Hello everybody. My name is Earl Hirsch. I am a professor of medicine at the University of Washington diabetes institute in Seattle Washington and I'm happy to invite everybody to our discussion today about improving overall resource utilization, reducing costs and complications and improving glycemic metrics with sensor based continuous glucose monitoring. These are my dualities. And the first question is do our home monitoring tools improve outcomes and costs? And I'm very interested in the history of diabetes and when thinking about this, Going back to the 1930s and 1940s, nobody thought about home monitoring tools and how it improves outcomes and costs. This was one of the first chemistry sets called the Chef Tell um urine glucose testing kit, which was literally a chemistry experiment patients had to do to test their urine glucose back decades ago. Eventually in the 19 forties and fifties, the urine clinic test was introduced. This was much more convenient and then home blood glucose monitoring in the 19 eighties. This was with the kEMP strip that you could read off the bottle. And then of course the large introduction of finger stick glucose testing. But until the really the two thousands, nobody paid much attention and to improve outcomes and costs. And so during the during last year's 1/100 anniversary of the discovery of insulin, do we have data to suggest that these older tools had an impact on acute complications and costs. And as it turns out not really, the answer is no, but that certainly has changed. When we look at our newer tools of continuous glucose monitoring. Now in the United States we have four different ways that we can look at glucose levels continuously. And we're going to talk about some of the data about outcomes and costs. Well the first question is with continuous glucose monitoring or C. G. M. Is will it reduce visits to the E. D. And I want to first show this data from 2020 Jama network Open looking at the association of multi morbidity glucose control and medication use with hypoglycemia related E. D. Events. This was a study of over 200,000 adults with diabetes. You can see that on average they were in the Medicare age group A one CS were pretty good at 7.2% over 96% had type two diabetes and almost 70% of them were on insulin with or without. So finally areas and that makes sense if we are especially looking at hypoglycemia. Now this is somewhat old in terms of when the state A. Was generated in the years between 2014 and 2016. But I want you to look at in this first graph Is looking at the age for hypoglycemia related E. D. visits and hospitalizations. We see that between 18 and really 60 for it stayed flat but then we had this huge pivot upwards After the age of 65 and even more E. D. visits and hospitalizations after the age of 75. And like we see in so many outcomes in diabetes and all of medicine that we see this relationship in socioeconomic and in this particular case economic issues with annual household income and specifically the higher the household income, the lower the risk of an E. D. Visit from hypoglycemia. Now in this next graph we look at the rates stratified by total co comorbidities. This again makes sense to me. The sicker you are, the more likely you have of having an E. D. Visit or hospitalization for hypoglycemia. But the one that is most interesting to me is the rates stratified by hemoglobin A. One C. And we see this over and over in the literature especially in the type one literature where we see this U. Shaped R. J. Shaped curve where we have the most hypoglycemia at the lowest A. One C. Levels. But then as the A. One C levels go up, we also see a greater amount of hypoglycemia speculating now. And this is pure speculation for these high levels of A. One C. It is thought that when they do check their blood glucose and they see these very high levels they give insulin but they are not checking their blood sugars frequently. So they give a large dose of insulin and they miss the fact that their blood sugar is getting low when they see the very high numbers the low the very low levels are a little bit more self explanatory except for the fact that with the very low levels these are also people who have more comorbidities with chronic renal disease with anemia from any ideology with advanced liver disease and so forth. And so these are people who may have low A one C's that are falsely low because of these comorbidities plus the comorbidities of themselves. That also makes sense to me. But we don't know that for a fact. Now not surprisingly having type one diabetes resulted in a 34% increased risk of hypoglycemia related E. D. visits or hospitalizations. And that makes total sense because these are individuals who are completely insulin deficient and it is more difficult to control their glucose levels. They have more hypoglycemia and more glucose variability. Now if we look at this in terms of resource use utilization. This is a actually a complex side looking at the resource use in $2013. This was a publication from 2018 and we're looking at this from the health care professional the E. D. The hospital point of view and total. And I just want you to look at the total and specifically the gray bar. The gray bar is looking at severe hypoglycemia and note That each episode of severe hypoglycemia when looking at all three aspects of healthcare provider E. D. In hospital. It's over $15,000 per event. Now that was a long time ago and we know that number is much greater now. Now. We've been focusing so far on hypoglycemia. What about ketoacidosis and hyperglycemia emergencies? Well we don't have as recent US data as I would like. This is from 2020 and we see the data from 2015. And whether we look at E. D. Visits for D. K. D. K. A. Or inpatient visits, what we see is between 2009 and 2015. We see this increase frequency Now we don't know what this looks like during covid. I'm hoping we get that data soon from the U. S. But you can see that this number had increased. And we know one of the reasons for the increase has been the expensive insulin. That's that topic always comes up. But I want you to look at this hospitalization in US population by age group, in this youngest age group. And this comes from Jama in 2019. In this youngest age group. The black is specifically looking at hyperglycemia admissions to the hospital. And we see this bump up in this 18 to 44 year old age group. We also see the bump up but not as great of a bump up in the 45 to 64 year old age group. But the bottom line is between 2009 and 2015. We saw hyperglycemia crisis whether in type one or type two diabetes. The increase was 81%. Now why would this be with all of the health care dollars we spend in the us. Why would we see such an increase? But this number is consistent in all of the studies that look at this. Now look at this when we look at specifically keto acidosis. This also comes from 2018. And what you see in the orange line is we see this increase in cost. We see an increase in cost. But what we see with the blue columns is a decrease in length of state. So an increase of costs and a decrease in length of stay. The cost actually went up to over $26,000 per episode. But even though we see this decrease in length of stay, the numbers went up. We went from 118. Almost 100 and 19,000 people in 2003, people in 2014. So we have an increased cost and an increased number of people and that makes the length of stay. Obviously not as important when you look at these numbers. So let's do the math with this. And I can convert this to $2021 In 2014. There were this many admissions from the previous side that's how much it costs. $5.02 billion us. That was in 2014. What I then did was converted to $2021 using a medical inflation calculator. I'm a little concerned what it's gonna look like for the 2022 knowing what's happening with inflation right now. But here's the bottom line it was $6.2 billion. If we convert that 2014 number that is how much money is being spent on D. K. A. In 2014. Now We're eight years past that now. And so we know that the cost of this is well in excess of the $6 billion and and that's you know I can't figure out the amount of money that is because once you get into billions of dollars that that is actually a uh a quantity that I can't figure out. Well on my own. So 2014 was a long time ago. How did covid impact DK admissions? What do we know? Well this is from the U. K. We have this U. K. Data and note that the blue is the first wave of Covid, March 1st to june 30th. The post first wave from july 1st to august 31st and 2020 is in the green. And then the red is the second wave November 1 to February 28 and 2021. The bluest type one. And I want you to notice that despite the fact that I would have thought that the D. K. Went up in type one, the DK actually went down and it then went up to the second wave. But by the end of the second wave the DK went down the point is the amount of D. K. S in the type one at least in the U. K. It was actually less than what we saw pre covid. But to me, the more interesting line is the red line which is the type two diabetes. And what you see is that in the first wave there was really no change in DK. In the second wave. There was no change. But look what happened in this second wave between november 1st and 2020 and february 28th 2021. There was this big increase before there was a decrease. Why would that be? And I don't know. I don't know if this was the covid. Maybe there was more S. G. L. T. Two inhibitor use in the UK. We obviously don't know. But that is fascinating to me. So D. K. Is expensive. And the question is can continuous glucose monitoring, reduce the cost. Well this is looking at D. K. A. With a freestyle libre. Going back to the U. K. Where we have so much data. This was a study with almost 2500 type ones from 30 hospitals in the UK. The mean age was 34 years Over half women diabetes duration 14 years b. m. i. 24. And the hospital admission for hyperglycemia and D. K. Was reduced from 5 to 1% for six months with the freestyle libre. This was an 80% reduction in DK 80% reduction. That's huge. Now let's look at another UK analysis with a different population. This was a prospective observational study. This is Type one diabetes. 900 individuals. They were a little older than the last population. The DK was reduced again from 10 to 2 episodes again six months. It was again 80%. It was an 80% reduction which was exactly the same as seen in the last U. K. Analysis. So what would happen if we reduce the D. K. D. K. Costs in the US by 80% using C. G. M. Well, if you just go back to the data and the $6.2 billion from 2021 if we reduce that by 80% we would save 4.76 billion dollars. If we saw the same results here with the use of C. G. M. And in this case specifically the freestyle libre rate. Well, there's more U. K. Data and it seems like we have more data from the UK than the US. Um This was a study that was recently published in october of 2022 in the new England Journal of Medicine. Looking at freestyle libre use In the UK now this was a randomized study with 156 individuals with type one diabetes, they were assigned to freestyle library that is intermittently scanned C. G. M. Or usual care which would be finger stick glucose testing. What you see at baseline. Is that the A one CS were about the same. A little higher in intervention 8.7% A one C versus 8.5 with the finger stick testing. But look what happens with the A. One CS. The A. One C. Went from 8.7 to 7.8 after 24 weeks compared to 8.5 to 8.3 with usual care. The adjusted difference here was a 0.5% difference in A one C. That was obviously statistically significant and more importantly this is highly clinically significant. And I want to point out these final A one C levels. This is in the steep part of the diabetic retinopathy curve where you see a great increase in diabetic retinopathy and therefore um this is extremely important data recently published Now. What if we look at the amount of time less than 70 or the amount of time greater than 300 at 24 weeks. Again. Recently published data and what you see less than 73.5% of the time with the freestyle libre compared to 6.5% hypoglycemia with usual care. So usual care had higher A one CS but significantly more hypoglycemia. I want to point out that the international consensus guideline on this from 2019 is the target is to be under 4% of the time hypoglycemic under 70. So the freestyle libre group made that guideline the usual care group did not when we look at time greater than 300. We also see great differences. We see 6.2% of the time above 300 with freestyle libre compared to almost 10% of the time with usual care using finger sticks. The data is overwhelming about freestyle. He bray and the intermittently scanned C. G. M. Compared to usual care. Now here's data that has not been published but this was presented in March of 2022 at the diabetes UK conference. The interment incremental cost of $6800 per quality adjusted life year. That is willingness to pay the threshold by the National Institutes for Health Care and Excellence in the U. K. It's called Nice. Their threshold is 26,271 for quality. Well the reason why that's such an unusual number is this is converted from from pounds to dollars but that is their metric. And yet the incremental cost for quality was 6800. Whereas anything is considered cost effective if it's under 2600 dollars per quality. So this is extremely extremely cost effective. Again, the caveat of this data is that it has not been peer reviewed and published. So even before the flash UK publication came out in October of 2022. Um when looking at freestyle libre fresh glucose monitoring for people with type one diabetes and looking at the budget impact analysis and this was published very recently in BMJ open diabetes research and care. The audit demonstrated that the freestyle libre system use is associated with significantly improved glycemic control hypoglycemia awareness and reduction of in hospital admission. I want to just point out, we know that if you can avoid hypoglycemia you can minimize hypoglycemia awareness. And this has been shown in the trials leading to a reduction in E. D. And hospital admission. Furthermore, this report noted that widespread adoption of freestyle libre system in Type one diabetes populations offers many benefits and has a relatively small budget impact compared with the total cost of glucose management. Small budget impact. I showed you the quality adjusted life years. But even without that this is a relatively small budgetary item when looking at all glucose management with everything else. So why is this important? Again, we are awaiting the publication of the cost effectiveness data. Well, if we go back to Nice, the National Institute for Health Care Excellence and they knew about that data that was published earlier. They knew about all their other data. They updated this august 17th of 2022 what Nice came out and said is that we need to offer adults with type one diabetes a choice of real time C G. M. Or intermittently scanned C. G. M. Often called Flash C G. M. Based on their individual preferences needs characteristics and functionality of the device is available. We are certainly seeing this here in the US more and more with our payers including Medicare but the british beat us to this where they are saying everybody with Type one diabetes needs to be on some sort of continuous glucose monitoring. So here's more data. Can see Gm improve outcome and cost. This is real world data in the U. S. My colleague Richburg install and I published this in the journal of the Endocrine Society. We define acute diabetes events. Is combining of inpatient and emergency outpatient events. This includes hypoglycemia, hypoglycemia, D. K. A. Coma from hypoglycemia and hyper osmolarity. Secondary endpoints all cause inpatient hospitalization. We looked at over 2400 people with type two diabetes receiving multiple injections before and six months after starting freestyle libre. These individuals with type two diabetes were 54 years old. On average. When we look at the data we look at the acute diabetes events, we see a split in the pre acquisition from post acquisition data almost immediately for either inpatient or emergency room visits. We are looking at six month data, we overall see this 61% risk reduction of needing to go to the E. D. Again this is very consistent with the data I've shown you but what surprised us when we look at this is that this reduction of needing to go to the er or to the hospital. It occurred immediately if we look at all cause inpatient hospitalizations. We see a similar reduction. It wasn't quite immediately but it was pretty close. And in this situation we see this 32% risk reduction of needing to go in the hospital. When freestyle library was used, we concluded that these findings provide support for the use of C. G. M. In type two diabetes patients treated with short or rapid acting insulin therapy to improve clinical outcomes and potentially reduce costs. We had to say potentially because we didn't have the UK data that we have. Now when we wrote this paper, more real evidence, well, we know there are two types of C. G. M. This comes from the IBM markets can research database. We reviewed both the decks calm and the freestyle libre. And this looked at over 3500 with type one diabetes. You can see there were more with freestyle libre, uh with the type one. Um Also more with the type two, there are almost 4000 type twos. There were more on freestyle libre. Again, these are all multiple injections. And what was done was with both of them. They were propensity score matched. Now for those who don't deal with statistical analysis frequently. What that is is a statistical matching technique that estimates the effect of the treatment intervention by accounting for the co variance that predict receiving the treatment. So this is a statistical method that can match the people the same even though there were differences in the numbers and differences in socioeconomic aspects and other things between the two groups. It is a well accepted statistical tool. They were propensity matched based on demographics, provider visits comorbidities, insulin pump use and baseline events. And what you see in Type one diabetes, whether looking at all cause hospitalization or acute diabetes events. You see no difference between the library and blue and the decks com in red for both of these. Furthermore, when looking at acute hype hyper glycemic events or acute hypoglycemic events leading to an E. D. Or hospitalization. No difference between the two. If we look at type two diabetes, we see wider confidence intervals because the numbers were not as great. But we see no difference between the two sensors with all cause hospitalizations or acute diabetes events. We see reductions with both sensors and finally matched acute hypoglycemic events and acute hyper hypoglycemic events. We see no differences in these acute complications. Both of them did the trick these. This again is us data. Um I also want to show this lifetime cost of C. G. M. Versus fingers to glucose testing. This is a Canadian analysis published last year. This is type one diabetes. What you see is the direct cost of the C. G. M. On the far left. The direct cost. But you see you have money saving with all of these complications with the greatest money saving with renal disease. And what this Canadian analysis concluded is that if we say away Willingness to pay is $50,000 per quality adjusted life year. And that is the metric that most countries say $50,000. They said that at least in Canada the willingness to pay is 99.7% compared to other therapies using this bar of $50,000 per quality adjusted life year. So the real world experienced conclusion is that patients with either type one or type two diabetes experience similar reductions in acute complication events and all cause hospitalizations when using either brand of C. G. M. Now I want to get into one final topic here with the freestyle library. This is the old ambulatory glucose report. Um time and ranges and the recommendations. All the metrics the ambulatory glucose profile and then looking briefly at each of the different days to take a look at where the spikes and where the low blood sugars occurred. We are now moving into something called glucose pattern insights and the glucose pattern insights are different than what we used to see on the previous dashboard because first of all, it's removing the time and range consensus targets from 2019. The A. G. P. Figure is different. It's now in color and this is really important when we are especially concerned about wanting to minimize hypoglycemia. You can see it much easier where the hypoglycemia occurs and shows critical A. G. P. Patterns. It can give you recommendations on medication or lifestyle. It's removed these daily glucose profiles and it's actually removed the glucose variability statistic which confuse some people. But the whole point here was to make this simpler and more easily usable, especially in primary care. So the problem is does the glucose pattern insight improve primary care provider decision making? Well, let's look at the report with primary care providers, not diabetes specialists, but they treat patients with diabetes. The care strategy is to identify and work on one pattern at a time. Thinking about this to simplify it. The most assessment and therapy change process by focusing on the most important pattern and I think everybody would agree the most important pattern is hypoglycemia. And then you can move further on to hypoglycemic patterns. So we address hypoglycemic first. If low pattern is mitigated, then we will go on to address the other patterns while addressing the high patterns. When the low patterns are mitigated. One needs to make sure, especially if you're adding insulin not to make the hypoglycemia worse when addressing the highs. Now, high variability may prevent addressing addressing the highs without making the lows worse. So we have to talk about lifestyle behaviors and I want to include in lifestyle, not just the typical exercise and diet that we talk about to mitigate some of the spikes, but for people taking Crandall insulin, the timing between when the Pran deal insulin is given and when the food is eaten and if needed, consider a different therapy that may better address the variability and you know the most common thing that we see is people on too much basil insulin or maybe they are only on basil insulin and they get to a very high blood sugar at bedtime and it's the basil insulin that's bringing them down overnight. Well in that situation they need something so they don't get so high at bedtime and usually that something is mealtime insulin even if it's just at that dinner time meal. So when looking at the effects of this new glucose reporting tool on therapeutic decision making and looking at a study with primary care providers. This was presented by my colleague jean right at the american diabetes association meeting this year. And what was shown was looking at these glucose pattern insight reports that identify patterns of sub optimal glycemic control, highlighting the clinically most important pattern the M. I. P. And offering therapy considerations to the primary care providers. To address that most important pattern. To assess the utility of these glucose pattern insight reports and clinical decision making. A reading study was conducted comparing it to the current standard glucose report which is of course the ambulatory glucose profile. The A. G. P. Clinical data from 10 patients with type two diabetes were used to generate these glucose pattern insight reports and ambulatory glucose profiles that non specialist primary care physicians. 35 of them evaluated each case and each report design alongside the A. One C. And medication regimen to make a therapy change recommendation. The therapy change recommendations were then categorized and whether they addressed the most important pattern or not with a priority on treating hypoglycemia. If it occurs coincident with other patterns within a given case. What was shown well the primary care providers addressed these most important patterns equally well with each report and cases presenting patterns other than hypoglycemia Across all cases and subjects. Therapy change categorizations were different in 79 instances with 67 of these instance presenting its hypoglycemia. That is a primary care physician recommending using one report addressed hypoglycemia while that using the other report did not with the subset all. But in one instance that is 99% of the primary care physicians correctly addressed low glucose within the G. P. I. Report when they did not for the same case using the A. G. P. So this shown an improvement an improvement with this tool compared to the way we were doing this earlier. These findings indicate that this new aid in identification and treatment of hypoglycemia that otherwise would have been missed. Using the current standard reports. And it's my own opinion, it has to do with seeing that red color of the hypoglycemia that otherwise would have been missed. Otherwise color actually is important. So to conclude acute diabetes emergencies are increasing. In the United States data continue to accumulate from around the world that C. G. M reduces the incidents of these life threatening events. I just showed you the UK data there's data from Italy, there's data from Greece, there's data from France. I did not have time to show you, but the data is all consistent. C G M also appears to be extremely cost effective for chronic complications in Type one diabetes and certainly also for the acute complications. More formal cost effectiveness. Studies assessing the impact of this technology in this country are in different countries and in the US are needed. I am hopeful. Very soon we see the cost effectiveness data from the Flash UK study that was recently published in the New England Journal of Medicine. And with that I thank you from Seattle Washington. I hope you found this helpful. Thank you very much.
Related Presenters
