With the advancement of technology in health care organizations are recieving an excessive amount of data. How can organizations manage that data to create actionable insights?
In this episode of Value-Based Care Insights, Daniel J. Marino is joined by Rick Howard, Chief Data and Products Officer for an analytics vendor and former Chief Data officer of Ascension Health. The two discuss the management of data within four categories: claims/finance, lifestyle, and social determinant data.
Key points include:
Using retroactive and retrospective models to manage data
Analyzing the four categories of health care data: clinical, claims/finance, lifestyle, and social determinant data.
Building causation models
Daniel J. Marino
Managing Partner, Lumina Health Partners
Chief Data and Products Officer
Daniel J. Marino: Welcome to Value-Based Care Insights. I'm your host, Daniel Marino. As we work with organizations around the country, folks are thinking about driving their business and overcoming some of the challenges that we've passed of the pandemic but thinking about how to improve some of their performance, improve their operating margins, and even be able to respond to a lot of the risk-based challenges that are faced with their organization. Data and information are absolutely key. And if you talk with a number of leaders, many of them will say, without information, we can't manage anything. It's the number one issue that is really facing our organization. And yet, when you talk with these individuals, when you talk with many of the leaders, I often ask the question, is it that we're data rich and information poor as we've heard?
In other words, we have quite a bit of data, more data than we need, yet we're just not creating information out of it. Or is it that we don't have the right level of data to develop actionable insights or the right level of information? Well, to help us today discuss this, I'm really excited to have a colleague of mine who I've known for many, many years, Rick Howard. Rick is a Chief data and products officer for an analytics vendor and is a former Chief Data officer of Ascension Health. Great guy, Rick, welcome to the program.
Rick Howard: Well, thank you, Dan. My pleasure to be here with you today to talk about this important topic.
Daniel J. Marino: Well, thanks. Let me run this question by you, Rick. Is it that organizations have too much data they don't know, they don't understand how to make sense of it? Or is it that they don't have the right level of data to create the information that they need?
Rick Howard: I think organizations have a sufficient level of data. The good news about the EMR adoption and the, the penetration of EMRs and health care is that we've been able to record data in a machine format that we can use and leverage. Now, some of the challenges that that's created is that we've got a lot of variation in that data, whether it be through data that is custom coded because that's the way the, the users of the EMR wanted it to be done, or whether or not it's introduced data from a third party that is using non-standard coding. So, we've got sufficient data to develop insights. The challenge we're gonna have is you've gotta put that data into a common data model with a common set of syntactical and semantical normalization steps in order to be able to leverage it repeatably and be able to get creative right insights out of it.
Daniel J. Marino: So, what I'm hearing you say is, is the integration of that data across the different platforms become really important, right? So when you have the clinical data, we're recording the clinical data one way, we have the billing data, we're recording the billing data a different way, all around the same patient, if you will, and then we may have claims data, right? Mm-hmm. <affirmative> that we're recording a different way. So the integration of that data is really what the challenge is for organizations, and then really understanding what it tells us is also the issue. Is that, is that correct? Yeah,
Rick Howard: That's correct. But let me expand your list a little bit, Dan. You're, you're correct. We've got, uh, we've got EMR data, we've got billing data, we've got claims data. But think of it this way. We've got supply chain data that gives us an indication of what materials we're using, whether it be procedural or quite frankly, even in a, in an annual wellness visit. We have HR data so that we can understand the cost of our human resources that are being applied to support patient care. None of those data systems are truly integrated and semantically normalized, where we can understand not only the EMR or the clinical activities that are being reported in support of that patient's overall health, but what it takes to be able to support, to be able to provide those services.
Daniel J. Marino: Right?
Rick Howard: And be able to create those cost numbers and those benefit numbers around that particular patient encounter.
Daniel J. Marino: Right? I'd like to think about it with what's the end in mind, right? So, you know, as a friend of mine used to say, if you don't know where you're going, any road will take you there. So if we're thinking about, if we're thinking about this level of information to solve our problem, let's say to understand the cost of care mm-hmm. <affirmative> for a particular service, let's use an example, I don't know, hip replacement and a bundle, right? We need to know what that cost of care is. So it's the integration than of the clinical data of the billing data. But as you said then, almost the supply chain data and then some of the other data as well too, to be able to tell us that that story, you know, that's easier said than done for a lot of organizations.
Rick Howard: Yeah. It's not, I'm not saying it's an easy task, but it is a necessary task. I mean, step outside of health care for a moment. Let's look at manufacturing. Let's look at any other industry. Each of those industries and I used to use this as an example when you pick up a pc, they know what the cost is, where it came from, why it's used, and how it's used for every bolt in or nut or screw inside of that pc. We don't have that luxury in health care because we've got these integration challenges. So if we're truly going to create a cost-effective model of care for our patients, that is not only looking at their beneficial outcomes, but allowing our organizations to stay financially healthy, we gotta start tying all this together.
Daniel J. Marino: Yeah, you're absolutely right. So I'd like to get your opinion since you work with data and you built, you know, in your, in your past tremendous amount of models, you know, when I think about a lot of the analysis that is being done in health care today, it's really trending analysis, right? Mm-hmm. <affirmative>, and I think too, to kind of look at it from a financial standpoint, you know, you would call that technical analysis where you're reviewing what's happened in the past. You're looking at trends to try to predict the future. But when we think about where we need to go and to really get out ahead of some of these issues that are dealing with patients that, that we're dealing with patients, we have to create predictive models. We have to understand what's occurring today, and then be able to understand the potential of what's happening tomorrow. And again, drawing a comparison to the financial field, that's really fundamental analysis, right? So it's technical versus fundamental analysis. It's the ability to be able to identify trending data versus proactive and, and predictive modeling data. Where do we need to go or where do organizations need to go as they start to think about building these predictive models so we can help, to get in front of these, especially as we enter into some of these risk-based contracts?
Rick Howard: Yeah, I think a fundamental first step is even if you are on a single EMR, it really doesn't matter because you do have data that gets introduced in the form of results into those EMRs that are, that are non-standard formatted data. So I think whatever the organization, some of the initial steps are models are gonna be garbage in, garbage out unless you standardize the data that is hitting those models on a, either a repeatable basis or on a one-time basis. So being able to aggregate the data from your EMR or from your other systems, as we just talked about, being able to semantically and tactically normalize that data, understanding contextually what it's telling you. And what I mean by that is think about a physician's note, how much rich clinical information sets in a physician's note that's unstructured. So you're gonna have to build technology that supports that unstructured to structured transition, right? To be able to leverage and utilize that data. So I think once you get a standard data management model in place and understand how you're both extracting, managing, and transitioning the data, and then take the utilization approach of once it's been managed and transitioned, now you can start to utilize it. There's an old saying in the industry, once you've been able to do this, the data, you do it once, but data provides returns repeatedly once you've been able to do this.
Daniel J. Marino: Yeah. Well, ike you said, once you create that normalized factor and sort of, the standards of how it should be entered, how you wanna extract it, how you wanna pull it together, I think that would take you a long way. You know, I could remember years ago when we did our EMR implementations, you know, we would go, um, before we, there was really a standardized approach of EMR. I mean, gosh, if you, if you implemented an EMR, one EMR into five practices, you can do it five different ways, right? Mm-hmm. <affirmative>, because <laugh>, that's correct. It was really built around the physician's practice style. Well, you know, that creates a, an information mess as you're thinking about aggregating that, that data. So you do have to create some level of standardization, so it helps you pull this data together. Otherwise, I would think you're not gonna be able to make sense of it, right? It's not gonna be able to work for you at all.
Rick Howard: No, and you actually bring up a great point, and it's really the point around workflow. Once you get a standardized data set and you understand how each physician is basically following a workflow through what the data you're, you're, you're collecting and understand outcomes, cost, uh, challenges, et cetera, then you start to go in and take a look at who has the best practice workflow, and you can start to then standardize those workflows for those physicians and create a set of standard best practice delivery points that in the end, when you do that, you'll find out is not only are you gonna get better outcomes, you're going to have far cheaper costs. Yes. Associated those better outcomes.
Daniel J. Marino: You are absolutely right. And I'll tell you, you know, where we, we worked, uh, recently with an organization that was entering into a risk-based contract, and they were using the raft score as their risk-based scoring methodology. Well, if you work backward, right? What drives a risk score is the access, where do the acts come from? It's, it's being able to see the patients and really around, you know, your annual wellness visits and so forth that you're capturing all of the codes. And it was amazing to me that this information was being captured in the EMR, yet it was being captured in different places based on the clinical workflows mm-hmm. <affirmative>, because, some of this was being captured by the nursing or the clinical staff, and some were being captured by the physicians. And our inability to create that standardized approach to the workflow limited the organization's ability to extract that data. And they actually underreported the risk categories based on the RAF score.
Rick Howard: Yeah. That, that, that's a huge distinction right there, Dan, because you're right. As we look at the variation that we built into our EMR implementations and allowing custom codes if you want to call them that, to make the physician's life easier, we've actually introduced our own layer of difficulty into the process that we're now having to figure out how to move around, how to make those changes, how to change those behaviors, how to change those workflows so that we can deliver the best care possible. So as you start to evaluate that, what you find out is your, your, your most visited patients, which are gonna be those with chronic conditions, are the ones that are really gonna benefit from understanding how best practice delivers these capabilities and your financial bottom line and how you risk adjusting for that financial bottom line is also gonna benefit from that activity.
Daniel J. Marino: If you're just tuning in, I'm Daniel Marino, and you are listening to Value-Based Care Insights. I'm talking to Rick Howard about the alignment of data with information to create something powerful for organizations so they can start to really manage, and inform their organizations as effectively as possible. Um, interesting discussion, Rick, I really appreciate this. I wanna touch on one other thing that you mentioned, and it was the categories of data. There are a lot of categories out there that drive our information. You mentioned, you know, the clinical data, which comes from the EMR. There's the billing and finance data. Many folks feel like the social determinants of health data is an important element. Um, what are some of the other categories or the things that you see as, as I'm important imperatives if you will to really drive an understanding of what's occurring with the population?
Rick Howard: Yeah, there, actually, I think you've hit the three big ones, but I believe there's one more. Because one of the challenges I believe we face in health care is, is that we are, the data that we accumulate is quite frankly, retrospective. So we're reacting to that data. A patient comes in, with a complaint. We draw blood, we get a lab test, and we make a diagnosis based on that lab test, right? But that's all in the retrospective. Um, as you start to think about value-based care, especially managing these patients proactively, how are you gonna do that with retroactive or retrospective data that you're receiving? So I think we need to open up the horizon of what we believe is clinically meaningful to start taking a look at data that might be more lifestyle driven.
Daniel J. Marino: Ah, yes, yes.
Rick Howard: Do we have patients that have high saturated fat diets, or maybe they are chronic kidney patient that has a high acid diet, and that's gonna be a challenge? So I think as we start to look at some of these things, we can get in front of either a chronic condition progressing or maybe even the diagnosis of a condition. And again, as you said before, through predictive modeling to understand not just the clinical lab result, and radiology result information, but also take a look at lifestyle information and say, what do we need to have a conversation with the patient to change in order to put them on a better track that's gonna be healthier for them?
Daniel J. Marino: You know, that's an interesting point because as you know, many researchers have said, look, if you truly want to manage chronic diseases, you can't just provide the interventions, the clinical interventions. You have to really look at the social determinants of health because there are so many influencers that are out there. But you drink bring up a point that I hadn't really thought of before, but it's really measuring the lifestyle and creating some lifestyle indications of what's happening with that patient. And boy, if you can, if you can really tie that into the equation, I agree with you, I think that could be, you know, quite powerful. Where do you get, what, what are some of the, you know, the categories of lifestyle data, where do you get some of the lifestyle data? I, I get these social determinants of health data, but what are some of the areas of lifestyle data that, that you've seen or you're referring to?
Rick Howard: Quite frankly, what I'm referring to is, and everybody experiences this, marketing organizations have so much information on what you buy, how you buy, where you buy, what your preferences to buy the types of you're buying. Well, then how can that, if you have all of that information, and that is a plethora of information, I get that. And a lot of organizations are not gonna wanna spend time on that. But I believe there are vendors entering the market today that are looking at that data to say, well, how do I extract clinical meaning out of that data? And then how do I turn that clinical meaning into an opportunity to engage my patient more appropriately with a very direct conversation? It's a little not as creepy. You wanna try to avoid the creepy factor there, but you wanna have meaningful conversations by tying all of that, what I'll call raw data into things that are clinically peer-reviewed. Most every doctor understands, almost every doctor probably had a conversation with their patient. Yeah. But they're having a general, not a specific conversation.
Daniel J. Marino: Right. And that's such a great point. You know, and I hadn't thought about it at that level. Are you seeing organizations when you're having conversations with folks, are they, are they considering lifestyle data as a separate category? Or are they just focusing on cost data, clinical data, maybe, you know, dabbling in the social determinants of health data?
Rick Howard: Yeah, even in my past, I've looked at this from the perspective of how do I build a model understanding that I've created a semantically normalized record for my patients, but now how do I take it to the next step? How do I create that, you know, that lifestyle data-gathering opportunity and pull that clinical meaning out? And I do believe now that we're starting to see new entrants into health care, you're starting to see Amazon through one move into health care. You're seeing Aetna and Walgreens get far more engaged in what they're doing in health care. Even Walmart is starting to take a harder look at what they can do from a single-store experience of providing care, providing your prescriptions, and even helping you fill your grocery order based on what they might understand your dietary restrictions are, for example. So I think there are gonna be opportunities for the, for the industry to start to leverage this data to really start making different, making, making conversation and engagement with a patient that's creating a different layer of dialogue and maybe bringing that patient in from an education perspective so that they truly understand the impact of an X factor for them.
Daniel J. Marino: Well, and especially as we think about how primary care is changing, right? So when you, you know, from a hospital or a provider standpoint, traditional provider standpoint, they're leaning their, their clinical care models from the clinical data, right? From the clinical side. Um, I think the Amazons of the world, and, you know, Walgreens now getting into this and Walmart, um, I think they're leading this from the lifestyle perspective to tell you the truth and primary care is an important part of what they're doing, but it's really managing the lifestyle, right? Um, you know, I mean, think about what, what Walgreens advertises themselves on at the corner of, of what is it happy and healthy. It's all about lifestyle. If you're, yeah, you're right. I think organizations have to really take that n there, in their equation as they think about, for instance, how they're gonna redefine themselves within primary care.
Rick Howard: Well, and think about what that is gonna be doing to the industry as we start to change out some of the, what we would consider the primary provider players into some of these, you know, industry participants, but not necessarily at the primary, at the traditional primary end of this. They're going to be looking for ways to manage the patient appropriately, and manage is the right word here. How do you take responsibility for life and manage it appropriately to the best possible outcome that you can come up with, right? So that's what every provider's responsibility is. You've got, I think Optum still is now the number one the employer of, of primary care providers in the country.
Care changing. Primary care is the quarterback. That's the initial interface of every single one of us when we enter, when we go to health care, we go in through the primary care channels.
Daniel J. Marino: So, yeah. Yeah. And you're, you're absolutely right. You know, I saw an interesting statistic on that with Optum. You know, they employ now 6% in total of the primary care physicians nationwide. That's huge.
Rick Howard: That is, that's, that's what I thought I had said I saw too. You're right. That's significant.
Daniel J. Marino: Yeah, you're absolutely right. So you know, I kind of think about when I'm working with organizations, different organizations are at different places, different, different, um, uh, let's say places in their analytics development journey, right? Some of them are pretty far along and, and probably can, can really resonate well with what we're talking about here. But some of those are very, probably rudimentary or in the early stages of developing their analytic capability. If we're, if they're starting to get into, say, a risk-based contract or even a performance-based contract, one of the things that you've spoken about in the past is creating these causality models that allow you to kind of understand how you can start to move that dial where, you know, for some of these organizations that may be resource constraint or just getting into some of their aggregation data, pulling together some information, but knowing that they've gotta do something, where do you believe they should start in looking at some of their data, pulling this together? Well,
Rick Howard: I mean, quite frankly, that's, that's with the entrance of some of these players in the industry. And quite frankly, this lifestyle data is the real driver behind building a causation model. I mean, we can build causation models because we know a patient has type two diabetes and they have an X B M I score, et cetera. So we can infer some of those behaviors from some of those metrics that we're, that we're gathering, gathering. But the reality of it is we don't know exactly how to pinpoint a specific behavior that could be sky, it could be, uh, creating that challenge within those behaviors. So I think you're gonna see the entrance of organizations, uh, I can't speak to one of 'em because they haven't launched yet, but one that's coming out in January that I believe we're gonna be attacking this particular area very strongly to try to create that realization model of do we have a patient that's overeating? Do we have a patient that has a high saturated fat diet we
Daniel J. Marino: Need? Well, and then building on that, tying it to some level of performance outcome, you know, and, and that end result, right? So mm-hmm. <affirmative>, if, if we need to change the habits of, say, diabetics and you know it, sure, it sure is, well, you know, the problem you're kind of solving then is, is reducing the overall cost to care for your diabetic population, maybe focusing on where they're at in an A1C level, but at the end, but at the end of the day, it needs to produce some type of financial outcome for you, right? Because then you're sort of connecting the dots. So I think pulling all of those together, doing this with the end in mind, um, something that I think FOC organizations have to focus on, otherwise, you're just building these data models, you're building this information sort of as nice to haves as, as opposed to must-haves.
Rick Howard: Yeah, agree. And, and, and I think as we get deeper into the value-based care modeling and we start to see a sharing of risk between providers and payers, I think you're gonna see a real transition towards the need to create these causation models, to create these understandings, to create these value models, including cost valuation, right? To be able, to deliver the type of care that we, that we, that our patients in this country should be getting. Um, but we've gotta do it away, in a way that the organizations providing that care are fairly reimbursed and quite frankly, managing the costs appropriately so that they can make the margins necessary to be able to move that organization and to continue to keep that organization, uh, valid. So I
Daniel J. Marino: I think that's the key, that's the means to the end, right? Doing it in such a way where it's creating that, that financial return and using data as, as a, as a tool to be able to get you there. But Rick, this has been, this has been great, and I'm sure, um, a number of our listeners appreciated the conversation and, you know, if they have any questions. Um, you, you're clearly a national expert in this, in this area. Any suggestions as to how folks might be able to contact you or maybe, uh, just connect with you in one way or the other?
Rick Howard: Absolutely. Feel free to send me an email. My, uh, email address is firstname.lastname@example.org, r they can look me up on LinkedIn. It's just Rick Howard there on LinkedIn. If they wanna send me a message via LinkedIn. Happy to engage with whoever has questions about this. We'd love to talk to 'em and start to build a consensus around what we think we need to try to do to support this type of model.
Daniel J. Marino: Great. Well, thanks again, Rick. This has been wonderful. As well to our listeners, I wanna thank everybody for listening today Until the next insight. I am Daniel Marino, bringing you 30 minutes of value to your day. Take care.
About Value-Based Care Insights Podcast
Value-Based Care Insights is a podcast that explores how to optimize the performance of programs to meet the demands of an increasingly value-based care payment environment. Hosted by Daniel J. Marino, the VBCI podcast highlights recognized experts in the field and within Lumina Health Partners