Moving healthcare data from one place to another is only part of the interoperability challenge. What happens when that data arrives but isn’t actually usable?

In this episode of the Informonster Podcast, Charlie Harp sits down with Laura Bush, Director of Product for PIQXL Gateway, to discuss her journey through healthcare technology and the experiences that shaped her perspective on interoperability and data quality.

Charlie and Laura talk about why fragmented data remains a challenge across healthcare, what it means for data to be fit for purpose, and why simply exchanging information doesn’t necessarily make information more useful.

They also discuss the growing implications for AI and take a closer look at how data quality programs can help organizations move from identifying problems to improving the data itself.

Follow Us

Have a question or topic idea?

Transcript

Charlie Harp (00:09):

Hi, I’m Charlie Harp and this is the Informonster Podcast. And today on the Infomonster Podcast, I have the one, the only Laura Bush. Good morning, Laura.

Laura Bush (00:19):

Good morning, Charlie.

Charlie Harp (00:21):

So one of the things we do, as I’m sure you’ve listened to the Informonster podcast many times.

Laura Bush (00:26):

I have.

Charlie Harp (00:27):

But one of the things we do is I like to have people tell their story of their journey. So tell the listeners about your journey into healthcare.

Laura Bush (00:38):

Okay. It’s probably similar to some folks that have been on here where it’s not necessarily a linear path. I started in physician practice operations when I was very young and realized, okay, what’s going on here? Why do we have so much paper? This seems weird when it’s 2008, 2009. And from there, I got into working with a regional extension center to advance meaningful use and really thought about helping the little guys who were bringing on these really big technology projects that really weren’t something that was in their wheelhouse. And so from there I was thinking, okay, so we’re now getting folks on electronic health records, but are we actually using electronic health records efficiently and optimally? And so from working in meaningful use, I started consulting with small to medium sized practices to essentially make their workflows more efficient. And that led me into value-based care.

(02:02):

And value-based care is really interesting to me because it was not only using electronic health records and digital quality metrics, but also it was payments. So are we using electronic health records and information correctly in order to get paid? And so that really was an interesting experience. And it really led me to the next phase of my career, which was implementing behavioral health, electronic health records, and psychiatric hospitals, which I found to be wildly fascinating because we were working with really sensitive data, but data that really needed to be digitized so that it followed the patient from their journey from one place to the next, community health centers to behavioral health hospitals, to outpatient clinics, et cetera. So from there, I got really, really interested in product management and how to actually build products efficiently in order to help organizations use electronic health records, move electronic health records, and then optimize those health records.

(03:26):

And so I was working with the Idaho Health Data Exchange and the Hawaii Health Information Exchange, the Puerto Rico Health Information Exchange. And I thought, okay, now we’re moving data from one place to the next, but can it actually be used for real patient care? It’s fragmented, it’s disjointed, it’s duplicated. It’s all of these things that maybe helped create a longitudinal record, but not a longitudinal record that could be efficiently used because a physician could be spending more time weeding through records stacked on records versus actually spending time with the patient. So that’s how I got into data quality and really creating an ecosystem of usable fit for use information. And so that’s how I ended up here at Clinical Architecture.

Charlie Harp (04:29):

When I first met you, I believe you were at Health Gorilla.

Laura Bush (04:32):

I was.

Charlie Harp (04:33):

And we were doing some projects together. I was doing the initial data quality review of the PIQI stuff back before PIQI was PIQI. And I remember having a number of conversations with you and Dr. Lane about data quality, about interoperability. So what do you think is…if you were to categorize the largest issue we have right now, not in healthcare writ large, but specifically in the world of data exchange and interoperability for care purposes, what would you say the single biggest problem is right now?

Laura Bush (05:15):

Fragmented data. What

Charlie Harp (05:18):

Do you mean by fragmented?

Laura Bush (05:20):

I mean, data that is being exchanged that may be very useful from the originating system because the originating people that entered in this information were using it for a specific use. But then when that data is being exchanged from place A to place B to place C, that fitness for use has been lost. And there may be gaps in documentation or perhaps nomenclature mismatches or even just completely wrong information for the use downstream. And so I think fragmented meaning it may be fit for use in one area, but not in another area. And so I think as we continue to move forward in interoperability, we have to think about what is the actual reason that we’re exchanging this information? And not only from source A to source B, but maybe downstream uses as well. So I think the fragmentation is probably what I consider to be the most concerning problem at this moment.

Charlie Harp (06:40):

I think one of the things too, when you look at what we’re doing with HIEs, with TEFCA and the QHINs is I still think we live in a world where we put a lot of energy into moving the data and exchanging the data, but I don’t know that anybody is really taking advantage of that in a meaningful way. I think that we put it into a mode where a human being can look at it with their eyeballs. But just anecdotally, and I’m curious if you have the same experience. When I talk to people and interact with people, it’s like, yeah, we get the data. We get it from an HIE, we get it from TEFCA, but our systems don’t do anything with it. Is that your experience as well?

Laura Bush (07:29):

Absolutely. I started seeing that way back in 2015 when I started working with value-based care. We would create reports from one EHR, one system and collaborate with another system, but that downstream system couldn’t talk the same language. And so all of this work that the originating system did, didn’t even feed into the downstream sources. And so I think, and even in the payer world, right? So when I was out in Idaho, we had a really cool provider-payer collaborative where we really tried to tie in the entire state’s ecosystem of payers and providers and really create that bridge between the two of them. But when we were creating the information from one place to another, that payer really couldn’t use that data because it was so holy. Not religiously holy.

(08:30):

Not religiously holy, but holy in a way that really presented gaps that you’re like, “Well, how are you paying a provider for the quality of care that they’re giving a patient?” Which we believe they’re giving them, but the data doesn’t back that up. And that is concerning in so many different ways, especially when you think about AI models. I am somebody who’s a huge proponent for AI, but you have to have the quality of data that is going to promote the program that you’re actually putting it into and the information you’re wanting to get out of it. So I always think, I’m going to use the term garbage in, garbage out. Everybody uses that, but that’s truly what it is. If you put something in a model and you’re expecting it to produce this result that’s going to be life changing, you could be way off because you just don’t have the wholeness of information in order to actually create that downstream positive impact that you’re wanting.

(09:41):

And I think that’s really been on, it’s been on my mind a lot lately because I think a lot of folks feel really excited about the possibility of efficiently using AI in order to really improve the outcome of populations. But if we’re not thinking about it in a more holistic way of what are we putting into this model to actually prove the fitness for use downstream, what’s going to happen to our patient care or our population care or payer models? There’s so many different outcomes that could be affected by it.

Charlie Harp (10:23):

Well, and I think that I’ve been making this argument for a while now that it’s really anything. I mean, I started out in healthcare in the clinical lab and then I worked for First Databank and Zynx and it was all around decision support. And decision support, for that to work in that world, they actually had to use the terminologies we provided to be able to use the decision support. And even then it was still a very broad brush of decision support and not always useful. That’s why people turn things off because it was too broad, too much. And as we start to create more data, I think healthcare data doubles every 73 days. And I think AI is going to accelerate that rate. I think that, and as we add more precision medicine data, that’s just going to add more granular and more kind of nascent terminologies into the mix.

(11:23):

I think it’s everything. It’s decision support. It’s digital quality measures. It’s old school business analytics, utilization of resources. And I think AI, the scariest thing about AI in that chain is the potential lack of a human arbiter that looks at the data and makes a decision. If AI is making the decision, at least with all this other stuff, you have a human to kind of catch the baloney, so to speak. Whereas with AI, you don’t.

Laura Bush (11:54):

Especially when you have an AI being the scribe and then the AI being the downstream model producer, you’re thinking one AI is making the information for the next AI. And is it just all hallucinated information? That’s scary.

Charlie Harp (12:18):

Well, and the thing about us as human beings is if we find a technology that allows us to be lazy, we tend to embrace that technology. It’s kind of part of our nature. And I think that the challenging thing here is it’s like data quality. People don’t want to deal with data quality. Data quality is hard.

Laura Bush (12:37):

And boring.

Charlie Harp (12:39):

I don’t know. I think it’s kind of exciting, but I’m weird. I think that one of the challenges is people want AI to just take care of data quality because they don’t want to do the hard work.

Laura Bush (12:50):

Absolutely.

Charlie Harp (12:51):

It’s kind of like, if I could have AI work out for me, I would do it because I don’t want to work out. Who wants to work out?

Laura Bush (12:57):

I think there’s some people, but –

Charlie Harp (12:58):

Oh, weirdos. Weirdos that aren’t excited about data quality.

Laura Bush (13:03):

Right.

Charlie Harp (13:03):

But I think it’s kind of one of those things where AI can help us in a lot of ways. It can even help us kind of monitor what’s happening with data quality. But I think that we have to figure out how to resolve the data quality because I would argue earlier you said that the systems that produce the data, the data works fine for them. I would argue that that’s not always true either. I think inside an EMR, EMRs were really designed to capture activity so that we could produce a bill so that we could get paid. And that act of capturing data to produce a bill is not the same thing as producing data for clinical care. And that’s the other thing too, is that people, when they build these AI models, even if you had the best quality data in an EMR, there’s still the uncalibrated uncertainty between what the physician was thinking about the patient and what they felt like they had to put into this tool they use so that they could get through the process of the visit or the encounter.

(14:02):

So I think that there’s a larger holistic question that we need to figure out as we really try to use this data. Is it fit for purpose? Right.

(14:11):

And I think that’s going to be the challenge that people are not going to want to deal with, but we’re going to have to, or we’re going to end up having AIs running amuck and operating against assumptions that may or may not be true.

Laura Bush (14:27):

Absolutely.

Charlie Harp (14:28):

So one of the things I also like to do in the Informaster podcast is have people talk a little bit about things that most people don’t know. Now, for those of you listeners who don’t work in the same office with Laura, you don’t know that when you go into her office, it’s like entering a tropical rainforest. In fact, I think I saw a Chupacabra in there. I’m not sure, but I think I saw one. So you have other things that. You’ve had a place where you sold plants or you did plants.

Laura Bush (14:56):

I did.

Charlie Harp (14:57):

Talk a little bit about that aspect of Laura Bush.

Laura Bush (14:59):

Oh, I love that. That’s so exciting. So yeah, I owned a plant shop for a couple of years and a yoga studio and it was super fun. I was also consulting with health information technology at the time. So I had this cool balance of being able to work on something that was really hard and challenging. I was creating the infrastructure for the Chicago Community Information Exchange. And at the same time, I was slinging plants to the folks of greater Howard County, Indiana.

Charlie Harp (15:36):

That’s a good way to find your Zen, right?

Laura Bush (15:38):

It was. It was really cool. Commune with

Charlie Harp (15:39):

Nature.

Laura Bush (15:40):

Yep. We had sound healing. People would come in and play singing bowls. That’s cool. It was really, really fun. It was quite the vibe. That’s what people say when they come into my office. Man, this is a whole vibe in here. And I said, that’s exactly what it is. This is supposed to be a welcoming and inviting space. So anytime anybody needs a little Zen moment, that’s –

Charlie Harp (16:06):

It is a very Zen space. Yeah. That’s absolutely true. It smells

Laura Bush (16:09):

Good. It looks good. It feels good. All of the things.

Charlie Harp (16:13):

So one of the things that we get to work on together is PIQI Framework and our PIQXL Gateway product. One of the things that I’m really excited about is this idea of quality programs. Do you want to talk a little bit about, so I know that there are data quality programs, like there’s the CMS digital quality measures. There’s the HEDIS digital quality measures. The thing that we’ve been kind of working on and we’ve been talking about in the PIQI Alliance community is this idea of a quality program. Talk a little bit about what a quality program is for the people out there.

Laura Bush (16:49):

Sure, sure. So in my mind, and hopefully in other minds, a quality program is a container that gives the use case for data to be fit for use for a specific purpose. So you just mentioned a couple of different situations like the HEDIS digital quality measures, CMS digital quality measures. There’s all sorts of different programs out there that fit into that container of a data quality program, even so much so as registry data quality. That was something that I experienced way back when I was working in the meaningful use world. We were submitting data to different registries as one of the meaningful use objectives, one of them being the American Heart Association. There were also some different specialty registries. And one of those things that I kept thinking about way back then was, oh my gosh, we’re submitting this data, but how is it going to be used downstream?

(17:59):

And so there are now, in my mind, lots of really cool ways that we can use the PIQI framework to actually ensure that the data that is going to these program authorities to ensure that that data is fit for use. So as we mature the PIQXL Gateway product, that’s one of the avenues that we’re going down to really make it super useful for lots of different reasons and really just kind of give it a new avenue to show that it can do all sorts of things with the PIQI framework.

Charlie Harp (18:41):

Absolutely. I think that we’ve done a lot with the folks at the PIQI Alliance around the USCDI version 3.1 aligned rubric. We’re now doing the partnership with NCQA and looking at their data quality plausibility rules. And one of the things that you’ve been working on with the folks at CRISP is this behavioral health. In fact, you were talking to Victor Lee about this just before we started this session. Yes. Want to talk a little bit about that?

Laura Bush (19:09):

Yeah. Yes. So ONC is piloting a behavioral health project that is encompassing about 10 different HIEs throughout the country. And we are measuring behavioral health data within the USCDI+ constraints. And so we are working with CRISP DC to evaluate their exchange of behavioral health data, measuring the completeness, the conformancy, the availability of the actual behavioral health data within their clinical data sets. 42 CFR part two data became a national regulation in January of this year where we are really encouraging that promotion and use of that data throughout the entire ecosystem with computable consent. And so now that that is being more often exchanged, we’re at this point grading the actual quality of that data. So I’m really excited to work with CRISP and give them the outcomes. I believe that that’s happening at the end of September. So we’ll have more data to produce then.

Charlie Harp (20:37):

Can you talk, because we did another thing with Connie. We did this, the REL-D rubric for them. Talk a little bit about what we did there because I think it’s kind of cool.

Laura Bush (20:47):

It is really cool. So one of the interesting things that came out of the COVID pandemic was that we really started finding pockets of inequity for different populations. And so Connecticut introduced a statute into their law back in, I think 2022, that hospitals and primary care practices need to record race, ethnicity and language in a specific way that is through the Office of Health Strategy in Connecticut. And so the Connie HIE is the statewide convener of health information. And so we started working with Russell Dexter and their whole

Charlie Harp (21:40):

Team. The one and only Russell Dexter? The

Laura Bush (21:42):

One and only Russell Dexter. He is their director of population health and really running the REL-D program. And so we have worked with millions of messages in Connecticut to evaluate the race, ethnicity and language quality and have downstream worked with over 20 hospitals within their state to actually improve that quality as we have started uncovering some of the gaps. Most of the gaps have been more of a nomenclature mismatch, not necessarily a lack of recording. It’s just a lack of recording in the correct way. And now we have started working with ambulatory physician practices to do the same. And he has an entire program that surrounds this data quality program where he is working with all of these practices to let them know what their state of their data is and then to actually help them improve that quality as we go forward.

Charlie Harp (22:53):

For me, the REL-D initiative at Connie was awesome because even though it’s kind of a narrow use case, because it’s just that race, ethnicity, language code set, the fact that it was kind of narrow meant that it was highly controllable. And as an experiment, I’ll put finger quotes around experiment. It allowed us to demonstrate what PIQI is about. PIQI is not just about scoring something to say you’re a bad actor. It’s about saying this is why it’s wrong. Because I’m a big believer that if people realize something’s wrong and you tell them what’s wrong, they’ll try to fix it. And so this whole closed loop of here’s your quality issue. You let people know exactly where the issues are. They fix those issues and you see the quality rise to the point where it’s usable data for what you’re trying to do. So I think that that whole process with Connie has been a great example.

(23:50):

And it also, the terminologies used in REL-D are not necessarily the same as USCDI 3.1.

Laura Bush (23:57):

Correct.

Charlie Harp (23:57):

Also. So it’s one of those where now when you pass data in FHIR or CCDA, you can have more than one coding terminology. But this is a very specific one where the terminologies required by REL-D were not the same as the ones required by USCDI 3.1. So that was also kind of an interesting spin on that.

Laura Bush (24:16):

It was. We got to exercise our chops on creating more rubrics that support those different data quality programs. And just like what you said, that closed loop system of discovering the gaps in data quality and then actually working with the downstream producers of that data was really, really eye-opening and exciting for me because that closed loop system is really what I wanted to see in real life. And doing it in a way that was not punitive, that was not degrading, that wasn’t critical. It was partnerships. And the downstream organizations were really, really engaged and really, really on board with actually improving their data as they discovered some of their gaps.

Charlie Harp (25:12):

What was your experience? What was the experience with the contributors to Connie in this process? When you guys gave them the feedback that this is what’s wrong and this is what your quality is bad. Did they say, “Oh no, no, our quality’s great.” And did they push back or did they appreciate having the information and they fixed it? Because a lot of people say, “Oh, no one’s going to ever fix anything. They’re not going to do anything.” What was your experience with these folks?

Laura Bush (25:42):

It was great. It was really, really exciting and fun to see it in real time because Russell has a great disposition and a great relationship with his contributing sources. And it was interesting to see that he would pull up their report and walk through step by step what we found, how we found it, where it was, what we encouraged them to do in order to fix it. And the folks on the other side were sometimes saying, “Oh no, we are definitely sending that information or we’re sending it like this.” And then you would hear them typing and they would come back and say, “Oh my gosh, no, we’re not.” And so it was almost real time that we were hearing the changes being made. Some of the situations were bigger than just interface issues. Sometimes when you’re working in a really large health system that has multiple hospitals throughout the state, you can’t just change your registration processes.

(26:46):

That is a huge undertaking when it comes to just organizational change. And so that was something that we heard also. It was like, okay, we’re really glad that you brought that up, but this is going to be a larger undertaking than just reconfiguring an interface to be able to send this. This is going to be us actually working at the registration level to say, we have to include this, this and this in the ethnicities, this, this and

Charlie Harp (27:20):

This

Laura Bush (27:21):

And the races, this, this and this in the languages. And we’re going to do it, but we can’t do it all overnight. So that was a really interesting concept too. I’ve worked in organizational change for years. My background in schooling is organizational leadership. And a large part of that is behavioral health changes, behavioral changes in organizations. And so I realized as that started bubbling to the top, oh, this is not going to be just an overnight flip the switch situation. It’s going to be, okay, how are we going to create a reason why people should do this and do the change in real time? So I’m curious to see as we continue to work with Connie, how long some of those processes take. But the greatest part of it was that every organization that we worked with was very positively responsive to our feedback and they were positively engaged in making those corrections and being compliant with the state statute.

Charlie Harp (28:33):

Yeah, because I think that a lot of these folks, they have a day job. They have stuff that they need to get done and they’re not necessarily thinking about the terminology bindings. They’re like, “Yeah, we’re just trying to move patients through. We’re trying to take care of them. We’re trying to do these things.” And so being able to do this is for the greater good. It’s beneficial. But like you said, if they’re not aware of it, I really think that data quality is, for all intents and purposes, bad data quality is unintentional information blocking.

Laura Bush (29:08):

Absolutely.

Charlie Harp (29:09):

I think because I can share data with you, but if the data is garbage, I might as well not be sharing data with you. And then I’m not sharing data with you. So I think that whole process with Connie, for me, as I was observing it, since I wasn’t as directly involved as you and Alex were, it was very much a, I’m really curious to see how people respond because I think that’s going to be a signal of how the industry will respond when we start rolling out PIQI in a big way across these different use cases. And so it’s very exciting.

Laura Bush (29:43):

It is very exciting. And I think Connie has been such an amazing pioneer in this work. I’m really grateful for all of the work that they’ve done with us. They’ve been wonderful partners. And because of that, their downstream contributing organizations have been really great partners too. And I think about it in the sense that it’s a cross-department change. It’s not just, oh, it’s our IT department that’s making this change. It’s literally from compliance, it’s legal, it’s IT, it’s patient registration, it’s operations, it’s billing. The whole system is involved in making these changes. And so when we start thinking about the actual impact that we’re having on organizations, it’s quite larger than what maybe we

Charlie Harp (30:37):

Originally

Laura Bush (30:38):

Considered.

Charlie Harp (30:39):

I also think, I was having a conversation with someone at CDC recently and we were talking about rubrics for public health. And one of the things they said, which I thought this was Abby Vail, people sometimes think that I want to have a rubric and these are the terminologies I’m using. But what Abby said in the meeting was, well, we want to use what people are using. We want to meet them where they are. We don’t want to create some new standard that is only applicable to us. Now, I think in some use cases, there might be fields or things like behavioral health. I think there are some things that are not part of the core USCDI 3.1. So I think there could be some variability there, but I think one of the things we’re going to find as we look across these rubrics, and this is another, I would say it’s an unintentional consequence of PIQI, is you can compare the rubrics and you can say that we as an ecosystem are creating confusing requirements where we say, “You have to give me this in this terminology, but for another rubric, you have to give it to me in this other terminology.” And every time we demand or ask for a separate terminology for something like biological sex or race, ethnicity or condition or procedure, every time we create a requirement that you have to give it to me in this other thing, you force a institution to figure out, how do I get you that?

(32:05):

And so I think one of the things that could happen as a result of PIQI is this alignment of requirements so that if we’re asking for something, if we’re asking for a particular data class or a field, that we have a place to go to, to say, “Oh, what is everybody else asking for?” So that we don’t create some weird bird terminological requirement that’s going to put undue burden on institutions. Because I can’t think of anything that’s existed like that before, where if I’m creating a registry or I’m creating a quality measure or I’m creating something where what’s stopping me from just demanding that you use a terminology that you have no other reason to use?

Laura Bush (32:48):

Oh, absolutely.

Charlie Harp (32:49):

So I think that could be another interesting thing that falls out of this.

Laura Bush (32:53):

Yeah. I mean, I think that that’s one of the things that I’ve always found with Clinical Architecture is rather than us creating additional roadblocks, we’re actually creating the flow and the bridges between the systems. We’re not trying to create more work, more stress, more problems. We’re actually trying to create a flow that says, like you said, we’re meeting you where you’re at and we’re trying to get from A to B without hitting a speed bump in the middle of it. So

Charlie Harp (33:30):

Creating those – Maximizing the effectiveness. Yeah, exactly. Hey, that’s our mission,

Laura Bush (33:34):

Right?

Charlie Harp (33:36):

So before we wrap up today, is there anything else you want to share with the listeners? Anything you think people should be on the lookout for or pay attention to?

Laura Bush (33:44):

Well, I think that TEFCA’s really ramping up. We’ve exchanged more than a billion records through the quality information, health information networks, and that’s just more cause for improved data quality so that we can use our information from point A to point B to point C. I’m really excited about the behavioral health work that’s coming down the pike. Consent is becoming more and more of a normal conversation that we’re having, and I love that. So shout out to Carol Robinson out in Washington, really paving the way out there with a computable consent at a statewide level. And those are the things that I’m paying attention to. I think that there’s probably more, but I’m trying to stay focused on creating the best product that we can possibly create.

Charlie Harp (34:40):

Absolutely. And for those listeners that listen to this on a timely manner, what conferences are you going to be at in the next couple of months? You’re going to be at Civitas. I’ll be

Laura Bush (34:48):

At Civitas and I’ll be at NCQA.

Charlie Harp (34:50):

Excellent. I’ll be there too. Awesome. What are the odds?

Laura Bush (34:52):

Oh, hey, look

Charlie Harp (34:53):

At that. All right. Well, thank you very much, Laura, for being on today. I really appreciate it. We’ll do this again.

Laura Bush (34:57):

Can’t wait.

Charlie Harp (34:58):

And for all of you out there listening, thank you very much. I’m Charlie Harp and this has been another exciting episode of the Informonster Podcast. Thanks for listening.