How Real-World Data Is Transforming FDA Approvals, Healthcare Decisions and Drug Development with Dr. Richard Scranton

The real world evidence allows us to look at so many other variables and so many other factors that could show an added benefit or harm or a whole host of other factors. You're listening to Data Nerds in the OR, a surgeons journey toward value-based care. The podcast focused on data and data science as the keys to a better healthcare system. Ideas brought to life by the vision and experience of host Dr. Bruce Ramsha. Each week Dr. Ramsha sits down with different players in the healthcare system to discuss how data is used today and what it could do if it were used effectively. How it could improve surgical quality, education, outcomes, and drive better patient care. Let's dive into the latest episode. Here's your host, Dr. Bruce Ramsha. Welcome to this episode of Data Nerds in the OR, I'm Bruce Ramsha, CMIO at Care Syntax. And our guest today is Dr. Richard Scraton. Richard's been a friend over many years and he's also a Data Nerd in healthcare. And I met him through consulting with Pacera Pharmaceuticals when he was chief medical officer there. And we'll talk a little bit about the work we did there. But first I want to give Richard a chance to kind of introduce himself. He's also got a background in epidemiology, really a storied career and lots of interesting things in his career. But one of the things that we did a lot of when we were working together was talk about data science. I want for you to give the listeners your background in anything you want to add. Sure. Well, thanks, Bruce. Thanks for having me on here today. It's also just great to have an opportunity to talk about data. It's been something I've had an interest in from my very beginning career. I started out as a practicing clinician in the Navy. And there's where I began to appreciate, you know, we had an integrated healthcare system there. One of the earlier ones that we could follow patients. And there were some really good physicians that were good with manipulating our computer system. We populate information about our patients. And so we were doing this before, became mainstreams and then taking that information and trying to find better ways to deliver healthcare was always impacted and collecting really good data and looking at trends and then taking that information. And in this case, making the argument to get maybe additional drugs added to the formulary. So that was my kind of beginning interest. But I found out very quickly I didn't have enough tools. Medical school doesn't really prepare you for that type of a residency to really think about data and that aspect. So I would have the opportunity to come up, go to Harvard where they had a great program to train physicians to be researchers. And I happened to be at a center in Boston, one of the VA centers there called Maverick. And the VA is this great epidemiology center is across. So they're always looking at data. So I happened to be there. We were one of the funded, what they call, Erick's, Epic Research and Information Center. But we were also a clinical trial center. So it was just a great time to get introduced to a lot of other people interested in data. Very large pharmacoepidemiology, research using the VA integrated data. So we had access to look at chronic illness. If you look at the VA, they have some of the largest cohorts of people living to 100 years. And you have all their clinical data and information so you can really begin to understand how medicines or changes in health style can impact the lives of patients. So that's really where it all began. And I continued to build that experience, and it went into the biotech world. And finally, you got the opportunity in working with you, Bruce, to take that earlier work and the love of kind of real world evidence or other ways of looking at data and applying that to our traditional clinical trial design, which oftentimes is my opinion, it's kind of archaic. And I find a group of people call them prepare time and it's really only a sliver of the information. We're only looking at one outcome and it's oftentimes really difficult to tease apart all the other factors that are impacting this outcome. And so what we were able to do was to try to marry both of best worlds to get better answers and better direction. And now I'm at a company, you know, it's the same idea. It's really pioneering how we take medications and we're already looking at ways to track patients and show the improved outcomes from just changing how people think about their medications, how they perceive their medications, all of this stuff that's going to require ability to gather that information. Yeah, that's great. I really enjoy getting to meet you because of our online thinking. And like you said, the concept that only clinical trials or clinical trials are the only way we can generate valid data is just so archaic. And so we talked a lot about the real world evidence, the clinical quality improvement method that we developed maybe, and I know you've had a lot of experience, probably much more experience than I have working on the regulatory process with the FDA, maybe even in the EU. Do you see that evolving there? I've seen it. I think real world evidence is playing a bigger and bigger role, probably still not nearly as big a role as it needs to, but what if you've seen in the regulatory process in U.S. in terms of the transition or the evolution of real world data being used for regulation? I think the only way that we've been looking at real world data from a regulatory standpoint is to look for the untoward side effects or harm. So pharmacovigilants, most of the time you're just looking to see if you see a safety signal. But this often not used to show what if the drug is doing something better, what if this dose is having a better effect and could you use that in faster decision making whether you're advancing a program through earlier phases? It was the same thing, though, in epidemiology, I would say when I was doing work with those large databases, when you try to get publications and you could get things published if you showed a drug did something bad. It was harmful, but one time I was doing work in the VA system and I was looking at statins and I had a funded career development ward to look at some of the side effects from statins, which is when you have a problem where it may impact your muscles and you get degradation of your muscle enzymes and that's a bad side effect. And I was trying to figure out why does that only happen in a fair, a small group of people. But from doing that work and doing my validation, I also found that statins and veteran males reduce the risk for fractures. And it was fascinating because there was actually a biological plausibility for why that would happen, the same kind of mechanism of how statins work also can promote bone. And so we demonstrated that and I had a hard time getting it published because people were saying, well, you're showing, you're showing this association is not proving because not a clinical trial. And what if you're wrong? I said, well, when we do that all the time and we say something, maybe associated with a harm, it's the same coin just on the head or tail, pick it. And I finally, I was able to get that published and the good news is it was picked up by the New Third Times. It was really well received, but it took, I would say, so much work doing rigorous methods of analysis to prove to the reviewers that this was a bona fide finding and that we should take it. Or anyway, so fast forward to the FDA. Same idea. Could we use real world evidence to guide decision making? And we did this actually at Pacera to get dosing for a local anesthetic to children. And we did that leveraging data being mined from a large university as well as looking at claims data to tell the FDA that this is the dose that's being used. So I don't have to go through this long dose finding studies in kids, which would be really hard to do. It's in the surgery center is trying to get some of the sign up for this trying to figure out what the right dose works. We actually did all this modeling and used real world evidence to say this dose is safe. Now let us use that dose to prove that in a clinical trial. And that's actually how we did it. And that allowed us to do one single study to demonstrate this indeed was effective. So I do think there's opportunities, but we've been stuck in the mindset that you can only use it in certain areas. And I think there's much broader applications. Yeah, totally agree. You bring up a couple of really, I think, interesting points. One is in doing the real world data, you're not locked into just one primary outcome. You're looking for anything that might be important. And you can learn so much faster. And it is just not a valid way of thinking to try to prove statistical significance of something that's generalizable when everything's in the real world is probabilities. There's nothing that's hard and fast, 100% guaranteed in every patient. And we try to present that way in a clinical trial with traditional research. And it's just not reality. The other interesting thing is I wasn't as much aware as the work you did in pediatrics with expral. We recently had a project with a different technology, a low pressure pneumo-peritium system that was being used off-label because it's only indicated for kids who weigh 20 kilograms and heavier, but it was being used off-label by pediatric surgeons because you would think lower pressure smaller kids make sense. And we did a CQAC project generated that data showed there was no safety or harm in the smaller kids, instead of the FDA awarded the expanded indication to all-size patients. So I think pediatric surgery or pediatrics is an area, as you know, where a lot of focus in companies in bringing technology to market is on kids. It's on adults because that's where the money is, that's where the larger volume is. But I think using real world data and helping to accelerate innovation for children makes sense. And hopefully your example, my example, will become the norm in the future of world data. It was really interesting. The low anesthetics, they have been around since the '60s, that yet they never did the work back then to actually get them approved in kids. So under the age 12, there were no low anesthetics approved, but obviously used because they're safe off-label. Then since we're coming later to the game, the tradition would be when you have to do a formal study. Well, low anesthetics work the same in kids as they do in adults. All we have to do is understand what the safety profile is. And absolutely what we saw was physicians were using an off-label, but safely because they knew, you know, they understood, you know, the dose and the adjustment. So that's how we're able to leverage that information and what it got sent. Other than that would have probably delayed, if we had to do the kind of traditional pathway, it may have never come. Right. Because it wouldn't be worth the investment, right? Absolutely. That's an argument that I think I hope will be stronger and stronger for the regulatory process. Any insight in the EU regulatory process you've got any experience there? Yeah. Only it's not in the. I mean, obviously they've been great for rural data in many aspects, but on this particular program, they also accepted that whole package in Europe as well. So they were very open and amenable to the same approaches. I have found them in other cases to be very amenable, to using data, to augment your filings and whatnot. But I think also the European do have some great data sources. I've only been able to work with a few of them, but many times we use that data to help inform, you know, a lot of times we're really using data for more marketing purposes and all of that, where we need to really get it is demonstrating those health outcomes are being driven. There was a study that was done in total need there and that was done with X-Pro and that was really leveraging their kind of integrated health care system to track variables that we weren't. We didn't have to capture our traditional way of having anyone fill out a form. So that's kind of another kind of a hybrid approach and we're running trials. Can you augment your data capture or at least your follow up with using these additional ways to collect information? And I think that's another way of doing this. If we want to do our longer registries and make it less cumbersome for the patient or the provider is to be able to track that information using existing data elements. Yeah, I think that's something we're working on with several companies right now that are pre-market. You know, that very strict protocols for the pre-market were, whether it's IDE or traditional PMA process, but we're working with them on collecting other types of data outside of the protocol in the pre-market setting, things like financial data, things that could help a company understand as they get to the market, how they can position a product to generate best value. So it's something where the real world data, hybrid with traditional clinical research can add value to a company and then hopefully obviously patients in the system as a whole. Yeah, I was just talking to a group of investors today and one of the other interesting thing where a data is being used. And I used this to some degree 20 years ago, but the much more advanced now, even targeting getting to the specificity if you're going to launch a drug, you might want to just try it in two cities. And you're going to try it in the cities in which you have data that ties you to these physicians, use medications in this way, you have a patient population that have these attributes. And so now you're going to that, so you want to get your first best benefit of the outcome because oftentimes, you know, when we're launching a drug, you're just putting it out there. And many times the physicians will use the drug on the most complex challenging person. And then when the outcomes look bad, and it's not so in this case, it's really targeted. So you get, and then you learn the other things you're getting feedback saying, well, this is what was working. That's what wasn't working. How do I change your education or training or what do we learn from this? So then you can be more successful in getting the right drug to the right patients and driving, you know, better outcomes from the onset. So I think all of those are things. The FDA has been open to responder analyses and things of that sort. And this is back to you're saying, when we run our trials, we don't have the right patients to know that this subset does better. And we can't capture all those variables because it's too expensive. But if you can start building support systems around the data information, you can hopefully get some correlations and some insights. And that'll describe better and better patient selection. Right. And that's one of the flaws. The output of a traditional clinical trial or even a registry is the average, you know, the average patient rather than understanding subpopulations and matching best value to different subpopulations. That's really a flaw in the current thinking and the current methods in healthcare. But hopefully this is changing pretty quickly. I'll pivot a little bit in that. We talk about data science. We're very passionate about data as you can tell. But something that's coming along with the data science is artificial intelligence, machine learning. That's a term that's being thrown all over the place. I wanted to get your input, your perspective of what is the real potential for AI, ML and healthcare? What's not really reality? So I'm now in a small startup company and what we're really looking for if you're looking at that business is people that can bring solutions to you that incorporate AI because the bigger pharma can do this. They're incorporating AI and a lot of all the work that they're doing. We don't have the ability to do that. So how can you get AI to the hands, people that could benefit most from it? That's the challenge right now. The second thing I would say from trying to incorporate any of this work into clinical trial research, what I find oftentimes is that our clinical trial sites want to go simple. I mean, if they could go back to paper sometime, I think they would just go to paper. And so whatever technology we bring, it has to be seamless. It really has to make their lives easier because what we what happened was with just explosion for example, how we collect data, you'd go in the trial sites and they would have like 20 laptops or 20 iPad because everyone has their own system. You got to use it this way and that's where a lot of errors occurs. I would love some systems that come out that are have broad applications across many systems for integration. Otherwise, it's just hard to get that initiation. But we would love to look at AI for a multitude of things. One of them is you're preparing and filing for your NDA. I mean, well, that's just aggregating a ton of data and putting a very, it's not like they're writing a novel here. You are, but no one would want to read it unless you're a regulatory person. So that's just the type of stuff we would love to try to use those applications. And we got to find the right providers that are bringing those tools. Yeah, there's a couple of things we're working on and we're learning. I think one of them is just the core foundation of data science and how to apply it to health care is always going to require human computing, interaction or symbiosis. So the computing in the AI is incredible power in terms of analytics and things that can do with the data, but they don't know what's important and what's not important. And so you need the human team, ideally a small diverse human team to really understand what's the most important thing to program into the computer. And then you still need that human team to interpret the analysis and then glean insights from that analysis and apply those insights. And that human computing symbiosis I think is not well understood yet in health care, but I think that's a future. The other area we're using some of the quickly evolving Gen AI and LLM technologies to do what you were talking about a little bit getting into the messy, fragmented health care data and acquiring and cleaning that data through automated processes. And we're seeing the ability to do that better and better as the technology evolves. Yeah. Now I was saying I was talking a little earlier, I met with a company and I can see the huge task what they're trying to do with they will need AI, but this company called Karnat, what they're trying to do is bring point of care diagnosis to chronic renal disease, but they want to do it from a population health perspective. So they're talking with rural areas in Africa where they're saying they're about those health care systems, understand that we could collect this data in real time. We could intervene in a prevention standpoint, which would save us so much money, but then we need the track it so that we're staying on the track. So all of this is exactly what we're talking about. They're doing that pioneering work to really prevent something that right today. We just don't know what to do with it because it's such a long time before you go to dialysis, but once you're on dialysis, that could have been prevented if you had a way to identify, diagnose and intervene. So I think that these are great applications for both data and AI. Yeah. Renal disease is one of many areas where we spend billions or more dollars and we just persist in doing the same thing rather than really learning quickly and improving. I think there's a lot of opportunity in those kinds of areas. Once we apply data science principles and measure value, we'll be able to identify sub-populations and match them with better care and prevention eventually because that's when you do enough feedback. It works your way back to prevention and then we can move from a sick care system to a health care system hopefully. Yeah. And that was kind of ingrained in the military too, you know, one thing in the Navy, we every year, even in my age, younger than, annual physical, but they had all these screening questionnaires and you're mental health and all that. We're really important. But the only time that I felt like we got the ball a little bit is that we could have done a little bit more prevention because my argument was more investment in prevention when your active duty means less cost in the veterans affairs, right? So we should just double down on all these interventions. But at least it told me that you can start really early in assessing these things. And if you have a system to track them, as again, in the beauty of that closed system as you do, more and more you can see the benefit of asking more questions following this discrete data, looking at things that perhaps you didn't think were correlated. You need a way to do that in a kind of standard way that you guys have been working on. Again, back to our work that we did very early on together when we were at Posera, it was really challenging, you know, and surgeons and mindset, look, I'm going to go in, I'm going to do this case. And if the person walks out the door, you know, I did a good job, right? And your work was pioneering, say, well, there's, but you can do more than just that. You know, you can look at other options. Could you do it faster? Could you do it with this intervention that would all set these and this subset of population? They don't have time to aggregate all that information and data. So when you look at us coming in with a new pain modality, the one of the biggest learnings I had from doing that was just the whole idea that we're doing minimally invasive surgery so they don't need this pain intervention because it's not supposed to hurt. But the data actually said, actually, this hurts a lot, but it can be easily treated with this approach. Right. So let me talk a little bit about that work. And for the listeners, the zero is the first company that brought a long-acting local anesthetic to health care. You know, local insects typically go to the dentist, you get the numbing in the gum and it wears off in a few hours, right? So the zero developed a technology where that effect, that numbing effect could last days. And so we saw the opportunity, or your company saw the opportunity to use that in terms of post-op, pain control. And like you said, you know, a lot of surgeons, we thought, well, if we're doing it minimally invasive, we're already solving the problem. Well, if you're the patient, it's still surgery. It still hurts, but you might be able to go home a little sooner. But it's not fun to be in pain like that. And especially one of the areas where we worked together on that was the laparoscopic ventral honey repair, where the mesh fixation is extremely painful, even though there's fallen citizens. So the challenge was to show the value as it was brought to market that I think was like $300 per dose and short acting low class X costs a few dollars per dose. So even today, I know there's like managers of pharmacy budgets that won't let surgeons have access to X-Brell. And we showed in some of our work where despite that $300 cost, when you look at the patient outcomes and the financial outcomes for the hospital, hospital can make a lot more money because the patient's getting out sooner. They're using less opioids or having less opioid related complications. So for the hospital, their net margin is maybe $1,000 or $2,000 better despite $300 cost. But still they don't look at their data that way, so they don't know that. We knew those margins already from that surgical billing standpoint were very lean. That was the other challenge oftentimes. And trying to take that kind of system that's broken up in pieces to say, it is all connected. And here's how we had to show that value. We were definitely pioneering the days of trying to integrate those kind of surgical cases and the patient outcomes. And even with the FDA, it was a challenge because I argued all the time doing pain research in surgery. It's really hard to do pain research because you can't randomize someone after a painful procedure to receive a placebo and just sit there and just ride in pain for days. It would be unethical. So but the outcome when you're doing a pain trial is the difference in pain scores. So how do you do that if the placebo is actually getting opioids because opioids are very effective in reducing pain. And so I made a many I testimony to the FDA and maybe arguments that we're looking at the wrong end point. The end point should be can I have you and not on myself have the same pain scores. But the people that got the non opioid are taking any opioids. So the difference is 50 or 100% reduction in the need for opioids. That's the outcome of interest. Now you know, I have patients getting nominated. You don't have in vomiting. They're not dizzy. They're not falling down. They're up and ambulating. Their bowels are then moving. Wow. All these outcomes that we never were really looking for are all associated on one decision. Use this intervention or this intervention. Yeah. And a couple of responses to that. And then we'll talk a little bit more about experel and the opioid issue. And I remember my career when the pain scores became the fifth vital sign and everybody was mandated to treat the pain scores. And before that, I would walk around the post op anesthesia care unit and see people in some pain. You know, they were paying that they got treated, you know, if they're in too much pain. But once they started treating the pain score, I'd go through the post op anesthesia care unit and everybody'd be puking and nauseated. That's a worse feeling than modern aims. So there was a real well intended, but off really unintentional consequences of that whole thing. The other thing I'll tell you, you may remember this during our project, you know, we were first analyzing the data for I forget which application was abdominal wall reconstruction or labential hernia repair. We were looking at the pain score data and it didn't make sense. We saw these anomalies like these patients had experel and it was post op day one. And the scores were zero, zero, one, eight or, you know, one, two, zero, seven. And we couldn't figure it out. So we went to the nurses on the floor and we said, well, what's going on with these pain scores? We can't figure this out. And they basically said, oh, we have to lie because the administration has put a rule in place. We can only give two pain pills if the scores six are higher. And when a patient wants to get up and walk after surgery, they're afraid they're going to be in pain. We want to give them two pain pills so they feel comfortable to get up a walk, but they're not in pain, but we're not allowed to give them two pain pills unless we put eight. And so basically we said, so the data is all crap. And they were like, yeah, so that's when we shifted to morphine equivalents, actual opioids that were given to your point. That was much better reflection of what we were looking for in terms of value based outcomes for that project. That's another thing on the data collection that we did, you know, with you and some other work. They taught us you needed to look, we saw the same trend in the shift change. And so the light shift comes on all sudden we're seeing pain scores and opioid use going up. And what the shift change we're going to like, I'm premedicating. So they're going to sleep the night and I'm not going to be busy. And they were also taught, as you said, I'm not following them all. It was get ahead of the pain. So that was the other thing they were just prescribed. So we had to change all these protocols to say no. And it was very challenging to figure the T's as a part while intended, those were caring for them. And that's really your studies. Yeah. We learned a lot, obviously. We learned a lot by interacting with real world data and having a diverse team work together. I mean, I really enjoyed, I know our team, our clinical team really enjoyed working with you and your team because we all had the same goals. Let's improve patient outcomes and improve value of care. And that's kind of what I wanted to talk about. We hear a lot about value in healthcare. How are we going to move to a value model in healthcare is still dominantly fee for service. We still don't measure value. We measure volume. And I was wondering your thoughts on if, when and how are we going to finally transition to a value model in healthcare because I know we're all passionate about bringing products and treatments and methodologies to healthcare to facilitate that. What are your thoughts on the transition to a value model? So I think the big change is going to be belief in the data, right? For all the reason you just said, people are still suspect of the data. If it's really good outcomes, but it's costly, people then suspect is not true. But what we found was also the reverse. And some of the data collection we were doing, we saw that some surgeons weren't having the same outcomes as other surgeons. Well, in part, that was on us for training, their technique wasn't as good. And so that's some of the other stuff that you can use data to the positive. You have to have the ability when you do these kind of value-based is to be able to both sides have access to data. And in our case, do better education, retraining of necessary. For some cases, just say, you know what, you don't get to do the robot anymore, okay? You take too much time and your outcomes aren't any better. So guess what? We don't have a ability, I don't feel like to do that right now. I can tell you when I was back in my very early, my first study I ever did in when I was a Navy resident was to look at the performance of residents undoing preventive things like getting tetanus shots done in their patients, doing at that time we were doing reptile exams where I'm doing more. And did all this survey data and it was really poor. So the first thing we did is we instituted feedback and we ranked everyone and we showed them their data. And I can tell you post that everybody improved in a preventive medicine. So you need data. You need to be able to bleed the data and you got to give that need back. And then you need to track it again and then probably incentivize all groups at it. Okay, if you're performing better, there's an upside for everyone. And I don't know where they're yet, but you need all those elements, I think. Yeah, I think you're right on target. This is exactly the methodology we evolved over time using data science principles. You have to have high quality data that's believable, you have to do feedback loops. And I do think another component alluded to it, you have to do it in a non punitive way. You have to share that data in a positive improvement way. I think the vast majority of clinicians, whether the doctors, nurse, whatever, we go into healthcare because we want to help people. If we see a data, we're not going to ignore it. We're not going to say, and I think from a surgeon standpoint, if we show a surgeon data around their outcomes, it's not good. They're going to do one of two things. They're going to stop doing that or they're going to learn how to do it better. And they're going to put some effort into that. To your point, we have to get good data in context, high quality, and we don't have that right now today. That's what we're trying to bring in the healthcare and what the whole CQI methodology is about. So I think the timing is finally getting there. I think there's enough awareness that the way we've been doing it isn't, and if we don't measure value, we're going to not be able to improve it, but you can measure something you can improve it. And the problem is we're not measuring value. We're not putting financial data with outcomes data for real patient care, for real clinicians in their everyday practice. And I think we can get there along those lines. We've talked about your company, Lindra, and the technology you're bringing to market. And I think there's a clear value story there for patients and for the system as a whole. So I was wondering if you could talk about Lindra, your work, and how you're bringing better value to patients and the system with your company. That, thanks, the Lindra is the company we're based out of Boston. The technology is out of Bob Langer's lab, who's a people know Bob Langer, he's world renowned engineers, brought many new technologies, who's the person behind the vaccine technology for COVID. So what we're really doing is trying to figure out how to take medications less frequently and deliver medications in the way that we, as clinicians, had hope medications would be delivered, meaning that we're trying to target the drug levels to be in the therapeutic range, and that go too low or that go too high. The way we had to do that though is come up with a technology that stays in your stomach for a week or longer, depending upon the application. And it's really been a high science, high engineering, half of my company are engineers and material science. So we're taking polymers that we commonly use in many medical applications to make a very small star-shaped technology that you take an oversized capsule, for the patient perspective, you're just taking a capsule. But once you swallow that capsule and it enters into the stomach, our technology deploys. And on this kind of star-shaped technology, we load the arms, there's another cool science where we actually take drug and we mix it with polymers so we could control the release. So not only do we need the technologies to stay there for a week, we had to figure out how do you release drugs slowly. And we release drug 24 hours a day, seven days a week, so you're not getting this typical very peak concentration and then a quick lowering level when you take a normal pill. All pills don't like that. It's just how they're manufactured. Unfortunately for us, because we've all realized that patients don't want to take a medication twice a day or three times a day, that we sometimes give a drug where you really exceed the drug level you need, but when you need that, so you will have this tail. So we'll cover you to the next day. So we minimize or eliminate that across all the drugs we're doing. And we focused on conditions in which we think that value to the patient could be huge from a multitude of reasons. People don't take drugs or consistently take drugs for, it's hard to remember every day, you get a busy day that day, maybe you're traveling, you can't think about that medication. But maybe you're also, there's a stigma associated with taking that medication. You don't want to take that medication with you and go to airport security or have someone and watch you take it or so we've moved towards making medicines less frequent, but it's often an injection. If you don't like injections or they have to be refrigerated or in some cases for schizophrenia, you have to go to the doctor once a month or once every three months and get an injection. I don't know about you. I have a dental appointment on Tuesday and there's a 50/50 chance I will have to cancel that by dental cleaning, right? My schedule is in my control. That's not really convenient to the person. What we're doing, if you could imagine after your patient was schizophrenia and you've got a caregiver involved, maybe every Sunday you come over, you have dinner and so just take your medication today and then live your life as normal for the rest of the week. Don't think about it. Your medication is there doing its job, but you don't have to do anything more than just get out with your life. There's many applications we think that has a great application to, but the second thing we did back to your point about data, we're using a drug very well known called resparidone. Resparidone's been around forever, very safe, very effective. It's actually the drug that's been put in those long acting injections. So we went to the FDA and said, "I know I have a new technology, but I'm delivering the same drug we know. So what if we just demonstrate that I am comparable to a daily pill? Can I get approved without doing what encounters the patients through a six month study showing that again, if you're on placebo, you're going to have problems. If you're on resparidone, you won't. They agreed. So we actually did that study with 46 patients studied for five weeks. So that's where the beauty where you can use data, we did all that modeling. The agency was open to that approach. Now we're going to do a large safety study just to show that the platform is safe, but what we then intend to do with that is when we go to do another drug, we will let the platform doesn't change. We just change what's in the drug arm. So I'm going to leverage that safety data and say, well, it was safe there, it's safe here. So we can, again, innovate quickly, bring more and more drugs to market. The other beauty is we can add more than one drug, we can put two or three drugs in there. So we can treat the entire whole morbid illness that the patient's suffering from from a single weekly administration. And then the program we're doing with the Gates Foundation is actually a monthly program for oral contraceptive for women in Africa. Again, think about the stigma there and the barriers to care and access. So a single once monthly would be all that would be required for treating women given that prevention. Well, I'm not surprised with you in the leadership position at Linder that you're having that kind of success and that kind of progress with innovation and being able to move things much faster and thank you for that because although I'm not any medications now, at some point in the future, I need a medication I'd much rather have to take it once a week or once a month than every day or multiple times a day. So thank you for that. So again, thank you for your time. I want to ask you one more question. It's just kind of a question I like to ask a guest and I'll ask Anya to see what else she would like to ask her comment on our world is chaotic health care is chaotic. It's a mess. It's over 20 years ago. I started to learn that health care is really not sustainable the way we're doing it tragically unintentionally. It's harming a lot of people. Are there some books or movies or resources that you've read or seen that are made you hopeful about the data or the future of health care or anything you can find? There's a new one that I'm going to mention. I actually spend probably most of my time reading about is how to make effective teams actually and looking at the outcomes there. I love the tipping point just to try to understand how do you get greatness and my takeaway from that was it's really just doing work. When you have an opportunity as we did go above and beyond and trying to understand and learn and work with great experts. If you have that opportunity, put it extra hours to learn. The other thing is really just how to manage teams effectively keeping them motivated. That's probably what I spend because I manage a very large team about 50 individuals and I find that's what that's my biggest challenge right now. I needed some more data analytics there to figure out if I'm what I'm doing is working or not. I tie a communication and on all my companies it's always communications. I've decided there's no long communication. It's comprehension. I mean, I'm communicating just fine, but they're not comprehending what I'm saying. I know the communication is too way street, but we spend a lot of time on doing that right now. That's probably most of the books I've been reading. That's an excellent theme. One of the things I've learned is if you have a complex problem, you can't solve that individually. You have to have a team and you have to have diversity and that's where some of the challenge of communication is when you have an engineer talking to a clinician, talking to a scientist and kudos to you, that's really, I think, going to be very valuable for you in your work and anybody who's in any complex system. The book that I recently finished reading is called Blind Spots by Marty McCarry and Marty's a friend of mine and I really write some interesting perspectives in healthcare. It's really about a lot of the things we've done in healthcare and promoted in healthcare that really haven't been backed with good data. It's just been kind of tradition and consensus rather than good data and so it's blind spots as the book, but thank you again for your time. Any questions, comments, I really enjoyed this dialogue. I knew I would. It's been too long since Rich and I have kind of nerded out about data and so that was fun. It was fascinating from a third-party perspective. I do have three clarifying questions that I think will help. The first one is really simple. Can you help explain what a statin is and what it treats? These are, that's just the abbreviated name for them. Those are the medicines that lower cholesterol and lipids. They've been around forever. They go by the names of lipitor or crestor, so there's many variations. But they've been demonstrated to not only lower cholesterol, but to reduce heart attacks and strokes and have a multitude of benefits as well as reducing the risk for fractures and men. That's the one that I was like, I didn't know that. Two more are a little bit more complicated, and one is, can you talk to me? You guys both mentioned the need for high-quality data. How do you build trust in the data that you're presenting so that, because we all believe that the data we're presenting is high-quality, but how do you build that, I don't know, brown swell to believe that the data that you're presenting is right and high-quality and good? And that's what Bruce was saying, unfortunately, people have been led to believe that the only way to show high-quality data is in a randomized clinical trial. That's what's been the gold standard. And now fairness, we have been burnt in the past by some observational studies showing some benefits. The best example of that was with the idea of hormone replacement therapy would reduce the risks of heart attacks, but we knew there was a concern that could also increase the risk of a woman's risk for breast cancer. But at the time, if you look at the data and the epi, the epi data would suggest that they're going to more likely die from cardiovascular disease, so do hormone replacement therapy. And long ago, then a study was done showing that indeed that wasn't the case. So I think that's where we have to get a little bit better understanding the decision you're making based on the high-quality data, and maybe again for some interventions, you do need to do some rigorous randomized clinical trials. It doesn't mean that all data isn't good or high-quality just because it wasn't in that structured form and either. Yeah, now I'll add a couple of things. Actually, one of the chapters in Marty's book, Blinds Thoughts, is about hormone replacement therapy, and some of the poor data and conclusions that were drawn that really were true. And I think at that level, I think the important thing is, even at the highest level of quality of a traditional clinical trial, you're never going to have something that's good for everybody or bad for everybody, and those kinds of conclusions, those one-size-fits-all conclusions are just not understanding real-world data science. So having a prospective randomized-to-child trial with perfect data that generates averages says, "Oh, this is good for people." Well, it's only good for subset of people, and it may be harmful in another subset that wasn't revealed by that trial because they were only looking at that primary outcome measure. It's really important to understand, even high-quality data has to be interpreted well. From a reality sample, real-world data, we've learned you have to interact with the data. The way that healthcare data is done is in fragments, it's poor quality. There was a study out of Pennsylvania where they looked at millions of records of a period of a decade, and found that over 50% were just copied and pasted. So it's extremely poor quality. So what we do is we go in, we get the data. We get rid of the noise that's one step in quality, is not getting all the day. For any context, for any patient process that we look at, well over 90% of the data collected has little or no value, and it's just noise, so focusing on what matters. And then understand, there's always going to be data gaps, data conflicts, data errors, data anomalies. You've got to do your best to resolve those in the best way we've found is working with the clinical team who's actually putting in that data and working on those patients. But we can pretty quickly resolve most of those conflicts and errors when we do a few queries and begin to build rules around the data for each of those processes in each of those clinical environments. When we do that, then we get a high quality data set. Yeah, I think it's done. I'll pick it back. I think it just takes work and the systems will get better, and you don't have to look for it. Even for that paper of the show, reductions of fractures, we first had to go in and validate the outcome of a fracture, because maybe an x-ray said rule out fracture. So we had a whole algorithm that you needed to have, if you had an x-ray, had a fracture, and then you went to see an orthopedic surgeon, okay, that was a real fracture, right? Then we had to go in and chart you. We took a 10% subset to see if that algorithm, what was the specificity of that? And that's how we then use that algorithm to weed out the noise to be able to look at that data. It takes work, it takes that type of rigor, and it takes back to, you have to have, you know, healthcare, physicians, clinicians, nurse practitioners, nurses, look at the data to interpret it, say, "Oh, this is what this means, and that's why it's noise." And that's why a lot of these companies are just getting aggregated data sets from like claims data, or administrative data, those data sets are full of errors, and they aren't able to clean those, because they're not going back to the source. You know, the algorithms are being generated by those aggregated data sets are being shown to have unintentional consequences, especially being harmful to marginalized sub-populations and minorities. And that's because they're not doing data cleaning appropriately, that's what we're talking about on you. >>Beautiful. And you got some time it danced around this, and I, so I'm asking you a question. Clinical trials versus real-world data. And I think, actually, Dr. Scranton, your explanation of clinical trials is great, and then Bruce, she followed up with real-world data, and that explanation was great. So between the two of those, I think we covered my question, but I wanted to make sure that people understood what a clinical trial was, and why real-world data is different. And I don't know if you guys have anything to add there. >>I would say there's a real-world data has come by a ton of different names, and I think that's the other thing. You can see companies built, whether you're doing observational studies or registries. These can be flavors of variations on real-world data, or it could be mining, as claims data, large databases. So they're all different, because they have different levels of, as Bruce was saying, of rigor of how you look at that data. On the clinical trial data, when you're running a clinical study, you are controlling as best you can, the patient population, you have a lot of exclusion criteria. There's a lot of people you don't allow into the study for a variety of reasons. We're trying to control, in cases when we do inpatient studies, we'll do phase one studies in a unit where we control what you eat, how much you eat, when you eat, when you go to bathroom. Right? So this is not the real-world, but we're trying to look at the drug effect and try to minimize because you're only to look in a small subset of patients. So if you have a lot of noise in there, you may miss the benefit or the harm for that matter. That's the challenge. It's good for its intended purpose, but the real-world evidence allows us to look at so many other variables and so many other factors that could show an added benefit or harm or a whole host of other factors. Yeah, thank you, Ian. Rich, thanks. I can't thank you enough. That was really awesome. We did a limited podcast, so this is the eighth one. This is our final one. I couldn't have had a better person to be on our ending podcast. You were awesome. All right. Well, thanks. Thanks for joining us for another episode of Data Nerds in the OR, a surgeon's journey toward value-based care. You'll find links in the show notes to any resources mentioned in today's show. If you're enjoying our podcast, please subscribe so you never miss an episode, and if you want more content like this, you can always sign up for our smart surgery blog via the link in the show notes. Or, if you want to ask Dr. Remshaw or the team a question directly, please send an email to [email protected]. This episode is brought to you by Care Syntax, the leading vendor-neutral surgical intelligence platform. The Care Syntax platform can ingest and analyze data throughout the surgical workflow, including clinical, operational, financial, and outcomes data to capture a complete picture of the surgical pathway. That data can produce actionable insights to enhance efficiency and innovation for surgical teams, hospital administrators and MedTech developers, driving innovation for the future of health care delivery. Care Syntax impacts over 3,000 ORs and more than 3 million annual procedures across the globe. Learn more at www.caresyntax.com [BLANK_AUDIO]