Geisinger’s Nick Marko, M.D.: What It Takes to Build a Data Infrastructure for Continuous Clinical Performance Improvement | Healthcare Informatics Magazine | Health IT | Information Technology Skip to content Skip to navigation

Geisinger’s Nick Marko, M.D.: What It Takes to Build a Data Infrastructure for Continuous Clinical Performance Improvement

May 7, 2016
by Mark Hagland
| Reprints
Geisinger’s Nick Marko, M.D., shares insights on the creation of data infrastructures to support clinical performance improvement

Nick Marko, M.D., a practicing neurosurgical oncologist, is chief data officer at the Danville, Pa.-based Geisinger Health System. Dr. Marko arrived at Geisinger three years ago, following the conclusion of his residency and two fellowships, and has been practicing neurosurgical oncology since then. Two years ago, after leading… he was named chief data officer at the integrated health system, which is renowned for its innovative work in clinical performance improvement, including its ProvenCare program, which was created and nurtured under the aegis of former CEO Glen Steele, Jr., M.D., Ph.D.

Dr. Marko spoke recently with HCI Editor-in-Chief Mark Hagland regarding the advances that he and his colleagues have been making in leveraging data to improve clinical performance, organization-wide, at Geisinger, which encompasses 12 hospital campuses, two research centers, 30,000 employees including nearly 1,600 employed physicians, and a 510,000-member health plan, and which as an integrated delivery system serves more than 3 million residents through 45 counties in central Pennsylvania. Below are excerpts from that interview.

Geisinger has a wonderful legacy as an innovator in so many areas around clinical performance improvement and evidence-based medicine. What is the landscape now at the organization for leveraging data for clinical performance improvement?

I think the appetite for using data and information for providing excellent clinical care and an excellent patient experience is as strong as it’s ever been. What has evolved over time is a couple of things. One, we’re seeing the evolution of our data-driven quality innovations going from a series of one-offs or use-case scenarios, to where we’re really trying to build at scale. So when you look at our ProvenCare programs and other population health programs, we’ve been pretty successful at using data to improve care and processes in individual situations. The next logical step is of course taking that approach and making it just the way we do business, right? And I think the first real hurdle for any organization that wants to do that is that you’ve got to have a culture that embraces and looks at data.

And that’s non-trivial in healthcare. A generation ago, the way physicians were trained was individualistic and worked against that. And to some extent, the evidence-based literature movement really pushed towards moving away from the individual experiences of physicians and towards the collective experiences of many. And becoming a data-driven organization is part of the fruition of that. We’ve taken a step beyond just evidence-based medicine and meta analytics, to taking raw data from patients and looking at the data in real time and on the fly. So it’s not just, what one answer can I get to one question based on meta-analysis, but, how can I do this all the time?

So medicine is evolving towards that, and the way that Geisinger is evolving forward, is evolving towards that. So a data-driven culture is really part of Geisinger’s DNA now. And that’s great, because it makes it much easier for people like me to do what we do. If you have to constantly convince people that looking at data is a good thing, then that takes up 80 percent of your life. Now, that doesn’t mean we don’t have challenges: getting the right data into the hands of the right people at the right time, and making sure to maximize our resources around data use, and how we do this at scale, all remain challenges. And the scale issue is one of my biggest challenges. You can end up with 100 one-off solutions [in leveraging data and analytics to do clinical performance improvement work]. But if you’re really going to have it be truly a part of what you do every day, you have to build a stable backbone, because you can’t keep pace with one-offs; managing one-offs doesn’t scale.

So we’re working towards using our own long history of data here, tying into the best available data.

In other words, a complete end-user-capable data infrastructure?

Yes, and a culture that supports that. And similarly, you can have a ton of people who want to use information, but if they don’t have the back-end capability of using the systems, they won’t be able to use the data. So we have to live and breathe information in everything we do all the time.

What has it been like being a specialist physician in a position that is most commonly held by a PCP? How has being a specialist been different for you?

Yes; the data-driven, evidence-based culture, has largely been dominated by people in primary care. And that largely has been logistically driven. Those people can do clinic on Mondays and IT on Tuesday, for example. For us, we kind of have to be available at any time. I’m lucky that people here understand that my patients have to come first, and my schedule will be unpredictable, and people don’t look down on that.

You’d think that they’d be thrilled to have a specialist involved, right?

Yes, they are, but it’s a little bit like marrying a doctor and finding it out it’s different from what you thought! So the organization has to be accommodating, and the other thing is, the organization that specialists are more expensive. So they have to be willing to subsidize you at the specialty rate, because specialists just won’t take a gigantic pay cut. And we’re all doctors, but we have to have a skill set that contributes here. So you’ve got to have that organizational physician.

The other thing is that specialists tend to have different mindsets. The summary of it is that specialists in general are often the end of the line in terms of people a patient has seen. We have to work with a great deal of uncertainty in our clinical practices, so I think that translates into how I do the data stuff, too, because that translates into my wanting to get us moving as fast as we can, while maintaining quality, and then we can make changes. Whereas sometimes people get a bit stuck in over-planning. And neither one of those things is right or wrong, and we have to balance things out.

So you’re saying specialists are more impatient than PCPs?

I think they’re more comfortable doing things under imperfect conditions. A PCP will want everything to be perfect, whereas a specialist will say, let’s just do this now and get it done. I’m pretty anti-meeting. I’m about getting in there and getting it down. And that can frustrate some people. I actually think it’s good, because there’s a balance there between me being aggressive and others being more conservative, but I actually think it balances out.

Tell me a little bit about some of the initiatives you and your colleagues have been involved in recently.

You know about ProvenCare. The next tier of things has involved more dynamic feedback to help us do care management in a more data-driven way. So whether it’s a population health management program with very timely data feedback, that’s got to provide us with some kind of dynamic view, that’s one example—things that are either population-centric or patient-centric that change on a fairly regular basis. Another example is leveraging new types of data that maybe weren’t so easy to use or accessible before. So we believe there’s lots of value not only in the structure data that comes out of the medical record, but all sorts of things captured in free text, unstructured text. That’s long been sort of a black box where information goes in but it’s hard to go out. So we did a proof-of-concept study involving using natural language processing to help us analyze radiology reports. And it was AAA—abdominal aortic aneurysms.

What made you and your colleagues choose to analyze radiology reports for AAA occurrences?

Patients will often get a CAT scan of their abdomen for other reasons. And the radiologist will often have an incidental finding of an AAA. And oftentimes, if physicians didn’t order the scan for that reason, they’ll just move on.

So that incidental finding will get lost, and you wanted to get it ‘unlost,’ essentially, correct?

Yes. So can we flag it automatically and make sure a case manager does something about it and helps bring to the physicians the decision of what to do about it. But that involves leveraging automated texted analytics that then feeds into a traditional informatics workflow, where this information comes into case managers. And then they reach out to patients and say, hey, we spotted this thing, I don’t know if anyone mentioned it or not, but somebody might want to watch that.

And then the case manager reaches out to the referring physician?

Yes, then they’ll reach out to the ordering physician, ask them whether they want to refer the patient to another specialist—they make sure the patient is connected to somebody.

How many occurrences of this kind of incidental finding around AAA are you identifying now?

We are identifying a couple of hundred of these a year. They all require some kind of follow-up. At the very least, you surveil them. And we’ve had multiple patients who have gone on to have treatment. About a year ago, we had a patient who decided to have their aneurysm electively operated on, it was about to erupt. They had surgery, which was wonderful, because the patient could have died. And it’s happening for something that might otherwise have been overlooked.

And what’s important about that story is that, based on that, we started doing that for other things. I’m a neurosurgical oncologist in my practice. And we started noticing here that particularly here in Pennsylvania, where patients might be seen elsewhere but having imaging here, patients with cancer sometimes get scans of the brain that show incidental brain metastases. And sometimes they receive only radiation. But we’ve got other options, and many times patients could do better with other treatments as well. So we’ve put together a similar program for incidentally spotted brain metastases. So we will have our brain tumor scan ordering nurses reach out to the ordering physician to offer to set up an appointment for the specialty resources we have available for them. The decision as to whether physicians will order that further specialty resources be offered is of course up to the ordering the physician. But if we can help connect people, that’s a good thing.

So these experiences have stimulated a broader effort, correct?

Yes. The follow-up to all of this is, based on a couple of these successful one-off use cases, we said, now that we’ve seen this work in a variety of domains, let’s switch from individual use cases to scale. So we’re creating a comprehensive text analytic framework, so that not just a select subset, but so that all of our text documents, on an enterprise scale, go through this same NLP process and content markup—so that anyone who wants to design a program like this, has an infrastructure on which they can do it. So it’s a matter of moving from solving specific problems to providing infrastructure at scale to allow people to do any number of things they might want to do. So this is the difference between building a ProvenCare pathway for cardiac disease or a data analysis program for AAA, to scaling all that up to a new level of comprehensive infrastructure. So lots of new tools and projects can come to fruition. And getting back to our initial theme, that’s the logical progression—from doing it right with clinical evidence to doing it right with data; and to moving from individual use cases to scale; and to making this just a part of what we do and how we do it. So our job is to make this available on the care side. And then the clinicians develop good habits. And that’s the difference to doing it once, and doing it on an institutional level. And that’s our progression.

What have been your and your colleagues’ biggest learnings around all this, on the journey so far?

I think the biggest one is that it’s really not terribly difficult to find lots of information generated in the course of what we do anyway, that will generate use cases for improving care delivery. And it’s not all that difficult to take that information we have anyway, take it off the shelf and wrap tools around it, and then bring all that into the way you do business. That's a cultural evolution, and the data is really the currency of that whole revolution.

What would your advice for CIOs, CMIOs, other healthcare IT leaders, be, as they consider building an infrastructure for analytics to support continuous clinical performance improvement?

My advice would be to think about your data as an asset, and something different from your technology, and your informatics. We’re really looking at it as that data, technology, and informatics, are really different phenomena. And it’s not the same skill set that makes someone a good informatics or technology person, that makes someone a good data person. So embrace the fact that each of those three things is separate. And that’s why I think the idea of a CIO who does everything is a concept that’s difficult to sustain. It’s like healthcare in general, right? Everything is getting so subspecialized. And in this case, technology, informatics, technology, and strategy, are all separate things. And how you start to put those things together, that’s when you really create change in an organization.

 

 

 


The Health IT Summits gather 250+ healthcare leaders in cities across the U.S. to present important new insights, collaborate on ideas, and to have a little fun - Find a Summit Near You!


/article/geisinger-s-nick-marko-md-what-it-takes-build-data-infrastructure-continuous-clinical

See more on

betebet sohbet hattı betebet bahis siteleringsbahis