Health systems are under growing pressure to increase operational efficiency, improve patient outcomes and adapt quickly to shifting care demands. At Becker’s 13th Annual CEO + CFO Roundtable, in a session sponsored by GE HealthCare, leaders from Duke University Health System (Durham, N.C.) and PwC explored how data-driven decision-making and integrated technology are helping large health systems orchestrate care across complex environments.
Here are four takeaways from the discussion.
1. Siloed systems can’t keep up with today’s care complexity.
Duke University Health System, a four-hospital system with more than 1,700 beds, once managed operations in isolation — each hospital using disparate data to make decisions. This fragmented approach created bottlenecks in emergency departments, extended wait times and hindered systemwide visibility.
“Before implementing the GE solution, we were looking at our hospitals in a very siloed manner,” said Kristie Barazsu, president and COO of Duke Health Lake Norman Hospital and vice president of Duke’s Patient Flow & Transfer Center. “We needed something that allowed us to look at the system as a whole so we could better manage care progression of patients and care for more patients, as demand was increasing.”
The answer was a command center powered by GE HealthCare. The system integrates data across entry points such as EDs, operating rooms and transfer centers — many of which fall outside of the EHR — to enable dynamic decision-making and predictive analytics for patient placement, staffing and throughput.
2. Change management is critical to success.
A centralized view of operations alone isn’t enough. As Kathryn Burg Plaza, CRNP, MSN, RN, a principal at PwC, explained, it takes a deliberate shift in workflows and team collaboration to translate insights into outcomes.
“What I’ve seen with the aggregation of data — which is so powerful — is not only do you know a discharge date and discharge plan, you actually know what is holding up the discharge from happening on a given day,” Ms. Plaza said. “With that type of insight, you can prioritize the x-ray and ensure the consult is completed in a timely manner … What is challenging about this is, how do you think through workflows in a different way, and how do teams collaborate and communicate differently to have this type of tool [command center] be effective?”
To facilitate adoption, Duke implemented two-week sprint cycles modeled after agile software development. Leaders determined the strategies each service line needed to focus on week one, front-line clinical teams developed processes to address automated alerts on quality, safety and care progression week two. During this 2 week sprints, the team would come back with opportunities to improve which the GEHC team took back and incorporated prior to the next sprint.
“We were able to roll one portion of the technology out at a time, make sure we got it right and build trust with our team,” Ms. Barazsu said. She added that aligning teams on “what matters most” — zero patient harm — went a long way.
“It’s been something that has worked really well, and we are actually spreading the way we roll out change in that manner to other areas that are not just tied to patient flow and our command center,” she said.
3. AI-powered placement improves quality, efficiency and equity.
Duke’s use of predictive analytics now extends far beyond bed assignments. The health system incorporates more than 20 clinical variables to determine optimal patient placement across its network, enabling timely access to subspecialty care while easing strain on capacity.
This approach allowed the organization to reduce lengths of stay, improve care coordination and cut temporary labor costs in half — saving $40 million. Variance in staffing needs fell from 7.3% to 1.1%.
“We were able to ebb and flow dynamically and had a report that would push out every four hours to show us how we did,” Ms. Barazsu said. “We were constantly learning from [the data/results] and [adjusting] how we staffed our units.”
4. Future-ready hospitals will be orchestrated, intelligent and automated — and bolstered by trusted partnerships.
As Duke expands, the health system is applying these lessons to new acquisitions. Ms. Barazsu emphasized the potential of combining real-time device data — such as CT scanner availability — with patient flow tools to inform scheduling and resource allocation in real time. She also discussed integrating urgent care and post-discharge pathways to further streamline patient movement and reduce unnecessary readmissions.
The command center may be a physical space today, but at Duke, it’s evolving into a new operating model: one where AI enables enterprise-wide care orchestration and everyone from case managers to ED staff plays a role in achieving zero harm and system agility.
With this type of transformation, panelists emphasized that health systems can’t go it alone. Ms. Barazsu and Ms. Plaza underscored the value of strong collaboration between providers and technology partners.
“You have to find a partner that’s willing to work side-by-side with you and is there in the moment,” Ms. Barazsu said. “As we’re rolling this out, we have partners that are there hearing what the users are experiencing in real time, trying to fix alerts and make sure that it works tomorrow. Find out what matters most, and then make sure you have partners that are supportive and work directly with you.”
