With two hospitals, a physician group practice encompassing more than 2,200 billing providers, and a total of nearly 800,000 outpatient visits a year, the University of Texas Southwestern Medical Center health system (UT Southwestern) in Dallas is a very big patient care organization, with a lot of moving parts. And yet leaders there have succeeded in creating a clinical quality initiative that defies the odds: it has succeeded as a broad, deep, fast-moving, substantive, and agile, initiative, all at the same time. And the genius of this initiative is in how replicable it is. For all those reasons, the editors of Healthcare Informatics have named the team at UT-Southwestern the number-one winning team in the 2016 Innovator Awards program.
For it is at UT Southwestern that clinician, clinical informatics, IT, analytics, and administrative leaders have come together to create an Ambulatory Quality Outcomes (AQO) Project that is improving processes and outcomes across 40 medical specialties, and using rapid-cycle improvement processes to improve care delivery and care management, physician collaboration, and the capability to participate in meaningful quality outcomes benchmarking. All of this is being supported by intensive development work on the part of IT and analytics leaders at the organization.
(l. to r.: ) Jason Fish, M.D., Mark Rauschuber, Duwayne Willett, M.D.,
Jacqueline Mutz, Ki Lai, and Vaishnavi Kannan
Here are the key highlights of what’s been taking place at UT Southwestern:
- A few years ago, Daniel K. Podolsky, M.D., the president of the medical center, gave a clear directive that he wanted his colleagues in the health system to be able to prove demonstrable patient care quality outcomes. Out of that broad directive, the AQO Project was launched in January 2015.
- Early on, a collaboration that spanned four key areas of the health system—clinicians, IT, analytics, and operations—came together under a governance group known internally as the “Quadrad”—meaning a quadripartite collaborative leadership composed of representatives from those four broad operational areas—for the initiative. Under that governance, clinical informaticists, informaticists, clinician leaders, and data analysts launched a strategy that involved rapid-cycle improvement processes, with two-week iteration cycles for clinical performance improvement, leveraging the capabilities within the organization’s electronic health record (EHR—which is the core EHR from the Verona, Wis.-based Epic Systems Corporation), and combining EHR capabilities with the development of an expanded enterprise data warehouse (EDW).
- Also early on, two key decisions were made that helped shape everything that has happened since then. First, the decision was made to engage the physician leaders from as many outpatient specialties (referred to as clinics) as possible, meaning, ultimately, 40 different specialties, rather than simply start with one or a few and slowly roll out processes. And second, related to the first, the decision was made to force process standardization, by creating a standardized data architecture for the clinical decision support mechanism to be leveraged within the EHR, across all specialties, for the initiative’s analytics purpose. At the practical level, this meant literally requiring all physicians within a specialty to document things in the same way and in the same place within their documentation within the EHR, in order to facilitate analytics and data-sharing. Crucially, Quadrad leaders agreed that the only way to spread innovation quickly across the health system was to avoid a repeated “one-off” approach, and instead, to compel physician leaders in all the ambulatory specialties to work collaboratively on a standardized data and documentation approach to the initiative.
- One very broad specific goal of the initiative was to create robust patient registries in all the specific medical specialties. To date, the leaders of the initiative have built 58 specialty-specific patient registries, created 134 process and outcome measures covering 77 primary and 44 additional medical conditions, created 111 new clinical decision support tools within the EHR, and created 97 new workflows across 40 specialties. Very importantly, 58 patient registries have been created, with over 16,000 patients documented to date, in those registries.
From the start, a broad-based, standardized approach
So why did the UT Southwestern leaders focus on the outpatient sector rather than the inpatient sector? The answer is simple, says Duwayne Willett, M.D., UT Southwestern’s CMIO. “We made the decision to build an initiative that we felt had to cross all specialties. And this was an approach that has applied across specialties.”
The reality, says Mark Rauschuber, the organization’s associate vice president and CIO, is that “We’re a proud institution; we have six Nobel laureates on staff at the university. But we wanted to show the data to prove our quality of care.”
Given that broad mandate to spread innovation organization-wide, “We had begun by doing two-week iteration cycles with our Epic EHR team a year earlier, so we had gained experience with delivering product on a rapid-turnaround cycle structure and make adjustments,” Rauschuber says. “And it was pretty daunting to look at 40-50 specialties. And in the past, everything had been one-off—we would tackle a project, do something in the EMR to support the project (clinical decision support), and then do a data warehousing element, so it would be many, many months for every project. So we wanted to create a repeatable process with a menu of items for registries in our EMR which is Epic, and then extract all that information in our data warehouse, so that this could be implemented broadly.”
The decision to move forward to compel physicians in all of the medical specialties to document in a standardized way within each specialty the key patient data from each specialty turned out to be the element that has turbocharged this clinical transformation and patient care quality improvement process. Not surprisingly, that approach separates the UT Southwestern organization from the small-bore, one-at-a-time clinical performance improvement approaches of so many patient care organizations nationwide.
UT Southwestern leaders are very conscious of the importance of that factor in their success to date. “From an analytics and IT standpoint, our culture had been one where we were struggling along, giving people some really cool one-off projects, but we shifted towards broad building of quality,” says Jason Fish, M.D., assistant vice president for ambulatory quality, and associate CMIO, for the organization. “So we had a lot of learning around that, to build things in a way that fit within this framework. Plus, we were pushing Epic: they had been taking a similar one-off approach. Instead, I asked, how do we strategize to build a replicable process that is rapid and can be replicated across the entire system organization?”
Fish strongly credits Vaishnavi Kannan, business analyst, electronic medical records, with helping to build the infrastructure needed to support the analytics work required to fuel the initiative. Kannan in turn credits the collaborative spirit of her IT colleagues at UT Southwestern, including that of Ki Lai, the organization’s assistant vice president of enterprise data services, with their collaborative spirit in working with her and her fellow data analysts to make it all happen.
How do Kannan and Lai frame the work that’s been done so far? “When Dr. Fish came up and spoke about the project, it really was about engaging the specialty practices,” Kannan says. “A lot of processes like PQRS [the federal Physician Quality Reporting System] really focus on primary care practices, and there wasn’t a lot [of foundational work established] for specialists, so we wanted to engage our specialists on quality, so to create our own quality incentive program, we asked leaders from each specialty to select top priorities. And the key guiding principle was to bring along team members from across the Quadrad, and to then constantly communicate with different focus groups, to then identify the key requirements for each practice group, and then utilize concrete resources in the EHR, and then come up with concrete practices.”
Says Lai, “This project has been very patient-focused, which has been very exciting for all of us. I’ve managed the IT development team. As exciting as the prospect was,” he says, “we also realized it was a tall order, given the number of registries we had to create, and we hadn’t done it before. So we knew we needed a new approach; there was no way we could do this in an old-fashioned way.” Out of necessity, Lai says, “We knew that we could not create individual data models, but instead, had to figure out how to create a generic data model for all the registries. And we defined a data set that included qualifier, denominator, numerator, and exclusion, based on industry best practices.”
Second, Lai notes, “We had to create design patterns to help us handle the data that comes from the EHR into the EDW”—enterprise data warehouse. “In other words, we standardize and we say, this we’re going to handle current medications, for example, and then we say, this is how we’ll handle the procedure groupers; or the problem groupers. We’ll always look at the same places, standardizing the places in the EHR where we’ll put things.” On top of that, Lai reports, “We created a transparent rules engine. We cannot be successful if all this code the programmers write is hidden somewhere that no one can validate or see rapidly what’s going on or validates the users’ stories,” he emphasizes.
With those key decisions in place, the Quadrad-based multidisciplinary team moved forward. The biggest challenge? On a process level, Kannan says that it was this: “As we started with a couple of clinics, we noticed that when capturing the requirements from our customers, we noticed that there was a conflict between the stakeholders and the analysts themselves in terms of understanding the requirements. Each group of specialists might understand the requirements differently, and within the quadrad itself, we might have misunderstandings about requirements. Also, we followed an agile approach while building these registries for the customers, and at the end of each iteration, we would do a demo; but then the customers would change their minds, and then would set us back. They might change their minds about the usability of a measure, or the understanding of a measure, itself, for example.”
Willett says that some of the key challenges, in his view, have included “technical challenges, when we were doing new things and in some cases, stretching the vendor; and certainly, there have been group dynamics challenges.” What’s more, he says, “We learned a lot about process, and the learning curve has been rapid.” When it comes to physician culture, he adds that “As you can imagine, simply exhorting doctors to use features or problem lists in the EMR only goes so far. Here we, appealed to the idea of specialists focusing on the condition or issue most interesting to them, and then working through a process where we said, we want the patient to enter as much information as possible, the staff to enter as much information as possible, and where we need your judgment, let’s make it work within your workflow; and they were deeply involved in the design.”
Is this kind of process of innovation replicable? Absolutely, affirm all the UT Southwestern Medical Center leaders. And what would they tell CIOs and CMIOs across the country? “I’d stress the teamwork, and not to be afraid of engaging clinical department partners, to work for the greater good,” Rauschuber says. “Yes,” Willett affirms. “This concept of co-development, where you’re all in it together, going through these rapid cycles together, not just throwing something over the fence—that is key.”