Oregon Health & Science University Hospital is a 576-bed teaching hospital, biomedical research facility and Level I trauma center located on the campus of Oregon Health & Science University in Portland. It is at the heart of OHSU, a large health system of hospitals and clinics across Oregon and southwest Washington.
THE CHALLENGE
Like much of the industry, OHSU has faced a persistent shortage of medical coders, hugely impacting coder case volumes and operational efficiency.
"We knew these issues would only accelerate – we had planned growth, including the expansion of additional beds, that would further perpetuate increased coder volumes and backlog," said Tammy Bickle, director of revenue cycle at Oregon Health & Science University Hospital.
"This backlog resulted in longer billing turnaround times, which increased coding-related denials," she continued. "Payers have timely filing periods, which is a deadline for getting the claim into the payer. Our shortage was such that we struggled to meet timely filing deadlines – so payers were having to adjust charges."
OHSU was losing money from reimbursement for those services, and coders were working an unsustainable amount of overtime.
"We had tried computer-assisted coding to help alleviate these issues, but it didn't have the impact we needed," she recalled.
PROPOSAL
OHSU decided to try AI-driven autonomous coding for a few reasons. First, the organization could not increase the number of coders on staff but still wanted to support work/life balance for the current coder team.
Second, to reduce the backlog. And third, to increase coder volume. So, OHSU turned to vendor CodaMetrix for AI-powered radiology coding.
"Medical coding is considered one of the most time-consuming, understaffed and error-prone parts of the health system revenue cycle," Bickle noted. "The vendor's proposal was to use AI to automate our radiology coding cycle, which helped us end our backlog of cases, provide more efficient and accurate coding, and relieve pressure from our coding team."
MEETING THE CHALLENGE
As the coding team looked at what was sitting in the work queue, the volume and backlog just kept growing. OHSU wanted autonomous radiology coding to reduce the backlog.
"We knew we had to reduce overtime for our coding team, not only for costs but because our team was putting in long hours and deserves work/life balance," Bickle said. "AI-autonomous coding helps reduce our backlog and overtime, makes the workload on our coding team much more manageable, and helps free up our team to code more complex work."
The AI-driven platform is fully automated and integrated with OHSU's Epic EHR.
RESULTS
The results from using AI-driven autonomous coding have been immediate and impressive, Bickle reported.
"For radiology, our automation rate has reached 92%," she said. "Our coder workload was reduced by nearly 28%. Coding-related denials for autonomously coded radiology cases are 70% lower than manually coded cases, and our automated denial rate was substantially lower – 0.33% percent for AI-coded cases versus 1.09% for manual coding.
"MR case denials, a high-cost imaging category, dropped 65% with automation, with an automated denial rate of 0.48% versus 1.38% for manual coding," she added.
ADVICE FOR OTHERS
Don't be afraid to try something new, Bickle advised her peers at other hospitals and health systems.
"All health systems have similar problems, but not everyone is embracing autonomous coding," she noted. "I was hesitant because we had previously tried computer-assisted coding, which failed miserably. My coding team and I were gun shy about using another new technology for coding, but AI autonomous coding truly has been such a relief.
"There was a lot of change management that took place to have key stakeholders embrace the implementation," she continued. "Understanding what the end result would look like was important for getting buy-in internally. I had to convince my coding team to give this a try."
When the team had tried computer-assisted coding – on top of an unwieldy onboarding process – the backlogs continued.
"It made it a bad experience for our team, and many of them lost faith in automated coding," Bickle remembered. "What's more, coders often are worried about losing their jobs to automation, so they're concerned about whether, if implementation goes well, it means they'll be completely replaced.
"It wasn't hard to convince the coding team, however, that there was still plenty of work to be done," she continued. "Even with AI autonomous coding helping in radiology, there are some complex cases where we need human coders to intervene. We told them no one was losing their job, particularly in the context of our growing health system."
Bickle reminded coders that each was just one person, and each could only do so much work.
"I don't like seeing people working 20 hours of overtime a week, it's not healthy," she said. "Once the program was up and running, and they could see they still had an important role to play in the organization, as well as their improved quality of life, they got on board.
"Not having anxiety around these backlogs is game-changing," she concluded. "Our coders were burning out, given the amount of overtime we were using. They were open to using this technology because they, too, knew the status quo was unsustainable. The success with autonomous coding in radiology led me to expand to automating coding with other service lines."
Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.
WATCH NOW: How can an IT exec become a Chief AI Officer – and then work with the C-suite
