycemic control. So some of the things that I think that's just very, very key to know is that the nurse caring for your patient at the bedside understands the glycemic target ranges and why that are there for your patient population. And I think that we have learned recently, you know, just through literature that has been published that one size probably doesn't fit all, and that as each population is being developed in your research protocols and your insulin protocols for your unit, define in what is appropriate for your specific patient population. And also I feel like for a nurse that if a nurse understands the why behind every component of care that the nurse does, the nurse will get it right 100% of the time. But if the nurse doesn't understand the why and we don't take the time and the energy to educate the nurse at the bedside behind the why, we're going to have those events that occur. So understanding those components of care between the surveillance of blood glucose sampling, what is hyperglycemia in relation to stress versus untreated, uncontrolled diabetes, preventing, identifying, and treating hypoglycemia is the key. And knowing the glycemic outcomes for your particular unit. So the nurse needs to know the insulin protocol for your unit. There are a lot of algebraic and computerized protocols that are out there, and those do seem to be more in line with more safe practice. But we know that there are a lot of institutions out here that are still using paper protocols. But the main thing is that your nurses at the bedside know the protocol. It's not, okay, I got this patient in. Okay, let me look at this checklist. I do this, this, and this. No. That's not the time to start learning the protocol. The nurse needs to know it. So spending that time so that the nurses are aware of all the components of that protocol, they need to be able to just reiterate it to you very quickly. And understanding that transition off that Judy just kind of described, you know, and Beth as well. Understanding that process of how we do transition off the infusion. And then learning new what's on the horizon. And I do think that that gleans us a lot of hope. So the glycemic target endpoints. Each unit, whenever we as a glycemic task force team develop the last guidelines, we kind of come up with a consensus statement of that maybe the glucose range should be within that 100 to 150 range. But I think if we learned anything that we know that that glycemic target range is definitely a moving target. And it's most appropriate to know that that one size doesn't fit all. But you need to look at your patient particular population and determine what is right for your patient. We know that it's very important that if you are not measuring or aware of your hypoglycemic events that are occurring in your unit, or the hyperglycemic patients, the percentage of patients in either category. If you don't know that, I challenge you each to take that back home as a to-do list. Because that is an absolute need for managing glycemia in your unit. But knowing that when your glucose is greater than 150, that probably should generate at least a thought process that maybe we should be adjusting our glycemic orders. And definitely if it's greater than 180, we definitely are doing something. We need to implement a process that we're treating those patients that are having the hyperglycemic events. So it's very important that we tailor those patients to the specific needs. And that we do have patients in the critically ill arena that does not have the need for IV insulin infusion, which we just kind of talked about. But it's very important that the nurse understand as well the three different types of insulin. The long, short, and intermediate acting. As well as the effects of the other antibiotics, not antibiotics, but anti-diabetic medications and the oral hyperglycemic medications that they're using. So the rest of the talk is not going to be designed specifically on the succutaneous. It's going to be more on the infusion patients. But those are just key components that the nurse needs to know, those three components of the different types of meds being used. So with the hyperglycemia, the nurse needs to be able to understand that pathophysiology. And I just kind of went over it here quickly. That when you have the decreased insulin production or secretion, that you have a decrease in the cellular utilization of that glucose. So that leads to your hyperglycemia that spills into your glucose. You have glycosyluria, osmotic diuresis. The patient starts dumping urine. And the patient can become hypovolemic. Well, the nurse is the gatekeeper of your patient, needs to be able to realize that this process is going on. And then that fats and proteins are being, you know, utilized for the cellular fuel, which can lead to DKA. And then you have your electrolyte abnormalities that occur. And then you have the compensatory mechanisms that kick in. Your nurse needs to know that. And so I strongly encourage you to take that back to your facilities, just to make sure that the nurse understands the why behind all these things. Because if she understands this pathophysiology that's going on within the body, that will make the nurse intervene quicker and hopefully provide the best care for your patient. So when a nurse gets an assignment of a day, this is just one aspect of what the nurse needs to do. But it's very important to know if you're treating a type 1 or a type 2 diabetic. And I can tell you probably nine times out of 10, we don't do that. And that's something that we absolutely should know. The second component is, is the total daily dose of insulin that's been required the previous 24 hours. If your electronic computer system doesn't have a graph format for the provider to quickly look at to see a high and a low and add up the total of insulin that's been given in the past 24 hours, you need to have some type of paper format that is easily addressed. Because the provider needs to know, do we need to make an adjustment? And if the patient has a hypoglycemic event, I surely hope that that results in a call to the provider and say, hey, I've had a hypoglycemic event, what type of adjustments do you want me to make? Because if we don't know about the instances occurring, we can't make those adjustments. The main thing behind that is, is if we quickly see that there was a hypoglycemic event, trying to identify the cause of that event. That's the key. And then what adjustments were made in the previous 24 hours to treat this problem. I do think that the hemoglobin A1C, you're going to find in the future that this is a component that we haven't been addressing in the past, but in the future we probably do. Because at our institution, one of the things that we implemented was checking the hemoglobin A1C on everyone that definitely had the insulin order set ordered anywhere within the facility. And then each of the different disciplines that developed order sets for their specific patient populations added the hemoglobin A1C as well. And initially, we did it predominantly to diagnose those patients that have, or to identify, not diagnose, identify those patients that have diabetes but has not ever been diagnosed. And so that gives us a follow-up for the discharge plan saying, hey, you know, whenever you go for your follow-up visit to your primary care physician, please let them know that your hemoglobin A1C was this on admission and that you need to be probably worked up for diabetes and litis. And so that was the rationale for us doing that initially. And now I think the future research will probably show us that there's more to glean from that hemoglobin A1C than we currently know. So as far as the components of care as well, we need to know the incidence of hypoglycemia. And less than 70, we know that the brain starts having some generic effects on changes that occur. But we definitely know less than 40 that we have the potential to have some serious detrimental effects that we cannot reverse. It's very important to also note that those patients that are maintained within our glycemic goal, whatever that is defined for your unit, and also to, I encourage you to get involved in the process that if your hospital is not doing this monitoring of the hypo and hyperglycemic patients, you know, get involved and be a part of the process. We know that hypoglycemia can occur any time throughout the critical time in the critical care arena. But one of the things that we know is that if you have a change in the things listed here without an intervention in place, the patient is going to have a hypoglycemic event. And that is if you have a change in the caloric intake, their nutritional status change. If they're getting two feedings and now all of a sudden those two feedings are off for more than two hours for any reason. If the patient was receiving a diet, now they're NPO for procedure. If the patient was getting your perinatal nutrition and now they're coming back with an electrolyte abnormality and you're having to take down that TPN and, you know, if you don't have a plan in place, the patient's going to become hypoglycemic. If you're having patients that are on vasopressors and steroids and those are being actively titrated up or then actively titrated down, if you don't adjust for that, that those changes and, you know, this can happen anywhere in the 24-hour period of time, you know, the patient has the potential set up for hypoglycemia. And then also with your CRRT, it's the, if they have the bicarbonate base replacement fluids that are changing, that can also impact. So I cannot emphasize enough that the glucose surveillance is key and the nurse is the detective at the bedside. So we need to empower her to be that detective. And if she's, he or she is in doubt, check it out. I truly think that, you know, we do a very good job, you know, when patients start looking like, oh, there's something going on with that patient, we do the basic, you know, vital signs. I think that glucose should be a basic vital sign. You know, with our rapid response systems that are in the hospitals now all throughout the United States, you know, that's one of the key things that they do when they go to a patient that's ultramarine status, lethargic, et cetera, you know, they do all the vital signs but they also get a glucose, most of them do. I think we need to look at that data and see how many of those patients are showing signs of hypoglycemia. But if in doubt, the nurse should be able to check it out. So with the blood glucose less than 70, we know that the brain does have specific risk. But if you have a patient that already has a brain injury, a blood glucose less than 70 could have seriously detrimental effects. And then the severe hypoglycemia categorized as that less than 40, you know, you have the plethora of problems that can ultimately lead to death. So it's very, very important that we prevent, identify, and treat hypoglycemia. But you cannot, I cannot emphasize enough that you have to know your data for your specific unit. And Judy mentioned this as well with the absorption of insulin. If you have patients that are in hepatic and renal failure, the absorption is not normal. And so given these patients, sub-Q insulin is probably not best therapy because of the absorption problems. If they're renally impaired, we know that right there along that it about half, prolongs about half, the half-life is prolonged of the insulin. And the insulin accumulates. And then the whole gluconeogenesis process in both the liver and the kidney is affected. So the patients are at much higher risk for hypoglycemic events in these patient populations, which how many of our critically ill patients don't have some type of renal impairment, right? So the use of subcutaneous insulin in this patient population is, needs to be very, very closely monitored. Now when a hypoglycemic event occurs on an infusion, the absolute first thing that needs to happen is stop the infusion, stop it. And then the nurse needs to know. She doesn't need to go to some type of, you know, book and look at, okay, for this amount I treat there. The nurse needs to know the protocol. And so the way I've always practiced in my 33 years in the critical care arena is, you know, give the nurse a basic idea of what the least amount to be, least amount of medication to be used. So for, to treat a hypoglycemic event, a D50 amp, anywhere from 20 to 40 cc is equivalent to that 10 to 20 grams of IV dextrose to replace. And that patient needs to be followed every 15 minutes until their glucose is consistently greater than 70. And then there needs to have some safety checks built into your protocol so that your surveillance, you're checking for surveillance at this point in time to see if the patient is going to have another hypoglycemic event. Now most of our patients have LONs, but there's occasional when our patients lose all LONs. We have the patient that goes crazy, they have LONs, you know, they're hypoglycemic. The nurse needs to know the options that are available. IM glucagon, obviously it's going to be a much longer absorption time. But if that's the only access you have, that's what you have. And then obviously, you know, if the patient has the ability to have an oral route, the carbohydrate options are also there. But the provider should, number one, be notified whenever a hypoglycemic event occurs so that an adjustment can be made. Now in regards to blood glucose sampling, it is very important that the monitors that were being used, that are being used, that we do those quality QA checks. Now you know on every unit, the nurse has about, you know, they have about 15 different checks that you have to do. You have to check the code card, you know, the glucometer, all the different things, you know, what's the temperature on the refrigerator. But this is key, you know, and it needs to be done every shift, every day, just routinely. And also, the validation of the way the nurses are obtaining the blood glucose sample, I think that needs to be, that's a competency that needs to be addressed yearly so that we ensure that all of our nurses are obtaining the blood glucose correctly and using the monitoring systems appropriately. We need to always be using whole blood. And I dare say if we did a check just here, the people who are here today, how many of you guys are still using finger stick blood glucose sampling? I bet over half of us do. So we do. But what we should be doing is I really feel like the venous or arterial sampling, and I'm going to explain why. The other thing is utilizing those blood conservation systems, devices, hooking those up to your A-line or hooking those up to your venous line so that you never waste blood. Because we also want to be very intuitive about our patient that's anemic. So from my standpoint is that, you know, finger stick blood glucose sampling should be our absolute last resort. And the main reason for that is because the patient who is hemodynamically unstable, their model, they've got, you know, they're hypotensive, they may be edematous, you've got your GI bleeder whose hematocrit may be very low. You can obtain a drop of blood. You can put it in a glucometer and it can give you a reading. But is it giving you the true glucose level in this patient population? Probably not. So, you know, it really does need to be a whole blood sample that you're measuring. So we know that blood obtained from an arterial line, that it is equivalent. Studies have shown that it does correlate with the plasma or the blood glucose analyzer values. And we also know that blood glucose samples that are obtained from a venous line are capable. But if the nurse doesn't put all the infusions on hold, draw back the waste, draw the sample, you've just got, if they do not follow that, you've just pulled back the medications that are infusing in the line. And so many times that results in hypoglycemia occurring because the sampling of the blood has been done wrong. So we need to just validate that the nurses are doing it as they should. We need to use a reliable protocol. And we all throughout, you know, obviously utilize different protocols, but the nurse needs to know the protocol up one side, down the other. And we do know that the key to success is that the nurse-patient ratio and the staff resources that you have in your unit are the keys that are needed for glycemic management. Because patients that are on insulin infusions, there's a question out there. How often should we be checking these blood glucose levels? And I would dare say probably a lot of us feel like probably that one to two hours is probably the max that you should go. But there are some protocols that go three to four to six hours, you know, for blood glucose sampling. So, you know, you just need to know for your institution and your protocol what is out there. But also work with your nursing leadership to understand and express what does the nurse need to be able to safely care for this patient population? Because if they do have a hypoglycemic event, they're going to be checking those blood sugars every 15 minutes until they're above 70. But those surveillance checks afterwards need to be in place to be able to detect a recurrent hypoglycemic event. So the staffing, the experts at the bedside, and the support for the nurse that's caring for that patient absolutely has to be in place for the protocol to be successful. And when you're transitioning off of the insulin infusion, understanding the different types of insulin is key and when to implement those. And I think both Beth and Judy covered that very well. What does the future lead us to? Now this is something I'm not very much aware of, but I'm going to give a little insight and then I want Jamie here to kind of give the detail behind it. But the Opti-Scan was FDA approved in 2017 and it's being used in a surgical ICU population. It's not commercially available yet. But it was not designed to screen or diagnose diabetes mellitus. But what it is being is in clinical research now and research protocols are being developed at Washington University and Stanford. Is that correct? And it is a single patient cartridge that's being used. It's an automated blood glucose sampling system. And it is a sensor placed on the venous port. It has not been tested in the OR. It's for patients that are 18 years and older. If you're a pregnant female, if you're undergoing MRI or CT, or if you're less than 18, it's contraindicated in that patient population. But essentially what it does is it's an automatic withdrawal of 3 milliliters of blood and returns the remaining portion of blood to the patient. It collects the venous sample via connection to the central venous port, the most proximal port. And it centrifuges that sample, analyzes the blood glucose, and gives you a blood glucose value that is equivalent to the plasma glucose concentration. So what it does instead of, you know, us checking blood glucoses every 15 minutes or every two hours or whatever, it's giving you that continuous trend. So it allows for the bedside nurse to be able to titrate those infusions more appropriately and time more quickly. whereas now, you know, we have to wait until we recheck with a glucose measurement. Now there are some restrictions or warnings that go along with it that, you know, there's a list of them here. But I do think that this is the future of good glycemic management in the unit. I think of it like a patient that's hemodynamically unstable that's on a vasopressor infusion. And we have the arterial line that we're there, we're constantly able to monitor their blood pressure and we titrate the vasopressors accordingly. That's what I see with the ability of this continuous glucose monitoring system is that when we see those patients trending down, that our protocols will be designed to make those adjustments needed to, number one, identify the trend down to prevent that patient from having a hypoglycemic event and to keep our patients in the glycemic target range that we have for our specific patient population. These are just a few more of the warnings or restrictions that go along with the device. But I think that more research is needed to move this continuous blood glucose meters from research to bedside application. And I'm very excited to see what these studies are going to glean and how it's going to affect us in the critical care arena. I think the main thing is is that we prevent the hypoglycemia and the hyperglycemia and that glycemic variance that happens with the patient on a roller coaster. We want to take that patient off the roller coaster and put them on a steady cork going straight forward. So, as far as the role of the nurse, glycemic management is huge. The nurse needs to know that target range. They need to understand the why behind and we need to take the time to make sure they understand. They need to know each component of care, focus on the blood sampling and the glucose surveillance. We can't put enough emphasis or attention to, if the nurse is in doubt, allow the nurse to check it out, check that blood glucose if needed. Ensure that the blood glucose sample is sampling whole blood. That's very important. I think the nurses need to start learning and be aware of what's new on the horizon. And obviously, this continuous glucose, blood glucose monitoring option is definitely on the horizon. Get involved, know your unit data, and make this a process improvement if it's not already in your institution. And, oh yeah, that's just one of the million tasks that a critical care nurse manages during a day in the group here right now. Thanks. Maybe one question or I'd like to. Thank you. Thank you very much, Karen. So, I'd like to introduce Dr. Ellie Hirshberg who's going to talk about glucose targets and literature that has evolved in the last several years that is going to point the way towards our new future. Thank you. Okay. I don't have any disclosures other than I am a big fan of decision support tools, electronic decision support tools. I don't think it's the same thing as cookbook medicine. And I hope this talk helps you agree with me. So, I'm going to talk a little bit about what we know, go over hyperglycemia, hyperglycemia, glycemic variability, and time and range. I'm going to talk a little bit about what the current recommendations are, and then about personalization of glucose targets and the case for electronic decision support. So, what we know, we all know that hyperglycemia is pretty common in critically ill patients. The stress hyperglycemia is bad. It causes a deleterious hormonal milieu. We know that it's multifactorial. There are cytokines, nutrition, and common ICU therapies that all contribute to the stress hyperglycemia in our patients. We know that acute disease, as well as comorbidities like diabetes or hemoglobin A1C are going to affect how our glycemic variables are being interpreted in the ICU, and we know that sustained levels, regardless of the patient population, are associated with certain mortalities. It just depends on relative. We know that hypoglycemia is bad for patients. It's also pretty common in critically ill patients. It's more affected in certain conditions like sepsis, liver and kidney failure. The liver and the kidney are both gluconeogenesis organs, and so when they're not working, we have higher rates of hypoglycemia. We know that there's actually a difference between spontaneous and iatrogenic hypoglycemia. Not all hypoglycemia is the same, and we haven't figured out how to actually predict which is which when we're taking care of our patients, but we do know, in some of the retrospective studies, that it would appear that spontaneous hypoglycemia is actually worse than iatrogenic hypoglycemia. At any rate, we try to avoid it because we don't want to cause more hypoglycemia. There's certainly an increased incidence with tight glycemic control, so the lower target you're trying to push for the more hypoglycemia you're going to have. And there's a stepwise increase in levels of hypoglycemia and the hazard ratio of death. Persevere hypoglycemia defined is less than 40 milligrams per deciliter. It's a 2.1 hazard ratio, and with moderates and only 1.4 hazard ratio. So we know hypoglycemia is bad as well. What about glycemic variability? It's also really common in our critically ill patients. We're often on a roller coaster or a ride. The definitions of glycemic variability are different in the literature. There's some that talk about fluctuations and percentages in a certain range, and then there's some that just talk about a patient who goes up and down on a given threshold being both hyperglycemic and hypoglycemic. What we know is that those variable levels cause an increase in oxidative stress. They cause apoptosis all over the body, and that glycemic variability is actually worse than hyperglycemia in some of the retrospective studies. There's an intrinsic tendency for some patients to have more glycemic variability, and yet treatment titration can also contribute to glycemic variability, and which contribution is from our titration methods and which is intrinsic, we have not quite sorted through yet, but it has an independent increase with mortality above just high levels of hyperglycemia, and we know that it's different in different populations. So non-diabetics have a higher association with mortality from glycemic variability than do type 1s and then do type 2s. So it's an important variable that we need to be paying attention to when we talk about glycemic control in our patients. And then we also know that the time and target range is important, and unfortunately willfully ignored in a lot of the prospective studies. It's by definition patient variation within the same target. So what percent of the time is the patients that I'm trying to control in the target that I have chosen? We know there's a couple of studies out there from Chase and Prizer and Dr. Crinsley have actually helped contribute to this literature that essentially if you have an increased time, so percentage of time in target range, you have an improved survival, regardless of what your target control is. So how do we know, I ask you, all of these studies are actually accurate if we're not achieving true time and target range? There's one study that showed that actually if you have a time and target range greater than 70%, closer to 80%, you have a dramatic increase in outcomes, and it's stronger for the non-diabetic population than it is for the diabetic. And what we know so far is that most of the prospective studies, the NYSUGAR included, really did not consistently account for time and range. I would suspect that these time and ranges also vary impacted by hemoglobin A1C or the patient's personalized predisposition to stay within a particular target. There's not a lot of data out there, but there's some preliminary data to suggest that that suspicion is probably true. So what we do or don't know from 2001 to 2020, essentially Vandenberg's study, like Dr. Crinsley talked about, really showed an improved mortality with tight glycemic control. NYSUGAR, which was a 42-center, pragmatic trial, really, if you look at the fine details, it's amazing what they were able to pull off, but it's not too surprising that it was a different method, because again, it was not a single center, and the protocol for insulin titration was not extremely replicable. Everybody did sort of different things, despite it being a rough guideline in terms of what to do. So then there was the VICEP and the GLU control, which were actually before NYSUGAR, which were also stopped early. None of these, however, really reported or talked about time and range, certainly none of them achieved a greater than 70% time and target range. So I'm not sure that we really know, based on this really good data, what glucose target range is appropriate for all patients. And over on the right, I think you can see that I've drawn how hyperglycemia, hypoglycemia, glucose variability, and time and target range all seem to affect mortality, and that these four things are really the things that we need to focus on as we think about what the appropriate target for our patients are, and that clearly one size fits all is not appropriate. After 2001, everybody went to tight glycemic control. After 2009, 2012, everybody went to 140 to 180, and in 2020, I think people are kind of all over the place, less than 180, less than 150, who knows. So the guidelines have gone in favor of sort of looser recommendations. In 2011, the ACP recommended a 140 to 200. In 2012, the ADA recommended a 140 to 180, but really that's being interpreted by many of folk to be just keep it less than 180. In fact, the trigger for starting insulin treatment is two blood sugars greater than 180. SCCM sort of recommended less than 150, 100 to 150. Again, the trigger threshold being a blood sugar level greater than 150, and then in 2014, Merrick actually came out with a study that looked at hemoglobin A1Cs and said we really should be looking at targets based on what your pre-ICU hemoglobin A1C is and recommended 140 to 200 for those with lower hemoglobin A1Cs or more in the closer to the normal range, and 160 to 220 for those with a poorly controlled diabetic status. All the guidelines, however, are saying just avoid hypoglycemia, and I think as yet we really have not reached a protocol that does a good job at avoiding hypoglycemia. So we know that dysglycemia in the ICU is complex and it's a very dynamic process, okay? There's control of several different glycemic variables, and time and target range to me seems like it's the unifying metric. It seems to incorporate both hyperglycemia, hypoglycemia, and glucose variability, and likely is going to incorporate some of the personalized mechanisms. It's a big part of what our blood glucose target and how that's going to affect mortality should be, but thus far we've had a bunch of different protocols in variable environments trying to sort of jam a square peg into a round hole. So I look at it this way. We've got a bunch of things that we're observing in terms of our patients, the hyperglycemia, the glucose variability, the hypoglycemia, and the time and target range, and then we've got some things that we kind of think we could control, the ideal BG target being the first, but that's impacted by the acute disease process. We know a little bit about that, right? We're going to treat our neurocritical care patient a little bit different than our CVICU patient, a little bit different than our mixed medical, a little bit different than our surgical. We know that it's affected by the individual response, I don't think we know exactly what the individual response is just yet. We know that it's affected by chronic disease, smokers, diabetics, the hemoglobin A1C status, and we know it's going to be affected by our insulin delivery. Is it going to be subcutaneous or is it going to be continuous? And that's the only way we can determine the ideal BG target is by controlling and observing each one of these factors. So the Just Right Protocol is ultimately going to decrease the orange by maximizing the ideal target range, and it's going to affect and control all of these mechanisms and limit all of these. So Just Right Protocol evolution is going to result in a high time and target range. It's going to limit hyperglycemia, avoid hypoglycemia, and reduce glycemic variability. It's got to have high clinician adherence because you've got to make sure that every single person that's doing it is doing the same thing in order to determine, oh, yeah, A, this set of rules works or this set of rules doesn't work. It should be flexible so that essentially the same rules are being applied to an individualized target, and it should respond to patient circumstances. A lot of people who listen and hear about protocolized electronic decision support sort of argue, oh, that's cookbook medicine. But the key is that if you are informative in how you come up with the formula for calculating the response, they actually respond quite well to individual patient circumstances. And then it should really incorporate both chronic and acute disease. So e-protocol insulin is an example of such a protocol. This is an insulin infusion protocol at our institution, was generation over several years. Dr. Alan Morris was sort of the leader behind the computerized decision support or the electronic decision support. It's an infusion protocol that, oh, can I go backwards? That was right click. Sorry. That focuses on the current infusion rate. It looks at current BG and target BG. It looks at rate of change of the BG, which incorporates the individual's response. And it recommends an insulin dose and an XBG timing check. It's fairly simple, if you think about it that way. But it does respond to the changing needs of the patient. It also has the nutrition recommendations and alerts if patients are not receiving nutrition. And then it's nurse executed. And thus far, we've enrolled it out in thousands and thousands of patients, hundreds of thousands of patients and has a very low hyperglycemia rate. This is an example of what the protocol interface looks like. And as Karen mentioned, it's key. It functions on the nurse being able to understand and know exactly what to do and exactly why this is important. And the instructions are fairly clear. There's an alert that says you need to check the blood sugar now. And then it says start the infusion at three units an hour or decrease the infusion or stop it entirely or give two cc's of dextrose or et cetera, et cetera, et cetera. So that type of interface, which allows the nurse to be clear on exactly what kind of information needs to happen also allows for every single decision that is being performed at the bedside to be monitored on the back end. So it's very good for research purposes. If we look at a comparison of e-protocol insulin compared to a bunch of the paper protocols that were out there, this slide was put together in about 2010. So it was after Vandenberg and after a couple of other paper protocols came about. You'll see that the insulin infusion rates that were recommended between e-protocol insulin and the Vandenberg protocol and this variability of all the paper protocols in between is the result of physician judgment. Because most of the guidelines or most of the protocols say increase by 10%, increase by 20%, try to reach this, try to reach that, which resulted in a lot of glucose variability and a lot less time and target range. Here's an example of just kind of how e-protocol insulin works over a range of certain patients. And this is just blood glucose measurements all within a particular target range. This was at the 80 to 110 target range. And you can see that compared to the computer protocol and the bedside paper protocol or simple guideline, that the bandwidth, the number of measurements that are reaching that time and target range is just superior with reproducible electronic decision support tool. In this particular case, the clinicians don't need to add the missing detail. In both the red and the green, the clinicians are adding the detail to help the protocol adhere to the guideline. And that makes it quite difficult because we all think differently. And as Karen commented to and as we as physicians know, it's very difficult to keep track of 150 things when you're in the ICU. So an electronic decision support tool actually allows the physician to focus on or the pharmacist to focus on exactly what the target needs to be and not on the finer details of what calculating the percent change that needs to happen in terms of the dosing. Instead, the rules calculate that for you. So here's an example of e-protocol insulin. Actually, those same institutions then implemented e-protocol. And you can see, despite the environment changing, it did a fairly good job of keeping the blood sugars in the narrow band range. And actually, it was then exported to Singapore, which doesn't even, it was translated into the appropriate language. And they were able to use it quite frequently with fairly minimal just-in-time training. And I think that that speaks to the importance of the nurses knowing and understanding that the instructions that they're getting are clear and that they know what they're doing. So we thought that e-protocol insulin was an opportunity for some good retrospective database studies. There's no clinician variability or very little clinician variability because 95% of all the instructions are adhered to. It's got a reliable protocol performance across several different environments. And so, therefore, it gives us the ability to actually compare different blood sugar targets in an identical protocol. And we can look at different populations. And we potentially could stratify by personal characteristics and see whether or not the target is actually making the difference. It's a true laboratory experiment where the only variable that's being changed is the blood glucose target. So in terms of just-right glucose homeostasis, we've actually looked at three different populations where e-protocol insulin was used across the board. And the first we looked at was glycemic variation. And the glycemic variation gave us the ability to actually look at the method of titration versus the patient's individual response. We started off with 6,000 patients roughly, 14 ICUs, 43% of them had diabetes. And the coefficient of variation was associated with increased mortality. The more variation you had, the worse mortality. Done and done. This was an independent effect from the presence of hypoglycemia, which is fascinating because a lot of the people up until this point had sort of said, hey, well the reason the glycemic variability matters is because people are dropping low. But it turned out that although they were collinear, there was an independent effect of glycemic variability or coefficient of variation on mortality. The other thing that we noticed is that this effect was different between non-diabetics and diabetics. Again, we were able to look at, you could imagine that if you had a patient in your unit that you were controlling who was a diabetic versus not a diabetic, and you were making a decision based off of a guideline, you might do something different. In this particular case, all the decisions were the same, regardless of whether or not the patient had diabetes. And it turns out because the protocol is accounting for insulin resistance in its mathematical equation, we were able to see what the effect was on patients who were either diabetic or not diabetic. And it turned out that the coefficient of variation was worse in non-diabetics than it was in diabetics. An interesting fact, and that makes us wonder about the NICE-SUGAR study, and did we really look at what we wanted to look at? Because was the population way too broad to really look at where the population, where the effect was going to matter? So then we took the same protocol, and we looked at 80 to 110 versus 90 to 140. Because our institution, just like everybody else, after the NICE-SUGAR came out, changed our target. But E-Protocol insulin has the ability of, all you have to do is just change the target range. You can use whatever target range you want. So as a group, we met, and we decided, okay, we want to avoid hypoglycemia, so we'll get rid of 80 to 110, but we're just going to shift it to 90 to 140. And this was led somewhat in part by our neurocritical care colleagues who felt it was very important to control hyperglycemia in a more narrow range, but really wanted to avoid hypoglycemia. So we then did a before and after, where we looked at all E-Protocol patients treated with either the 80 to 110 protocol or the 90 to 140. 3,500 patients, roughly, 1,500 in one group, 2,000 in another group. About 40%, again, were diabetics. And it turned out that 80 to 110 was favorable in the non-diabetic population, and 90 to 140 was favorable in the diabetic population. Interesting. Again, we looked at hypoglycemia. Our hypoglycemia rates with E-Protocol insulin are actually pretty low, which is how we were able to get our hospital to buy off on not changing to a different protocol entirely. So our severe hypoglycemia rates are about 1% of patients, and that's how they were sort of okay with the 90 to 140. So it turned out that hypoglycemia also had an independent association with mortality. And it turns out that based on BG target, the hypoglycemia within a BG target did not affect hypoglycemia. Now, that may be because our hypoglycemia rates were so low, it's very hard to sort of determine causality. But I think a lot of people make the jump that all of the harm in blood sugar control is off of low blood sugars or hypoglycemia. And I think that this data helps prove that that's not necessarily the case. So we learned a lot about BG target from that. Then we did a time and target range at 70 to 140 milligrams per deciliter, up to about 9,000 patients. We collaborated with Dr. Crinsley on this. And essentially, 84% of the time and target range in the non-diabetic patients. And if we were achieving greater than 80% of time and target range, we had a significant decrease in mortality, particularly more profound in the non-diabetic population. The diabetic, about 50% of them did have diabetes. And we were less good at achieving our time and target range there of the 70 to 140. But if the diabetic patients were in the target range greater than 80%, they also had a decreased association with mortality. It speaks that in terms of how, where, why, and what essentially the diabetic patients are responding differently than the non-diabetic patients. And how we decide where to treat them going forward, I think Dr. Crinsley has really added a lot of good information. So from e-protocol insulin, we were able to look at individual response, the effect of diabetes, yes, no, and the comorbid condition, and how that impacts the BG target. And then we lastly looked at the acute disease process. We looked at a group of patients on cardiovascular surgery, essentially 80 to 110 versus 90 to 140. Again, because we've been using this protocol for upwards of 15 years, we were able to go back and see, okay, how many patients were managed in the 80 to 110 era and how many were managed in the 90 to 140 era. You can see that we achieved a pretty good separation in our blood glucose targets, 104 in the 80 to 110 group and 122 in the 90 to 140 group. They were all pretty much cardiac ICU patients with low hypoglycemia rates. The less than 60% was like .34% and the less than 40 was 1.16%. And we found that an 80 to 110 target was associated with significant benefit in both the diabetes and the non-diabetes group. And this speaks to the different population, right, because I've just gotten through telling you that, oh, well, sometimes diabetics, they can handle a little higher range. They do better in the 90 to 140 range, but maybe not if they're having cardiac surgery. We know that the myocytes, when bathed in high blood sugar, it actually impacts their ability to squeeze appropriately. So e-protocol insulin and clinical care, up until this point, I just kind of asked my unit director to run essentially what we've seen from 2017 to 2020. And he has access to all of the data from all of the 14 units. And you can see that from the last three years, we've taken care of 9,311 patients, over 140,000 measurements. Our percent measurements in range is 50%. That's not time and range because you have to look at the amount of time. And so the time and range is probably a little bit higher but we didn't have time to actually calculate that. And our percent of patients less than 60% is 3.5%. Our percent of patients less than 40% is 0.32%. So this is e-protocol insulin is a detailed and reproducible instruction, set of instructions. This lessens the cognitive burden on all of the clinicians at the bedside, your pharmacist, your nurse, and your physician. It's got consistent rules for monitoring. And those rules change based on the rate of change. And it allows for clinician alerts. So what we did find over the years when we started doing it is that almost all of the hypoglycemic events occurred with trips to scans. And that's because people were forgetting to turn things off. So by putting these alerts in place and allowing the protocol to be actually nurse run instead of clinician run, we were able to reduce our hypoglycemia rates in the early stages. So I hope I have convinced you at this point that as simple as it may seem, the control of blood sugar in the ICU is really not a simple thing. It does take a team. And it takes a very good, wise protocol that is reproducible. We want it to optimize all of our time and target range, as well as look at our individual responses so that we can come up with the ideal target range. And it doesn't have to be e-protocol insulin. But any kind of protocol that's reproducible and that takes the cognitive burden away from the bedside clinician I think is going to be effective and helpful. And help us move towards a more personalized blood sugar target in medicine that will hopefully reduce mortality across all of our patients so that we can now say, oh, this patient has this pre-admission hemoglobin A1C with this set of acute diseases and this kind of insulin delivery response. Here's what target I'm going to choose for them. So a decision support tool is just that. It's support you at the bedside and free you up to make those very challenging decisions of what target I want to choose. It's explicit and it's reproducible. It's an automatic part of workflow. So it's easy for people to use. And it's an integrated team approach so that essentially we can treat everybody as though they're Goldilocks with the appropriate blood glucose target. Thank you. Thank you for that outstanding presentation. What you didn't hear was 140 to 180. Intermountain does a fantastic job with blood glucose control and they've been on the stricter side of glucose control over the last 15 years that I'm aware of. They use the e-protocol system as an electronic decision support system. But there are three commercially available systems that you should mention, well, you can mention that are available for, you know, as a software application that your unit can purchase. And that might be helpful to you. In terms of targets, I think as a take home point, I'd like you to go home thinking about the status of pre-admission glycemia and the importance of perhaps thinking of different targets for people who are well controlled when they come in. Personally, if you have a normal A1C, 80 to 140, I think is a reasonable broad enough target to shoot for. For diabetes patients who are poorly controlled, I don't know what the right target is yet. I really don't know. I mean, we're doing 120 to 180 right now. Maybe it should even be higher than that. We don't know. Like it might be trading off mortality against infection. So. I think what we do know is you need to avoid variation. Avoid variation. Avoid hypoglycemia. Measure frequently. Measure frequently. And have a data system in place where you know what your glucose outcomes are. You need to know what your hyper rates are. You need to know. You need to be able to create a histogram of your values on a regular basis. Otherwise, you're driving this car blindly. I mean, you have to know what the glucose values are in your ICU on a regular basis so you can, you know, so you can make a change. That's how you get nurses to make changes at the bedside and clinicians is you show them the data. I just want to ask you both because you have so much experience. I want you to put your future thinking hats on. Because I've always been intrigued. Like you're talking about this time and target. What's our target? And that we need to personalize. I'm wondering if when you have these e-protocols and you're having all this data, if we will be able to do that side algorithm that will do a little machine learning for a specific patient so that we may be able to have moving targets by the patient. Like that patient's own personal target may be different from that patient's personal target, right? So that we can look at what you're already starting to kind of look at in your e-protocol. But do you think that's going to be in our future that we'll be able to personalize targets, you think, with some machine learning? That's a great question. I think we need to get to closed loop efficient management before we get to that degree of machine learning. There's, and I guess what I would say is that the piece that we as clinicians can still provide is knowing the kind of flying the plane, right? It's really very similar to a pilot. I'm going to set the destination. But then there's going to be all these factors that are happening in the plane that are helping me reach that destination. So I don't know if AI is at the point where it can choose the destination for a person. But I think we can choose the destination, program it in to a sophisticated decision support tool. And that it doesn't have to be the same for every person. Because you don't need to fly to the same destination at every time. Yeah. I think our future is multiple targets in the ICU. I don't think there's any doubt. And the other part of your question was, will that target be changing as the patient recovers from acute illness? Fascinating question. I think that might be a little bit further off. Yeah. Hi. Greg Manky. I'm a vendor working in this space. And first of all, I commend you as a great presentation. Thank you. So I have a statement and a question. So the statement is a follow-up to Dr. Crinsley's comment. There are four other, there are four commercial products out on the market. And they have all been deemed as a class two medical device, non-exempt. Which means they all have had to file a 510K and get clearance from the FDA to be used. So my question is, what's the status of e-protocol relative to regulatory affairs? Well, we have not used it for commercial purposes. So we did get an IDE to use it for experimental purposes. And it gave us approval for that. And then our, within Intermountain Healthcare, because it was a homegrown system, we didn't really need to get any other proprietary approval beyond that. Okay. I understand. I just want to make sure the audience understood that as well. Thank you. I just want to make sure that Jamie addressed the OptiScan and what knowledge you have about that. They're not here. They're not here. Okay. OptiScanner is a pretty interesting device. It draws blood out of a central line, puts it into the box, centrifuges it, and it doesn't use a chemistry analyzer. It is a mid-IR spectroscopy device. So you have a spectrometer inside a box at the bedside. Sends most of the blood back to the patient. You get an output every 15 minutes. So you get a glucose reading every 15 minutes. The two institutions, Washington U and Stanford are doing research projects on it now. It's not really commercially released. It's only approved for surgical patients because the landmark study was a 200-patient study that was multicenter, and 160 out of the 200 patients were surgical ICU patients. So they didn't clear it yet for medicine, but that's coming down the pike. The other interesting thing about this device is because since it's not a chemistry analyzer, the spectroscope can measure other analytes. And so the FDA is about to make a decision. We, the company believes it to be favorable on lactate. So now you have glucose and lactate every 15 minutes. You don't necessarily need lactate every 15 minutes, but you could be checking it intermittently. And after that, the company plans for hemoglobin and mixed venous O2. So glucose, lactate, hemoglobin, mixed venous O2, well continuously every 15 minutes, 72-hour cartridge. You plug it in. It doesn't have to be calibrated after, for the entire 72 hours that it's running. So it's really a unique and exciting technology. I wonder what all of you are doing relative to the nutritional support when it's held for procedures and meds. Are you ignoring it? Are you pre-calculating your nutritional support to account for it? Are you doing catch-up? And then associated with that, what are you doing with the insulin? So I can speak to the e-protocol insulin. Actually, before you started and at every interaction, it asks you A, what your nutrition is, and B, if there's a change in the nutrition. So if you're stopping the nutrition for a study, it will stop the insulin and prompt you to check a blood sugar in 30 minutes. And then same thing when you're restarting. It'll make sure that you've restarted either at the previous rate. Or if you're anything less than 50% of the caloric amount that you were doing before, it'll wait an hour before it recommends an insulin dose. And if you adopted a practice. there. It's a detail do you catch up? At our institution, we specifically developed what's called a nutritional change protocol. And so if that patient is on two feedings and is going to be off, whatever the two feeding rate was going at, if the sodium is greater than 135, we hang D5 at that set two feeding rate that the patient was going at. And if the sodium is less than 135, we do D5 normal saline at that rate. If the TPN is taken down for any reason and we're not hanging D10 back, we do the exact same thing. We hang the D5 or the D5 normal saline at that TPN rate. So therefore, the caloric intake, you know, does not change. And that way, you prevent the patient from having a hypoglycemic event just because his nutritional status has changed. As soon as, you know, they start back the two feedings, immediately the infusion of D5 or D5 normal saline is discontinued. And we haven't published that data, but the incidence of hypoglycemia around those nutritional changes drops significantly whenever we implemented this process. And you're not doing anything to catch up with protein time that was going on? No. Made it very simple for the nurse to know, okay, my two feedings are going off. I'm going to this procedure. I hang either D5 or D5 normal saline at the infusion rate of the two feedings that was there. You know, something very easy that can be replicated over and over and over that the nurse doesn't have to think about calculating, you know, oh my gosh, which fluid, what amount, what formula, you know, just something very simple. And we do that, too. It's just that when we do start, we do start with 10, I think. We have to do it. Any other questions? Thank you very much. Thank you.

nurses

glycemic management

target ranges

components of care

surveillance

blood glucose levels

hypoglycemia prevention

insulin protocol

expected glycemic outcomes

ICU challenges

personalized targets

electronic decision support tools

improved outcomes

machine learning

glycemic control