Video Technological, Behavioral, and Practical Aspects of Using Real-Time Glucose Alarms Play Pause Volume Quality 1080P 720P 576P Fullscreen Captions Transcript Chapters Slides Technological, Behavioral, and Practical Aspects of Using Real-Time Glucose Alarms Overview Hello, everyone. My name is Earl Hirsch, uh, coming from the University of Washington in Seattle. And I am honored today to be able to talk to you about optimizing and individualizing alarm based notifications to minimize risks of hypo and hyperglycemia. This is a topic that has been of great interest of mine over the years. We've had C g m. And as you will see, a lot of what I have to discuss today is opinion, because there's not as much data as we would like. These are my dualities and let's get started where to set C G M alarms. And this is not a new dilemma at all. In fact, this I think, very interesting. Article title. Turn it off. Diabetes device alarm, fatigue, considerations for the present and the future. Note that this comes from 2013. Many of these authors are well known names in diabetes technology, but this is not a new problem. So understanding this very delicate balance give you an example. There is an alarm that is set for an individual at 90 mg per guest leader instead of 70. And when you have an alarm set that high, you're going to have more true positives. That is, when the glucose levels get low, it is almost for sure going to be low if you're going to have an alarm set at 90 so the sensitivity is increased. But there is also more negative. And this is the problem that when that alarm goes off at 90 it is unlikely more often that the blood sugar is really low. Um, and some of this is obviously based on the accuracy of the sensors, which in the early days was a bigger problem than today. But it's important to note that more alarms going off when the glucose is actually safe and stable. This has been historically a major problem for the arm fatigue, which in the days of the less accurate sensors was a major issue. But I will tell you it is still a major issue today with our better sensors. So this is a delicate balance. Sensors receiver Operating characteristics curve reflects the fact that as more true offense of hypoglycemia are detected, there will be more false alarms on top of that, which is why where one puts that alarm is not a trivial decision. The problem is that people respond to relatively fewer alarms if they perceive the false alarm rate is too high now. This doesn't come from the diabetes literature. This comes from a an engineering article looking at alarms, and so where one has the alarm. If you're having a lot of false alarms, people stop paying attention. And I think I still see this every day in my clinic. And it gets worse because frustration with sensor accuracy has resulted in many patients patients to stop using C g M altogether. And this, of course, was more common in the earlier days of C. G. M. I don't see people stopping now like they did a decade ago. Still, over the years, alarm fatigue has been a major problem, and we do have data about that again. This is not a new problem. Note the date here. This was 11 years ago in 2010. This comes from the JDRF continuous Sensor study. Now, this was a study that we were part of here at the University of Washington, and we published our data in the fall of 2000 and eight. But when the authors went back to look at the data what was shown was that alarms going off too often was a significant barrier for both youth and adults in the JDRF sensor trial. Again, these were the earlier days of C g. M. But we still see this today. Why is this important now? Fast forward. This study comes from Just last year 2020. This was a study with 85 Children and adolescents. Mean age of 13.5 years mean a one c was 7.8%. Why do why does this population stop? C g m well, when one looks at the percentage of answers for strongly agree and agree, one sees that alarm fatigue is the number one reason along with inconvenience why this population with an average age of 13.5 years Um, why they stopped cgf. So this is not from 2010. This is from last year. Alarm fatigue is still a problem. Another, uh, discussion about alarm, fatigue and youth, their parents. But in this study also adults, this comes from 2011 and here the investigators were looking at the benefits and the barriers of C g. M. The dark bars are the youth, the white bars are their parents, and the gray bars are the adults, and you can see why they like the benefits of C. G M here and all the different reasons. But I want to look at the barriers for a moment. And when you look at the alarms, what you see is that for the youth and the parents, that is the number one reason. About over 30% about a third of the time. The alarms are a barrier, interestingly, not as much. So for the adults now, one has to also realize that as much as we have tried to standardize everything we do in C. G. M, this is certainly not standard, especially when we look at the alarms. Let's talk about the Medtronic guardian. First. One can set the alarm between 50 and 90. There's no hard alarm. Preset can also be alerted for falling rapidly, which is called the follow alert or approaching the low limit, called the Alert before though similar high alarms for hyperglycemia as what the Medtronic sensor shows for hypoglycemia. And there's the decks Come here. There is a hard alert that patients can't change it set at 55 There is an urgent low soon, which is optional, which predicts hypoglycemia less than 55 20 minutes. There is also a low alert that patients can set between 60 and 100. And then there is a repeat alert between zero and four hours. The high alert has a range of 1 20 to 400 high repeats are similar. And one of the things that I do with I would say every single patient is I look at the alerts on this sensor and I'm curious what I see. And it's very interesting how patients may turn alerts off because they're in a work situation there on a zoom call. They don't want their alert to go off, and then they forget to turn it off. And the other real common thing I see is the alerts are on, but the alert repeats are not turned on, and this is especially an issue in the middle of the night. If someone doesn't hear it the first time, there's a good chance they will hear it the second time. So there's a there's a whole science behind this that I don't think we discuss enough and now Of course, we have the liberty to with alarms and with the library to, it's an option to keep the alerts on or off. The low alert is set 65 to 100 the high alert at 1 80 2 400 mg per deaths leader. So all three of the major sensors have these alerts. So how does one personalized this? Well, it's a critical discussion topic with the health care provider and the patient, and it needs to be reviewed. And it's sort of an terminology that, um maybe is overused, but it's extremely true here. It has to be personalized, and it has to be personalized on so many different issues in terms of that person's glycemic targets their risks for severe hypoglycemia and so forth. But it needs to be reviewed regularly because people's clinical situation changes and we'll go through what some of these issues are. And as I noted, patients will often change or turn off the alarms and forgot that they did so depending on which sensor they are using, will need to either observe on the report or actually look at the reader itself so you can always review these alarms with the patients. So what are these factors? Will age? I mean, a lot of the teenagers do not want to be bothered with the alarms and many of these teenagers who have not had problems with severe hypoglycemia. Having the alarm is not as critical or at the very least, as an example. You could keep the alarm at a lower rate, depending on the person duration of diabetes associated with age. The longer ones duration of diabetes, the less they are going to be able to appreciate their hypoglycemia. And for somebody, for example, with long durations of diabetes, Um, you will want to have the alarm's set at a higher level. Ruth Weinstock published in 2013 that after 40 years of Type one diabetes without C G M, one can see about 20% severe hypoglycemia per year per year, and that comes from the T one D exchange. So one has to look at that. Yesterday I saw an individual with 71 years of Type one diabetes 71 years. That person is definitely going to have a high alarm. Um, if you have somebody who's not taking insulin or a cell final, Yuria Well, then, the alarms are not critical at all. History of severe hypoglycemia, the alarms and higher alarm. Um, level, um 80 or 90 is not unreasonable for some of these patients. Um, glycemic targets, um, where we see more hypoglycemia. But we also see, um, the need to keep the blood sugars in the normal and low normal range pregnancy. I see Mick targets is actually something I've thought a lot about. But I know of no data that actually suggests what should that be? And we'll do a case on this in a moment. And then, of course, there are professional and social challenges. There are many patients who don't want people to know they're wearing a sensor. They don't want people to know that they have diabetes. Um, but not everybody is like that. Former Prime Minister Minister Theresa May. She warned her library proudly on her arm. Um, this was a photo that she had done. And this is the shoulder. I can tell you. This is the shoulder of President Donald Trump during this, uh, photo op. Other factors to consider for high alert targets. Well, of course, if somebody has higher targets because they've had their diabetes for 40 or 50 years. That's going to impact both the low and the high alarm, professional and social situations. As we noted, Key point of taking mealtime insulin. And this is my opinion. If, for the high alarm alerts, one probably requires insulin with that alert, especially if greater than one hour after the last meal. Now there's a huge controversy with my colleagues about this in that if one gives a bullets of insulin with a pump or an injection of insulin with one of the mealtime insulin and the alarm goes off within a few minutes or within an hour, should you wait at least until the hour to see what happens? And and it's interesting because sometimes somebody may have just miscalculated the amount of carbohydrate, especially when they are eating out as a rule of thumb. I think that for the most part, you should wait an hour before you start giving insulin, because by definition, as we will see in just a moment, you are stacking the insulin. Having said that, if after an hour and certainly after two hours, the point of the high alarm is that when that alarm goes off, you are going to do something about that. And usually what that something is is give insulin Now. I saw a patient, um, earlier this week and his alarm hardly ever goes off on the high side. It did go off after an anniversary celebration. He does not take insulin. And what he did was he went out and took a two mile walk at night in Seattle. I hope he lives in a good neighborhood. I don't know about that, but he did, and his blood sugars came right now. So the alarms aren't critical just for people who take insulin. And I see a lot of people who do it this way. So the trend arrows are also important when thinking about these high alarms, because if the alarm goes off and the arrow is trending upward, you are more likely to need a correction dose of insulin for people who take Crandall insulin. And then the question is, do I need to wait an hour? Well, it also depends on where you have that alarm set. If you have an alarm set at 300 mg per guest leader and the alarm goes off 30 minutes after eating. It's very safe to take extra insulin, I think. But if that alarm goes off at 1 80 not so much. And so it's hard to give specific recommendations, given the variability in the patients and the variability of how we have the alarm set, the alarm repeat, which is not seen on all the sensors. Um, you don't want to have it too frequent, but you don't want to have it off either. And what I do on this one on the high alert with the alarm repeat is I have it at three hours. Let's say the alarm is on at 2 20 and yet the blood sugar is now at 3 10. If you have it going off every 30 or 60 minutes, the patient is going to get alarm fatigue very fast because it's going to take hours to get that political glucose level down to below 2 20 in that example. So again, this is a science and all these different parts that have to be considered, So let's do a few cases. This is an 18 year old college freshmen type one diabetes for four years on multiple injections, a one C is usually around 8%. Never a history of severe hypoglycemia. This is a very active individual place, intramural basketball, soccer and baseball. Long discussion about alerts for his c g M. He does not want to use any alerts, but has agreed to use a low alert only. And with this history, what level for a low alert, would you start N. Y? And this is This is a rhetorical question. Um, he plays a lot of sports. Um, he's a freshman, Um, and he's a freshman. He's only had his diabetes for four years, so, um, he's not at a huge risk, and he's never had a risk of severe hypoglycemia. But I want him to have his alarm on. But the other thing I know that goes without saying he's an 18 year old college freshman. So the other concern I have, of course, is alcohol, alcohol and sports. And insulin is an interesting combination, and so one could make an argument for almost any level. But as a rule of thumb, my default alarm for most patients like this would be 70 because at a level of 70 you have time to treat it before you get into any real trouble, but it depends on the patient. Depends on the doctor, but that's how I would think through this one. What about this woman? Her name is Sharon. She's 28 with Type one diabetes. Now she is planning for pregnancy. She has just started to wear her C G M a month ago, but she's been wearing a pump for many years. She last required glue gun when she was 14 years old. She's now 28 so you don't want to get too cocky about it. You know, she she has not had any recent severe hypoglycemia, but she's now in a different world as she's trying to bring her blood sugars down to normal. Her most recent, a one c, is 6.8. She's new to see GM, but you see, on the last 14 days her averages 1 42. Her percent percent coefficient of variation is 34 a half, and her g m. I matches her a one c, which is, which is a good thing. She doesn't have any discordance there, and so this is where she is. And the question is, besides adjusting her insulin dose and improving her diet. What do you propose for high and low alarm and why? Again? This is sort of rhetorical question. Um, I see pregnant women who want their alarm, their high alarms set at 1. 40. Now, I will say anecdotally, and I don't know if this has ever been studied that I don't think we see the same alarm fatigue and a woman planning for pregnancy or a woman who is pregnant. But if her average is 1 42 I don't think you want to start start her high alarm at 1 40. You want to put her high alarm at a level where she's going to do something about it. It's also going to teach her about food, since she's new with C G. M. Um, I would probably put her high alarm at this point, knowing her average is going to come down. I would probably put it right now at 1 60 maybe 1 80. But it's a discussion you have to have with the patient. The low alarm, Um, we call time and range in pregnancy 63 to 1 40 based on our consensus conference, I think, at least at first and again she's new to see GM. I would put the alarm at 70 and get her used to the alarm. Um, but again, there's no right or wrong answer for this particular question, but it really deserves discussion with the patient. What about this gentleman? This actually just happened. So 50 54 year old man, Type one diabetes for 4.5 decades, he has never wanted to wear a pump. He takes insulin, Degla deck and a spurt. He uses pens and he comes in emergent Lee to see me with his wife due to frequent recent severe hypoglycemia. Despite wearing a C. G. M. I will tell you he is not hard of hearing. We do have patients who do have problems with their hearing. This is not him now. His low alarm was set at 80. He has no alarm. Repeat on his on his sensor. What would you suggest for his low alarm settings seeing this and is there anything else you would suggest again? This is what this is one of the problems with these virtual recordings. We can't have discussion or chat, but what I did was I turned his low alarm up to 90. Number one. I obviously turned on the alarm repeat for him. But the other thing I did was that I prescribed for him a smart pack. Um uh, right now we have just one on the market. We are about to have three. I'm hoping later this year, but for a way for him to stop potential insulin stacking to see what's going on with the insulin disappearance curves. Um, I wanted him to have that he has not had it yet. And the other thing I will say is that, um, I did have a discussion with him about automated insulin delivery, but again, he was not interested in going that route. Now this is Jane. This is what her sensor looks like. She's 34 with type one, trying to lose weight with lower carb diet and more exercise. Her insulin to carb ratio is 1 to 15. Her sensitivity factor is 50. Her target is 1 20 she often exercises after dinner after her evening exercise. And this is what was shown after evening exercise. It's about midnight here. What would you suggest? And obviously, um, she needs to eat and and this is very scary, getting ready to go to bed with a downward arrow and an 86 blood glucose. Um, and and it's interesting because, um, we don't see here what's going on with her basil insulin. Um, but as a rule of thumb, um, Basil, insulin is an entire other discussion, depending on if you're on multiple injections. If you should change it, if you should change it long term. If you change it short term as an example. If this woman is on bedtime guarding, maybe at least just for this one night and maybe consistently she should go down on it. You can make adjustments on a daily basis with charging. You can't do that with Dagbladet as an example, or even you 300 guarding because of the long half life. But this woman needs to eat and what the big concern is. When patients see the arrow and the lower blood glucose is, they overeat, and then they end up getting an alarm overnight because their blood sugar is high and that happens a lot. But what she really needs more than anything is she needs less Crandall insulin. If she's exercising after dinner, and this is something that his easy to do. But it's not. If the exercise is spontaneous, because if you're exercising with the big insulin load after, um, eating, this is going to happen each time you do it. And the rule of thumb is, um, if you know you're going to exercise as a place to start and it is admittedly trial and air, you may need to go down by about 50 50% on the mealtime insulin. What about this gentleman? Peter? He's 45. He's Type one open loop pump. His insulin to carve is 1 to 10. Sensitivity is 40. His target is 1 20. His blood glucose before eating was 1 35. He gave five units of insulin. It's now 45 minutes later and remember, remember what I said. As a rule of thumb, we generally don't give insulin if it's less than an hour. More importantly, the glucose isn't all that high. It's certainly within target, and I think more importantly, this is the reason why you wouldn't have a high alert at 1 50 or even 1 60 because there's a rule of thumb when you have that high alert. You want to do something about it? Here's the same patient, same postprandial glucose. And now he's 45 minutes again after eating. Now, would you give any extra insulin given now that he has this arrow going up on his Lee Brae? Now it gets a little bit more tricky. Should you give insulin the, um, as a rule of thumb, I would say no. Let's at least wait for more of that insulin to dissipate. But here we are again, the same postprandial glucose. Now we have an arrow going straight up, and I think before we answer this, we have to think, What do these arrows going at 45 degrees compared to 90 degrees? What do they mean differently? Well, um, here's the same patient. And now instead of 45 minutes later, he's two hours later, and I think most people two hours later. Yeah, we are now going to give insulin, but the point is, you have to figure out how much insulin to give. And you have to know what these arrows mean and you have to know about insulin dissipation curves. So, first the decks, com trend arrows and what's important and I'm not going to go through. All of these is that the trend arrows are different than, for example, Medtronic or from the freestyle library or library, too. Um, and it's the library to, of course, that has, uh, the alarm. But note that you can have two arrows, which means it's increasing by 3 mg per investigator per minute or two arrows going down with the same thing. But if we now look at the library, you get one arrow up or down, which is greater than 2 mg per guest leader per minute, and there are no two arrows up. It's only one arrow up or down, but you also have the 45 degree angle arrow, um, going up or going down. And let's get back to our patient now because this was published in an endocrine society journal a few years ago, it's really a consensus on how to deal with these trend arrows thinking about our patients. Now this whole table is dealing with labour a arrows, assuming the patient is more than four hours after eating when that insulin is dissipated. Now, if you remember, our patient had an insulin sensitivity factor of 40. So what that means with one arrow going up, um, is that that patient needs 2.5 units of insulin, But there's a big caveat. This is assuming more than four hours post meal so you don't stack the insulin. So the calculation suggests 2.5 units at four hours. But in our patient at two hours after the meal, our patient Peter and five seat I would only give a fraction of this insulin. I wouldn't give the whole insulin because you're going to be stacking the insulin with that arrow going straight up again. This is just for the library now, for the other situation where there was an arrow going at 45 degrees up. This is the recommendation at four hours. So at two at 45 minutes, I probably wouldn't give any insulin because there's still so much insulin on board. Now let's look at Linda, 71 year old trainer, yoga instructor. She's on deck to deck and lies pro. No severe hippo in 30 years. Her glucose levels are flat overnight on the basil insulin Degla deck. She's quite active. Her C g M. Four hours after breakfast shows this. Okay, that looks great. She now plans a four mile hike up a steep grade. It's two miles up and two miles down. I've done this hike. This is Mount Side right outside of Seattle. This is a very aggressive hike, and the question is she decides not to snack it the at this point before the hike because it's four hours since her breakfast and her insulin and her trend, as you saw, was flat. Do you agree with that decision? And the answer is, it's the wrong decision we put in insulin action times and all of our bullets calculators. But let's look at the actual data. This is the data that most people remember. This is the pharmacogenetics data. This is simply measuring insulin in the blood. So if you're measuring insulin in the blood with a typical rapid acting analogue, this one happened to be lice pro. You can see it peaks in an hour. At two hours, you're pretty much gone, and at four hours you're completely gone. And regular insulin, of course, is a much longer far Mikel kinetics. But I'm not interested in when the insulin is in the blood. I'm interested in Once the insulin gets to the blood and it finds an insulin receptor, and then it finally binds the receptor. And all these post receptor, um, challenges have to occur for the glucose to get into the cell. That's what I'm interested in. And that's the pharma co dynamics of insulin. And this is a You guys seem a clamp where we are measuring how much glucose needs to be given when 0.2 units of insulin are given. In this case, it was a spurt and regular insulin. Now, this is important because this is the true insulin activity. Um, not the kinetics, which is just measuring the insulin up. And what you see here is after two hours, you're having your peak of insulin activity, and after four hours, you're having quite a bit of activity. Still, with glucose infusion having to give quite a bit, and you still have activity here six hours later. And in fact, when we first started using bullets calculators, this was the only choice. We had six hours because that's how long this lasts. Um, and even if you put in 23 or four hours into your bullets calculator whether it's a smart pen or a pump, it doesn't matter what you put in. This is still what it's going to be, Um, and that's a whole other interesting discussion. But for our patient Linda, the yoga instructor, at four hours, she still has quite a bit of insulin activity from her mealtime insulin. So if she does this hike without a snack, she is going to get hypoglycemic. And it's not because of the basil insulin. It's because she still has insulin on board from Herp Randall insulin. So with all of that, I would like to conclude that the trend arrows and alerts on C G. M are important, if not critical, for the best outcomes in preventing hypo and hyperglycemia. The levels of the alerts need to be individualist with consideration for not burning the patient with unnecessary alerts. General understanding of trend arrows are important. Yet understanding the details and the pharma co dynamics of the insulin can be of great assistance to the patient, as we saw in our last lady. So with that, I would like to thank you very much. These are the These are the mountains we see this time of year This These are the Olympic mountains, overlooking a very cloudy Seattle, and I want to thank you all for your attention. Published March 31, 2021 Created by