This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Today on Town Hall

I think, we measure things sometimes in healthcare and in general simply because it's measurable. Are we measuring the right things? Are we measuring things in the right way?

And it's back to that point about weaponizing data. Just because we can measure something doesn't necessarily mean it's the right thing. And how we use that measurement we want to do that for good and not for ill.

My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of This Week Health.

Where we are dedicated to transforming healthcare, one connection at a time. Our town hall show is designed to bring insights from practitioners and leaders. on the front lines of healthcare. Today's episode is sponsored by ARMIS, First Health Advisory, Meditech, Optimum Health IT, and uPerform. Alright, let's jump right into today's episode.

  All right, here we are for a town hall episode and I'm joined by David Vaudrey. He holds the position of Chief Data and Informatics Officer at Geisinger.

  David, welcome. Welcome to the show. Thanks Bill for having me. Well, I'm looking forward to the conversation. Data and informatics at Geisinger, I think is really at the cutting edge of where healthcare would like to go.

So many healthcare organizations are mired in fee for service. Not that you don't have fee for service, but you guys have a lot higher percentage of value based care. I think a lot of organizations are trying to figure that equation out. And so I'm looking forward to the conversation. Let's start with this.

Tell us about Geisinger a little bit, so people get an idea of, the geography, who you're serving and that kind of stuff. And and then if you could go into your role a little bit and what your role does at Geisinger, that

would be great. Sure. Geisinger, for those that aren't acquainted, is an integrated health system in central and northeastern Pennsylvania.

By integrated, I mean we have both a clinical enterprise that takes care of a million plus patients. And a health plan that has about five to six hundred thousand members. So as you can imagine, there's a considerable overlap in those populations. But we're not a completely closed system. we're pluralistic in the sense that our clinical enterprise cares for patients, to your point, in a fee for service and other value based types of models and then our health plan members of our health plan can seek care, of course, in other venues as well, but where there is that overlap, we tend to find a lot of opportunities to innovate, and I think that's one of the unique things about Geisinger, and my role specifically as the Chief Data and Informatics Officer, it really streamlines a lot of data collection and data use.

I think as you said, I like the word you use, Bill, that a lot of places are mired. In a model of care delivery that

It's, yeah it's sick care. It's not, we're not trying to keep people well because you don't benefit if people are well. I'm not saying that people, that health systems don't try to keep people well, but it doesn't benefit them when people don't come through the front door.

But we've, we've talked about this for years. It's, it's, It's such a challenging place to be. They don't want to be there. But the payment system forces a lot of health systems to put their focus there, and Geisinger has been able to really break

out of that. I think that's right. I mean, we're learning like everyone else, and certainly there are other great examples of integrated systems across the country.

But I agree with you, and I think we're trying to unmire ourself, and another way I would use that phrase, you know, we're also mired in data, and that's just not just Geisinger unique. I think a lot of places struggle with being data rich, but information poor. And as I look at my role, trying to help our organization be more data driven.

Make better decisions using timely and accurate data. That's a big challenge even in an integrated model where we tend to have advantages of being able to bring the claims data from the payer side together with the clinical data from the care delivery side. The truth is, like, we're drowning in data, and I like to use the analogy of, the data tsunami that people talked about, how do we leverage this data?

How do we translate this? TS Elliot wrote, how do we translate the knowledge to wisdom, if you will. Where's the information that's lost in knowledge and where's the knowledge that's lost in wisdom he wrote? So I think we have that challenge but we're doing our best to to overcome those challenges of drowning in data, if you will.

All right. So let's, we will start as unsexy as we possibly can, and we're gonna talk about data governance. And that, that's about as unsexy a topic as we can come up with. But if you're going to do value based care, how important is data governance? And how do you establish data governance at an organization like Essinger?

Yeah, that's a good question. Part of it, I think, is understanding where your data come from, the various sources. understanding how those systems talk to one another, and then try to figure out, what the best source of truth is. How do you define the metrics that matter to you as an organization?

And how do you simplify? I think, again, one of the challenges is too much data, too much information that the knowledge, the wisdom hasn't been extracted, the insights, as some people will say. And so I think governance is a key part of that, but that word means a lot of things to a lot of different people.

My focus here in the last five years, as I've been at Geisinger, has really been to simplify, standardize. We've got an effort, as a lot of places do, to rationalize the number of applications that we use to try to forge deeper relationships with the public. Perhaps smaller number of vendors as opposed to the proliferation that can happen if you're letting you know, a thousand flowers bloom, as it were.

you know, you talked about the single, the source of truth for the data and I was fascinated when I came into healthcare, 'cause I came from outside of healthcare and they said, Hey, can you generate a report that has, I don't know, length of stay? Let's just do length of stay. That's one of my favorites.

And I said, well, sure. Yeah, we can generate that. And I went back to the team. They're like, well, you know what? And they asked me like 55 questions about what life is. I'm like, what do you mean? What's like this days? Like when they came in, when they left, it was like no. There's like, 50, how do you handle all those different definitions for, something that from the outside looking in would look very.

Like, oh, well, there's just one definition for that, but in healthcare, there tends to be multiple definitions depending on, who you're talking to, how you're measuring it, who you're reporting it out to, and

that kind of stuff. Well, that's another great point. And I think one of our approaches in that vein of simplifying and standardizing part of that, to me, ties back to how we visualize the data and are we putting good Metrics good labels on our metrics.

it's appropriate, perhaps, to use different definitions of length of stay, a geometric length of stay, an arithmetic length of stay. They're valuable in different areas to different people. But what I think really confuses us oftentimes is when we don't identify what we're meaning in terms of those metrics.

And that's when I think it leads to a lot of confusion. You're absolutely right. Someone might ask for a report. If you ask from two different groups, you get two different answers. That's a scary proposition, especially again when you're trying to leverage those data to make timely and good decisions in an organization like ours.

What's changing in the world of data right now? I, I was talking to a CIO this morning, and he's been with the same health system for 25 years, I'm saying. he was over, when he started, he was over data and analytics. I'm like, man, 25 years, data analytics. Even when you go back 25 years, I'm thinking, were we even talking about data warehouses?

Were we still on client server? Like, what were we doing? And then we ended up talking about how we were wrestling with for the last 20 years, like consolidating and normalizing that data in healthcare. But someone like guys here, you guys have been on a common EHR platform for quite some time. And that's, I'm not, it's not all the data, but it's a bulk of the data.

around the medical record and whatnot. Do you feel like you guys are far enough, I mean, are you far enough down the path? are you now looking forward to, oh, now we can do things that we couldn't do five years ago with the data? I

don't know if I'd characterize it that way. You're right. We've been using the same electronic health records since 1996.

We were the 11th customer, I understand, of Epic Corporation, and so we've got a long history with them. When you think about nearly 30 years now of data in the same infrastructure, obviously that has evolved tremendously. The health care needs of the country have changed and as technology has advanced, but having that core and that history has really been advantageous, I think, to us.

There are a lot of places you talked about what things look like 25 years ago. Most people are on paper charts 25 years ago. Believe it or not, I mean, it's only been in the last couple of decades through the Meaningful Use Program, that widespread adoption of EHR. in the country has has been achieved.

And so, yeah Geisinger, again, is, was an early adopter and has been a leader, I think, in this area. And it's allowed us to take advantage of those data. A lot of people, I was on a call the other day with a vendor who was complaining about EHRs and how difficult it is to get the data out.

And I had to interrupt them and ask our team, I said, is that our number one problem? And most people on the team said, no, I, we're pretty facile with. extracting the data and using those data in various ways. And so I think it, part of it is just, it's a culture number one, but it's also having the history, the experience been able to do that.

And again, the technology keeps changing. I don't know how many data warehouses we've gone through, how many vendors are, homegrown versus, engaging partners. But we've got a robust Platform, and we've got, as a result, very longitudinal data. So not only, covering the payer or the provider side, but also going back.

More than a couple of decades and that's very advantageous if you're trying to do long term tracking of your population and it's also advantageous at the individual patient level to know that you know, I can have confidence that when I see a provider here that they've got my entire close to my entire health record in most cases.

  📍   In the ever evolving world of health IT, staying updated isn't just an option. It's essential. Welcome to This Week Health, your daily dose of news, podcasts, and expert commentary.

Designed specifically for healthcare professionals like yourself. Discover the future of health IT news with This Week Health. Our new news aggregation process brings you the most relevant, hand picked stories from the world of health IT. Curated by experts, summarized for clarity, and delivered directly to you.

No more sifting through irrelevant news, just pure, focused content to keep you informed and ahead. Don't be left behind. Start your day with insight at the intersection of technology and healthcare. This Week Health. Where information inspires innovation. 📍 Increase 

 

I

remember talking about a story where Geisinger was piloting, I don't know how far you've gone with this, but you were piloting with primary care visits doing genetic mapping. And I mean, has that continued, that project?

Yeah, it's called MyCode. It's one of the Largest and I would argue most successful precision medicine programs, personalized medicine programs in the country.

I think we've got today over a quarter of a million people who have consented and most have provided their blood samples. For sequencing, and we've got, I think, close to 200,000 full exome sequences that are being used today for not only biomedical research, which is, a powerful tool for discovery of new drugs and new therapies.

But also that we feed that information back. It's part of my. Medical record if variant is uncovered on my genetic testing that might impact my health. That goes right back into the EHR and my primary care provider can bring me in. We have a large population of genetic counselors that are fantastic, that can help.

People that have a a variant of concern, and they can get themselves further testing, they can get their families tested, and it's a program that, that literally can save lives. I,

well, that's, I was going to ask you, like, You were talking how it's a successful program and the sheer number of people that have been tested is fantastic.

Saving lives is fantastic. What what are the metrics that you're looking at to determine if that project is being successful and really touting that success?

Well, I think it's that. I think it's the, again, the volume. I believe there are 60 to 70 genetic variants that are screened for in our MyCode population, and it's a voluntary program.

I mean, no one's obligated to participate. The fact that we have such a high number Rate of participation, I think, suggests the level of trust that people in this community have with their providers. And so, that's one metric of success. People vote with their feet.

Secondly, I think there's been about one to 2% rate of people with one of those. Variants of concern that are identified, and that information, again, is fed back to them. A lot of programs like this across the country, and I come from an academic background other great organizations are doing these types of programs, but often with anonymized, discarded samples and focus solely on the research side of things.

There's no connection back to an individual if something is, Discovered in their in their genetic code that will be impactful on their health. I mean, indirectly, it benefits all of us but there's no direct connection back to me or my family if something is identified. So I think that's one of the unique things about our MyCode program.

It's a, again, it's a fantastic program that Geisinger's really let in.

Yeah, it's pretty, pretty cool. You said, you know, your team looked at you and said, hey. Getting data out not, our biggest challenge. curious what are the challenges that you're facing today? If you could just share maybe one, one of the challenges that you're facing today.

Yeah. And I think we touched on it. It's strounding in information. It's how do we become,

is it more sources of information, or is it because you're integrated and payer provider that it's just, there's just a ton of, I mean, just the normal sources of information, or are you seeking out other sources as well?

You can make an argument that more is better. And I think that's true to a certain extent but if it's unwieldy, unmanageable if you're not using it, think of the challenge of alert fatigue, which is common in healthcare. I think it's known to all of us. The Boy That Cried Wolf scenario we have so much data that we're generating all sorts of alerts.

We have, pages and pages. If you ask somebody, and this is not disguising it, but if you asked for a printout of your medical record if you had any substantial care, let's say you're in the hospital for a week, your medical record is hundreds of pages long. And, who can make sense of that?

Like, you know, we need better algorithms, we need computers, a lot of talk these days about generative AI, the ability to synthesize and summarize information. All of that is important but I think, one argument that I like to make to our colleagues here at Geisinger is that having more data does not necessarily make you data driven.

more is not necessary unless it's being used effectively. So you could argue that it's a failure to fill, it's not that we have too much information, maybe you say that it's a failure to filter out the right information. And I think our job as informaticians is to get the right information to the right people at the right time, in the right context, so that they can make better decisions.

And that might be a physician at the bedside, that might be a patient who's trying to make a care decision, that might be a boardroom decision as well, so I think it really spans The gamut there. Yeah, I love

the idea that being able to amass more information doesn't make you data driven. Or even saying the words data driven doesn't make you data driven.

But it begs the question What does a data driven healthcare organization, like, what are the characteristics of it? Is it what they can do once they get the data that makes

them data driven? Yeah, you know, one of my colleagues here, he's been here for a long time likened our analytics work historically exactly where we started, with fee for service versus a value based approach.

He said fee for service analytics means how Again, more is better. Like the more you do, the more reports we generate, again, the more data driven. I it's a bit of an illusion. I think it's wrong. I think if you said, if you counted the number of bytes or the number of, terabytes or petabytes in your data warehouse you count the number of reports that you generate every day or every month, like, I'm not saying those aren't somewhat interesting and somewhat valuable metrics.

But really it's about the use, and if you say, value based analytics is a phrase that he coined I think that's an important way to frame it, is are we gleaning value, are we are they being used, number one, like, tree that falls in the forest if we're producing reports that nobody's looking at, or if they're wrong, or if they're, discordant or if they're not adding value, then, what's the point?

I can, I can make a million of those and not add any value to the organization. So I think that's a hallmark to me. If you really are data driven, it doesn't mean that you have the most. It means you've figured out how to use it the most effectively.

What does the world of analytics look like post report?

I mean, the way we're talking now, it just feels like report can be old. It can be stale data by the time you get the report and that kind of stuff. I'm sure you've had conversations around this. What does life look like post the report, post the document floating around with toner on it?

Oh, great question. I, again, I think that you have to start with the problem you're trying to solve. And I think there's an element here of timeliness. There's an element of accuracy and data quality, and there's an element of simplicity in decision making and so I don't think there's anything wrong, frankly, with printed documents.

Thank you. piece of paper with a few numbers on it. Like if that's what solves the problem. I don't

want to get the emails. I'm not anti paper or anti toner. I don't, no emails.

There you go. Yeah. And speaking of email, I mean, again, you're thinking of what I would call fee for service analytics. Well, when I started at Geisinger, those were the metrics, how many reports do we.

Send out how many emails and you can even look at how many emails get opened. Now you're starting to get more into value based, I guess, analytics, if you want to think of it in those terms, but what's the right tool for the job? And I might need a real time alert on my phone.

I might need a piece of paper on my desk in the morning. I might need in the context of the electronic health record. A nudge, a reminder put the right information in front of me in the context of the workflow that I'm in. Great examples of that are, classical decision support in the EHR and order entry.

Tell me if I have a drug interaction or a drug allergy or something along those lines. Increasingly, we're building predictive models that help guide decisions. If someone has an elevated risk of something, and those risk factors can be mitigated. Good example is, I was just on a call before this with colleague who's built and deployed a model around stroke risk, identifying people who are at risk and turns out there are some modifiable factors.

There's some that aren't, and like If your predictive model says, David's at high risk of stroke because he's old, and he's, got certain conditions that are not easily influenced such as my genomic factors okay, your model's not very helpful. But if you said, David's at high risk of stroke, in addition to those reasons, here are the modifiable factors, things like my lipid management, or my blood pressure control, or my diabetes management.

Or even the fact that I haven't had a flu shot this year. Those are things that we can, in fact, modify, and we're aggressive at Geisinger about reaching out. Again, back to that value based model, we can focus our attention on preventive care, on getting people flu shots, on getting people screenings, on managing their hypertension or their cholesterol or their diabetes.

And when we do that, we can actually reduce the risk of some of these significant events. And to your point, like, in the fee for service model, like, you're financially incented for every stroke patient that comes through your doors. If you think of it from the payer point of view, you're financially disincented.

And so, where your financial incentives lie, that's

So you guys are kind of schizophrenic. There's part of you that wants them in, there's part of you that wants to keep them out, but the beauty of value based care is you'd rather keep them out because if you keep them healthy you're, I mean, the beauty of value based care is you get paid either way.

You're getting paid to keep them healthy, and it's just a better business

model. I think it's, you've said it well, it's a much better business model from a patient's point of view, I would argue. It's a much better business model from a community's point of view, where, you have a an entity like Geisinger in the community that is trying to keep people healthy.

When I got here, I came from a place that was much more focused in the fee for service world, and the metrics that we tracked naturally were things like volume and cases, how many people did we have in the ORs, and those types of things. It's not that we don't track those things at Geisinger, but we also look at utilization in our emergency departments, utilization in the inpatient setting.

And, we want to appropriately keep people out of the hospital if they don't need to be here. If I can prevent somebody from needing to come to the emergency department by getting them timely and high quality primary care services, if I can get an appointment within a day at a primary care doc, maybe I can save that trip to the ED, which costs a lot more and is frankly a lot less convenient and a lot less productive oftentimes for somebody's overall health.

So, David, you,

you mentioned, you got to start with the problem. What problem are you trying to solve? what kind of problems do your clinicians want to solve with data

today? I think like everyone, they strive to be data driven. We keep using that term. But they want to do the right thing by their patients. And again, a lot of that is closing care gaps. Care gaps, again, being basic things like getting screenings done in a timely way. Whether it's colorectal cancer, breast cancer screenings diabetic screenings.

Even something as simple as having your blood pressure checked. How are we doing at those things? And how can we do better? Because it's obvious, I think, to anybody it's, homespun wisdom that an ounce of prevention is worth a pound of cure. And I think that's true financially.

It's also true from an overall health perspective. And so I think our physicians and nurses, other clinicians here, want to do the right thing by their patients. It's a very mission driven organization, and I think we want to use data to help them do that, to understand how they're performing and to make adjustments as possible.

One thing, Bill, I will say is we've made it a very, one of my colleagues says it's a cardinal rule here that we do not weaponize data. And I love that phrase when I heard it. And I think about how data can be used in a punitive way, in a judgmental way in a comparative way that, that can lead to Anxiety and frustration and burnout.

I mean, you have all sorts of problems, like something being attributed to me that maybe I had no control over. there's a famous quote, I'm sure you've heard it, that not everything that counts can be counted, and not everything that can be counted counts. I often talk to my team about there's a simple story that you may have heard.

It's about the drunk and the lamppost, and it goes like this. There's a man, he's inebriated. And he's standing next to a lamppost looking on the ground. The police officer comes to him and says, Sir, what are you doing? He says, I'm looking for my keys. I dropped my keys. And the police officer says, Well, do you know where you dropped them?

And he said, Yeah, over there, in the dark. And he said, Well, why are you looking here? And he said, Well, this is where the light is. And I think, we measure things sometimes in healthcare and in general simply because it's measurable. Are we measuring the right things? Are we measuring things in the right way?

And those are things that we think about a lot, I worry about a lot. And it's back to that point about weaponizing data. Just because we can measure something doesn't necessarily mean it's the right thing. And how we use that measurement we want to do that for good and not for ill.

One last question.

I'd be remiss if I didn't talk about futures. So, and it's surprising. We haven't mentioned the word generative AI on the show at all. And if I allowed this episode to go without mentioning it, this would have been the first one of the year that we hadn't mentioned it. What do futures look like in analytics?

What are the technologies? What are the What's the direction that we see use of data and how it's going to be impacted by technology moving forward?

Yeah, I am a believer in artificial intelligence and I think we've made tremendous strides as a field in the adoption of generative AI. I would say with that caveat that I'm a little bit of a skeptic.

I don't think we're going to solve every problem for every use case. Simply by throwing a large language model at it, and I'm sure others might some people would agree with that. Maybe others might disagree. There's a lot of potential. I think it's going to take some sorting out. But the things I'm most excited about One is a pilot that we're kicking off here shortly around ambient documentation, which I suspect you've spoken about with others in the past, so we won't go into detail unless he has specific questions.

The idea of generating text, I think is valuable, or generating other types of outputs. I'm excited. A future world where and I'll give credit to one of my neurosurgery colleagues, Dr. John Slotkin, who said, imagine if I beat an x ray a CAT scan into a large language model and in the context of surgical planning.

ask the machine to provide some suggestions about different surgical approaches. That's powerful stuff. I mean, couple that maybe with 3D printing and like it unlocks all sorts of new and exciting capabilities, I would say, for our clinicians to provide better, safer, more effective health care.

And so those things are not, fantasies like they might have seemed a few years ago. On the other hand I don't want to fall into the trap that. Now we have this hammer let's start looking for nails, right? And I think it's a pretty general purpose tool. I think you'd say that about artificial intelligence in general.

And I, I just don't want to fall in the same trap. I often talk with my team at least about The hype cycle for new technology adoption, you have this early rise into what's called the peak of inflated expectations, and then a precipitous decline oftentimes to the trough of disillusionment, and I think we're right in that hype cycle.

I don't know that I could speculate on exactly where we are. I don't know if we're still where we are. Going up and the hype is increasing, but there's certainly a lot of hype around AI specifically. At some point, I suspect we'll realize that, maybe it's not going to solve every problem we had hoped.

And at some point again, after that trough of disillusionment, you find your way back to some plateau of productivity, I think it's referred to. And I think this won't be much different from the other hype cycles that we've ridden in the past, including, by the way with AI. AI in healthcare is at least a 50 year old field at this point, and one of my good friends I co chair our, what we call our AI steering committee here with my colleague who chairs our radiology department, Alpin Patel.

And I joke with him that it was, So I think back in 2012, 2013 that Jeff Hinton stood up and said, silly for us to be training radiologists, that whole field is going to be obsolete to be replaced by putting words in his mouth. I mean, a brilliant person, but here we are six, seven years.

Later, maybe more than that. And, we still need radiologists. We still need, humans in the loop in most things, including self driving cars, for that matter. We're supposed to all be Driving cars these days that drive themselves according to some prognostications.

So I think there's a real critical element of us to understand, how we can use technology to improve health care, but not necessarily replace physicians and nurses. I think we're far from that.

Well, David, I want to thank you for your time and appreciate the conversation and look forward to staying in touch.

Hearing how things progress.

Thank you, Bill. It's a pleasure. Appreciate it.

Thanks for listening to this week's Town Hall. A big thanks to our hosts and content creators. We really couldn't do it without them. We hope that you're going to share this podcast with a peer or a friend. It's a great chance to discuss and even establish a mentoring relationship along the way.

One way you can support the show is to subscribe and leave us a rating. That would be really appreciated. And a big thanks to our partners, Armis, First Health Advisory, Meditech, Optimum Health IT, and uPerform. Check them out at thisweekhealth. com slash partners. Thanks for listening. That's all for now..