Duke University Hospital

Structured reporting, concurrent abstraction, and embedded performance analytics drive clinical and operational excellence at Duke University Health System

In the 10 years since the passage of the Health Information Technology for Economic and Clinical Health (HITECH) Act, the U.S. healthcare industry has seen a lot of change—most notably in the now ubiquitous use of electronic health records (EHRs) to document and manage clinical care. But one thing that’s stayed the same is what is documented: Basically, large blocks of analog text that are relics of the dictation paradigm. While EHRs are doing their job well, capturing and storing patient information digitally, the technology is falling short when it comes to turning that information into data that can then be used for registry reporting and analytics.

To bridge the gap, traditional data abstraction is generally considered to be the best option—but it’s expensive, time-consuming, and inefficient. In the cardiology department at Duke University Health System (aka Duke Heart) in North Carolina, they knew there had to be a better way. According to Dr. James Tcheng, practicing interventional cardiologist and professor of medicine at Duke University, it was structured reporting and concurrent abstraction.

When they first started collecting clinical information as data, Dr. Tcheng says there were maybe 35 elements. Over the past 20 years, however, the demand for data has grown exponentially, making a “homegrown” approach to data management unsustainable. Through it all, he says, “the data has remained sacrosanct.”

Duke Heart has used Epic as its EHR since 2013 and continued to use the structured reporting platform already in place for EP, Cath, and Adult Cardiac procedures. But it was clear that it needed a healthcare IT company to partner with to replace the legacy system. As a trusted cardiovascular registry software provider for more than 25 years, LUMEDX was an obvious choice, and the new Health View Analytics platform meant that Duke Heart’s data would continue to be put to optimal use.

Aligning and integrating its clinical workflow with data collection was the first step. With the whole team involved in the acquisition of data at the point of care, structured reporting enables concurrent analytics that makes real-time insights easy. Benefits have included improved data quality, a boost in efficiency, and better patient care.

“Basically, we have the tools so all evaluation and data capture and cleanup happens during the case at the point of care, or soon thereafter,” said Joe Kelly, Director of Cardiovascular Informatics and Quality Improvement, Duke University Health System. “Data then becomes timely, efficient and significantly more useful.”

But before any of that could happen, there had to be a change in the mindset. “Previously, team-based documentation was actively discouraged,” said Dr. Tcheng, co-chair of the ACC Clinical Quality Committee that authored the ACC/AHA/SCAI 2014 Health Policy Statement on Structured Reporting for the Cardiac Catheterization Laboratory. “The nurse’s note was the nurse’s note, the doctor’s note was the doctor’s note, and never the twain shall meet.”

With the goal of more accurate and complete data, they ignored the skeptics and proceeded with implementation of this new workflow solution. “When it was happening, it was gnashing of teeth and ‘the sky is falling, the sky is falling,’” Dr. Tcheng said, “but it didn’t take long for that to turn into, ‘Why didn’t we do this before?’ They loved it.”

The new workflow was implemented in the cardiothoracic surgery department under the direction of Jill Engel, Associate Vice President of Heart Operations, Nursing and Patient Care Services; and Dr. Peter Smith, Chief of Cardiovascular and Thoracic Surgery and Mary and Deryl Hart Professor of Surgery, School of Medicine.

“We had to help the providers understand that no one was asking them to do more documentation, they are just being asked to document differently,” Engel said.

“The clinical data elements entered by the most qualified provider at the time of the event is the key concept,” Dr. Smith said. “Using the intelligence, skill and knowledge of perfusionists, nurses, and other providers is critically important.”


To optimize operations and ensure accuracy, Duke used electronic rules to validate discrete data points—also checking for missingness. In 2014, prior to LUMEDX deployment, Duke Heart’s missing data elements in the CT surgery department, for a variety of medication data elements, ranged from .8 percent to 47 percent. On the go-live date in 2015, Duke eliminated almost all of these missing data elements overnight.


Unsurprisingly, collection of data at the point of care within the CT surgery workflow and the cath lab significantly reduced the time it takes to produce the final report. What used to take 4-5 days now takes 4-5 minutes. “That’s a remarkable metric in and of itself,” Dr. Tcheng said, adding that concurrent abstraction reduces the need for cleanup weeks or months after a case has concluded.

Best of all, it frees up doctors to do what they’re trained to do. “This really did focus our physicians on the cognitive work— making recommendations for how to manage care,” he says.


Using HealthView Analytics, which is embedded within the workflow, they now also have the ability to extract near real-time insights that can then be used to make data-driven decisions and change clinical practice—not at the end of the quarter or next year, but now. For example, an unusual increase in post-procedural complications can immediately be flagged and once the cause is identified, it can be stopped.


Targeting metrics of interest in near real-time—for example, blood product use in CABG patients—makes it easy to send out weekly summaries. “The sooner you have the data, the sooner you’re able to act on it,” Kelly says. “That is the power of real-time analytics.”