 On behalf of the McLean Center for Clinical Medical Ethics and the Center for Health and the Social Sciences, Dr. Meltzer and I are delighted to welcome you to this lecture in the 2018-19 series on improving value in the US health care system. I'm so happy to introduce our speaker today, Dr. Albert Wang. Dr. Wang is professor of medicine here at the university and serves as director of the Center for Chronic Disease Research and Policy. He's also the associate director of the Chicago Center for Diabetes Translation Research. From 2010 to 2011, Albert served as a senior advisor in the Office of the Assistant Secretary for Planning and Evaluation at the Department of Health and Human Services. Albert is a primary care doctor who studies clinical and health care policy issues at the intersection of diabetes, aging, and health economics. Using methods for health economics, he's performed seminal translational diabetes research studies, including a well-known 2008 widely cited study that forecasted the growth in the diabetes population and related health care costs over the ensuing 25 years. Albert Wang has also led economic evaluations of community-based diabetes initiatives, clinic quality improvement programs, new diabetes technologies, such as continuous glucose monitoring, as well as the application of basic scientific discoveries like genetic testing to clinical diabetes care. Dr. Wang has received numerous honors. They include the research paper of the year award from the Society of General Internal Medicine, and he's been elected a member of the American Society for Clinical Investigation. Today, Dr. Wang's talk is entitled, as you see behind me, Finding Value in Innovations in Diabetes Care in a Brief Conversation that I had with Albert just before this introduction. Albert tells me that this is a talk, and do I have it right, that's never quite been given in its entirety before. So we're very excited and looking forward to it. Join me in giving a warm welcome to Albert Wang. Thank you, Mark. Can you hear me? Great. So the outline for this talk is kind of a march through the history of diabetes care and the associated economic studies that we've been part of along the way. So I'll review sort of the general public health and public policy challenges that we face using diabetes as a model chronic condition. And I'll talk about basically march through a review of innovations in diabetes prevention and care that have happened over the last 15 to 20 years. And then I'll review efforts to characterize the economic value of these innovations over time. So bear with me, you've seen this a million times, but it's still dramatic every time I see this. So this is the prevalence of diabetes starting back in 2005, and we basically marched through time. The darker and deeper the red, the more diabetes there is in the country. I'd say 20 to 30 years ago, we had a prevalence of diabetes around maybe 3% to 5%, and we're now nationally at a prevalence of around 8% to 9%. So nearly one out of 10 people in the United States has diabetes. And let's see if I can speed up through time. There's 2010, and you can sort of see that there's geographic disparities and the prevalence of diabetes. Here's 2015, the most recent year. A lot of the country's deeply red with diabetes prevalence rates above 9%. And this is from the Centers for Disease Control. So there are a lot of epidemics. I mean, this is actually concurrent with the opioid epidemic, but the obesity and diabetes epidemic has been going along for over the same time period. So a fairly devastating growth of this common chronic disease. And part of this has to do with a steep rise in it. So that was prevalent diabetes. So how many people have diabetes at a given point in time? And the reason that the prevalence of diabetes has ridden so much is because the incidence of diabetes, that's the rate of new cases of diabetes has also has shot up, in particular in the sometime in the 1990s to early 2000s, there was a steep increase in the incidence of diabetes that you can see from here to here. And it appears to have lessened a bit, the rising incidence has lessened a bit since 2010. So in the late 2000s, Anurban Vasu and I actually developed a forecasting model that accounted for the new entry of people with diabetes in their life course. So it included their exits from the system as well and tracked overlapping cohorts of people as they entered with diabetes over time. And this is our projection of how many people will have prevalent diabetes from 2009 to 2035. And we're right now around here, around 30 million people around 2020 and we'll see whether or not our numbers are about right. Typically these forecast studies are historically always wrong. They're right typically in terms of the direction of growth but they're always off in terms of the magnitude of growth. But these are, we projected that the population size would nearly double from around 23 million to nearly 40 million people over the next 30 years. And these are the associated costs of care for different generations of people entering with diabetes at different time periods and causing this dramatic increase while the population size is projected to double the cost of taking care of these people is projected to triple during that same period of time. Now, when we did that forecasting study, we actually didn't even, we assumed that rates of diabetes and the rate of entry into diabetes would be frozen in time. What we saw in 2007 would just continue. And so a lot of this growth is actually due to just the aging of the population and shifting demographics. And this assumes also a fixed rate of obesity of around 30%, which we may be, and the problem with all of these is we may be wrong about each of these assumptions. We also assumed the cost of medications would be roughly what it was in 2007 and that things would not change. So you can sort of guess if you think through those three assumptions, we are likely wrong about the actual, eventual costs of diabetes in the year 2030. So we're actually not that far off. So this is a snapshot from the American Diabetes Association from 2018. We have around 30 million people in America, around 10% of the population with diabetes. The vast majority is type two diabetes. So I'll talk a lot about different classes and types of diabetes. But 95% of people have what we call type two diabetes. And they cost the country around $327 billion annually. That is a mixture of direct and indirect costs. If you just look at the direct costs of medical care, it's around $237 billion per year. David's in the audience and he'll tell you, this is actually not that big a number. Because actually the overall size of the healthcare budget in America, healthcare economy in America is closer to $2 trillion, is that right? So this is like, it's around $3 trillion. So this is something, but $300 billion is a lot of money, but it's all relative. So this is what I alluded to earlier. So in that forecast that I showed you from around 2007, 2008, we assumed that the costs of medications would be really static. We did not know that this would happen. So this is a little research letter in JAMA that we published a couple years ago that used the medical expenditure panel survey to just estimate how much the per capita costs of diabetes would change over time. So this accounts for not only the dose of medication people are receiving, but the kind of medicine, and it accounts for the changing price of medicines over time. So the average person taking insulin is depicted in the orange line. And you can sort of see that the straight line is actually the per capita cost of diabetes medications over time. It doesn't look too bad. It's around $600 per person. And actually declines a little bit overall because it's a mixture of all the people with different kinds of medications. But what's striking is the rise in the annual per capita expenditure for insulin over the same time period. So it goes from around $200 per person per year in 2002 to around $700 per person per year in 2013. So this is actually due to not only changing prices, which I'm gonna talk about, but it actually has to do with people. Actually, the actual person using insulin is actually using more insulin today than they were in 2002. In other words, the average daily usage of insulin in terms of the units that were dispensing per person per day has increased. But what caught a lot of the people's attention is that the price of insulin for the same exact unit size has actually tripled during that time period. So the price of insulin went from $4.34 per milliliter in 2002 to $12.92 per milliliter in 2013. So insulin pricing is one of many examples of a dramatic increase in price that we don't fully understand and I'll end the talk about how there's an increasing scrutiny over why these prices are going up and how they're affecting people. Part of this is due to the shift from human-based insulin to insulin analogs, which do we have endocrinologists in the audience? Raise your hand if you're an endocrinologist. No, none of them showed up. So endocrinologists love the insulin analogs because they're modified versions of human insulin that allow us to specifically time the duration of effective insulin and when their peak action is, gives us a lot of control over how the insulin works. They're dramatically more expensive than human insulin, about, in some cases, 10 times more expensive. So that's part of what's going on up here. So rising prices, epidemic of disease. Now, all of that is quite scary, but if you have diabetes today in 2019, you're actually quite lucky to be alive. I mean, it's not lucky that you have the diabetes, but to have the diabetes now is a lot better to have diabetes from the 1970s or even earlier, a century earlier, when people died routinely. I have a very young death from diabetes. So this shows you the rates of different complications. Once people have diabetes, what has happened to trends in the rates of complications, major cardiovascular events like heart attack, stroke, amputation, renal failure, and in general, the cardiovascular outcomes have undergone a steep decline in terms of you're much less likely to have a heart attack or a stroke than you were in the 1990s. And the rates of other complications like renal failure have unfortunately been a little more stubborn and have not declined as much. So while we have experienced an epidemic, boom in the number of people with diabetes, the average person that we encounter who has diabetes is more likely to live longer, to have a lower rate of cardiovascular outcomes than they were decades ago. In fact, actually people with type one diabetes, these are individuals who have to produce no insulin whatsoever and are dependent on insulin to survive, actually now have a life expectancy that's very similar to someone who doesn't have diabetes. David, you had a question? Yes. Is it more about managing sort of? Yes. That's a very good question. So I've actually seen other studies of the general population and rates of cardiovascular outcomes and they are overall, so I think the improvements in the diabetes population are part of overall improvements in cardiovascular prevention. But this bottom graphic is actually an attempt to, this actually is a mixture of people with and without diabetes and the rates of, there is some decline in, I think MI, but it's not as steep as what we see in people with diabetes. But you can guess what likely happened during this time. I believe it's very likely this is actually the statins. And probably better management once people had lower smoking and better management once people had coronary artery disease, better procedures, all that contributing to the decline. I don't, this is too early to be attributed to better glycemic control, I think, but possible. So if you have diabetes today, it's a lot better than it was several decades ago. So what do we do with this? We have a huge growing population of people with this chronic condition. We have very expensive care, but at the same time, it does appear to be producing benefits for the people who have the condition. And how do we judge the value of these innovations or changes in care? And so we've been, at University of Chicago, we have actually a long and proud tradition of having a connection with the theoretical foundations of cost-effectiveness analysis in the United States. And this is a picture of three leading figures who have connections to University of Chicago. Will Manning, who's unfortunately passed away, was actually on the original panel on cost-effectiveness in health and medicine. And basically that was the original rule book that we used to conduct cost-effectiveness analysis. Prior to the publication of that panel report, it was a real Wild West of analysis where people kind of made up metrics and outcomes and decided whether or not something was cost-effective or not kind of on their own. Will helped to establish the rule book for doing those analyses. And more recently, David Meltzer, who is here, and Anurban Basu, who is a graduate student of these two, they were co-authors of the recent second version of the panel of cost-effectiveness of health and medicine, and that's published in this issue of JAMA from 2016, and there's, I think, a more extensive report elsewhere. So hopefully we follow the rules that they set out in our evaluations. So you're gonna have to, so I'm gonna explain very briefly the cost-effectiveness analysis method and the basic metric that we, the outcome metric that we use in interpreting all the subsequent analyses I'm gonna show to you. So the basic synthetic metric of cost-effectiveness analysis is the incremental cost-effectiveness ratio, which is a simple equation. It's the difference in costs of two different options that you're comparing. It could be two treatments, it could be two policies, over the difference in an outcome for the two different options. And in a version of cost-effectiveness analysis we call cost-utility analysis is expressed in differences in costs over differences in quality just in life years. So I'm gonna use this repeatedly during the entirety of this talk. To interpret this outcome, it's actually typically best to look at both the difference in costs and the difference in outcomes separately in order to interpret them, because if you just looked at the number you wouldn't know where that result lands on what we call the cost-effectiveness plane. On the cost-effectiveness plane, the y-axis is the difference in costs for the two options and on the x-axis is the difference in health outcomes for the two outcomes. And we talk about the four quadrants of the cost-effectiveness plane when interpreting the result of a cost-effectiveness analysis. Most interventions, most innovations in healthcare land in the northeast quadrant. And in this quadrant costs are higher with the new innovation, but health outcomes are better. So we are typically interpreting results for things that make life better, but at higher costs. And that requires some sort of threshold in order to understand whether or not something is cost-effective. And that's where people talk about the $50,000 per quality threshold. Below that threshold, things are considered cost-effective. Others, including David, have written about how the real cost-effectiveness threshold that we use in the United States is probably closer to $250,000 or $300,000 per quality based on the way we spend money already. In this quadrant, the northwest, costs go up and health outcomes get worse. So in this quadrant, if something lands there, it's what we call dominated. It's really not something we would ever do. In the southwest quadrant, these are interventions that lower costs but make health worse. There have been some really interesting articles about how maybe we should be thinking about interventions that do this. Not very harmful, but just a little bit harmful. But if we can save money, maybe it's worth considering these southwest quadrant interventions. And in the southeast quadrant is kind of the nirvana of cost-effectiveness analysis. These are interventions that save money and improve health. And we would love for everything to land here. I'll show you some examples of cases where we have found examples of things that have actually landed in the southeast quadrant, but that's how to interpret the cost-effectiveness ratio. I talked about the rules of the road, which are set out by people like David. And the tools of the trade are really, and certainly in diabetes economic evaluation, are these, what I'm showing you is a cartoon model of a simulation model of diabetes complications. This one comes from Nita Lightirapong's recent paper in Animal and Internal Medicine. It's actually based on the United Kingdom Prospective Diabetes Study, OM2, Outcomes Model 2. And these models typically follow the life of an individual. They're at risk for developing different complications during their life. They follow each individual every year. If they survive, we also account for the likelihood of death. If they survive a given year, we go on to the next year and we model their entire life with these simulation models. And these models are built on typically cohort data for long-term follow-up studies of people with diabetes, where we know from epidemiology the rates of complications, and we know the major predictors of those complications. And we assume that if we modify those predictors, and this is a big assumption, if we modify those predictors, that we change the course of disease. So I'm just gonna march through, it's a little bit of a history of the major trials of diabetes, major innovations in diabetes, but I'm gonna talk about prevention, the value of intensive diabetes care, like basically treating people intensively with lower glucose, blood pressure and cholesterol. I'm gonna talk about the value of designing an intervention to just implement and deliver diabetes care better in a healthcare system. Actually, we're trying to do this right now in our own health system. And I'm gonna veer a little bit towards some more specialized forms of diabetes. I'll talk about devices in the context of type one diabetes, where we actually are in an era where we have actually now the artificial pancreas. And then I'll talk about personalized medicine, how we've done studies about, and this is actually where we've really been a very different kind of a health economic or where we've evaluated innovative ideas like personalization, both in terms of setting goals for diabetes care, treatments, but also in terms of targeting treatments to the genetic profile of an individual. So believe it or not, many of the innovations I'm gonna talk about, they've all occurred since I was a resident in, since I finished residency. So it's only been in the last 17 to 18 years that we've had actually the evidence available to study, to prevent diabetes and also treat it. Prior to that, we were flying by the seat of our pants. So in terms of diabetes prevention, this is a large-scale random, multi-center randomized controlled trial that was published in the early 2000s, around 2002. This is, the University of Chicago played a key role in actually enrolling most of the African-Americans in this trial came from University of Chicago. And actually, if you go to the M200, M corridor on the second floor, you'll still see some of the DPP patients there in the outcome study that's still ongoing here. In any case, this is a major randomized controlled trial that enrolled people with pre-diabetes. So they didn't have diabetes, but they had a little elevation in their blood sugar. They were randomized to a lifestyle intervention, which included an effort to reduce weight by 7%, 150 minutes a week of exercise, just brisk walking, and a better diet. They got a lot of, they got free gym shoes and a lot of other things along the way, a lot of encouragement to do their lifestyle change. Another arm received metformin, which is actually a early stage treatment for type two diabetes as it is, but at a lower dose. And then a third group was the placebo arm that got nothing. And during the course of the trial, the individuals who received the lifestyle intervention had a 50% reduction in their risk of developing diabetes. So this is a, I think this is, this is what your mother would have told you to do, but in a randomized control trial, the metformin arm didn't do as well, but did reduce the incidence of diabetes. And the placebo arm, of course, did the worst. So this has actually served as the evidence base for all the subsequent work in translating diabetes prevention into the community. And it has also served, actually is now actually a covered service in Medicare. So, and actually that's a more complicated story, but Medicare actually now covers diabetes prevention, the diabetes prevention program. And so what Health and Economics group, I think as a combination of the CDC investigators and Bill Herman from the University of Michigan, they attached the end of the findings of the DQP to a simulation model of diabetes and its complications. And here they forecast out the rates of incident diabetes for the different arms, for the placebo arm, they go on to develop more incident diabetes than the metformin arm and the lifestyle arm. So they're essentially just extending the findings, assuming everything holds true from the end of the trial going forward, this is what would happen to the rates of diabetes. And then they account for the complications and costs of having diabetes decades later. And because of lower rates of diabetes, lower rates of complications, individuals in the lifestyle intervention arm had the lowest lifetime costs compared to, these are the intervention costs, but overall costs are lower in the intervention arm versus placebo. So there's a little increase in costs with implementation of the lifestyle intervention because you have actually the cost of the intervention, but you have cost savings from lower rates of complications. They have better quality of life in the lifestyle intervention and higher at 0.57 over a lifetime. Metformin also increased quality of life, but not as much. And the overall difference in costs over difference in qualities for a lifestyle is $8,800 per quality. So it's in the Northeast quadrant, very cost-effective by most conventional thresholds as is the metformin arm at $29,900 per quality. And so this actually is, so these are both interventions that we would consider cost-effective, they're not cost-saving. Does anybody know the follow-up of what really happened with the subsequent follow-up of the patients? Can you guess what happened with the lifestyle intervention people? Rita? Yes. So they had a difficult time maintaining the lifestyle intervention and they actually stopped adhering to the therapy. And the metformin arm, they actually ended up doing, they actually in follow-up, they actually do pretty good. So it's kind of sad, nobody wants to hear this, but taking a pill a day was actually easier to adhere to. And long-term, this result actually might be flipped. This model, this assumed that everything would hold stable over time, but actually things likely flipped, David. Oh, I think life expectancy goes up a little bit. Yeah. I'd have to look up the paper, but I don't remember exactly. Yeah. David, I'm pretty sure they didn't do future costs. What? I'm pretty sure they didn't account for future costs. I'm pretty sure they did not account for future costs. They broke your rule. Yes. Exactly. Yes. Why should placebo be effective? Is this because a lot of you did placebo, but you know other things. And maybe at the same time, the person you knew for the placebo is also conscious of a lifestyle and all of a sudden- Sure. Yeah. Sure. Of that maybe. So he also had that. So we added this because valuable because you do much credit to placebo. I see. Yeah, that's a good point. The placebo-armed individuals, they definitely had freedom to do whatever they wanted. They were not in terms of their lifestyle. So you're right. This is not really a comparison against nothing because the placebo-armed is actually doing something. And actually at the end of the trial, if you think about it, they learned about the results and could actually take out, could adopt the lifestyle intervention that was done in the trial. But in terms of the actual execution of trial, they literally received a placebo versus metformin. So they literally had a sugar pill that they took every day. But that's a good point. But it's very complicated. And just another side note, so this would show you, the result of these analysis suggests that there is actually no true cost savings for the system with diabetes prevention. And that doesn't really fit what public health people want to hear. They want to hear that something saves money. So what happened was this study was basically repeated in some form under the auspices of the Center for Medicare and Medicaid Innovation. It was an innovation award in a Medicare population. And within a two to three-year period in a Medicare population, that program managed to somehow save money. And with the innovation program, the Secretary of Health and Human Services has the ability to implement anything that saves money across the entire system. And that's why Medicare now covers diabetes prevention. Yeah. Yeah, yeah. We won't, the CDL said we won't score this as cost savings. And so there was a memo written by this under... He basically said he just thought he'd do a large amount of things for the Secretary of Health. Well, that, that, that, that... But Jen, I'm not sure... Yeah, absolutely. Absolutely. So that's a great, thank you for that insight because it didn't, even the original actuarial result doesn't, anyway, it doesn't quite make, certainly doesn't make sense in terms of our understanding of how diabetes works. So if it was saving money, where were they saving money and how was it happening? The other kind of unfortunate sad story about diabetes prevention as well is that it's actually, even when it's a covered service. So you may have heard about United Health Care Covering Diabetes Prevention and now Medicare covers diabetes prevention. Well, it turns out that once United Health Care adopted diabetes prevention, there has actually been almost no uptake of, even though it's a covered service, it's almost, it's actually not even, it's rarely used by anybody. And in terms of the Medicare covered service, if you've actually looked to see where you can send somebody to get diabetes prevention services, there are actually almost no service providers in Chicago because it's very hard to make money delivering diabetes prevention. There's actually been a lot of, a lot of requirements tied to getting payment. You have to basically get the patient to lose weight before you get paid and so it's very difficult. So it's incomplete story, but very interesting. I think preventing diabetes, almost no one would argue with it, but there's a lot of complicated details in implementation. Yes. Well, it has to do with the fact that any kind of form of prevention, and I'll just show this later in genetic testing is actually even a better example, but you have to basically assume that you're implementing something across a population of people. And among, so the costs of a program are borne by everyone who's exposed to the program, but the actual event rates are only experienced by a minority of that overall population. So that it's the difference between, it has to do with the cost of the program and implementing it, how many people get touched in the denominator and then the numerator of people who actually benefit is smaller than you think. So that's actually the general problem with prevention. Okay, so in the same year that the diabetes prevention program publishes results, yes. I wanted to try, but I wanted to talk to the question that I heard it. Gained? It took away half the... Yeah, it does really well. Right. You're right, right, right. All right. Well, the, I mean, the uptake, the lack of enthusiasm for the uptake of diabetes prevention actually, I think hints at some disutility associated with lifestyle intervention. So at the same time, the EPP came out in the early 19, in the late 1990s around 98, the first trials of intensive diabetes care were published. This is a combination of the United Kingdom prospective diabetes study, which provided the evidence for preventing, for blood pressure lowering and glucose lowering, but also statin trials were coming out in the late 1990s. So in 2002, the Setters for Disease Control cost-effectiveness group published this, actually fairly, I would call, in the field of cost-effectiveness analysis and diabetes, a landmark study in JAMA, describing the cost-effectiveness of each of these components of routine diabetes care, lowering sugars, lowering blood pressure and lowering cholesterol with the statin. And I think I may have just gone straight to the bottom line here, which is, so they did the same thing that the DPP folks did. They attached the trial results for these three different therapies to a forecasting model of diabetes that they had built. And for intensive glucose control, they found that the incremental cost-effectiveness ratio was around $41,000. It improved quality-adjusted life years by about $0.19. Total costs increased a bit, similarly for hypertension, for cholesterol as well. The ICERS around $52,000. Again, it improves quality-adjusted life years by $0.3475, but at increased costs. The one intervention among these three therapies that had a very different result is hypertension. And actually, if you know the history of the UK PDS, it produced both cardiovascular benefits, but also microvascular benefits. So in the case of hypertension, it's actually a cost-saving intervention. It not only increases quality-adjusted life years by 0.3962, but also reduces overall cost because the number of complications prevented is so numerous. And an interesting sub-finding of the study is that, of course, there's an age effect. So the age in which you receive the intervention alters the cost-effectiveness of the therapy. So in particular, for intensive glucose control, you'll see that for a young person in their 20s, intensive glucose control has an ICER of 9,600, but if you're an 80-year-old, the cost-effectiveness ratio is around $2.1 million. So there's basically an issue of life expectancy, how much time you have to benefit from a therapy that affects the economic value of interventions. So that's great. And so around the same time, the UK PDS came out and at the same time, the JAMA papers came out around the cost-effectiveness of diabetes care, there were a lot of efforts underway to try to disseminate throughout the country best practices with diabetes care. How do we get more people to achieve A1Cs of 7, blood pressures below a certain threshold, and how do we get more people on statins? And one of the efforts was something called the Health Disparities Collaborative, which was a national quality improvement effort in federally qualified health centers to try to improve the quality of diabetes care. And it included at least three major components in the intervention, included continuous quality improvement, included basically teaching around the chronic care model, which I'll show you in some more detail from Ed Wagner. And then there were active learning sessions for health centers. So health centers did a lot to form quality improvement teams. And actually, Marshall Chin was involved with this national evaluation of the Health Disparities Collaborative. This is the typical plan-do-study act. You basically develop a plan to improve diabetes care. You try it for a while, and you move on to the, you see whether or not you've made any improvements, and you iterate on the intervention itself. And one point to make about the Health Disparities Collaborative, it took place in an era without electronic medical records. So if you wanted to know if A1Cs were better, blood pressure was better, or if there was more statin use, you had to actually pull charts to basically abstract that data. So when I refer to the cost of this particular program, there's actually more human costs associated with quality improvement in this era than we have, I think, today. And this is the classic model of the chronic care model developed by Ed Wagner from Group Health Cooperative. But you basically have health systems within communities. And then you have the health care organization. You have the activated patient interacting with prepared proactive practice teams, producing productive interactions. And then that leads to the functional clinical outcomes that we have for our patients. So there's just more details about the Health Disparities Collaborative. There was actually a special randomized control trial that Marshall led, which included things more teaching around shared decision making. But you get the point. This was a collaborative effort where they actually learned from each other using, they actually had a listserv. They had regular calls with each other, leaders of the quality improvement teams across health centers throughout the country. So this was not conducted as a randomized control trial. So we have, but this is what we know from the chart abstraction done over time, which is that in terms of processes of care, it appeared that diabetes care did improve in these health centers from 1998 to 2002. People who had an A1C measured increased from 71% to 92%. Lipid assessment also increased. There was more prescribing of aspirin during this time. There was more prescribing of ACE inhibitors during this time, going from 33% to 50% of patients. And actual laboratory outcomes and other bio-measures also improved during the same time. A1Cs decreased from an average of 8.6% to 7.9%. LDL cholesterol also declined. Most likely due to higher rates of statin prescribing. And blood pressure remained roughly the same during this time period, even though there was greater use of ACE inhibitors. So we, as part of this, incorporated these results into the simulation model of diabetes that we had at that time, a type 2 diabetes model. And simulated, basically, if we were to compare the bio-measures and the therapies prescribed in 1998 to the bio-measures and therapies prescribed in 2002, what's the incremental gain and what's the incremental cost of changing diabetes care in that way? And what we found is that if you believe that those trends would remain separate over time, there would be reductions in blindness, renal failure, lower rates of heart disease, and improvements in quality just in life years with the changes in diabetes care. Overall, with more cost, of course, and this included both programmatic cost but also the increased use of therapies. And with this combined implementation of diabetes care, we had an incremental cost-effectiveness ratio of 33,000 per quali. So this program, if it were to remain in place, would be as good as a new drug. So overall, what I've shown you in the last series of talks, series of papers, is that diabetes prevention, either through lifestyle or metformin, is very cost-effective. In terms of diabetes treatments, hypertension management is actually cost-saving. The other two, glucose control and cholesterol control, are cost-effective but probably not cost-effective among the very old. And in terms of implementing diabetes care in a health system, it's also very cost-effective. So sort of a shift around, there's basically a separate sort of trajectory of innovations going on in the field of type 1 diabetes. And around 2005 or 2006, we were approached to be the health economics team for a major randomized control trial of continuous glucose monitors. Continuous glucose monitors at the time in 2006 look like this. This company is still in existence, Dexcom. They're still the leading manufacturer of continuous glucose monitors. But these monitors have a little straw that goes basically under the skin and detects subcutaneous glucose levels continuously. And you can see in this graph in this little picture, what the person is doing is they're seeing their sugar levels in their subcutaneous fat continuously over time. And this device, along with the measurement of the sugars, has associated alarms, can tell you when your sugars are too high or too low, and also allowed you to prepare for meals to anticipate how much you'd have to adjust your insulin dose with these measures. So it really was an incredible advance to have continuous glucose monitors. And the Juvenile Diabetes Research Foundation organized a multi-center trial of these devices. This was one of several devices. The other device was the Avid Navigator, which doesn't exist anymore. This was the probe that would be inserted into the skin and taped to your skin. And this is the little pager you would carry to see your numbers. And the actual randomized control trial was published in the New England Journal. What they found was that people who were randomized to the CGM basically have fairly dramatic. This is actually a 26-week trial. And these are the glycated hemoglobin levels, the A1C levels, at week 26 at the end of the trial. And these are the three different subgroups. These are adults over 25. These are adolescents 15 to 20 and young adults 15 to 24. And these are young children 8 to 14. So they had a whole range of people, the whole age spectrum. And you can see for the adults, there's actually a shift in the A1C distribution at the end of the trial with the use of CGM. Whereas you don't see much of a shift at all for the adolescents and young adults and for the children. And do you know what correlated with these differences in the subgroups? Well, it relates to what we talked about earlier with the lifestyle. It's basically simple adherence. So can you guess which group wore the devices more? So this is basically completely tracks with wear time. So the adults were more willing to wear the device regularly. The adolescents, according to any pediatricians in the audience, so apparently they're quite difficult. And wait, I actually have one of those. And they don't listen to you, they don't wear the device. And then for the children, even though you would think that they had the oversight of their parents, there was actually at that time a technical problem which was the children didn't have enough skin landscape in order to glue the probe on. But in any case, if you use the device routinely, you can improve your sugars. And we did this additional, this will never be repeated again. But what we did was we actually made people in the trial do time trade-off questions about their current experience at 0, 13, and 26 weeks. The reason I'm saying we're likely not to repeat this again, it was very arduous. We actually caused some emotional distress among some of the trial participants when they answered questions about trading time off. But it was invaluable in many ways because we were actually able to detect a utility change separate from the glucose improvements. And you can see that patients who were adults with high A1C's above 7, they had higher utility ratings at 13 and 26 weeks than those who were randomized to standard glucose monitoring. Yes. Oh, they're insane questions. So imagine you have in your current state of health, let's say you have a choice between 10 years in your current state of health or 10 years with diabetes. Which would you choose? That's a standard opening question that you should answer. I would choose life without diabetes for 10 years. Then we alter the math. And we say, what would you do if you had a choice between 10 years with diabetes but five years in perfect health? Which would you choose? And we go back and forth. And we find a point in which the person says it's no different. And in these individuals, look at the ratings they gave from the utility ratings from 0 to 1. The ratings are very high. These people did not want to trade off life. Their quality of life was actually quite high by multiple measures. It's actually indicated here as well. But there is a difference between those who were using the CGM versus those who were not. Actually, the difference in quality of life is even greater among those with good glucose control. A1C is below 7. And this is likely due to the fact that the CGM helped them prevent hypoglycemia, alleviated fear of hypoglycemia. And so you can see here they also have higher utility ratings during this time frame. Very difficult to execute. We somehow persuaded the trial investigators to do this. I think I remember David and I going down to Florida to persuade them to do this. They relented. It's actually, it was the first data collection of time trade-off utilities in type one diabetes that had been done. So the other thing that happened was, so we combined the utility improvements and quality life with the improvements in sugar and again, attached to trial results to a forecasting model of type one diabetes. And again, with CGM, there would be improvements in reductions in rates of complications across the board because of improvements in glucose control. And life expectancy would also be, would overall be higher with CGM versus control. In terms of quality adjusted life years, there would be a fairly sizable difference of 0.6 years for those with A1Cs above seven, but for those with low A1Cs below seven, there'd be even a higher quality adjusted life improvement. And here is the overall point estimate for the ICER for both therapies. So landing in the northeast quadrant, you'll notice that there's actually quite a bit of uncertainty around, there's actually a big cloud of uncertainty around these results, but these point estimates were then later used to make coverage decisions across the world. And so CGM is now, it's not totally easy to obtain, but it is generally a covered service for type one diabetes. So fast forward 10 years later, these devices, you'll remember the old clunky Dexcom. This is the Dexcom, like I think, yeah, Dexcom five, G five. It's the next, it's sort of like they try to copy the iPhone, you know, iPhone is seven, eight, 10. So Dexcom G five looks like this. It also has a little probe that goes underneath the skin with a little straw detects subcutaneous fat, but the detector is now in your phone. So you can have an app in your phone, that's where the software is for detecting, to for tracking your sugars, or you can wear it as a watch, or you can have it as a freestanding device, like a pager, like before. So this is the kind of the modernization of the continuous glucose monitor. And the diamond trial was done very similarly to the JDRF trial, and the diamond trial, they randomized individuals with type one diabetes who were using multiple daily injections. The JDRF trial was done in people, many of whom were using insulin pumps already. So most of the people in the diamond trial were receiving multiple injections per day. And basically the same thing happened in the diamond trial as in the JDRF trial, which was that there's a shift in the 24 week heme globin A1C distribution, shifting downward for those using the continuous glucose monitor. And again, we looked at the within trial analysis. In this case, we were not brave enough to do time trade-off. We used more indirect methods for eliciting utilities. We did not find any change in utility. I think we used EQ5D, is that right, Ben? And we did find, of course, that the costs of using the device increased costs overall for these patients. It's about, I think, for annually, it costs about $5,000 per year to use a continuous glucose monitor. And again, we attached the trial results to a simulation model of type one diabetes, now using a more updated version of the model from Sheffield, England. And in the end, with improvements in sugar and also improvements in prevention of hypoglycemia, there's an overall improvement in quality in life years, as well as quality of just life years and a difference in costs with an ICER around 98,000 per quality of just life years. So very similar to the ICER that we produced from the JDRF trial. One thing of note is that many people with these devices do things to prolong, the device requires replacement of parts routinely. And when we did the modeling, we assumed that the patient would follow the rules and replace the sensor, for example, every week. But in real life, what people do is they cheat and they prolong the use of the device as long as possible, sometimes extending the life of replacement parts two weeks, three weeks. And if you account for that adjustment of real world use, the actual cost-effectiveness ratio actually comes down to as low as 33,000 or 42,000 per quality just life year, making the intervention even more cost-effective. One thing I'm not showing you is that there was actually a second trial attached to this first trial that was actually much, and I didn't have time to assemble it, but actually has even more interesting results. The people that were randomized to CGM were randomized further to an insulin pump or not. And the addition of insulin pump actually worsened the quality of life, which gets at this idea that it's possible that the sequencing of these devices may affect their value to individuals or their quality of life, but also that what the CGM manufacturers believe is that the CGM is the driver of the better outcomes. It's not the pump. But I'm happy, and that's a separate publication that I'm happy to share with you. So in terms of these devices and this whole world of the artificial pancreas, which I've not even fully talked about, the CGM is cost-effective in adults with type one diabetes and using pumps, adults with type one diabetes using multiple daily injections. And with real world use, the CGM becomes very highly cost-effective because of longer-term use of supplies and lower replacement costs. We faced a lot of challenges in doing these studies and measuring increasingly things that patients don't even know about. So one of the challenges in the more recent diamond trial is that people are experiencing hypoglycemia that they don't feel. So how do you assign a quality of life value to that? How do you monetize that? They're not waking up, they're not feeling it, but we know that it's probably not good for them. So that's one methodological challenge that we have faced. And really where this field is going is that because we now have full-on devices that are a combination of pumps and sensors, it's really the entire artificial pancreas that's going to become the next, should we pay for that routinely? So I'm gonna shift gears towards studies of more kind of personalization or individualization. And this is a recent study that, if you, Hennig-Gran-Rowes recently, Nita Leiterpang described, it's published in Anals of Internal Medicine. And what she, and this is the team of investigators, including Rochelle Naylor from Endocrinology, Philip Clark from Oxford, Reza Skandari who's now in Imperial College, Jen Cooper, a project manager, and Erin Wynn who's now at the Medical College of Wisconsin. And what Nita wanted to study was, in the 2002 study I showed you from the CDC, they just assumed that everyone in the world would just adopt intensive glucose control. But we know that this is not likely to be good for everybody, that there likely needs to be some individualization of the glycemic target based on how sick people are, how old they are, what other conditions they have, where they are in the stage of disease. And this is a graphic that comes from the American Diabetes Association showing what they proposed are different clinical criteria for titrating up the sugar target from more stringent to less stringent. And what Nita did was, she basically operationalized the entire protocol for individualization and shifted people's targets over their life course as they lived with diabetes over time and then used the UKPDS model to model its long-term effects. Also accounted for medication side effects including hypoglycemia. And the model that we use again is this United Kingdom Prospective Diabetes Study model. This model doesn't account for things like the legacy effect that we now know exists in diabetes. Nevertheless, what happens with individualization is that more people are allowed to have higher sugars. And so what Nita did was, she led basically people when the individualization protocol had higher rates of complications because they have higher sugars. So this could be in the Southwest quadrant. Maybe we're letting people have more complications. So they have higher rates of myocardium infarction, stroke, blindness and so on. At the same time, the individuals had better quality of life even though they had higher rates of complications. And that's because in the model she also accounted for hypoglycemia which affects quality of life from day to day. And people in the individualization strategy had lower rates of hypoglycemia. So you have slightly higher rates of complications but less hypoglycemia. And overall, she compared the costs of implementing a uniform lowering of sugars as like what we did in the CDC study compared to individualization. And basically overall costs of care of delivering individualized care is actually lower than it is for the basically making everyone, mandating everyone to achieve an A1C below seven. And a lot of it's due to lower use of medication. So that's where the cost savings arise. And with that, again, this balance between quality of life and longer life, life expectancy was found to be a little bit lower with individualization but quality of life was actually improved due to these changes in avoidance of medications and avoidance of hypoglycemia. Overall, the quality of life improves if you account for all these things at the same time. So individualization of glycemic control in the United States might save money. It might save up to $234 billion. It would reduce life expectancy. So maybe it's in the Southwest quadrant but quality of life is better. So it's a difficult trade-off. But I think this paper also highlights the fact that and actually in other studies we've shown is that people's preferences for different states, these analyses are very sensitive to your assumptions about people's preferences. So in the end, for the individual patient, we have to account for what they think is important and what's valuable to them. Maybe it's okay for them to remain on medications even though they're in their 80s. So another variation on individualization is this study that is pretty unique to the University of Chicago. I don't think it would happen anywhere else because what we did was basically partner with Graham Bell and this is a picture of Graham Bell and Ken Polanski and this is a picture of Stephen Fiennes. These are scientists that discovered monogenic diabetes. So these are investigators in the 80s and 90s who discovered these rare forms of diabetes that are due to a specific genetic defect. And this is a family tree of a real individual with a pedigree of an individual who has monogenic diabetes with a genetic defect entitled HNF for alpha. And I think this is the picture of the family. And so what happened was I think I literally ran into Graham Bell in the hallway and started talking to him about cost-effectiveness studies and we came up with this hatched this idea of studying the economics of genetic screening. And that began a series of papers that we've done in neonatal diabetes and monogenic diabetes and I'm gonna feature a study that was recently led by a medical student. Actually on the second year of medical student, Matthew Goodsmith backed up by Rochelle Naylor in pediatric endocrinology. This is just a graphic showing all the different genetic forms of diabetes that we now know of and what parts of the system they affect. This is a picture of the beta cell. Some of these genetic defects can actually, the genetic defect can dictate what therapy is best for the individual. So this is a picture for individuals with this particular form of monogenic diabetes HNF1 alpha in which they have a defect where their potassium channels on the surface of their beta cells remain fixed and open. And it turns out that the treatment for this channel is actually an old-fashioned diabetes drug called sulfonate aureus. If the person takes high doses of sulfonate aureus, this potassium ATPase channel closes. Thus starting membrane depolarization, opening up of calcium channels and then release of insulin. So the beta cell becomes revived with the use of sulfonate aureus. So if you know that the person has this genetic defect, you can target the treatment to them correctly without knowing this information. You may be noticing that they have higher sugars and treating them blindly without knowing about their specific form of diabetes. So what Matthew did was he developed a basically a giant simulation model of diabetes that accounts for four or five different forms of diabetes. One form being HNF1 alpha. One is the GCK modi in addition to models of type one and type two diabetes. And what he did was say, let's assume that we can assess the whole population of children with diabetes in America. And let's say we can institute uniform biomarker testing and genetic testing in this population and change their therapies according to what form of diabetes they have. Compared to what we have today, which is we're blindly treating children without knowing this information. And in the case of HNF1 alpha, as I told you earlier, we assume that they would take on sulfarnia urea treatment, which actually not only, it actually improves their glycemic control, but also lets them avoid therapy like insulin. For GCK modi, these are individuals who actually don't develop complications from diabetes. So they don't need treatment at all. So without the genetic testing, they would be treated. So we did this additional step, which was to add on cascade testing. So if you know an individual has modi, you can actually then go on and check individuals and their family to see if they have that genetic form of diabetes. And that therefore extend the benefits of genetic testing beyond the original person who was tested. And so this is just a picture of the complex model that Matthew built that tracks, these are individuals in the control arm where we don't know anything about the genetic status and we know how they're treated currently based on observational studies of pediatric populations. And then on the upper part of the graphic, we have the institution of the screening, the biomarker testing, and then the genetic testing that follows, and we have the targeted treatment of all these individuals to match the therapy to their genetic form of diabetes. And what Matthew found is that if you institute this biomarker and genetic testing strategy, you actually save money for the system a little bit per person, only a relatively small about $191, but you don't increase costs. You basically lower costs a little bit and you improve quality of life a little bit overall for the whole population. So for this rare condition, which is around one or 2% of the population, if you institute this program, you may care better, but things are about neutral or slightly better. If you add cascade testing, the cost savings are even more accentuated and the benefits are actually also accentuated. So these are examples of a strategy that lands in the southeast quadrant of the cost effectiveness plane. And the challenge is that the majority of people don't benefit from the testing, only a small minority of people benefit from the testing. And this just demonstrates how much these individuals with monogenic diabetes have a better life with the appropriate therapy. In the case of H1 and Alpha, these individuals have life extension of half a year quality of life extension about half a year and they save about 100,000 over the course of their lifespan. Individuals with GCK, Modi also save a lot of money and have smaller improvements in quality of life. And these are just the results of the sensitivity analysis showing that if we add cascade testing it, this is actually expressed in terms of net monetary benefit, we more to the right the point is the greater the net monetary benefit is. If you add cascade testing, you accentuate net monetary benefit. So the key issue with genetic testing and economics is that we have to reduce the number of people who undergo the testing. And so the biomarker screening allowed us to narrow the number of people who would get the genetic testing and then combining that with cascade testing makes this a really valuable intervention. And at this point, it's actually very difficult to get genetic testing. It's not routinely covered by insurance, but I believe that these kind of analyses could change that. And this paper is really powerful because it shows the potential population health benefits of personalized medicine. So most innovations in diabetes prevention and care, they improve health, but usually at a greater cost. And these innovations, what I've done is I showed you a whole series of cost-effectiveness analyses, but that doesn't really solve the budgetary problem we have, which is that you could have a, we have every year an increasing number of cost-effective therapies that insurance can cover. But just because something's cost-effective doesn't help solve the budgetary limits that payers face. So the more things that we find to be valuable, the more it compounds the problem of the budget. And, but I've tried to show you in a couple of examples, there are some innovations where we can actually save money or keep costs relatively neutral by making and make health better. And that's in the form of personalization related to personalizing diabetes care by disease history, comorbid conditions, and in the future, I think with the genetic profiles of individuals. And then I didn't even talk about how things like having access to care at the right time in your life, like through the Affordable Care Act can also change, can actually be something we can also monetize in terms of economic value. And all of this actually is, you know, have to be viewed also in terms of the broader policy landscape that is happening at the same time. And I just want to feature a few things that are happening because we live in a country where we don't actually use cost-effectiveness results to make decisions, at least not overtly. We're actually in some cases banned from using cost-effectiveness analysis. But I'll show you some examples of where it's being snuck in. But I don't know if anyone, everyone aware that there's a change in the House of Representatives since January? It's really quite different. So the House Committee on Commerce and Energy sent this really awesome letter to the three major manufacturers of insulin, I think over this weekend. And this letter is awesome. It basically says to them, how do you justify the increase in price of insulin? And they have this particular quote that actually is related to the JAMA, it actually cites the finding from our JAMA letter that I showed you at the beginning. Despite the fact that it has been available for decades, prices for insulin have skyrocketed in recent years, putting it out of reach for many patients. For instance, insulin's price nearly tripled between 2002 and 2013, then nearly doubled between 2012 and 16. Medicare Part D spending has also risen as has out-of-patient cost spending. Diabetes patients who do not have insurance are particularly vulnerable to price increases. So what this is an example of is basically there's increasing oversight from government, from Medicaid programs, from these congressional committees that is, what they're trying to do is they're trying to use the force of the bully pulpit to change the discussion around drug prices. The Trump administration is also trying to change through a lot of action to try to change, try to, they're targeting in particular the pharmacy benefit managers and trying to reveal all the rebates that they seem to benefit from. They're interested in that part of the system, but the manufacturers also are, I think, playing a role in increasing prices and Congress is targeting the manufacturers. In the last year, I was named to something called the Midwest Comparative Effectiveness Public Advisory Council, which is part of something called Institute for Clinical and Economic Review. I mentioned that the United States doesn't use cost-effectiveness analysis in informal decisions around coverage decisions, but this was a organization started about 10 years ago that actually is trying to serve as a pseudo-governmental role to actually formally act like the United Kingdom's NICE. So on this panel, we review the cost-effectiveness analysis results of different new therapies. The day of these meetings is really quite wild. The morning session starts with the presentation of the analysis just like I showed you. It is followed by two hours of basically direct commentary from the public, from individuals who frequently have the disease, or parents of individuals who have the disease. And they're not always happy that someone is asking about the economic value of the therapy that they're benefiting from. Yes, it's done in the nicest way possible. It starts with a careful discussion around the benefits of a therapy. And it's only later that we talk about the cost-effectiveness analysis, but they do have cost-effectiveness threshold. They use 150,000 per quali, and the panel has to vote. In public, on whether or not they believe the therapy is cost-effective or not. Yeah? Who decides what? I don't entirely know how the therapies arrive at ICER's attention, but I think it's possibly ICER itself identifying novel therapies that are very expensive. But I think the insurance, I wouldn't be surprised if there's some role of the insurance plans, actually shunting ideas to ICER. But the afternoon meeting is attended by not only payers, but the manufacturers of the drugs. So far, the therapies that I have seen, one cost $1 million per quali. The other was had annual costs of $300,000 per year for the patient. So the therapies I've seen so far have all been extremely expensive. Nothing like, in the case of diabetes, we're talking about relatively inexpensive medications. The problem is the diabetes population is so large that it becomes very expensive for the population as a whole. But I have told my team in my office that if I get shot at a public event, it will be at one of these meetings, because I have seen some very angry people at these meetings. And as a country, we have to decide whether or not we're gonna continue rationing care at the bedside as we are. Doctors are having to make prior authorization requests and having to fight for individual patients to get coverage. Or do we have some central role by the payer? They're probably making decisions as well, not explicitly. But this is kind of an amazing opportunity to see what it's like. And actually, if meetings are ever in Chicago, I'd welcome you to attend. They're open to the public. And I'll probably be there scared. But this is an attempt to operationalize and turn cost-effectiveness analysis into real policy action. And one of the evaluations of cystic fibrosis drugs actually did lead to a formal hearing in front of the New York Medicaid insurance panel where they're instituting an annual review of high-cost drugs requiring the manufacturers to explain themselves why the drug has to be as expensive as it is. And actually, our work in this area is really, I think it's just taking off and what's really exciting is that in the area of diabetes, that Nita Leiterapag is leading this large team of investigators to build really what is gonna be the first American multi-ethnic model of type II diabetes. All the prior examples I showed you before are based on British populations, 90% white. So it really odd that we've been using those models to estimate the value of therapies in a multi-ethnic country like ours. And so she's going to account for Asians, Latinos, African-Americans, and also try to account for all the new drugs that we have available to treat diabetes. So, you know, our... How do you know about the conditions that they're able to cover your pulse if the diabetes is half in your... Actually, I did not talk about that. That's another topic, but actually conveniently, Dr. Leiterapag also leads our behavioral health interventions too, so you could actually talk to her about that. Maybe there's some way to fold that into this analysis. But I did not talk about behavioral health interventions at all. And these are all the investigators in Chicago, at Norrick, University of Washington, and Oxford, and Kaiser, Northern California, that are part of this larger team evaluating diabetes and its costs. So thank you, and I'm happy to answer questions. I don't... Sorry. Should I go ahead? A quick question about that number of seven for the hemoglobin A1C. Did I ever hear you talk about an evaluation of whether that was the right number, or does it depend on a person's point in their life, or is that... Yeah, I didn't show that, but yeah, we've interrogated that quite a bit. And so certainly for older people, we have a paper where we looked at older people who are by age and by how sick they are, and the sicker somebody is, the older someone is, the less likely that A1Cs of seven are actually valuable, beneficial or valuable. And that actually has led to change. That actually is partly why the diabetes guidelines for older people actually changed in 2012 because of that. I wrote them. Great talk. What? What are they now? For the sickest, less than 8.5, middle health, less than eight, and for the healthiest older person, less than 7.5. That's a great talk, Albert. I was wondering, I know the Wall Street Journal did an analysis or published an analysis recently looking at single payer universal healthcare. I'm just wondering in diabetes, has that ever looked at so where the government is buying in bulk these expensive medicines and what the price point would be if the government was buying in bulk and covering citizens with single payer? I'm not gonna answer your question directly, but I'll answer it by talking about just what other countries do in terms of drug prices. They basically have far more government intervention around prices. They've completely, the government plays a big role in setting prices for medications. And the United States just does not play a role at all. We let the free market, as it is, decide the price. So the prices of things in Japan are ridiculously low. The prices of drugs in Canada, that's why people are talking about just going ahead and bypassing United States and buying drugs from Canada. So I'm not sure your question about the conversion to a single payer would, without other changes, would result in lower prices. The VA is actually a good example of a universal health system where they're actually able to negotiate with, and actually they get much lower prices because they are able to buy in bulk. Medicare for whatever reason is, well, we know it's legislated to not negotiate on price. I don't know why, but if it could, it might actually lower prices. And the last, I'm amazed that you stayed as long as you did. The last thing is that we host a monthly workshop on these sort of topics. So please contact me or Viva Nathan if you're interested in attending the monthly workshop. But thank you for your attention. I went crazy. I went crazy.