HealthLeaders Media: Healthcare Executives Discuss Different Ways to Measure Value
Creating the Data Foundation Needed for Value-Based Care
Healthcare leaders face enormous challenges as the industry transforms to a value-based model, and developing the right data capabilities is critical to success. With the value-based care transition upon the industry, all healthcare organizations strive for a similar goal—obtaining data-driven insights to make better decisions and improve patient outcomes. In this Roundtable discussion, we explore developing a strategy to acquire, aggregate, and curate data, as foundations to discovering hidden stories within that data that inform organizational decisions. By investing in a robust analytics strategy, insights can be transformed into actionable plans.
HEALTHLEADERS: What are the ways you are measuring value?
SAM BAGCHI, MD: Outcomes. What are our patients experiencing with our care? Whether it’s in the hospital or in the clinic, or whether it’s just your overall health.
MOUNEER ODEH: The Triple Aim objectives are a really great framework for thinking about value. The patient experience is obviously a critical driver of that—the health outcomes and then, obviously, the financial cost. We are at the very beginning in healthcare of this transformation, this data-driven culture and analytics. I spent my career in analytics in different sectors within healthcare, originally in pharmaceutical, life sciences, and diagnostics. Providers are by far and away the most interesting situation. And the opportunity for analytics to drive transformational change is just tremendous, but there’s such a huge gap between our current capabilities and the potential that exists. We’re understanding how involved patients are in their own healthcare services. You think about Medicaid patients: How do you address that? It’s really engaging them even at the community level, so it’s not just the physician-patient interaction. It’s really taking that out to places where they live, not just those interactions that you have with the health system.
HEALTHLEADERS: It’s also that follow-up with them that asks, “Did you get better?”
ODEH: It’s the follow-up. It’s the care coordination. It’s the engagement with their family members. I’m a big fan of what Disney does. They have an operations team. The magic of Disney is actually an operations research team that analyzes. They create these little data points to understand all the various touchpoints that make up the Disney experience, and they analyze that. They look at how long the lines are. When they get too long, they send somebody to give the free tickets. They think about the trash loads. The entire experience is mapped out digitally. We don’t know that as consumers of that service. In healthcare, we’ve got these digital footprints. All these digital engagements that they have, they’re time stamped in our systems. So we have the ability to really understand and optimize that patient experience. So when they are coming in and waiting in the ED room, how long are they waiting there? How long does it take them to go through the health system? What’s going on afterwards? How long is it for the follow-up services that are needed? That type of engagement is really something that healthcare needs to get to. The second dimension is obviously the cost of care. I think we’re all realizing that in order for healthcare to really survive and create value, we’ve got to bring the costs down.
STEPHEN ALLEGRETTO: It’s the battle of utilization versus value, and short term. You’ve got to have a 3% margin, otherwise there’s consolidation. If we don’t understand the value of what we’re providing to our patients, Medicare is going to take back X number of dollars for every patient that we serve. I think it’s ‘How do each of us define value?’ That is one of the basic questions. So you say, ‘How do you measure value? How is it defined?’ We’ve tried to look at value as quality over cost. Again, what does that really mean? You take a look at, ‘Is it a process or an outcome on the clinical side?’ So for orthopedics, we measure the process outcomes that are in our EHR. We also measure the outcome of whether you got a blood clot. And so if you measure those two quality items, we also measure the cost for these patients, but we can’t yet measure the process outcome for every patient population yet, but we’ve been able to at least try to identify that if you can understand that value is equal to quality divided by cost, you have a starting point. How do you look at it over different organizations, because we can’t do it yet? You need to do that across the school of medicine and then our physician entity. And you had mentioned the Medicare Shared Savings Program. We’re in that too, and that same patient is an episode patient as well. There’s no way to merge that data together. I don’t want you to think that we don’t have this all connected yet, even for Medicare. We get all the claims data, but merging that claims data up with our physician organization, with the hospital organization, and it’s the same place and it’s the same episode. Then you have to look at our penalties that we get: the penalties for the Medicare spend per beneficiary, which is now 25% of the penalty. This is complicated, and we have to be in both the utilization side and the value side.
BAGCHI: Another way to think about it is how any competitors think about us as health systems, and so value to a patient or to someone who is in a Medicare Advantage plan may not be whether or not I can get the lowest-cost hip or knee replacement and where I can get it, but it may be how to avoid the surgery altogether. When we think about value in our hospitals, we really have to think about there’s nothing magical about a hospital. I am a hospitalist; I spend a lot of time in hospitals. There’s very few things that you can’t do somewhere else. So what are the things that you have to do in the hospital and what [are] the things that you need competitors whose bottom line depends on not using the hospital, who literally wake up and go to bed every day thinking about hospital days as the primary driver of cost? I think that’s not a 10-year or 20-year proposition. That’s a two- to five-year journey depending on your market. If you’re in a market like in California, it’s already happening. Hospitals are closing. You’re winding down inpatient utilization because the Medicare Advantage multispecialty groups, who don’t have a stake in the hospital game, are managing your days every day. That’s, I think, the big part of the value equation is what’s meaningful to the people that spend money on healthcare, and how do we fit into that. I think you have to be thinking about your efficiency and your market share in your area. I think that’s a little more disruptive way to look at it, but that’s literally how I think unique competitors, multispecialty groups that are getting stronger, value-based contracts are thinking about hospitals. It’s becoming a commodity.
ODEH: This is as much about a shift in the mindset of administrators of health systems as it is changes in the business model and the workflows. The mindset under a fee-for-service is you’re really reacting to, as a physician, what’s in front of you. Now in a value-based world, you have to prove your value, so you’re much more proactive.
HEALTHLEADERS: How do you find actionable data to drive and improve care, and what strategies are you employing to do so?
BAGCHI: You have to have a sense of where your issues are to begin with, and then try to find the data sources you need to solve the problem. And then it’s an analytics problem. For example, perinatal birth weight is a core measure, and these CMS quality metrics frequently seem tantalizingly close to being able to be automatically abstracted. The ECQMs aside, if you just look at things like throughput times or perinatal birthweight, we’ll take the birth weight, the first birth weight we get in the EMR. That is almost always not the correct birth weight. It’s usually the first birth weight that somebody bothered to put into the EMR, which generally is a few hours or can be several hours after birth, which generally is lower than the actual birth weight. To understand that problem, you have to go really understand the clinical workflow that generated that data. It could lead to all kinds of wrong assumptions about the care of our patients if you don’t understand the workflow. So starting with the data in that situation really does not help you solve the problem. You really have to understand what’s happening with patients before you can solve the problem with the data.
ALLEGRETTO: We recruited a clinician from Vanderbilt that came up to New Haven, Connecticut, around managing sickle cell patients. I didn’t know how many sickle cell patients we have in Medicaid—we had 175. But we didn’t identify it as a major area. He said, “I’m now going to coordinate all resources starting in the ED. When they come in with their pain, no one in the ED knows who they are. They don’t go to their appointments with their primary care doc. Then when they’re admitted, no one knows them; are they chronic abusers of the system?” So he redesigned everything. He put a notebook together profiling every one of those patients. It’s only 175 patients, right? And he put that down in the ED. He reduced bloodstream infections. Those quality variation indicators went down. We went from spending over $5 million a year on 175 patients to $2.1 million. ED visits are down; their primary care visits are up. We avoided 300 inpatient admissions over the last three years .
BAGCHI: In the Shared Savings model, you’re right that this year’s performance is our new baseline. It’s a short-term strategy. There’s interventions we know work. And we know also that those interventions aren’t consistently applied to patients that need those interventions the most. In my mind, that’s the gap we’re trying to close with some of the value equation that we talked about.
ODEH: The power of the data is when people understand the strengths and limitations, and it’s really important to educate on those limitations. Then, if you ask the right question, there’s tremendous value from it. Who asks those questions is, I think, the key. And it has to be operational, whether it’s clinical or business folks. They need to be the ones driving that, because it’s a hypothesis-driven process.
ALLEGRETTO: We all have limited time. And I think you need a balance of “here’s what the data is telling us and here’s what the clinicians are telling us.” Because what are we trying to find? Avoidable variations, right? CMS is going to telegraph where some of the places we need to go are, and we need to go after that, because otherwise we’re going to get dinged. Because you need data to help us understand that, and that’s coming through that lens. And then we need the clinical intelligence to say, “Here’s some avoidable variation that the value added for that quality and cost is going to be three times the benefit, and it’s low-hanging fruit.” That intelligence is what we need to include with the intelligence of what everybody else is telling us we have to do. It’s got to be both sides of that. That’s the difficulty in every organization, including ours. Because everybody has a great idea here—I don’t want to go on just my opinion. So everybody has got really good ideas here, so we take five or six avoidable variations here, and then here’s what Medicare and Medicaid [data] is telling us. Or the data that you’re able to pull from the claims. If we can find the low-hanging fruit on the patient care side, because we know 83% of all the costs are on the patient care side, which equates to the clinical realm, it’s not our fixed costs. And again, understanding and engaging clinicians with that data around avoidable variation that they inform is the key to clinical redesign. And we could be at this forever, because it’s every day the variation occurs. That’s why you need the real-time data to understand variation as it occurs. And then what happens if the physician champion leaves? I hate to say it, but in many cases the variation comes back to where it was before, because that specific clinician is passionate about it—they care about it every day. And when you do make improvements and reduce variation, what is the clinician’s biggest fear? That we do away with the baseline. What I mean by that is we have been measuring the improvement over the last five years, so our trend has a baseline from 2012. The physician is concerned that we will replace the baseline with a more current year, so he says, “When you leave, are you still going to measure me for my improvement from my baseline five years ago? Because if you take my budget now, I can’t reduce that variation anymore.” Our ultimate goal is that he uses the data to maintain his level of improvement.
HEALTHLEADERS: Then how do you attract that kind of champion? How do you keep them? How do you attract more?
ALLEGRETTO: This is the key: Physicians and surgeons are not trained to think that way. They’re trained to take care of patients and do surgery. We need to focus on both education and hiring physicians that look at value through that lens of quality and cost. I think it’s one service at a time. And it’s one person at a time.
HEALTHLEADERS: Do you approach the data analytics mission as one of building the ultimate data warehouse for all needs, or do you take a more incremental approach based on discrete population health or cost control objectives?
BAGCHI: That’s a question for the data team.
ODEH: The time of creating a data warehouse is five years, and having the organization wait for that, I mean, that just doesn’t happen anymore. It’s all really use case–driven.
KEN TARKOFF: If we were here two years ago, everybody would be talking about meaningful use. The interesting dynamic that's going on with a lot of legislators is they're saying, “We gave you all this money to put EHRs in, and you told us that our costs would come down once we had all these EHRs installed.” None of us here said that. [laughter] But I’m just telling you that somebody told them because that’s why they ended up creating that. They say, “Okay, well, wait a second, costs aren’t going down. They’re going up. Okay, well, we spent all that money, now what?” Now they’re putting a lot of pressure as you’re seeing MACRA and MIPS and all the other things that are coming in, because now they’re saying, “We spent all this money and it didn’t get the outcomes we wanted.” And I think the discussions, at least that I’m in with a number of different organizations, is the dialogue you’re talking about. Everybody is now realizing this is a serious problem they’re going to have to address.
HEALTHLEADERS: In our September 2016 Buzz Survey, 47% of respondents said interoperability and time required interface is still a major barrier, outranking workflow and EMR integration. I’m guessing that everyone here has their challenges with interoperability as well.
BAGCHI: The industry challenge is there’s a lot of important information about our patients dispersed amongst competitive healthcare providers in any given market. It’s a willpower issue. Do we really want to solve these problems for our patients, who expect when they come to see their doctor that the doctor has more information than they actually do about their care at other sites or at other access points? In general, with some exceptions, data sharing, HIE collaboratives, and things like that have been blocked by competitive forces in markets like Chicago and Dallas, where the EHR does not want to share data with Baylor or Methodist; or [where] Advocate doesn’t want to share with Presence or Northwestern. We worry more about some of the competitive risks of that than the value for patients. I don’t. I think we should share the information.
ODEH: There’s a big gap between where we are today, and the theoretical potential for insights in healthcare is staggering, and it’s really exciting. One of the big reasons for that gap is the complexity, and I think that interoperability is a huge driver of the complexity that is holding us back. So for us it is a challenge. The technology does exist, but a lot of it is a governance issue; even internally there are governance issues. Externally, it gets so much more complicated because of the competitive factors and whether it’s a different health system or it’s our HIE. It creates a dynamic there where you’ve got the providers and the payers also participating in a HIE. Organizations get very, very conservative about the use cases that are permitted for sharing that information. It’s a very long negotiation to expand those use cases. I think that that’s a tremendous challenge for us.
TARKOFF: The interoperability problem of getting the data that you need to be able to manage your growing network is not solved by the HIE, [which] was designed with completely different use cases, and it’s definitely variable across the country. The direct transport mechanism doesn’t solve this problem either. That was a meaningful use–based transport mechanism that is not workflow to solve a business problem designed for a typical health system. You have to be careful about defining that problem, because everyone will say, “Yes, we do this. We do that.” If you say, “How does interoperability work in your expanding network with movement outside of the hospital? How do you get the data that converts to information to manage it effectively?” We all would agree that is an industry problem right now. The next problem is, how do I improve the quality of the source data, improve the quality of what’s captured? How do I provide it in the right way? How do I gain insight to it? How do I get data that’s not documented in the system? Just getting an interface up is the beginning of the problem. Then you iterate, and then you make it better, and then you mix the clinical and financial.
HEALTHLEADERS: At some point, it becomes a common interest of two competitors to share so they can both be successful in the new more value-based MACRA world.
ODEH: Yes. So the permitted use cases tend to be very, very specific to a CMS quality measure. When we talk about things like general population health use cases, everyone freezes because that’s just too broad. We’re not there where we have the trust or the mechanisms in place where we accept that that’s the right thing to do.
HEALTHLEADERS: It’s something really specific like, say, sickle cell.
ODEH: Something like discharge notification. Then it’s permitted: It’s easy, and we agree on very narrow use cases. We have a lot of difficulty thinking about things more broadly than that.
ALLEGRETTO: It’s funny we discussed our sickle cell patient populations. None of us are making money on sickle cell patients. I hadn’t thought about the whole competitive issue getting in the way of data sharing when it comes to Medicaid populations. All of us are losing our shirts here. Of course, we’re going to have to share the data for these patients.
HEALTHLEADERS: I want to ask you about getting to near-real-time, real-time, and predictive analytics. How do you view that migration or evolution to catch things before they happen or as they’re happening?
BAGCHI: I like where some of the EMR vendors are going with the ability to bring more intelligent multifactorial clinical decision support to bear for patients, instead of just that they are a patient on this medication. They are a patient with these conditions, with this age, with these comorbidities. And now they’re bringing forward more differential decision support. I’d love to see our experiential data, whether it’s based on physicians that struggle with certain things. We talked about trachs or respiratory failure. I’d love to see my decision support those best practices applied differentially to physicians who need it the most, for patients who need it the most. We’re not there yet. For me, that’s what real-time analytics is. It’s not just the data, because the data real time doesn’t help your clinicians change their behavior.
HEALTHLEADERS: In the Buzz Survey, only 5% of respondents to our internal data strengths question felt they had a very strong capability to acquire the right data at the right time. Clearly our readers feel that they’ve got a long way to go as well.
ODEH: So our vision is to make health, health education and discovery smarter, the way that we know that [what] we’ve done that is delivering the right insights to the right people at the right time in the right context. That’s the precision analytics concept that we’re bringing to healthcare. Retrospective analytics, they do create learnings that inform workflows, best practices, clinical pathways. There’s a lot of value in that, a lot of low-lying fruit. Also, next-day type of information for many care coordination activities or business activities is very, very valuable. But delivering real-time insights to the clinician, it’s that last mile requires an integration or an ability to deliver that through an EMR. Or through something that somebody is willing to log into, which can be a challenge if you’re asking people to log into multiple systems.
BAGCHI: There are some nice algorithms that are being developed out there to bring 20, 30, 40 different clinical variables to bear as results come in, and then start to predict things that are beyond what normal clinicians will predict.
ODEH: Earlier on we talked about needing to be hypothesis-driven, asking the right question, and the clinicians or business owners are the ones to ask that. Predictive modeling is one area where it’s interesting to take a departure from that. I still think you need to bring the clinician in to synthesize that. Sometimes there are some interesting signals that a datamining, data-scientist approach will start to call out. And then a clinician will look at that and say, “Oh, I understand the mechanism for that, but I wouldn’t have thought. It wasn’t intuitive to me until you showed me that.”
EDITOR’S NOTE: This Roundtable transcript has been edited for length and clarity.