Hello, everybody. Welcome to the last but not least general session of the summit. My name is Olivia Gilbert. I'm an advanced heart failure provider in Winston-Salem, North Carolina, and also the vice chair for this year's summit. So very happy to be here with you all today. Very excited about this last session. It will build on the session from yesterday. We sort of did an introduction on inequities in healthcare, and this session will expound on that by focusing on social determinants of health and patient-reported outcomes to better address and talk about some of those. So you might notice some overlapping themes with yesterday's general section there. We are very pleased to be able to host Dr. Tom Maddox for this presentation. He is the vice president of digital products and innovation at BJC Healthcare and Washington University School of Medicine. In his role, he provides strategic oversight and direction of the BJC digital team and the BJC Washington University School of Medicine Healthcare Innovation Lab. He's also a practicing cardiologist, a professor of medicine and cardiology at the Washington University School of Medicine, and a health services researcher. His research interests have focused on healthcare delivery innovation, learning healthcare systems, and quality of care in cardiovascular disease. He's authored over 250 peer-reviewed publications and received multiple grants exploring optimal care delivery. He's also translated this research to clinical practice by designing and directing a national cardiology quality and safety program, CART, at the Department of Veterans Affairs. He's currently a trustee of the American College of Cardiology and has held multiple national leadership positions at the ACC as well as the AHA. We are so pleased to have him with us today. Please send in your questions as he's talking. We've arranged for there to be a 15-minute session of interaction with the audience at the end, and I've seen a sneak peek of his slides, and I know that they're very thought-provoking, and there should be a wonderful conversation to follow. So very much looking forward to it. Honored to have him. Thank you very much. Good afternoon. All right, so today I'd like to spend the next 45 minutes or so speaking with you all about disruptions on the horizon, so innovations in both patient-reported outcomes and social determinants of health, and what we're sort of seeing on the horizon in terms of new ways to collect, analyze, and use of this information for the betterment of our patients. And I think we were going to have a presenter view for the slides. I don't know if that's still available from the technology folks. There's going to be a presenter view of the slides. Is that available through the laptop? Well, I'll get started. Maybe they can adjust it as we go. So like I mentioned, we have two types of data sources that we're going to be talking about today, both of which I think are absolutely critical in, there we go, I think we're getting some of it now, that are absolutely critical in providing quality of care for our patients. Now we're flipped off the slides, so we'll get it figured out. I'll focus on this and what you guys are seeing. So over the next 40 minutes, I want to talk about sort of four general areas. What are these types of data? Why do they matter? Why do we need to be talking about this? What are the both current and then emerging methods for their data collection, analysis, and use? And then finally, how will the NCDR and other ACC programs benefit from these innovations? And by the time that we're done, I hope you'll be able to agree with me on these statements, that patient-reported outcomes and social determinants of health data are really important for both health and healthcare, that digital tools can really revolutionize how we collect patient-reported outcomes in particular, that collecting and addressing the social determinants of health is a really key part of truly providing holistic care, and then finally, that the NCDR use of both PROs and social determinants data can enhance the quality of the care, the research that we do, and innovation in bettering the health of our patients. So let's start with patient-reported outcomes. We're slowly getting caught up here. This is looking better and better. Nice work. All right. So we'll start with patient-reported outcomes. So what are they? So one definition is that they are standardized assessments of a patient's or person's health status that provides a global perspective of their overall well-being. And the key aspect of PROs is they are defined by the patient. No one else but the patient can actually know what they're experiencing and what they need to report. So we know that patients themselves can tell us their experience, their definition of perhaps pain or fatigue or anxiety or any other type of patient-reported symptom that they may be having. And we can only get it indirectly as clinicians. We can certainly see if somebody's in pain, if they're grimacing or something like that, but in all cases, all we can do is have an indirect assessment of that. So it's absolutely critical that we have a method, a mechanism to get patients to tell us what they're experiencing. And by doing so, it prevents our misinterpretation, well-meaning as it may be, about what a patient's subjective experience is, which is ultimately one of our goals in optimizing as a care team. And in many ways, I would argue PROs is the ultimate patient-centric outcome. So PROs measure across multiple domains, and there's been quite a bit of measurement science applied to how we do this. We typically think about it in four different domains, which you can see represented here. So one is sort of the overall global health, the overall sense of well-being and how somebody perceives the summation of their overall status of well-being. Mental health is obviously a key aspect of this, and we talk about things like anxiety, depression, anger, et cetera. Physical health, of course, is a key part of health as well. So somebody who is able to function as they wish physically, they're experiencing fatigue, pain, pain interference, sleep disturbance, sexual dysfunction, other parts of physical functioning. And then finally, one piece that I think is also important to bring is social health. So what is a patient's ability to interact with the world around them, and the communities, the family members, other people that they need to have an overall sense of well-being? Another sort of important thing about PROs is that we often think about them in terms of a generic assessment of their overall well-being and then disease-specific patient-reported outcomes. So like I've mentioned, we wanted to see how somebody's doing overall in areas such as fatigue, sleep impairment, depression, or physical activity. And then in cardiovascular care, we're often looking for different aspects that we know are related to specific cardiac conditions, whether or not they're having dyspnea or edema, for example, with heart failure, whether or not somebody's experiencing angina if they have coronary disease, or if somebody with AFib is experiencing palpitations. And often what we find is that in good PRO assessment, it's a mix, that we're often asking people for both assessments of their overall generic health status and then applying different disease-specific assessments on top of that. And generally, it's the intersection of these two that's going to give us our most holistic view of health care. Now there's been a lot of work in sort of standardizing how we do this. There's a lot of psychometrics and important aspects to ensure that we're accurately collecting the degree of patient-reported outcomes and that they're consistent and measurable over time. And a lot of the work has actually come out of the NIH. Back in 2004, they actually started developing an overall systematic approach to this. The Patient-Reported Outcomes Measurement Information System, or PROMIS, is the acronym for that. And this now has built into a very robust library of standardized series of PRO surveys, typically in the generic realm. They're disease agnostic, but they've been applied to multiple health disciplines, cardiovascular care, cancer care, mental health, all the other aspects of health care. And they are widely available and used both in clinical care and in research studies to characterize it. They're also freely available in multiple languages and multiple literacy levels so that we can access them in a variety of our patient populations. And we find that using these tools is often a very valuable way to start assessing PROs in our patients. We also need to think about cardiac-specific PRO assessments. And there's a lot out there. I've listed a list here of 16 or so. And you can see the domain that they tend to measure and then the aspect of cardiovascular disease or cardiovascular procedures that can give you information here. And you can see that a variety have been done both in coronary disease and structural heart disease, typically valve disease, as well as ways to measure before and after states in various cardiac procedures, most commonly here, PCI, but we can also see it in various procedures in congenital heart disease and peripheral vascular disease. And what's also important to notice here is it's a mix. They're primarily disease-specific PROs, but there's also some generalized assessments as well. For example, the HRQL SF36 is an overall measurement of quality of life. And we know that being able to sort of mix these up and provide multiple assessments from different domains is the best way to get a holistic view of our patients. Now once PROs are collected, they have a wide variety of uses. And I think this graphic is a nice illustration of how we can use them. So for example, in value-based care, they are becoming a more frequently asked for domain in how we're assessing the value of care that we're providing. Is our care meaningfully affecting outcomes that are important to patients? In registries, and you guys are very familiar with this, in CDR and elsewhere, it's a key measure to track the natural disease progression of patients participating in the registry, as well as the treatment effectiveness as we try to administer it to them. It's also a measure of real-world evidence that determines the impact of treatments in the general population. It's a key parameter in patient management plans and allows us to track patients over time and see how we're doing in terms of optimizing and maintaining their optimal symptom burden. It's a measure of effectiveness and cost-effectiveness analyses. We now know that the so-called hard outcomes, such as mortality and re-hospitalization, are only a small part of what a patient is experiencing. And we're recognizing more and more that true cost-effectiveness needs to measure the impact of symptoms from their point of view. It's also a parameter to track patient experience and opportunities for improving our service as healthcare delivery systems. And it's also a way to engage our patients in tracking and managing their health. We know that without their optimal engagement, none of our treatments will reach their maximum potential. So being able to do that in partnership with a patient is something that PROs can facilitate. And then finally, of course, it's a key outcome in the research world, as we look at research studies of both therapies and then health services to improve those. So what about NCDR in particular? So we, right now, do collect in various measures PROs in various parts of our registries. Five, to be exact, you can see them listed here on the left. And you can see the primary PRO tool or tools that are currently available in the case report forms. You can also see where in the course of evaluating the patient we collect them, be it pre-procedure or a follow-up, both. And you can see that right now we do have variability. As you can imagine, if we're requiring the data element, then we see high rates of its collection. For example, in AFib ablation, we see that the quality of life assessment specific to AFib is collected in almost 100% of cases. But in other cases, for example, in Cath PCI, it's not currently required. And as a result, obviously, it makes sense we're not going to be collecting it systematically yet. But by having that available on the case report form, we now have the beginning of the platform to be able to systematically collect it. And it may be something that you want to evaluate at your own particular center. So all of this is great. I think to me it's clear, hopefully I'm convincing you, that PROs are an essential part of providing high-value care. And that they can really open our eyes to important pieces of providing that holistic care. But now we need to be thinking a little bit about how can we do better? How can we make sure that this really is a key part of all the health services we provide? Not only in our registries, but just in our overall approach to health care in general. So what we wanted to do first was what do patients and clinicians want from PRO collection? And AHRQ actually did a survey, this was published back in 2020, that asked not only patients and clinicians, but also our partners in technology and IT, about how can we start thinking about collecting this? What do we need to collect? So with patients, they were fine engaging with us and collecting their outcomes, but they really wanted it to only be around their active health issues. They certainly were quick to say, don't overwhelm me with assessments. We don't need to do this all the time. But if I am actively managing my coronary disease or AFib or other aspects, then I am interested in tracking that to make sure that I'm achieving my optimal outcomes. They really wanted us to focus on usability, make it easy, make it quick, make it comprehensive, but don't burden them unnecessarily. They did want bidirectional communication. They were often finding that they may send something and never hear about it again. And so they wanted to be an initiator of a conversation rather than just something that goes into what they perceive to be a void. And then finally, of course, this is private information. And when we're talking about things like symptoms or particularly sensitive subjects like sexual function, they want to feel good that this will be protected with the same care that all their health care information is as well. When we talk to our clinicians, they too were not interested in the burden. Go figure. I think all of you are familiar with that. So they wanted to make sure that it was the key patient populations that we were asking for this information and that it would provide actionable information. None of us want more information that we do nothing with. We're already overwhelmed. So how do we both receive the information, understand what it means, and then know what next positive action should flow from that? We need it integrated in the EHR. We all know that we have a million different electronic inputs of information. We need that to be aggregated and connected to the ways that we work in our electronic health records. And then finally, usability is really important too. It needed to be an intuitive way to access the information, visualize it, and interpret it. And I think we'll talk a little bit about how we can do that. And then finally, our technology colleagues also brought up some really important parts. They wanted open standards so that we could see this be similar across different platforms, that they needed to have the data sharing. Just because you're interacting with one EHR, one health system, and something else next year, we need to be able to move this data. It needs to be liquid. And that we need both leadership support and EHR vendor support to make this work well. So for these as-for needs, and I would argue their personalization, ease of use, and EHR integration, it suggests that, in my opinion, digital tools really do have a lot of key advantages in PRO collection. And fortunately, we're starting to see the emergence of more and more tools and companies and approaches to PRO collection that I think are really promising. You know, importantly, digital tools can start to collect this information outside of the traditional healthcare encounters that we have. You don't have to wait for the clinic visit. You don't have to wait for, hopefully, not a hospitalization or coming in for a procedure. It also allows for a dynamic checking over time to see how the trajectory is going in terms of symptom burden. It allows for so-called computer-assisted testing, where we can just intelligently ask questions. And if it turns out you're not having an issue in a particular domain, digital tools can sort of move you to another part of the questionnaire and really make it a much more efficient process in collecting where you're having problems and honing in on those issues in particular. And then finally, of course, the surveys could be easily adapted. If we're seeing that a patient has a certain literacy level we need to respect or a language preference, we can edit these tools accordingly. And the good thing is that by doing all this, by personalizing all of this, that we are likely to improve the engagement and compliance that we're going to see of our patients in participating in this. And it's only if we have this data that we can act on it. And so really maximizing their willingness to engage in their ongoing partnership with us is key. And these tools obviously do help a lot with the privacy, not only the data privacy, but it's a private way to ask for sensitive issues. And so I think there's a lot of potential there as well. So let's talk about a couple of examples. So I'll give you one example. Like Olivia mentioned, I'm at Wash U in St. Louis, Missouri, and we have a very active sort of entrepreneur community. Often a lot of our med students actually are big drivers of that. And this is a company they've built. It's called CareSignal. And what they were realizing is although digital tools are really good, a lot of people are uninterested in yet another app. And actually a lot of people have access and are familiar more with SMS texting and would prefer to do that. It also doesn't accidentally exclude people who can only afford an SMS only phone as opposed to an overall smartphone. So what they do is, depending on the patient, their needs, the care team needs, and the particular diseases they're experiencing, they push out very customized PRO assessments where really the answers are just single numbers going back to the patient. On a scale of one to five, what's your angina burden? Scale of one to five, how are you feeling today? And what they're finding is it's so short, it's so quick, and they're able to deliver it at opportune times for the patient, that they're getting very high adherence rates. And often the patients are feeling actually more cared for, that somebody is reaching out proactively and asking them how they're doing. And if it turns out they give a signal that they need additional help, there's a nursing pool that's ready to respond on that often in the same day, sometimes within the hour, to reach back out to them and ask them how else they can help. And we studied actually the impact of this approach in our primary care ACO and we found high rates of engagement up to 80% of people were responding consistently and we were able to start to reduce some unplanned readmissions as well as overall total cost of care. Another example is a company that some of my colleagues at Cedars-Sinai and the University of Kansas have started to use. It's called Noteworth and this is an app-based platform. You can see some examples here both on the phone and a representative laptop screen. And a couple of things that I think are important about this factor is for the clinician's use. So you can see not only are we reporting single points of data, but we're actually looking at their trajectories and we're cross-correlating it with particular changes in management plan or changes in what that patient has been doing. And it allows us to start to think really strategically and in a personalized way about where do we need to intervene to really change what this patient is experiencing. If they tend to be a bit more active and it's at high activity levels, they're experiencing some angina, then it helps us think a little bit more about maybe some PR on nitroglycerin on top of their background therapy or other ways to adjust and optimize PROs based on the trajectories and trends that we're seeing. A final example I'll show you here is a different company called CareSense. Some of my colleagues at Houston Methodist have been using this one. And what they have done is taken that PROMIS library that I talked about earlier and really imported it into an entire library that sits inside this app. So clinicians really can sort of puzzle piece together the collection of surveys that really are going to give them the best sense for that particular patient. And what they've been using it for is both quality of care assessments as well as bundled payment requirements. So in all these situations, what we're seeing is examples of how digital tools can be personalized, how they're integrated both within their own ecosystem and then back into the EHR, and that they're providing a lot of what we were hearing our patients and care teams wanted when it came to PRO collection and use. So once we do have the data, we do need to think about how it flows back into our source of record in the EHR. This is an example that we have actually from our orthopedic clinics at WashU. So we have a chronic pain clinic within our orthopedic department. And as you guys know, there's a high co-occurrence of chronic pain and depression and anxiety. So what we were hearing from our patients is we needed, we were missing the opportunity to both detect depression and anxiety symptoms and then address them as part of their overall pain control strategy. And so what some of our teams did there is they used the PROMIS questionnaires. In this particular case, they were using a tablet-based interaction that was in the waiting room of the clinic. They're talking about moving it to the patient's home, but at least right now, when they're coming into the clinic, they're filling out their scores for both an anxiety questionnaire and a depression questionnaire. And then it's being tracked. You can see some examples, if you're familiar with the Epic interface, flow charts, graphics, other things that show trajectories over time. And what this enables the care team to do is kind of see where are we, are we seeing problems, and do we need to start enhancing our care plan, not only to address the source of their pain, but some of the mental health ills that might flow from that. And we saw in our initial outreach to our clinicians huge amounts of excitement in adoption of this. And that our orthopedic surgeons, let me say that again, our orthopedic surgeons were asking about depression and anxiety. That's a huge win right there. Another example I'm going to show you is a platform that we're just now putting into place in our hospital system in St. Louis. This is OneView, and it uses a tool in Epic called Epic Bedside. This now is starting to allow us to be a little bit more proactive in using some of the in-room technology for hospitalized patients and move beyond just showing them a TV program or ordering something from the cafeteria. And now what we're doing is being able to show them where are they in their course of their hospital care and have them give us feedback on are we making any improvements on the symptoms that brought them out. And so as we start to build this in, it's yet another example of digital tools integrating into our overall data flows and care plans. All right, so that's digital tools. Let me talk about a few other techniques that I think are on the horizon and I think will be really impactful around PRO assessment. So this phrase, ecological momentary assessments, EMAs, is probably something to keep in mind. And this has been a technique that's been used for a while in a lot of research circles and I think is maturing to the point now that we might see it in the clinical arena. So EMAs are a method for acquiring the repeated collection of a person's momentary experience in daily life. And so this allows us to understand sometimes down to the second level of what somebody is experiencing. And so you can see an example of what this looks like in an app format. And you can see these questions. What were you doing right before the phone went off with the alert to fill out this EMA? What kind of physical activity were you involved in, assuming you checked physical activity? If you said I'm doing another activity, it says what's this other activity, gives you some options. And then asks a couple of other environmental questions, were you sitting, for example. And what this is meant to do is be a very, very quick hit. It's a very low burden to the patient and just very quickly allows them to say today while I was cooking, while I was picking up my kids, while I was out for a run, this is what I was experiencing. And what it allows us to do is start to pair that with other environmental assessments of where they are and really start to get even more personalized in both the experience the patient is having and then how we might treat that accordingly. And we have one example from a pilot that was done around obesity treatment. So these patients had the EMAs as part of the phone represented in the middle of the slide. And then they had an activity monitor, an accelerometer to track their movement, and actually a bite counter. So this was able to actually track how often they were eating. And what they were able to do is sort of pull together all the information about what were the triggers for somebody when they might be overeating. Was it more in the evening? Were they watching TV? Were they out and about with their kids and the snack bar was right there? What were the sort of environmental state that was correlating to symptoms and activities that would help guide a very personalized care plan? And so as you start to give insight, and often the insight was coming to the patient even before the care team member, to say, ooh, boy, my trigger is putting down that bag of Tostitos while I'm watching TV at night. Or other things where they realized, oh, these environmental cues, maybe unconsciously, are leading me to some activities that aren't going to help with my health. So how can I start to think about modifying that? So again, there's always going to need to be a balance between tracking, privacy, and burden. You can imagine that can be over-rotated in certain situations. So it's very thoughtful to use this strategically. But done well, and we've seen this play out both in this pilot and other examples of it, can be a really powerful tool that gives us information that we really haven't had before. The sort of final area that I'll talk about is indirectly assessing PROs. And indirectly is a tough word in PROs. Like I said, the patient has to tell us how they're feeling. But at the same time, we know that we can take advantage of the fact that we're all glued to our phone with its accelerometers. And we have other ways of being able to track different types of activity. In this particular pilot, these were heart failure patients, and the investigators were looking at ways to use an actigraph, which is just an external belt that measures movement, and then the iPhone-based accelerometers to say, can we start to correlate to some of our more traditional activity assessments in the clinic, the six-minute walk test and other elements as suggestive of a patient's overall physical activity. And what you can see here is they were able to make comparisons to sort of the traditional measures in the blue with some of the digital device-gathered information in the red and show high degrees of correlation. And so what we can do is by pairing this with direct inquiry about the patient's experience, start to understand where are they from a physical activity point of view and what can we do to optimize that. We can also start to see a lot of indirect information from interaction patterns on a phone. And this was a really interesting study coming out around Parkinson's patients. And so they looked at the five different domains you see here. They would basically have little almost games on the phone and ask patients to do certain sort of physical activities that could give a sense of how active are they currently and are they seeing any changes over time that might herald a change in their, in this case, Parkinson's progression or improvement as they use Sinemet or other Parkinson's meds to improve it. So being able to tap and understand some of the granular aspects around the dexterity, the speed and any abnormalities and kinetics of finger tapping. By being able to record voice, you can start to measure a lot of granular changes in the loudness, the pitch, the breathiness, the roughness, and any vocal tremors in the voice. And then, of course, walk activities, the gait, and not only the gait but the balance. Are you favoring one leg over the other? You know, are there other aspects that we need to be thinking about that might have an impact on their health? So all of these can be correlated, obviously, to PRO domains around things like physical activity, quality of life, and symptom burden. And then finally, once we have this data, can we start to use it more intelligently? And I think we're all aware of the increasing sophistication of our models and using machine learning to really pull out insights that traditional modeling approaches haven't had. And this is just sort of a current summary example of how PROs are enhancing the accuracy a lot of our prediction models. And you can see here on this graph, we looked at sort of a collection of studies that looked at post-surgical improvement, depression, pain management, hospital readmission, or oral health. And they wanted to see if the introduction of PRO measurements improved the ability to predict trajectories of these various types of health outcomes. And you can see that they measured it over time, either short term, within a year, or longer term, more than a year. And they were able to validate it both with internal and external data sets. But in all cases, we saw significant gains in the accuracy of these models, because we know that patient symptoms, as reported by the patient, is a unique and complementary data element to all the other more traditional health care measures that we tend to assess. So how will all these potential innovations inform us here at the NCDR and at the college? So I think from a data collection point of view, I think we are on the threshold of starting to introduce more PRO tools in data abstraction, and importantly, to have them directly integrate into the EHR. We also think that with the digital tools around app-based tools or texting tools, that we're going to be able to have an easier way to do data burden, and not just rely solely on the manual abstraction methods of the past. And we think through, both with EMAs as well as indirect activity collections, that we can really start to improve our holistic assessment of a patient's status and how to improve that. I think from a data analytic point of view, we can greatly improve our risk prediction. We can start to bring in time and location-based assessments to further personalize our care plans. We can do trajectory-based analysis to understand these changes over time and with management plans. Start to do layer data analysis of their clinical state, their physical activity, their symptom burden, and sort of correlate all these different aspects of their health together. And then ultimately, to be able to use it in a variety of ways. We're going to be able to understand better the quality of our treatments. We can potentially enhance some of our accreditation programs by distinguishing those top-tier programs that are reliably collecting and using PROs. Our new patient models that are emerging are going to start tracking patient quality of life, and so we need to have this data to inform that. We'll be able to provide more in-depth post-marketing studies for new devices, new procedures that come out, and really see what their short-term and long-term benefits are on patient symptoms. We'll be able to tailor the appropriateness of our care and management plans. For example, if we see really high burdens of angina in subsets of patients, maybe we target them initially for elective PCI. Risk adjustment using baseline PRO status will be really important to benchmark our care and measure the quality of care and the delta change that we can have by our treatment plans. And then finally, for clinical trials, PROs can be used for both primary and secondary outcomes of these trials, and they can also assist us in adverse event reporting. So that's PROs. I hope I've convinced you of the first half. So let's move now to social determinants of health. So what are social determinants of health? And you've heard this said several times, both at this conference and I'm sure in your own daily lives. My definition that I've been using is social determinants of health are the non-medical factors that influence health outcomes. And this is just one way to categorize it. There's a variety of ways, but I think an easy way for me to think about it is these five domains, economic stability, jobs, et cetera, education, and that includes general education as well as health literacy. Health and health care, and this is primarily do you have access to it, do you have access to high quality care. The neighborhood and the built environment is critically important. This not only is the home in which you live, but the neighborhood and the social environments in which you interact. And then finally, related, of course, is the social and community context and the overall world in which you live, work, and play. And how is that either helping or harming your overall health? So the World Health Organization also defines social determinants of health as conditions of circumstances where patients are born, grow, live, work, and age. And that these conditions are shaped by a huge number of factors well beyond health care, including political, social, and economic forces. But it's important that we understand this context because it is within that context that we can think about what is, what are the realistic expectations and the modifications we might need to make in our care plans and interactions for these patients. There's also an emerging focus on the impact of social determinants and quality reimbursement and health outcomes. So for example, in the hospital readmissions reduction program, we see that there's increasing attention on the effect of social factors on outcomes. And we're starting to include this in how do we hold hospitals accountable for their outcomes as a part of their quality of care? Are we taking into account the social context in which those patients are receiving care? And then how do we reimburse accordingly? We've talked about penalties for no longer rewarding suboptimal quality of care. But do we need to think about increased reimbursement if we're seeing a patient population that needs more investment in the social safety net? There's emerging payer interest. This is also in the commercial space as well. And they're starting to test models about how they address social factors because what they're seeing, and they're on the hook for the cost of care, that their costs are being driven a lot by social determinants rather than just health care. And it behooves them to improve that as well as the overall care delivery path. Patient outcomes obviously are hugely affected by social determinants. Like in my hometown of St. Louis, we had a report come out from some of our social work community called For the Sake of All. And they were able to show that there was up to an 18-year life expectancy difference when you were north of one particular street in St. Louis compared to being south of that same street in St. Louis. And it spoke to the market differences in the social determinants between the two parts of the population. And then finally, this is a key quality domain. I think most of you are familiar with the 2001 Institute of Medicine, now National Academy of Medicine report about the six domains of quality, and equity was a key one. In some ways, I don't know that it got as much attention as some of the others initially, but it is now front and center as you've seen both in this conference and in other venues. And then finally, I always find this statistic, even when I see it, just really, really surprising and a little humbling as a clinician. So we know that social determinants of health drive health outcomes much more than direct health care delivery. And in fact, in many estimates, we find that 80% of the health is driven by non-health care factors, organized here by socioeconomic factors, physical environment, and the patient's own health behaviors, which we know are often a factor of their own environment. So health care is only 20%, which I think maybe knocks us all down a peg a little bit on our relative importance. But I think it also speaks to where do we want to have the impact and how can we start thinking about that 80% and that we are realizing more and more that zip code matters more than your genetic code. So I think one thing we need to think about is how do we approach the patients that we're taking care of. So what I presented to you is just a very classic way that I think we as clinicians talk about and present and think about our patients. And so this is an example, a 50-year-old African-American man, diabetes, hypertension, has frequent urination and fatigue, and due to hyperglycemia. In this case, sugar is up to 600. Presents a week ago with fatigue, malaise, tooth pain, had an abscess that caused the elevated blood sugar, the elevated A1C, elevated blood pressure. So we did what we do, right? And this is all really important. I'm not diminishing any of this. So you've got antibiotics, fluids, insulin, discharge with Scripps, long-acting daily insulin, short-acting mealtime insulin, antibiotic. Good job. We checked the list. He has a history of prior ED visits and admissions for this before, and that's maybe your first sign that there's probably more to do. So that's our point of view. What about his point of view? So his pharmacist told him his insurance now prefers a different type of insulin, so now he's a much higher co-pay. His spouse is trying to help him eat healthier, but finds it difficult on the budget because it turns out healthy eating is incredibly expensive relative to McDonald's. He's trying to exercise more, but the sidewalks around his house are broken, his work shifts are irregular, and it makes it really hard for him to have a routine. And then finally, he and his family share a car, which makes it difficult for him to get to his appointments regularly, often causing him to neglect some of the ongoing maintenance care and ends up landing back in the ED. And then finally, his family, depending on his income, he's the primary breadwinner, so he doesn't like to take off work even if he did have transportation to go to the physician. So what's important to him? Social determinants of health. The cost of care, the cost of healthy eating, a physical environment not conducive to physical activity, and transportation issues that prevent regular medical center contact. So why does this matter? If this is visible, if this is part of how we think about our patients, is it going to change your management plan? It's certainly going to change mine. And is it going to change the composition of the care team that we need to bring to this gentleman? Absolutely. I don't have expertise in many of these areas, but I know people who do. And so I think being able to think about that team and organizing around both this kind of perception as well as the more traditional healthcare is likely where this data will help us head. You know, the geographic distribution of health outcomes and social determinants of health further illustrates this very tight relationship. And so this is an example. This is a map from the CDC 2017 to 2019. And it shows the spread of cardiovascular deaths around the U.S. The darker red indicates higher rates of death. And you can see that it's highly concentrated in the Midwest and the Deep South, the so-called stroke belt. So I think we've all seen variations of this map. I know I certainly have. But what's interesting is when you start pulling in other maps around social determinants. So here's the percentage of population living in poverty. And you can see that the distribution with darker colors meaning higher rates of poverty looks very similar. Here's the distribution of unemployment by county around the same year. You might recognize some similarities. Percent of patients without a high school diploma. Again, darker colors representing lower education. And percent of population under 65 without health insurance. I've got many comments on this slide. But not enough time. But as you can see, all of these are indicative of the tight correlation between social determinants, our overall context, and health outcomes. And that's why it's so important to focus on this. So fortunately, I think our community, our cardiovascular community, is starting to wake up to this as well as the overall health care community. And you can start to see in the emergence of some of our guidelines and recommendations for treating these patients to start thinking about social determinants of care. If you look at the top orange boxes with risk factors on the far right, all of the risk factors have to be evaluated in the context of social determinants of disease. And in the risk prediction to the left of the gray oval, you'll see second from the bottom, social determinants are a key input into risk prediction so that we can understand both the risks and how we start to address it. It's also important to realize that social determinants of health are not a static situation. They're dynamic, and they have impacts throughout the lifespan of our patients. So this is just an example of, as we go from fetal growth and development to when somebody's in utero, we know even then some of the social factors are starting to have implications on the lifespan of, in this case, the fetus. And then as they move through childhood, adolescence, adulthood, and the next generation, you can see examples under each of them of health behaviors or physical environment factors or of our socioeconomic factors that are causing problems and will have impacts not only at that time, but over the lifespan. And so having a clear view of this and an ability to follow it over time is a key part. So I think I've convinced you this is a needed component of understanding cardiovascular risk and disease and treatment, and it's important to reduce it. So how do we first collect the information so we have visibility here? So right now, most of the data collection is manual and often regulated to our social work colleagues and sometimes our community health workers if they're available. It's important to collect this information directly from the patient. In many ways, that's similar to PROs. How somebody would describe their social environment, they're best equipped to do so. So that's a key piece of data collection. And it's also sensitive. It's embarrassing to talk about your homelessness or your struggles with substance abuse or maybe that you live in an undesirable neighborhood. And so being able to ask this with sensitivity and compassion is a really key part of this. And it requires training on how to get this information appropriately. It even calls the questions we've been asking about this. Is it going to get anywhere? Do our patients even want us talking to us about this? And I'd say the answers are resounding yes. We've seen studies where we've specifically asked patients, we're thinking about talking about these areas of healthcare. What do you think about that? Do you want us to or is that an imposition and something you don't want? And what we heard is that overwhelmingly the patients felt like social risk screening was appropriate, that it's okay to include it in the EHR. They expect all of it's private in the EHR, but it's important to have it in there. Screening was really important. They believe social risks impact health, so they understand the value of it. They obviously wanted their privacy protected. And then I thought this was the most interesting part. They get that we can't fix it all. They don't believe that just because we're asking that we're the salvation for everything they're facing. But rather it's more that they just felt seen and heard and recognized so that their ability to take insulin or do the other things we're typically going to ask them are being asked about and accommodated in their own personal environment. So as I mentioned, community health workers really know social determinants of health. This is their stock in trade. And the few that I've interacted with are really gifted individuals. In general, they tend to come from the neighborhoods of the patients that they're working with and can often speak to personal experience with some of the social determinants that they have. And they also have training in supporting patients as they start to navigate the social safety net and the healthcare system to get through it. This is an example of a program, the impact model, that has a lot of data efficacy. It's out of the University of Pennsylvania. And it's shown a very effective way to both train social workers, have them access both social services and the healthcare system, and you see the outcomes that have resulted to date on the lower right hand. Reduced likelihood of hospital readmissions, increased quality of discharge communication, which has long-term impacts on overall trajectories, and then increased access to primary care. Now currently, as I said, most social determinants of health data is manually gathered and stored. But increasingly, we have EHRs that are starting to provide ways to capture and use the data. And this is an example of what EPIC has done. We call it the EPIC wheel in my health system. But it's representing here, you can see the different domains. They've categorized about 12 of social determinants of health. And then use color coding to sort of understand where patients may have particular needs and be able to focus the team on providing them. And organizations like Boston Medical Center and Mainline Health in Pennsylvania have standardized the screener in EPIC, and it's now part of the intake process, part of the medical history and overall social history. There's also a wide variety of publicly available databases that largely the government collects around various aspects of social determinants of health. And they're freely available. You can download them today. And often what we're finding is both health systems, social service organizations, and even companies are starting to use this data to try and triangulate where there may be populations that have particular social determinants needs and start to target solutions to improve that. And you can see various examples of the data sets here and some of the data elements they collect. Well, one thing I'll point you to is the third from the bottom, the Social Vulnerability Index. It's one of the most common ones and something that I wanted to orient you to just because there is some work in the NCDR going on here. So the CDC runs this, and it's an index that takes into account 15 different factors around social determinants, poverty, lack of access to transportation, crowded housing, and it basically organizes the risk in a given census level, so subdivisions of counties and where there are particular limitations in some of these social determinants. And then each track gets a separate ranking in one of four themes. They take those 15 and organize them into four themes. And you can see those four themes represented here. And this is an example from Gwinnett County, which is in Georgia. So socioeconomic status in the green, household composition, which is largely a marker of how many people in a given house, race, ethnicity, language, and language in particular for people where English is a second language, and that introduces a lot of barriers, of course, in both social services and healthcare, and then finally housing and transportation. So what this is able to do is allow us to get an ecological sense of where a patient lives. It may be that patient is doing okay. It may be that they're not representative of their census track. But because they live in a high-risk census track, we may want to at least make sure we're not missing an opportunity with them to improve some of these factors and their deleterious effects on health. So we've started to see all this starting to come together in a couple of really interesting use cases that are actually coming from the private market where a lot of interesting solutions can happen. So this is an example of a growing, a fairly large now, social services company called Unitas, and what they have done is started to provide a social screening tool that starts to integrate information from the patient, from the EHR. They do connect to the health system data, but they also start to bring in some of these public databases, including the Social Vulnerability Index, and really start to give an overall view of a given patient panel or an overall population seen by a health system. So what you can see here is the number of patients up top, over 80,000 individuals, members in this case of a health plan, and then an average social needs score that importantly they start to break down by specificity. So you can see in the dark blue bars in sort of the middle of the chart, the high needs in this area are health literacy, food insecurity, and loneliness. So that starts to give you a little bit more actionable information, starting with what do we talk about? Like what's the conversation with the patients? And then if we're confirming that they are experiencing one or more of these, then we can start to think about linking to the services that can start to address some of those. And the other side of Unitas' health business model is they have a growing directory of social service organizations by community, so they can start to inform the care team and the patient about what's available to you to address to your own particular set of challenges. Another example is a recent company coming out called N1 Health, N of 1 Health. Same idea, they've got even a broader span of consumer data and it involves things like purchasing decisions, other data traditionally used for marketing, but in this particular case can be predictive of social determinant needs. And so they're able to provide some personalized insight both to patients and their care teams to start to address some of this. So currently the NCR does not collect social determinants of health in a systematic nature. We do collect some indirect assessments, sex, zip, race, ethnicity, and insurance. But we actually do want to get more granular. We want to move to those actual domains. And it may be that the answer here is not to add yet another set of data elements to the CRF. In fact, I'd argue that's not the answer. But rather start to think how do we connect to these other sources? If a health system is collecting social determinants of health and putting it into their epic wheel, how do we start to connect that to the NCR? If we have publicly available databases, and by the way we have publicly available databases, how do we start to connect those? And there may be opportunities to pull these into the registry information and start to feed back to our teams, not only here's how you're doing from a healthcare delivery point of view, but here's the environment that your patients are living in. And can we start to add to your management plan to ensure that they'll get the best possible efficacy from your treatment, as well as any social needs they're experiencing? We wouldn't be able to necessarily provide the individual level information, but by simply highlighting with some degree of specification, this potentially could be a real value gain for the consumers of our NCR information. So let me give you an example of what this might look like. So some of the ACC's data scientists have started to map what the social vulnerability index would look like in some of the maps they have for NCR hospitals. So this is an example, and I'll get it bigger for you. So this is from St. Louis. This is where I am and most of our hospitals participate in one or more of the NCR registries. And you can see in sort of the little red aspect right here, that's to the right, it's the Mississippi River. So that's right where downtown St. Louis is abutting the river. And you can see it's largely red, which in this particular legend indicates higher or worse social vulnerability scores. And that tracks with what we understand. We have a couple of hospitals. Our biggest one, Barnes-Jewish, is in that area of the red. A little bit north is Christian Hospital. And we know they have a high preponderance of poor patients with low social resources. And it would suggest that those hospitals and our cardiovascular divisions that are getting data from NCR would benefit from knowing about this so we can start to think about our management plans post-PCI or post-device implementation or post-MI. In contrast to the far west is all the green. And this is where our wealthy suburbs are. And we have some hospitals there, too. And it's not that you have to ignore that. Everybody has their own challenges they need to manage for their own health plan. But it suggests that on average they won't be facing some of the social needs. And so the management plans can be tailored accordingly. So just an example of how we can start to think about the integration of some of social determinants of health, the ways that we collect it, and ultimately how we might inform NCR and our overall membership. And then how can data be used? So we can use the data to risk stratify and target those with higher scores, like I've said, for further inquiry and intervention. At BJC we use the readmission risk scores to target those kind of conversations. And it's shown some early success. We can use the data to personalize our interventions, get back to our example. Perhaps we need to find a transportation program to help him out to be able to bring him in for his clinical care. Maybe we move him to virtual care where it's appropriate. Social determinants of health is an equality reporting metric. We're seeing it more and more. I showed you with the hospital readmission reduction program, and we'll see more coming down the road. Obviously population health would be greatly benefited by having this kind of insight to tailor their plans. I mentioned aspects of accreditation. This could help encourage data collection, show off best practices, maybe even provide a health equity or social determinants gold star to really show those exemplars in the field that can show us the way. Obviously this is a key way to address disparities. And with it currently front of mind in so many parts of our society, this is yet another tool to help us do so. And then finally, of course, clinical trials and sort of understanding what is the relative value of various therapies that we're testing when it's modified by people's environment and what social factors are affecting their health. So hopefully I've done what I set out to do, and that is convince you that PROs and social determinants data matter for health and healthcare, that digital tools can revolutionize PRO collection, that collecting and addressing social determinants of health is part of holistic care, it's incumbent upon us to do it, and that the NCDR use of this data can enhance the quality of our care, research and innovation. I appreciate your time. Thank you for that wonderful, wonderful talk. Very much enjoyed that. Very insightful. We have a couple of minutes to take some questions that have come in, so I'll pass that to Barb. So one of the questions was around the PROs. Are you finding any barriers with the aging population and using these digital formats? So two observations we've made about older generations, and we'll say 60s, 70s, 80s, that yes, it is true they have less engagement with digital tools than younger populations. But what I find is interesting is the absolute numbers. So we see that in general it's around 85% penetrance of smartphone use in populations younger than 65. It's around 67% in over 65. So even though we need to talk about the third of people and how to engage with them, I think we have found that the absolute rates of digital tool use in the older population is still pretty high and less likely a helpful tool. And I thank their grandkids for teaching them how to do it. There's a question here, I'll kind of paraphrase. Are these tools all sort of created equal? Are there some better than others? Are they available nationwide, regionally? And then I'll add, is there a cost to either the patient or the hospital or health system? Yes, there is. And this is a really important point, that we don't want to accidentally introduce yet another divide. The so-called digital divide is a real risk as we start to think about moving things more and more to digital platforms. And I think in some ways we might want to think about this, about how do we prevent divides for medications or healthcare access? And so can this be built in as part of a benefit about when you get a prescription, do you need a prescription for an app? Do you need a prescription for a data plan? If you're a particularly poor population, but we know that having this tool will be a key part of achieving your health outcomes, maybe we need to pull it onto our costs and provide that as a benefit. But whatever the solution, I think it's a really important thing to pay attention to so we don't accidentally introduce a new problem. I think we have one final question. Are we seeing any correlation between the use of these assessment tools and hospital admissions or readmissions? So well, yeah, but I'm going to draw a line to that. So there's the use, then there's the activity that occurs because of the use, and then there's the outcome. So where you have all three occurring, and we saw this in our primary care ACO, that is people were reporting their symptoms and having that bidirectional communication with their care teams and being able to notify them and get action taken. Then unplanned hospitalizations dropped and total cost of care dropped. But what we have found is just simply putting the tool out there and assuming it's going to work without any of closed loop activities or that bidirectional communication often doesn't impact the outcome. Wonderful. Well, we very much appreciate you being here. Absolutely wonderful session.

digital tools

patient-reported outcomes

social determinants of health

high-value care

engagement

privacy protection

ecological momentary assessments

care plans

data integration

health disparities