UNC Health Care shares how it reached peak analytics maturity

Its success with the HIMSS Adoption Model for Analytics Maturity highlights the value of investing in sound data governance practices and intuitive visualization tools.
By Mike Miliard
10:59 AM
Jason Burke

The HIMSS Analytics Electronic Medical Record Adoption Model has long been a valuable methodology for hospitals and health systems to assess their implementation and use of information technology.

More and more providers have matured in their capabilities in recent years, and now hospitals are routinely reaching Stages 6 and 7, the top rankings of the framework.

But the EMRAM is just one adoption model developed by HIMSS Analytics. Another valuable tool is the newer Adoption Model for Analytics Maturity. It aims to help measure and advance providers use of data – beyond mere clinical decision support and toward more transformational clinical and operational improvements.

The model has eight stages, designed to progressively measure and advance providers' analytics capabilities and track the gains they're able to make from more effective use of technology and efficient care delivery processes:

Stage 7: Personalized medicine and prescriptive analytics

Stage 6: Clinical risk intervention and predictive analytics

Stage 5: Enhancing quality of care, population health, and understanding the economics of care

Stage 4: Measuring and managing evidence based care, care visibility, and waste reduction

Stage 3: Efficient, consistent internal and external report production and agility

Stage 2: Core data warehouse: centralized database with an analytics competency center

Stage 1: Foundation building: data aggregation and initial data governance

Stage 0: Fragmented point solutions

More than 350 hospitals have successfully attested to Stage 7 on the EMRAM, but so far just two  organizations have earned the top rank of the Adoption Model for Analytics Maturity.

UNC Health Care & School of Medicine is one of them. At the HIMSS Big Data and Healthcare Analytics Forum in Boston on Oct. 22, Jason Burke, UNC's system vice president and chief analytics officer – speaking alongside Philip Bradley, regional director, North America for HIMSS Analytics, during the session "Milestones on the Road to Analytics Maturity" – will offer perspective on his organization's analytics journey.

He'll share AMAM’s value to providers aiming to make the most of their data and how UNC managed to make headway with its own efforts – prioritizing analytics as an asset, embracing sound data governance practices, disseminating data visualization techniques across the enterprise and more.

To start with, Burke's department at UNC Health Care, Enterprise Analytics and Data Science, is its own business unit. They partner with the IT team, known there as ISD, to develop and deliver advanced analytics capabilities across the enterprise.

"ISD came to me and said, 'Hey, HIMSS also has this analytics model, would you be interested in participating?’” said Burke. "I thought it was a great idea for several reasons. One, I always value having an outside perspective coming in and providing some feedback on where you're at and what you're doing. Our group is reasonably new to the health system, we've only been here for the past two or three years, so it was a great opportunity."

It was also a compelling way to "showcase some of the partnering that had already happened, between the folks in IT who focus on analytics and the folks on my team who focus on analytics," he said.

"All those years of working with HIMSS" – UNC also attained Stage 7 on the EMRAM earlier this year – "had already helped develop a set of core infrastructure capabilities that put us well on the way with this. We weren't starting from scratch. We already had a tremendous amount of energy and momentum and real capabilities that had already been achieved."

In 2014, UNC had made a series of commitments for IT strategy – going system-wide on Epic, pooling data in a centralized warehouse and, crucially, making a "fairly strategic commitment to enterprise analytics and data sciences," said Burke.

His team petitioned the health system's board with a pointed request, he said. "We think our health system has to be different, more sophisticated, more compelling as it relates to data science and analytics. We think the future of healthcare will be data driven, that analytics will be a foundational capability that every health system will need to deliver the right care to the right patients at the right cost.”

"It was a fairly big ask," said Burke. "But our leadership really got behind it. And over the past two-and-half years we have considerably grown and invested in a lot of the capabilities that are in Stage 7: developing predictive analytics, data governance, quality and resource management orchestrated together to drive better-octane fuel for analytics."

Good governance 'no longer optional'

Data governance, as has been often noted, is an essential first step.

"We certainly think it's no longer optional," said Burke. "It never really was, and it's definitely not now. We're pursuing what I call a multi-pronged strategy around data governance – we’ve hired data governance professionals into a team, called Data Governance, under an Executive Director of Data Governance."

The domain covers several different categories, such as data stewardship across the organization, "trying to help the business establish and maintain accountability for core data elements that become foundational building blocks for the healthcare system," he said.

"It also includes aspects of standardization, where we're helping the organization reduce the variability – not just in the data but even in the definitions of some of our performance measures and metrics," Burke added.

It's a frustration across healthcare that oftentimes organizations "can't even agree on basic definitions of things like length of stay," Burke added. "And in my world, I often describe trying to get the health system to a common source of truth. There are many different ways people consume information, but if we can get to one source of truth that will help people understand how to interpret information, how to use it in analytics."

As such, much of UNC's data governance efforts are "really focused on transparency," he said.  "We've found that one of the biggest issues people have is around data governance is just not having access to information about the data: What is it, where did it come from, how should I use it, what's an appropriate and inappropriate use of information?"

To fix that, the health system spent a lot of time working to "create and expose documentation to end-users to help them contextualize the data," Burke explained. "My group is called EADS, so we have a website called 'EADSpedia,' where you can go to and type "LOS" in the search engine and get all of the definitions about length of stay – what they mean, where they're used. It helps people make more informed decisions."

UNC has also developed a "certification process for data," he said, where members of his team work with different business units to certify the integrity of their data with levels of certification: gold, silver and bronze.

"Each one of those levels establishes greater levels of confidence and clarity," he said.

Data visualization and self-service enablement

When EADS was first formed, it did an assessment of how the staff at UNC Health Care were consuming information.

"Most of it was in more in the way of traditional business intelligence applications, as opposed to data visualization," said Burke. "So we initiated a process to bring in additional data visualization capabilities – in our case we opted for Tableau – and established a data visualization center of excellence within the healthcare system, where my team and ISD use Tablea for the creation of a lot of assets that traditionally would have been just a table or a report somewhere."

Self-service enablement is a centerpiece of UNC's enterprise analytics strategy, he explained. "A lot of what we've been doing now is trying to get data and training assets into the hands of federated analysts around the health system so they can create the data visualization applications that they need themselves."

There are several ways EADS is doing that, but one of the most popular is called Orbits – an acronym that stands for Operational Reporting and Business Intelligence Teams.

"They're business and clinical subject matter experts that pair up with EADS and ISD counterparts to develop and standardize the ways a particular department or clinic or team needs to consume our enterprise information," said Burke.

"So, for example, we have an 'orbit' that's focused on perioperative services, one that's focused on revenue cycle, on that's focused on hospital quality, etc," he added.

Tableau and tools like it are big helps when it comes to self-service initiatives, Burke explained. "They tend to be the model of choice because they take a lot of things off the table. You don't have to be a PhD in any sort of analytics discipline to know how to leverage those tools."

3 tips for getting to AMAM Stage 7

While there's only a small handful of providers atop the HIMSS AMAM right now, Burke expects the number of Stage 7s to increase substantially in pretty short order.

"Every organization is in their own place in their journey toward this," said Burke. "But again, I don't think it's optional. This is the way medicine should be practiced, and we're finally at a tipping point."

He offered three pieces of advice that worked for UNC Health Care – and may well work for other health systems when similarly applied.

"One is this idea of business-IT partnership," he explained. "Most of the work that's involved in data and analytics is not technical. It sounds very technical, but most of it is about getting people to a common understanding of what our data means and how we can best leverage it. That's a collaborative dialog between business and clinical and technology staff. I just don't think you can do this work unless you focus on that."

The second, he said, is that UNC has invested extensively in governance – not just data governance, but "governance around creating groups that are empowered to make decisions around what our data should be used for and how do we characterize it and what do we standardize on. Those groups inevitably help us to develop priorities. There's so much pent-up demand here that it's impossible to do it all at once – you need governance to facilitate the decision-making and the details of the project but also to oversee the investments."

The third piece of advice he offered is to "challenge the status quo," said Burke. "We're very intentional in really pushing people to ask the five whys – what decision are you going to take based on the insight you're asking for? And as those conversations unfold there are all kinds of opportunities for innovation that emerge, and all kinds of opportunities for challenging the status quo."

Big Data & Healthcare Analytics Forum

The Boston forum to focus on effective pop health management, AI and precision medicine Oct. 22-23.

Twitter: @MikeMiliardHITN
Email the writer: mike.miliard@himssmedia.com

Topics: 
Analytics