Real-time visualizations improved care delivery at UVA Health

Dr. Claire DeBolt speaks about data visualization at the HIMSS23 global conference in Chicago on Wednesday.

Photo: Jeff Lagasse/Healthcare Finance News

CHICAGO – Patient care in the intensive care unit generates huge volumes of complex data. This influx of data can be a lot to handle, and overwhelms staff people, who are trying to aggregate and summarize pertinent information. University of Virginia Health Medical Intensive Care Unit (MICU) providers and data scientists pursued a tool that would help, employing data visualization to make sense of the information.

According to Dr. Claire DeBolt, clinical informatics, pulmonary and critical care fellow at UVA Health, a number of data science projects have been implemented in MICUs across the country, but their success has been limited. 

Addressing an audience at the HIMSS23 global conference in Chicago on Wednesday, DeBolt said this lack of success has been due to a number of reasons, including mistrust, a misunderstanding of the project and difficulty meeting the high standards that are required. Software and hardware challenges, as well as training barriers, certainly don't help.

"We needed to get the right message to the right person at the right time, on a 24/7 cycle," said DeBolt.

The COVID-19 pandemic made this need even more pressing. The virus stretched staff resources, decreased care compliance and increased both healthcare acquired infections (HIEs) and mortality.

Historically, the UVA Health MICU leaned on checklists and bundles – standardized care packages that were pulled together to reduce patient harm and improve outcomes. But there were a number of reasons why this approach was prone to failure, not least of which were communication barriers, limited customization, poor design and duplication of work. 

Pulling information from the electronic health record in this way was occasionally resulting in missing or incomplete information. 

"It was a pain point for data scientists," said DeBolt. "The team was filling this out every day, and every day it was getting erased. We can't track that data over time."

The theory behind data visualization is that it decreases cognitive workload by helping the brain more easily recognize patterns and trends in a data set – and engage in what DeBolt called "higher-level situational awareness."

There were examples of this kind of technology in use, including iDashboard, a rounding dashboard that used a data visualization scheme entailing color schemes and graphs. eSimpler was another existing solution, which could place a physical checklist into the EHR.

"We hypothesized we could create a continuously updating digital display summarizing nursing documentation," said DeBolt.

A project team led by data scientists, quality improvement coaches and intensive care units began a development phase to concoct data visualization technology appropriate for the MICU. Relying heavily on input from MICU nurses, the team came up with a solution called the ICU Bundle Board.

The software uses color coding – red, yellow, blue and green – so the staff knows where documentation needs to happen. The dashboard filters the information so that red and yellow items are displayed first. It auto-updates every five minutes, providing a nearly continuous display. 

Importantly, it doesn't require any additional documentation, and instead merely reflects the data that's already in the EHR. 

"The tiles can be clicked into so you see why that color is showing up the way it is," said DeBolt. "It spurs people into action. All of those conversations can happen because we know about it."

Under the hood, the ICU Bundle Board pulls data from the EHR's HL7 data streams and the enterprise data warehouse. A RAMP platform reviews the patient data every five minutes and provides the updates.

While it's still early in the project's life, the results so far have been encouraging: a 10% improvement in complete maintenance care and documentation, as well as a significant reduction in red tile duration. 

In the future the team plans to expand the concept to encapsulate respiratory data, with an eye toward other uses down the line. 

"We're continuing to develop the Bundle Board," said DeBolt.
 

Twitter: @JELagasse
Email the writer: Jeff.Lagasse@himssmedia.com