The 2025 Project: Market transformation
Exploring AI, analytics and the future of health care
- So welcome, everyone. Thanks for joining us for this session. My name's Kyle Rose. I have been in and around healthcare data, specifically market data, for 20 years. I started with The Advisory Board and now, of course, I work with Optum. During my, kind of, tenor doing healthcare data, I've seen a lot of changes in the market with regard to the data that's out there, the data we've had at our fingertips, and frankly how smart we are with how using the data. I helped to put together Crimson Market Advantage, which was our market analytics product back in 2009 and then, since then, I've been chasing the white whale of data, as we're trying to find better and better data sources for everybody out there. For those of you who've been around a while, who remembers looking at HCPCS data with a MedPAR Standard Analytic File back in the late '90s, 2000s? Yeah, I mean, we've been there. We've also seen state hospital data, state hospital association databases, we've seen them improve but we're far away from the fact that, at one point, New Jersey didn't have any state data at all. Now, most states have some data. It's available and it's getting better, but it's usually pretty facility-focused. So a lot of what we're gonna be talking about today is outside the four walls of the hospital because, frankly, I think that's where the action is. In terms of just little bit of road mapping, I'm going to go through kind of a setup of why we're here today. A couple of the big macro trends and, specifically, what data demands these market trends have of us. I mean, we have a number of things that we're facing. We have a number of imperfect data sets. We're always chasing the perfect data set. But I think this is always gonna be an asymptotic situation, where we're kind of approaching perfection, but never achieving perfection. So alas, we're sitting here with five, six, seven different forms of data, all with different data definitions. Our bosses are saying, "Well, give us the numbers." Well, what numbers, right? It's all very an oblique kinda situation. I think that the good news is is that more data is at our disposal. The bad news is, of course, is that more data is at our disposal. It's hard to take a stock of what actually is out there and how valuable each of these data sets are and how they interrelate. So that'll be my first piece of business, is just to talk about where the market's going. I'm gonna talk a little about three or four big initiatives we have going in analytics, in Optum analytics, around better understanding our market and some of the strategies specifically with data. This is not a data talk. We're not gonna be talking about data specifics. They let me outta the basement, but I refuse to talk about data basement stuff in front of a group like this. So I am gonna talk about some of the more exciting features of what we have planned over the next couple of years. If we all are at 2025 and we're still using the data that I'm talking about today, God help us. Hopefully, by then, it'll have change. But I think it's useful to be aspirational. In terms of macro trends, these are not new. We've been facing these for a really long time. Payment and regulatory shifts are pushing business out to lower cost settings. It was HIPD, the hospital inpatient setting, to HOPD, to ASC, the physician office. This trend has happened in service line and service line over and over, and this is a 30-year-old trend and we can actually see all these shifts. I think a little bit of what's different is that it's paired with the two bullets at the bottom. One is that there is this kind of vague word out there around consumerism and price transparency. I'm not really sure what to make of them yet. I think there's much to do about them in the national press and the national media. People are saying, "Well, if you have a high deductible plan, "you're shopping around for business." I've seen various polls that suggest that 85 plus percent of business is not chosen by consumers still. It is directed by physicians. We probably know that, intuitively, working in hospitals, working around hospitals. But I think it's hard to deny that there is some kind of consumer horizon where consumers are gonna be making more of these decisions. And then price transparency, who knows how this is gonna play out. I felt like it was off and now it's on. I saw just yesterday in "The Wall Street Journal" that the Trump administration's pushing again for price transparency. Who knows where that'll land. If that'll just be about inpatient and outpatient or will that be about all these ambulatory settings. Will a patient, at some point in the future, be able to say, "I wanna get a hip replacement "and I can pay X in a hospital, "Y in an outpatient hospital setting, Z in an ASC," and who the hell knows. Maybe A, they're paying for it to do it at home or in a physician's office. I mean, it is hip replacement surgery, but who knows where the technical advances are going. But to have this real choice and to sit there and look at all the scenarios I think is liberating for a lot of, I think, people that think about healthcare economics. But from a healthcare administrator's standpoint, it's really difficult to pin down where the demand's going to be and how lucrative these procedures are gonna be in the future. So that's, I think, one of these things underpinning and why kinda the shift from the four walls of the hospital out into the ambulatory settings is more acute now than it might have been in the past. And then, finally, value-based care is nigh. I have a slide on this in a minute, so I don't wanna steal my thunder. But we've been talking about value-based care for a long time. It was part of the Affordable Care Act. It was part of the subsequent amendments to the Affordable Care Act. It's been before that. We've talked about capitation in this industry for 20-plus years. But it's very difficult to pin down the timing on this and I think it's gonna be market by market and system by system. So it'll be interesting for us to follow it. I think what is an absolute truth is that, in some point in the future, you're gonna wake up and you're gonna look at your market and you're gonna say, "Well, listen, I have a chunk of my business "that is under value-based care, under risk. "Do I understand how those patients "behavior differently, if at all, "from patients when it was a fee-for-service model? "When we were getting paid for surgeries "and surgeries alone. "Is it going to do things to the demand "that I hadn't forecasted "and maybe I need fewer ORs "and more ambulatory care settings." So I think that that future is ahead of us, no matter how fast from a market-by-market perspective value-based care actually comes. So what are the demands on us? I mean, if we have these trends kind of at our head, these are the headwinds, I think, can data or any other kind of intellectual property or solution come to the rescue? Well, the first thing we need and this is the same thing, this is a decade's old need, it hasn't changed, it just might be coming more acute, we need greater visibility into patient journeys, especially as it relates to the up and down stream from the surgery. If you think about it, the way that we all value market share today is really kind of middle-out approach. We're looking at a hip replacement surgery, a cardiac event, but there's a lotta stuff that usually happens before that and there's a lotta stuff that happens after that. We just happen to be valuing the most lucrative piece in the middle. And under a fee-for-service environment, when we are hospital administrators, not system administrators, perhaps, maybe that's all we care about. Growing orthopedics by 4% every year is probably a recipe for all of us getting a bonus. But I do think that as this world comes into focus, where we're having to think about care settings that are less expensive, maybe not inside the four walls, where we're having to think about managing a piece of care across the entire continuum, all of a sudden, the upstream, so before the event, the surgical event, or inpatient stay, the visit, the things before that and after that become, perhaps, even more important than the stay itself depending on the structure. So having visibility into the entire patient journey is going to be really important over the next couple of years. The second piece is patient origin and patient access. If you think about your market, you're only as good as your catchment and we've always been priding ourselves over the last several decades on you have a big hospital system. I mean, this is the traditional thinking, at least. A lotta people are more progressive now. But the traditional thinking is you have a big hospital with a big hospital brand. You have physicians that are loyal to that hospital and everything else kinda takes care of itself. The patients will find their way into your network. They'll undergo their surgery at your facility and I get it, there's a lotta moving parts there. There's a lotta different departments. There's business development and physician liaisons and the direct-to-consumer marketing folks at organizations who are all in charge of making sure that that is as focused on your system as possible and away from, for example, your competition. But it has largely been on autopilot, I would argue. I think moving forward, I think we're going to have to be more proactive with where we think about, where are patients actually originating? And do we have the data that actually tells us that because we just haven't had the data before. I mean, you guys all get zip code data for your inpatient stays, I'm betting. But do you get zip code data on where primary care visits come from? Probably not, I would argue probably not. Unless it's coming from your own systems and then, you're only seeing the physicians that you touch. You're not seeing everybody else. So having that understanding of where patients originate is going to be important. And then, subsequently, thinking about access. So where do you place your primary care physicians or your free-standing ED or your urgent care center, all of those places that, essentially, are at the top of the funnel? So these kind of catchment mechanisms that funnel everything through to your acute care hospitals or I guess there's a world where we're saying, No more acute care. Let's do more conservative therapies and obviate the need for an acute care episode altogether. Either way, we're going to need a way to capture those patients at the point of access and understanding where they come from is step one to do that. And then, finally, quality metrics. Not just for your own physicians and this has been kind of another white whale of ours for a long time and I'll tell you a little bit about the history of a couple of our false starts. But wouldn't it be great to have some sort of standardized, normalized quality data for all physicians in your market so that you're not just going out there and you're saying, Hey, listen, I'm gonna get busy docs and PCPs with really big panels and that's all I care about. Those are the only things I'm gonna index on is business? Then, you get them into your system. Again, you wake up at some point in the future and you have part of your lives under risk and you realize that you have a lotta busy docs who are pretty crumby docs and they're quality data is pretty crumby. And so, the outcomes are pretty crumby and you're losing the bonus payments associated with the value-based care or you're losing the ability to control care. So it has to be more than just focusing on loyal docs or busy docs or docs who see a lot of patients. It has to be more than that and I think that getting a sense of quality data for the entire market is, again, step one to doing that. So these are our challenges. These are our needs from a data perspective. Let's get started. You wanna solve some of these problems? I do, as well. So what are the tricks up our sleeve? We've got four cards up our sleeve. Maybe one ace, one king, two queens, maybe a jack. So, these are all kind of great ideas, but they can't all be the same kind of impact. So I think I've striated them here from left to right in terms of the top impact I think that they can have right now on an organization. The first is the idea that, just as I mentioned, when we're building a network, you can call it whatever you want. You can call it a CIN, an ACO, a group of affiliated physicians, your employed physicians. However you define network of doctors, there is an us and a them in a market. There's an occidental and an accidental, so we are thinking about this world where we are us and everybody else is the field. And so, when you're designing your network of providers, and by providers here, I mean physicians and facilities and access points to the entire kit and caboodle, when you're designing that kind of thing, what you need to look for is do you have the right kind of data at your disposal? Are you really just looking at the loyalty and that busyness of the doctor? Or are you considering some of these other elements, like quality, because a high-value network now is no longer just a lot of surgeries pumped through the system. High value means something means very different and quality has to be part of that. So that's the first thing that we're doing. We're gonna take a look and I'll explain how we're gonna do it in a minute. We're gonna take a look at bringing in a data setter, a set of data that allows us to score every single physician in a market on a quality standpoint and from a business standpoint. Things like loyalty, efficiency, how busy they are, how many surgical volumes they do. So that's point one. Point two is demand forecasting. I think underpinning all of this is that we've all been doing demand forecasting, I mean, my first gig at The Advisory Board 20 years ago was I created the ortho, neuro, spine service line forecasts for The Advisory Board. And the way that we did it then, in 2002 and '03, is exactly how we, at Optum now, do it today, and all of our competitors do it. It's actually very basic. You take local demographics. Florida, there's older people and the mountain states, there are more active people, et cetera. These are kinda generalities about local demographics. And then you apply national utilization rates. It's very simple. You put those two things together and you get demand in a market. But what do we know right off the bat from that? We know that national utilization rates are bunk. They just don't mean anything. I mean, there's a whole project called the Dartmouth Atlas Project that was designed essentially to refute that very premise, that there is any kind of standardized utilization across the country. There's enormous amounts of regional and local variationing, so why are we all using national utilization rates or national prevalence rates or national incidence rates when we're trying to do demand forecasting? It doesn't make sense. So there's gotta be a better way. The other thing about forecasting is that it's all straight line-based. You look at the last five years, there's a line. Then you look at the next five years and you just keep drawing that line straight. Well, what do we know about the future? We just talked about how value-based care is a little bit of a Crazy Ivan. It's gonna come in at some point in the next two or three years and really shake up how demand occurs at our organization. And so, why not try to predict or at least have of different scenarios where we're looking at demand, but demand under the ejus of value-based care. If value-based care is really executed very well, we would be trading, I don't know, 20, 30% of our surgeries for more conservative therapies. What does that mean? Do we need fewer ORs? Maybe, we certainly don't need to build more. Does that mean that we need more capacity on lab or diagnostic imaging, capacity on the post-acute side, SNFs and LTACs, home health, more capacity on telehealth, access points? It changes everything. So this idea of doing straight line forecasting is, I think, getting more and more problematic as we speak. But everybody does it the same way. I nearly guarantee you that every single forecast that you have at your organization is based on, A, national utilization, prevalence, or incidence rates, and, B, employs some kind of moral straight line forecasting. Both of those things we know, just statistically, are unlikely to be true moving forward, so why not try to get ahead of that? So we have a project design and overhaul how we're thinking about demand forecasting, so that's point two. Rightsizing the ambulatory footprint. A lot of what I've observed in working with hospitals over the past 20 years is that we all operate in our little silo departments and sometimes there's a bat phone between departments. You know, marketing can call planning. Planning can call, I don't know, master design or master planning facilities. But, usually, those mechanisms organizationally don't exist. Everybody lives in an existed silo. One of those silos or two silos that I'd like to break down, and this is the first and then the last leveraging, the gold consumer record, that'll be the next. One of them is this divide that usually happens at organizations between strategic and master planning. There's no reason why if you put up a five-year master plan that many or all of the elements of the strategic plan are incorporated and I'm sure you'll say to yourself, "Well, we do that. "We have in our master plan, "strategic planning feeds those things." But is strategic planning actually sitting at the table when they're doing site placement? Are you looking at a variety of sites and saying, "Well, listen, I know that this is the site "that the physician prefers or the board member prefers," which is usually how it works, right? But if you actually located this primary care physician office, medical office building, in a different geography and maybe had a slightly different focus for that facility, we could see 2x the returns of this physician or board member preferred location. Wouldn't it be great to have the ability to do that? And moreover, wouldn't it be great to have the ability to scenario plan in real time? For example what if, and this is an example that came from an actual kinda client that we were talking to, they had pinned all of their hopes on putting one of their cancer units in a property that the city owned. It was a five year plan. They had recruited physicians. A number of administrators had kinda made their career on this thing. At the last minute there's an administration change in the city and the city says, "That's not any longer gonna be a cancer center, "It's gonna be a park." This hospital is just devastated. They have no backup plan and no ability to scenario plan in real time. They couldn't say, okay if not there then what? Do I have options? Could I split it up? Could I divide it among my existing facilities? Could I find an alternative placement sight? Everything just kinda ground to a halt and frankly a lot of those administrators just kind of went into the ether. They went into other jobs or went to other departments and the whole thing fell apart. I just think that there has to be a better way to be able to do this kind of thing and I think that have A, more of a continuous set of communication between strategic and master planning, the idea that master planning is just all about facilities probably has to go out the window because master planning in this new environment is a lot about right size and care, and they probably should be plugged into network architects at your organization as well. And then finally is there a way to have a dynamic platform, maybe a software platform, maybe a consulting platform but a dynamic platform where we can actually make these dynamic scenario plans in real time. So that's kinda point three. So that's one silo that I love to break down which is this divide in hospitals typically between master and strategic planning. The other is that bat phone that I mentioned before between marketing. So the folks that you direct to consumer work, and planning. So folks that do other kind of work. It's astonishing me that if you have a campaign for orthopedics growth at 3% or 4% that oh, gets us all bonuses that I mentioned before, that comes out of strategic planning that's fueled by a lotta data, a lotta hard work. You look at the demographics, you look at the latent capacity in the market, you look at the facilities and you say okay, we can grow by 4% if we hire another doctor or something. But there isn't always a straight line communication between that campaign and calling the marketing department and saying okay, we would love to have marketing support. Now, hospitals and systems are getting a little wiser here and they are having more of a coordinated effort between planning and marketing and they come together and even physician liaisons and the physician outreach. One of the issues that I've identified is that it's really difficult to do an ROI calculation if everybody sharing in kind of the rising tide floats all boats. Well, was it because of marketing's billboards and their digital director consumer stuff? Or was it because planning highlighted the need to employee another physician or was it because the physician liaison team who did, hustled and worked their butts off and did 50 new visits for those orthopedic physicians, or he refers to the orthopedic physicians in the last six months. Who gets credit for that? Everybody's always worried about who gets credit and it get it. Finished checks are being made and these, a lot of times these departments are call centers, but I think it is, it behooves us all to join in and break down those silos and break down those barriers, and I think one of the keys is that, we can begin now to use data, in this case kind of a, this idea of a golden consumer record. We have a number of data elements that are very patient focused for the first time and I'll introduce them to you more in a little bit. But those data elements are going to allow I think marketing to have a conversation around the cost of acquisition of a patient, the total life time value of a patient, and that same data set with using the same definition and the same nomenclature planning will have the ability to use that same data and say well, did I understand when a patient came into my network and when they fell out? What is it at a physician or was it a facility? Is my network designed or the path or journey designed in a way that reliably produces a set of visits at my organization? So wouldn't it be neat for planning and marketing to have and be speaking the same language? So that means singing from the same hymn book. And I think we have again a trick up our sleeve that'll allow us to do that. So these are four big priorities. If we could nail these four things and help you do these four things, essentially plan for the transition of value based care, which is the first two things. So, design a high value network that takes in quality into an account, and to do good demand based forecasting where value based care is taken into account, and if we could help you right size what your care footprint looks like in the market, which is again a way for us to help you transition into value based care, and in the end if we can be more impactful with how we design and execute our campaigns, both marketing and planning and physician outreach, I think we all probably walk away in three years winners because you'll win on behalf of your organization. So yeah. I mean this is not, it's easier said then done. This is one slide. It took me about 40 minutes to put together. So this isn't the totality of it. There's a lot more work to do, but I think that this is the beginning if we're all kind of walking in lock step here and we all do our jobs and we do our job and you do your job I think that we can really kind of like I said, win in our markets over the next four years with these strategies. First, let's talk about value based care. Does anybody recall the story of Tantalus? Do you guys know Tantalus from mythology? Remember your Edith Hamilton or Bufinch's little books that they made you read in seventh grade? Well, one of the stories was, there was this story of Sisyphus, right? He was pushin' the rock up. Right before he got to the top the rock kinda rolled back down. The story of Tantalus was a little different. He's sitting in a pool of water. He was punished. He did something, I think he served like a human to the gods and the gods got pissed so they put, they did this to him. So he's sitting in this pool of water and he has like some sort of fruit tree right above him. Every time he reaches down to get a drink of water the water recedes. Every time he reaches up to get a bite of the fruit tree the branch removes. So he's always kind of tantalizingly out of reach. This is where we get the word tantalizing. I think that for a lot of people, for a lot of us I think value based care, at least the shift of value based care in our individual markets is just like this story of Tantalus. The closer that we get to this purported finish line the further I think it moves. This is a class study looking at 13, a survey from 13, 16 and 18 of the same administrators, 40 administrators. Take a look at it. They become a little bit more optimistic about the, it happening soon. So in the three to five year time frame, and they think that under two years are already there. That number declines. The number increases on six to eight years and the number of nine plus years does go down but the point is here is that people are basically saying that every two, three, four years this value based care transition is still two, three, four years away. And I think that the issue here is that it's not going to be a thunder clap moment. It's not going to be that for any given market where a big deal is inked and you hired the two big employers in the market and all of those lives are under risk. Ipso facto, 80% of your lives are under risk. It is not going to happen like that. It's gonna come in kinda fits and starts. It's starting for everybody right now in the bundle payment programs for Medicare but that's usually just a very small slice. There are markets like in the Pacific Northwest and other markets where there has been a lot of adoption. I live in Austin, Texas. There's no Texas city that has more than five or 10% max of their lives under risk. So, this is going to be A, maybe slower then we thought, B, it's not going to be binary. You're kind of in risk or not. It's going to be everybody's in transition, and C it's gonna be market to market So you're gonna have to figure out where your market is on each of these plot points. So I think one thing that we're working on right now is developing a value based care transition maturity model. And the only thing that this model will do is it'll, you'll be able to plugin some basic inputs about your organization, things like employers at your market, your market dynamics, and it'll place you on kind of a left to right schema of are you ready for risk but not quite there? Are you ready for risk and you've already begun? Are you not ready for risk? Are you basically just fee for service? And so stand by for this 'cause I think it'll be just a little calculator but I think it'll give you and your kind of fellow executives and your administrators just a rubric for evaluating where you are on this transition and frankly, I think that this is important. It's important for us to all know exactly where we are in our competencies. Many of you may say, "Hey Kyle, "my organization is such in a way, "in a market where I don't even care about risk. "I'm just gonna let risk happen to me." That's fine and that is a very valid strategy I think for some markets. But some of you might be saying value based care is a top priority. Maybe a top three priority in my organizations. All we talk about at our board retreats. But fewer then 5% of our lives are under risk and I don't even think we have a really good sense of where we are. I have a feeling that most people are gonna fall kind of in that camp. So wouldn't it be neat to essentially just benchmark you, yourselves against where kind of similar organizations who have adopted more risk are versus similar organizations who have not committed any risk at all. So I think that watch out for this maturity model. But kind of everybody's gonna be somewhere and the somewhere's I think almost certainly going to be a transition. Nobody's going to be all fee for service. Nobody's going to be all value based care. Everybody for the next five, dare I say 10 years might be in transition. And so it is our job I think to appropriately size where we are in this transition and then bring data tools and other tools to the table to be able to help us make this transition a reality. So, if you think about the kinds of things that you need and forgive me for the slide being so text dense. You can, like I said you can download it. But there are different needs for organizations at each of these steps along the way. I've just divided it into four big categories. There's probably countless subcategories within each of these. But if you're all fee for service, you're still addressing core physician alignment issues. You're still growing orthopedics by four or 5% but that still requires you to have a primary care strategy, a medical specialist strategy, a surgical specialist strategy. This is the old business. Things that we've been doin' for a long time. But that still requires data and having better market intelligence to do those things is better than not having better market data. The second phase is that you're aligned but your network isn't optimized over on the value proposition. I would argue that probably if you're looking at kind of a curve, this is kinda the bell part of the curve. This is where I think most people probably find themselves now, and you've been largely reacting. Maybe you have some bundle payment programs, maybe you have a CIN, maybe you were part of the ACO demonstration project, but you're not fully in because who would be fully in at this point? It doesn't make a ton of sense, but you're also probably not looking at a macro market level for your market, whether or not you have the right composition of physicians and facilities and access points and are you in the right geographies for the right products to be able to serve a population under risk? 'Cause as I just mentioned, if you have a very well managed population the whole point of it is to reduce the amount of surgeries and increase the amount of almost everything else. And so that's just not how we built our markets. We haven't built our markets with the indexing on the ambulatory, the pre, the up and down stream before things hit the four walls of the hospital. And I'm not here to alarm you. I mean your hospital's not going away any time soon. I guarantee you that. Maybe in 30 years but not 10 for sure. And even so there will still be a requirement for a buncha hospitals across the country, but it does make you think twice about do I have my market optimized in a way that I feel satisfied with? And if we all left today, our organizations today and we had a legacy that we were gonna leave on your market, did you leave your organization with the right composition of things, assets in your market to do well over the next five or 10 years? I think most of us would probably say no and one of the hindrances of course has been the data. The next phase kind of in this maturity model as we've done our research and all this is by the way research based and we're trying to be as quantitative as possible. The next phase is coordinating care management across the continuum. So it's not just about building the network, it's about actively managing patients within a network. So there are tools that you can use to do that. The Optum has a tool for example. But I think more importantly it's this idea that you're transitioning from building a network to kind of running patients within a network. And that's a big mental decision, a big shift that hospitals have to have, and what's fascinating for me is watching the organizational dynamics of hospitals play out here because a lot of times what'll happen is that the network stuff happens on the ambulatory side of the house. So the physicians. So there's an executive that is in charge of the physicians. They're also usually in charge of building out a network and when you make this shift from kind of managing patients within the four walls of the hospital to managing patients across the entire journey, I have found that a number of hospital administrators who were kind of top dog, they were the grownup table at Thanksgiving at the hospital, all of a sudden they're paying second fiddle to some of the ambulatory administrators because all of a sudden that's where a lot of the growth and emphasis is for the organization. I don't think it should be like that. I really don't think that we should be relegating our position as hospital administrators to anybody and in fact I don't know that it's that useful to think about there being a divide organizationally between the ambulatory world and the actual four walls of the world, the health system world however you determine this organization. I think it's much more valuable to be thinking about this entire journey and to think about the journeys as the units. Are you optimizing around the journey? So the journey may or may not have a surgery but the journey always has an access point. The journey always has an exit point. Are you managing it during that entire life circle? I think that that's the shift that needs to happen. And then finally on this last piece, once you've got your network built out, once you have the basic tools, building blocks to manage the patients within the network, are you humming at this point? Are you keeping trains running on time? I mean, we'd all love to be in this state and I think that there are certain markets where I would say that some service lines might be in this state but there's no single market or no single system that is all the way far to the right right now in the country. I guarantee it. But at that point you could argue that our job, how do you grow when everything is under value based care? You grow by signing up employers and you grow by optimizing your bonus payments by keeping patients within a network and getting, having them have good outcomes. That's really different. That's a really different way of thinking about growth and growing orthopedics by 3%. And so the skill set that I think it's going to take to manage growth in that far right environment, this far right environment is probably different then the skill set frankly of people managing growth now. So as administrators are you going to adapt? Are you going to just seed the business of what you're doing now? Kind of growth in the "old fashioned world" quote unquote to this growth in the new world? I would argue that none of us needs to do any of that. I think we can all transition with how our markets are going to transition and I don't think that these are skill sets that are kind of may center or kind of inherent to one individual or another. I think we can all develop these things but it, I think one of the first steps is recognizing that this is a different way of thinking about growth if we actually execute value based care correctly. So let's talk about how we're gonna use big data to facilitate this transition before fee for service and value based care. And let me just start by telling you that we tried at least three times or at least two times and have failed, pretty spectacularly at the Advisor Board and offed them a couple of years ago. The first came in 2012. We called this illustriously the two by two matrix. And on one axis of the matrix you can see here it is on the y axis you have physician revenue share. So this is the busyness metric that we got from our claims database. So I mentioned I ran a program or still run a program called Crimson Market Advantage and plenty 2020. It uses hospital claims. You can get a lot of busyness or volume number from the claims and that's what's on this y axis. On the x axis we kinda tapped a sister program at the time of the Advisor Board called Crimson Continuative Care, and for those physicians that worked with the hospital they had a quality score for each of those physicians. And what we did was very simple. We just put these, we plotted these two points on a grid and so you essentially get four quadrants. You get at the bottom poor quality, not very busy. You're gonna avoid those docs. You get at the top, they're really busy but they're poor quality. You may have an effort to reform those physicians. At the bottom right you have their high quality but they're not really that busy, you might wanna recruit them because you might think to yourself well, I can bring my resources to bear to grow this business. They're already good quality partner of ours, why not grow their business? And then in the top right hand quadrant you're gonna keep these folks at all costs. They're busy and they're good doctors. This was pretty simple. The problem and we made some inroads. We did about 40 markets like this, everybody loved it but then the very next question every single time was you just told us the quality stuff about the physicians that already work with me. I would wanna know the quality statistics, data on all the other doctors in my market 'cause I wanna be able to recruit them. I wanna be able to benchmark my guys and gals versus the other physicians on the market. And what we said was we don't have a source of that data. There is no source of truth for, at least at the time published source of truth for quality data beyond some very basic outcomes, measures and things that are published through like Leapfrog and other things. So, we just, we're out of luck and that was kinda failed. That was a false start. The next project was a little bit more detailed. We at this point had that clinical database and we had had it for almost 15 years at the time and we were kind of aggregating all of this data up. We called it Project Felix. I'm guessing that somebody is a fan of that weird Japanese cat. I'm not sure where that comes from but it was called Project Felix internally, and what we were trying to do is say okay, across all of the data, the clinical data that we've gotten over the years for all of the clients that we worked with, can we build a big enough database that covers 90 plus percentage of the physicians in the market? The answer was still no because even if we were able to get the kinda coverage, the data was old, it was difficult to do risk adjustment, kind of retrospectively and so that program ultimately failed as well but the idea, the concept was almost identical. It was to say okay, not just for the docs you work with but every physician in the market, if we knew how busy they were and how loyal they were and we knew how good they were in terms of clinical quality or some kind of proxy for that, we would be cooking with gas. Then we could truly build a high value network of providers and we could begin to facilitate this transition but it didn't work. So, the Advisor Board was acquired by Optum two years ago. Optum is a big organization and big organizations has a lotta data. One of the assets that Optum had that we realized, hm, maybe we shoulda realized it a little bit earlier but we realized it about a year ago, was this technology called symmetry. Probably meant, anybody heard of symmetry? So what symmetry is, yeah all the, yeah. You're funny. All the people that work on my team just raised their hand. The idea of symmetry is, it's pretty simple. It's really to give payers a standardized way to do episode treatment groups, and episode procedure groups. So essentially you take all the codes from all that outpatient, inpatient codes and you say what codes represent an orthopedic treatment group or an orthopedic procedure group? What represents a heart attack? So if I have a joint replacement patient who gets a heart attack then I can rule the heart attack out of the joint replacements. Simple as that, right? But think about the exercise here. This is taking hundreds of inpatient codes and literally thousands and thousands, almost 100,000 or more outpatient codes and clinically mapping them to relevant treatment and procedure groups. This was an exercise. This technology called symmetry does it. They hire nurses and physicians that kind of review this every six months and they put one code in one bucket and one code in the other and never the twins shall meet and that's what they did. As part of this technology they've also been capturing some cost and care kind of standards for each physician in the market. And so what that has begun to allow us to do is say hey listen, we've been sitting on this. Not only are we defining pathways, we're defining journeys, but we're also saying that there is a standard journey and a physician may or may not be within that standard. So maybe that physician is 2x as expensive as all of his similar peers or her similar peers and we can risk adjust that. Now that, then you start to think well maybe if I use that as a proxy for care quality at least in this first cut, then I can begin to build this two by two matrix again. I can have on the one side, they're doing a lot of business. They see a lot of patients. On the other side we can see that they are either in line or out of line with the kind of cost and care pathways that we've kind of seen in this symmetries worth of data. So I think that this, we're now piloting it. We have two markets that we're working on and on. I think it'll be really interesting to see because if we do this right and if the data is as promised, we're gonna be able to score every physician in a market with this combination of efficiency and business and whether or not they kind of conform to a version of care quality. And I think that this'll be kind of gold for people when they say okay, there are 40% doctors in my market are independent. I have limited amount of capital. I know I need to recruit a certain amount of groups and a certain amount of independent physicians. Which physicians do I target? This data set, having these two data sets over laid on top of one another is going to be the key too scoring those physicians and rank prioritizing them for entry into your system. I don't see any other way we can do this. We need market data on both the efficiency and busyness side and some proxy for quality if not quality itself. That's how we're going to do it. I'm just thinking folks in the organization, maybe raise your hands if you have this data today, is this, are your jobs set up in a way where you're the ones who are doing kind of the selection. This doc or this physician group is in or out or so just raise your hands and say are you directly involved kinda choosing who's in and outta your network? Yeah, not very many of us. I'm guessing because there's this selection bias of folks who came here. Most of you actually recognize you from other things that I've done with you all but a lot of you are planners in strategic planning and business development. A lot of those departments don't actually have a lot of control over selecting the physician and for groups for inclusion or not into a network that's usually kind of on the ambulatory side or the physician practice aside, and I just think that as administrators, if you had access to this data, you could become part of the conversation. You could become part of the engine that's driving inclusion or not and wouldn't it be great to bounce that up against what geographies are important? And whether or not you have the right facilities and in particular geography and that's important. You'd want all those things into account as well. So not just which physicians to add in or out and that's gotta be the smartest way to do it. It's gotta be the smartest way to add all of these different variables together so that when you're making this multi variable decision you have all the inputs that you need. The problem now is that we have all of these inputs lie with three, four, five different departments at the hospital. There's no single executive who's corralling all of them up to one another. I feel like I'm preaching to the choir. I see a lot of nodding heads. So I mentioned the next thing we need to do. So, if we solve this let's just take for granted that the symmetry thing works , that we kinda get our butts into gear and we kinda produce this product there. We've solved one problem. We've solved the problem of identifying and scoring which physicians to include or not. The second though is to figure out what demand is going to look like. So you can figure out how many dots you actually need. What kind of facilities you're gonna need in the future. And this is this kind of overhauling of how we think about value based care and how that influences demand. The first trick of all of this as we've sat down with our data science team across Optum and they have a lotta data scientists, is to figure out what a well managed market looks like. I mean that term means nothing. Well managed market, right? There's no definition for that. There's not even a really sense of argue in a well managed market or not. I never heard anybody talk like that. It's not a natural way of talking about anything. But there has to be a sense of what kind of dynamics of market and when I say well managed here I mean a managed market under risk. So these are patients that are really tightly managed under a risk bearing contract, and so you're really focused on outcomes more than volumes. That's what I loosely mean by well managed here. So, how do we define a well managed market? Well, we had a couple of ideas and we're kinda chasing these down one in turn, in kind. The first is essentially the doppelganger method. So this is to say you look like, Cincinnati looks like Austin in terms of its demographics and I made that up. I don't know if they actually look like one another. Cincinnati looks like Austin in terms of the demographics, their size, the composition of the providers maybe if it's a three horse versus a five horse town, that kind of thing. There's kinda dynamics that you can standardize and say that one city looks like another. And if one city has had more lives under risk then the other, then you look at the last three years and you say well how has that better managed? I'm making a value judgment here that I don't really try to out on it but it's not better just a more well managed market. How does that more well managed market look differently from the market that was mostly fee for service? And the difference, the delta between those two data points or the delta between those two markets is essentially your model. You begin to say well, what happens is, is that cardiac surgeries go down, intro ventral cardiology procedures go up, the need for lab goes up, the need for rehab goes down. I made some of those up but this is the kinda dynamic that you would like to see when you compare one market versus the other. So that's the doppelganger method. The lookalike approach and then you just put a quantity and you quantify all of those differences and you feed them into a mud hole and then you essentially have like, well what does a well managed market look like? Well, it changes this way. That's method one. Method two is the canary in the coal mine model. Or so named. And so what this is doing, it's saying that there are certain populations like Medicare advantage versus Medicare fee for service for example, or populations that are under various employee sponsored programs. So they're just populations that exist out there in the nature in United States that we tend to think of as being better managed then other populations. So what you could do is you could track and follow those populations, figure out how they behave in nature and then use that to say well maybe this would be a way, the theory is is that it's likely that if you managed a larger population in the same way it would have a similar kind of effect on things like demand. And so you take those little tiny micro populations, you figure out how they're different from the rest of the population and that becomes the delta that you then infuse with a model. That's how then you figure out moving forward. Both one and two have a lot of problems . There's not a perfect solution there. I mean, data folks in the room are probably screaming through clenched teeth because there's just, how do you control for bias? Are you doing all the recession analysis that you need to actually identify which things are important and which things are ancillary to the conversation. You need to do a lot of statistical stuff to these to make them work. Or maybe you don't. Maybe you just say hey, listen this is better then anything we have here so let's just try this and see what would happen. So that's what we're doing. That's what we're testing now. I think it's a combination of being really rigorous where we can with our resources and the kinda just accepting that maybe this is how you were gonna model in the future. I think the third is again coming back to this technology about symmetry, it's a little bit different 'cause what we can do here is that we can begin to say okay, we know what physicians actually have very little cost variation and care variation. So they're very either at the exact mean or maybe head of the mean. They're better than average. And if we follow just the patients that those physicians saw by service line and we adjust it by severity and all the things that you would adjust for, could that begin to tell us a story about how if all of the physicians were managing patients really tightly like this, how they would behave in the future? So this is, I mean you could call this a variation of the canary in the coal mine methods but instead of choosing kind of like set populations like Medicare Advantage versus Medicare fee for service here you're saying I'm gonna just follow the patients associated with physicians that I know to be very tight in terms of their clinical management. And those were the ones that become essentially the markers for how we evaluate demand in the future. So, stand by. We just launched this project this summer. We have three different data science groups at Optum working on this, various parts of it. The quotes that I get for running symmetry across the country, I don't even know what that means really but we're gonna run this symmetry model across the entire country and it's gonna take several months and hopefully at the end of this they're gonna spit out a buncha gold but we'll see. I mean this, I think stay tuned because this is interesting stuff. If we could figure this out then we could really build a demand forecast that was smart. Essentially it was just took into account different scenarios. And I think that the way we're gonna present this to you all is in form of a scenario calculator. So we don't wanna presume that we know everything about your market. So we don't wanna say that listen, we know for a fact that in five years you're gonna be 40% at risk. What we'll do is we'll say based on a market of your size and your characteristics, if you are 20% at risk this is what it looks like. If you're 40% at risk, 60, 80, 100% at risk, this is what it would do to the underlying demand in your market. So you could plan and have maybe like a 40% model and have kind of a set of scenarios that you need there and then maybe you have like kind of a rip cord model which is the 80% model and say if the market really swung what would that do? And how kind of SOL would I be if that happened tomorrow? Because we're almost all in fact SOL if that happens tomorrow. None of us have enough ambulatory clinics. None of us have enough primary care and all of us probably have too much capacity on the in and out patient side if this all really did happen tomorrow. The other thing we're doing, this is a little bit less sexy, maybe I should've led with this but this is, we're kind of over hauling how we do demand forecasting as well. I mentioned that demand forecasting, everybody does it the same way. Like I said, we do it this way. We've done it this way for 20 years. It was literally my first job 20 years ago was develop this methodology or help the company develop this methodology. All of our competitors do it this way 'cause there's really no better, smarter way to do it kind of a national basis. So you have to build a client market. Well, so again just a reminder, we take local demographics, age, sex, sometimes some socio economic factors. So if you're older then average that is taken into account for example. But it's national idealization rates which the thing about national averages is that it's wrong for every single person, right? Like that's by definition. An average is wrong for any one individual 'cause none of us are average. Or at least my mom tells me I'm not. And so you take local demographics and then maybe you do some market level scenario planning. You say again, if we feel like we're more technologically advanced then normal or we feel like we're going to be more ahead of the curve on value based care transition then we'll just apply some sort of factor to this forecast in the future. That's how we do demand forecasting, that's how the market does it right now. The smarter way is to do is based on what your market actually looks like. So you still use local demographics 'cause they're not going away. They're yours. But we're not using national utilization rates. We're looking at market level utilization rates. We're looking at what happened in your market in the last five years not what happened in the nation. It seems simple but that's a change that we're implementing on all of our demand forecasts to get to the market level. And we're doing it by populations as well. So you're not just saying okay, orthopedics in my market grew because of demographics by 2% and because of utilization by 3%. You would also say well, that is true but I also know that there's a set of populations. Maybe obese patients or patients with metabolic syndrome or other kind of populations that you identify that consume orthopedics in a very specific way. So you're actually doing two at the same time. You're growing it based on the historic trends of your market at the market level, but you're also trying to smart about populations in your market that aren't necessarily tied to demographics, 'cause you're not gonna get how many metabolic syndrome patients here in your market from looking at a demographic sheet. That doesn't have that. That has age, sex, how much they spend on credit cards. That kinda thing. But the claims data that we have does tell you how many people have metabolic syndrome or how many people have gotten a conservative therapy. Like, let's do spine surgery. Let's say everybody that's gone to a chiropractor had some kind of conservative therapy for your spine, you actually have twice as many as those people then the national average. So I'm not just gonna grow your market based on how it's grown in the past five years and how old and heavily female or male it is. I'm also going to grow orthopedics or spine in this case because I sense that you had this kind of latent demand that's been building up, because you have all these conservative therapies that are gonna flip to more aggressive therapy in the near future. This is kinda the smart market level and population level for demand forecasting that we probably all needed for the last 10 years but didn't know enough. I mean, I've been in this world for a long time and we've just been talking about demand forecasting and doing it differently over the last six months. I think it's because it's not really that sexy to talk about demand forecasting and the dirty little secret is that everybody to an organization and mine, they're all used to looking at demand one way. So it's like screwing with the major league baseball rules. If you mess up the rules and change how the games is played, then all the records in the past are bunk. Like that's what the purists say, right? Like then how can you possibly compare Ty Cobb to Albert Pujols? Like it doesn't work 'cause they're playing different games. I think this is the big fear about doing demand forecasting differently is that if we begin to do it differently and we begin to say okay, there's an opportunity here, how do you bounce that against what you thought three years ago or the investment you made last year? I do think that there's going to be a pain moment that we'll have to accept but it's in service of better data and a more right answer approach. I think we kinda have to bite the bullet, but I guarantee you're gonna get static from your administrators and your colleagues about introducing a new kinda demand forecasting 'cause they just wanna say it the same way over and over, year over year 'cause that's kinda how we think. So, the combination of these two things is a project we're calling Forecasting 2.0. It's a really lame name. Hopefully we'll come up with a better name that's also better then Felix. But in this world we're gonna do demand forecasting at the market and population level, not the national utilization level, and we're gonna allow you to do scenario planning based on how mature your market is on this transition from fee for service to value based care. I know that we've been talking a lot about demand forecasting and maybe you never thought that you'd come to this talk and here me talk this much about demand forecasting. I mean it's definitely not sexy but this stuff has teeth in terms of how we're planning for our markets and if we don't figure out what our underlying demand is both latent and present, then we're, whatever we do is probably misinformed at the worst and probably wrong at the, or misinformed at the best, wrong at the worst. So I think that this, getting this right is pretty imperative. The next thing that we're going to do is this idea of breaking down the barriers between master and strategic planning. We've recently entered into a partnership with an architecture firm called Array. I've always been very frustrated at the advisor board and now Optum where we say listen you should grow spine absolutely in the northwest corridor and recruit three physicians and then they'll say and probably add an ASC and then the, our clients will come to us and say, "Okay that's great. "What does that ASC look like? "How many doctors should it be? "How many square foot? "What's a mix of services in there?" We throw up our hands and say hey, listen we're not architects, right? So, one of the kind of attractive things about this partnership is that you can look at this entire spectrum. You can see everything from the planning and the demand side all the way through like what the facilities look like. More importantly then that though, the Array came to us with a really good idea which is this idea of having a dynamic scenario based model, almost like a video game. Do you remember "Sin City" from the past? Where you could essentially drag, if you built a school but they didn't have roads this school wouldn't do well. But you also could essentially drag a facility from one part of your world or your land to the other and you could see how the people would interact with that thing. This is the same idea that we would do for facilities. If we move the ASC or the medical office building from the northwest suburbs to the southwest suburbs, what would that do to demand? Would there be more patients that would ultimately access those facilities? Would those facilities generate more downstream revenues for us? Wouldn't it be great to have this kind of dynamic scenario planning right at our fingertips? This is the project that we're building with Array right now and I think that it, frankly we wouldn't have been able to have a lot of the facility based competencies unless we'd partnered with an outside firm, an architecture firm. So that's exciting for us as well. And then the last piece is just this golden consumer record. The idea is here again, there is no bad foam between marketing and planning. We wish there were. I think at least, at very least using the same patient data, this longitudinal data we have and by the way we have three major payers worth of longitudinal data now, by having those data assets and letting marketing and planning use that common data asset, maybe they'll be able to organize campaigns in a more cohesive collaborative way. That's our vision. We have a sister organization that it deals with the planning side, sorry, the marketing side of the house called consumer acquisition services. So, we are in the phase of kind of working with our sister organization to make this concept come alive. Thank you very much. Have a great day.
Applying next-generation data and methods
Moving beyond the decades-old outmigration to ambulatory settings driven by demographics and utilization patterns, we discuss how new data and methods predict how markets will transform around changing regulations and evolving patient access dynamics.
Speaker: Kyle Rose, VP of Analytics, Optum
Access the presentation materials