Your Practice Made Perfect

This podcast series provides support, protection, and advice for today’s medical professionals. Brought to you by SVMIC, a mutual insurance company that is 100% owned and governed by our policyholders.


Oct. 12, 2018

Episode 037: One Size Doesn’t Fit All

Surgeon Dr. Bruce Ramshaw and Brian Fortenberry discuss problems in healthcare and how new models for healthcare based on systems and data science could be applied. Dr. Ramshaw’s goal is to learn the real structural reasons that the global healthcare system is failing and what can be done to fix it.

Have a question about this podcast? Contact us.

  • Transcript

     

    Speaker 1: You are listening to Your Practice Made Perfect, support, protection and advice for practicing medical professionals brought to you by SVMIC.

    Brian: Hello and welcome to this episode of our podcast. My name is Brian Fortenberry and today we're going to be talking about problems in healthcare. And, boy, that can be a long list but to help us kind of narrow that look down and talk about some of the problems, and hopefully how we can get to some of the solutions, is Dr. Bruce Ramshaw. Thanks for joining us.

    Bruce: Thank you very much. It's a pleasure to be here.

    Brian: Before we get into this highly charged topic, tell us a little bit about yourself and about your background.

    Bruce: Sure. I'm a general surgeon by training. A little over two years ago, I started here as the Chair of Surgery at the University of Tennessee Medical Center in Knoxville. Prior to that though I've had kind of a journey. I was in established academic medicine at Emory University and University of Missouri. But also I've been in private practice for almost half my career so I really have experience in both worlds.

    But about 14 years ago, I began as Chief of General Surgery at the University of Missouri and I began to study healthcare because I wanted to do a good job as Chief of General Surgery. I hadn't been chief of anything in my life before that. As I studied I began to realize that our healthcare system was absolutely guaranteed to fail the way we were doing it. And that was kind of an overwhelming thing but I felt an obligation again as a division chief to try to do a good job. And I began to look at new models for healthcare, ways to redesign how we deliver care. And then I continued to study and learn what are the real system structural reasons that our global healthcare system is failing and hopefully what can we do about that.

    And so that led me to getting kicked out of two jobs along the way. I spent about five years in a startup because no one would hire me to do this. My clinical practice is really in hernia disease. And abdominal hernia disease really gets reimbursed about the lowest of any surgical procedures so it was very difficult to get an organization to hire a hernia surgeon who was trying to change all of healthcare and it just wasn't time for that yet. But we learned a lot in our startup about how we could apply new models for healthcare that really is based on systems and data science and really a core of it is being able to measure the value of care we provide. And that's what I'm working on as far as implementing those concepts in an infrastructure here at UT.

    Brian: This problem seems to, as we were talking about just a couple of minutes ago before we actually started, it's a hot topic out there today about the healthcare system being broken. But you were talking about this years before it became this hot of a topic.

    Bruce: And that was part of the problem of the timing of not getting a job. Yeah, 10-12 years ago, this was not common knowledge. People weren't suffering as much as they are today. People weren't recognizing the system failures that more and more people are recognizing today. So I think the timing for this is good. And like any disruptive change, you have to go through that startup period where you learn how to do it outside of any established organization because you have to go through some trial and error and that's what we did in our startup down in Florida.

    Brian: When you got started with all of this and when you started looking at it, what were some of the things right off the bat that you saw and you're like "this is a problem and it's going to keep perpetuating"?

    Bruce: Yeah I think a lot of people are recognizing this is that, we're incentivized in healthcare to do stuff. We do stuff. We see patients, we prescribe medicines, we do surgery, we do procedures but we're not incentivized, and so we don't have any infrastructure around measuring the impact of what we do, on outcomes. And especially most critically the value of care we provide. What is the value of a drug device, a procedure, a surgical treatment? What is the value of that in the context of real definable whole processes of care? And for different subpopulations? And in different local environments? These are all complex variables that we haven't paid attention to because we have a scientific paradigm and system structure in healthcare that doesn't accommodate it at all.

    Brian: If we have this healthcare data management system that has been broken the way we're doing it, what are some of the flaws that we have out there right now that really need to be addressed?

    Bruce: So this is not my idea. This is based on science, the science of systems and data. And in data science there's understanding that, number one, data requires context. So when you have data pooled in an electronic medical record across many different types of patient processes, it's all noise. There's no context. So you have to pull the data out into definable processes. So the data for a hernia process, which we've been focusing on, is going to be very different than the data for a breast cancer process or for a trauma process. And you have to understand that context matters. The second thing is having data across the entire continuum of care and a measurement of the outcomes in terms of value for that entire continuum of care.

    And the other thing is the need to decentralize the data into each local environment in context. We're very reductionist minded where we try to centralize the data and get averages. And when you look at centralized data and you get averages that is very predictable for the whole population but, when you try to apply that to any local population or subpopulation of patients, it all falls apart because it goes against data science. In data science, you have to decentralize data into local areas in context.

    And there's a book that I read recently, it's called “The End of Average” by a faculty member at Harvard in their College of Education. And he talks about the early days in the U.S. Air Force where planes were crashing regularly every day and it was not in combat, just in regular exercises. And on the worst day, 17 planes crashed. And they couldn't figure out why. They looked for human error and didn't find it. They looked for mechanical failure... didn't find it. And finally someone realized, at the beginning of the Air Force when they were designing the planes, they took all the pilots that had been recruited and they did over 120 biometric measurements. They measured their heads’ circumference, arm length, leg length, waist circumference. And they built the perfect cockpit for the average pilot. The problem was there was no average sized pilot. They're all different shapes and sizes. And it made it very difficult flying the plane. In fact, in some cases the pilots fit so poorly in the cockpit, that that was what was leading to all the crashes, the poorly designed cockpit.

    That's what we do in healthcare today. We try to push the average treatment for all the patients and it doesn't work. And we need to understand the complex variety of patients' biologic variability we can't control but understand, through systems and data science, we can identify subpopulations and patterns and match optimal treatment to the right subpopulations.

    Brian: Since it isn't a one size fits all like you say, what is some of the real solutions out there? What are some of the real tangible things that we can put our arms around?

    Bruce: It's gonna require us to understand the need to manage data and this is where patient centered care really should be. Its around patient centered, meaning the patient process should be the center of care. So we need to mobilize the data around the entire continuum of care around definable patient process. Once we can free up that data and we can measure value of care for that whole definable process, we can begin to use nonlinear analytical tools to gain insight into what factors in that process are really driving those outcomes that measure value. Then as the clinician, as a clinical team we can do something about that.

    Brian: So this patient centered approach to measuring the value of care... tell us a little more about that. Tell us how that's going to look when it comes to the data.

    Bruce: One of the critical things again against what we're doing in our reductionist model today is that really should be driven by each local environment and each local clinical team. And we have a problem structurally there, we don't have clinical teams for the most part. Although organically, we're kind of moving in that direction. In my field of general surgery, we saw that first with obesity and bariatric programs. We do see bariatric clinical teams now built around the complex problem of obesity and the surgical management of obesity. So we're beginning to see that. Here, we have a soft tissue team that's organically kind of evolved at U.T Medical Center and that soft tissue oncology team has really come together. And when you begin to build clinical teams around definable patient processes, you begin to realize that clinical team can define what are the important factors in the patient care process that matter to outcomes that measure value and, when we gain insight through analytics, we can identify opportunities to improve value. And that's really the original intent of machine learning or artificial intelligence, it's always been required to be a human computing symbiosis. Where the human team programs what goes into the computer, there's a variety of analytical tools that are available and then the analysis of those analytical outputs is what the human team can do to then apply innovative quality improvement ideas to improve the process.

    Brian: You hear in healthcare today at least from an outside perspective it seems to be a more consumer driven business now than it ever has been before.

    Bruce: Part of the discomfort to hospital administrators and doctors is that transparency is coming faster and faster. Investigative journalists are spending efforts into... because of the suffering, you know the financial and the physical suffering that's occurring to all of us; patients and doctors and hospital administrators alike. We're all suffering under a system that's broken and not sustainable and so that transparency is happening faster and faster and what it does is it puts the focus on really the patient as a customer and people don't like the terms of consumerism and customer but, ultimately the good thing is, we should be focusing on the patient process and measuring the value of the care we're providing so we can learn from it and improve it.

    Brian: And you hear this push toward quality improvement and reimbursement models based on quality improvement and these types of things like that. In your opinion, is that a good thing? And if so, how and why? Or if not, give us a little back on that.

    Bruce: So the language is good. The current understanding and application is very reductionist and not appropriate. We're trying to use quality metrics that are mainly process metrics. And there really is no value measurement and it's not in context. So to apply these tools, we have to understand the science behind these tools and how we would need to apply them in each local environment across the whole context of a whole definable process and, ultimately, the most important thing to measure and improve is the value of care. Because if you can measure the value of care, you can decrease costs at the same time you improve outcomes. The corollary is, if you don't measure the value of care, you can't do that. And we've been proving that in healthcare continuously.

    Brian: Even though the trajectory of the healthcare system, not only in the U.S. but worldwide, has had issues and problems and it seems like this inevitable crash course. It's kind of like sitting back and watching two trains headed at each other. Is there a way to avoid the big crash?

    Bruce: I think there absolutely is but it's going to require us to all learn a new way of thinking and to learn a new scientific paradigm and apply it. And learn how to apply it. This is happening in Australia and China and the UK because I go to these countries and I give talks about this, invited because they're having the same struggles we're having. When you look at per capita spending U.S is way higher than anywhere else. But if you look at the slope of increase of per capita spending, it's essentially exactly the same everywhere and it's not sustainable. Because we're all doing the same reductionist science based care model.

    Brian: So when you hear politicians and other people go “we spend more on healthcare than any other organized country”... which may be true but the percentage-

    Bruce: The slope of increase per capita spending is the same in every developed country because the healthcare is delivered in the same reductionist system model. Nowhere in the global healthcare system is anyone measuring the value of care in the context of whole definable processes.

    Brian: Is there anywhere that has got it right? And if not is there anybody that's close that we should kind of look to as they're getting an idea of heading in this direction you're talking about?

    Bruce: No one has understood and applied these systems and data science concepts but, I've seen in the last couple of years, almost everywhere in the world is recognizing the way they're doing it is broken. So the good news is the awareness has come. The difficult challenge is to learn and apply this new kind of thinking in terms of systems and data science applied to actual patient care.

    Brian: What are some examples out there of this improving value that we keep talking about?

    Bruce: The main examples that I know of are primarily from the startup we did down in Florida around hernia disease. Because we had a small clinical team, we had a business analytics unit that worked with us to measure value and to learn how to improve value. Some of the things we learned, there were many available things, drugs, devices that we could implement into the process that cost less than what we were currently using. And if our outcomes didn't suffer, we could kind of easily replace one product with a less costly thing. Now this is a team approach, we would do research, we would kind of as a team learn you know is this really as good a quality as what we're using but it's lower cost and if that was the case we could implement it and then we had lower costs, same outcomes, better value.

    But really the discovery that became more valuable was identifying what are the factors that really impact outcomes in terms of value. And one of the biggest discoveries we had, at least in our own hernia program, was we had a CQI meeting about a year into it. We were looking at our outcomes, we were looking at the patients who had complications and it was actually our Patient Care Manager that said "look, those patients who are having complications, they're the same patients that tend to send me more e-mails, give me more phone calls pre-operatively. They have unrealistic expectations. They have anxiety. These are people who are really struggling before surgery with the emotional state." So we began to measure emotional state. And when we did the analytics, it was the highest modifiable factor predicting outcomes.

    Brian: Really?

    Bruce: So kind of fast forward over the last five years, we worked on developing better measurement tools specific to hernia patients with anxiety and other emotional issues. And it continues to be the highest modifiable factor. So we started in implementing, about 2 years ago, preoperative pre-habilitation to better optimize patients prior to surgery, including for patients who have emotional issues, cognitive therapy preoperatively and our outcomes have gotten remarkably better.

    Brian: How do you get hospital systems or communities or physicians in general to buy into this team approach? Because I'm assuming that probably costs money. So how do you get them to buy in?

    Bruce: So this is where the research around disruptive innovation really helps guide how this happens and the first step is recognition, the way we're doing it, is not sustainable and is broken. And so I think we're getting to that point. There is some initial cost in any transition but the long term value for any organization that does this recoups the ROI more and more over time. Because once you're decreasing costs / improving outcomes, your hospital margin can go up. At the same time your costs go down and your patient outcomes improve. And so it's a systems and data science approach that takes some time and the initial ROI doesn't give you immediate return but, in the long term, massive return on investment. Because if we can get patients better optimal treatment based on their specific factors and needs in the context of each local environment, we can reliably decrease costs and improve outcomes. Again back to System Science 101, if you can measure it, you can improve it. So if you get a structure in place where you're measuring value and you're providing tools and analytics for the clinical team to lower costs / improve outcomes. And when I say clinical team that includes the patient and family, that includes everybody who's part of that process.

    Brian: It's not just the healthcare providers on the team.

    Bruce: Correct, yeah. The team is everyone who touches that patient in that process, including the patient themselves. We've had several ideas for process improvement that came from the patient. For example, the science is all about measurement and improvement and we were measuring wound infections the way that everybody measures wound infections because the CDC defines how to measure wound infections and every hospital has to report it. It's based on the location of the infection, superficial, deep or organ space. But when you talk to patients who had wound complications and you asked them, "Is that a good definition?" They say "That doesn't make any sense." And we ask them "Well how should we define wound infection?" And they'd say "Well how invasive is a treatment required to heal my wound? And how long did it take to heal my wound?" And we started measuring it that way, it was a much better reflection of value than the CDC definition. So as we continuously try to improve our measurement and we improve our process, we can get massive improvement in cost as well as outcomes.

    Brian: So at the end of the day we find ourselves in this, and people put different names around, healthcare crisis, healthcare problem or whatever. And you have politicians trying to invent a solution to a problem. And most of them probably have no idea how to fix the issues because they're not in healthcare. Where do we go from here? What do we do?

    Bruce: So again I think there's a couple of things from the disruptive innovation research. If we can rally around the concept that the way we're doing it is broken and we got to think differently then we can start using examples. And that's what we're doing here at UT Medical Center is being able to pilot examples together. We're going to start with hernia disease and bariatric and maybe colon and vascular this year but we're showing pilot examples based on patient process context where we can actually measure and show how to measure value. We'll hopefully show how to automate data polls from the existing system so we can have the data in one patient process place where we can actually make it usable. And when we can do that, we can show examples of how we can use data and analytics to gain insight into lowering costs and improving outcomes.

    And the neat thing around this science is, it's not one size fits all. So an improvement in another local environment, let's say at Vanderbilt or some other system in Tennessee or anywhere, they'll have different ideas and different insight into how to improve. And if we can collaboratively share that knowledge we can continuously improve forever rather than having kind of a law of diminishing returns if we only do it in one local environment. So the opportunity for team approach, measurement of value, continuous quality improvement, data analytics, gaining insight into improvement, ideas and then collaborations across different local environments that's the future of healthcare if we have a good future.

    Brian: Well I tell you it does me some good to know that we have people like you that are out there that are really in healthcare, analyzing this, realize there's a problem, and are able to sit here and tell me today it's broken but it's not a complete and total destruction path here. And I commend you on being a pioneer 14 years ago before this was even a hot topic. Thanks so much for being with us today sharing all this information.

    Bruce: Thank you very much. I appreciate the opportunity.

    Speaker 1: Thank you for listening to this episode of Your Practice Made Perfect with your host Brian Fortenberry. Listen to more episodes, subscribe to the podcast and find show notes at svmic.com/podcast. The content of this podcast are intended for information purposes only and do not constitute legal advice. Policyholders are urged to consult with their personal attorney for legal advice as specific legal requirements may vary from state to state and change over time.

The contents of this Podcast are intended for educational/informational purposes only and do not constitute legal advice. Policyholders are urged to consult with their personal attorney for legal advice, as specific legal requirements may vary from state to state and/or change over time. All names have been changed to protect privacy.


About our Host

Brian Fortenberry is Assistant Vice President of Underwriting at SVMIC where he assists in evaluating risk for the company and assisting policyholders with underwriting issues. He has been involved with medical professional liability insurance since 2007. Prior to his work at SVMIC, Brian worked in the clinical side of medicine and in broadcast media.