Photo: Morsa Images/Getty Images
Small and rural providers often don't have the resources to roll out sleek new operational enhancements, especially artificial intelligence capabilities. But those that do are finding that AI is helping improve key operational metrics and gaining staff enthusiasm – whether they implement the tools quickly or gradually.
We spoke recently with two rural hospitals that structured their AI integrations very differently. If financial considerations can sometimes be a barrier to adoption, their experiences and early results prove that even lesser-resourced organizations can successfully implement leading-edge technology in a way that works for them.
Patterson Health Center is a critical access hospital (CAH) in Harper County, Kansas. It focused on a gradual culture shift to implementing AI to improve staff work-life balance. By making AI tools in its Oracle EHR optional – and actively communicating with providers about its use – the rural provider helped reduce "tech anxiety" and achieved adoption across its small organization.
CGH Medical Center, meanwhile, is a municipally owned, 99-bed hospital located in Sterling, Illinois. It used the opportunity of a major migration to Epic to make use of that vendor's Launchpad AI Starter Kit. By upgrading to a new EHR and introducing AI at the same time, the rural hospital was able to achieve reductions in staff documentation burden while lowering the learning curve for a new system.
While a slow and deliberate experience worked for one hospital, an all-at-once approach worked for the other.
"You don't have to necessarily pilot it in areas and start small," explained Jolee Parks, CGH's director of health informatics. "You can just kind of go 'Big Bang.'"
Steady approach builds buy-in
Still, earning rural patient trust with newer AI technologies can be plenty challenging for some smaller, rural organizations.
"Oftentimes in rural communities, we’re afraid of that pushback," said Sarah Teaff, Patterson Health's CEO. "We used it as more of a wellbeing-type strategy for our providers rather than a return on investment."
Patterson, a 16-bed CAH, worked with Oracle Health's CommunityWorks to implement the vendor's Health Clinical AI Agent and help automate the "hard work" of charting.
Through a gradual introduction and the optional use of AI for patient notes, the rural health center used a "long runway" approach to AI to reduce physician burnout and improve face time with patients.
AI removed the barrier of "clicking through boxes," allowing providers to be more "one-on-one" with patients, Teaff said. The tool’s ability to "weed out" non-clinical conversation – small talk about a patient's child's basketball game, for instance – and target the relevant medical data helps maintain the quality of AI notes the staff reviews.
Adding the clinical agent into its previously upgraded Oracle EHR was a "flip switch" experience and did not require a steep learning curve or extensive retraining, though the CAH's staff was also provided frequent demos.
During these sessions and other staff communications, the center repeated the message to staff that using AI is both the doctor's and patient's choice. Despite initial fears that a rural community might be skeptical of AI, the rollout was also successful from a patient perspective.
Tech training stressed the need for providers to obtain explicit consent from patients before using ambient listening and documentation tools during visits, Teaff said. To date, Patterson Health has achieved 84% agreement to use AI during patient encounters.
While Teaff said the most measurable success was using the new clinical AI agent to record and summarize visits, an Aha! moment came when providers using the AI tools realized they could also dictate notes during their commutes, Teaff said.
Per patient, documentation time dropped from 20 to 12 minutes, while time spent in the EHR per patient decreased from 15 to 9 minutes.
Use of AI has also given back to Patterson Health's doctors using the AI tools roughly one to two hours each day, significantly reducing their after-hours documentation time.
"We can see the timestamps of when they're in the [EHR], those that are having to do things after hours," she said.
Teaff said they are hopeful that all seven of its providers will adopt the tools.
"Hopefully with these first small wins with our providers who are utilizing it, we'll slowly get 100%," she said. "They need to be doing what they're good at, which is having that face time with their patients."
'Rip the Band-Aid' for faster gains
CGH Medical Center, which is about a two-hour drive west of Chicago, launched a new Epic EHR with Microsoft's Nuance DAX AI on Nov. 3 to enhance operational scaling and solve staff shortages.
While Epic offers smaller community and rural hospitals a chance to share an EHR and its costs through its Community Connects program, CGH opted to migrate from a highly customized, 20-year-old legacy system with components from Cerner and NextGen to a new stand-alone EHR with the vendor's AI Starter Kit that would easily scale across the hospital's various departments and specialties.
"There was a lot in terms of the documentation requirements, things that we're asking them to do regularly – more discrete data fields," explained Ben Schaab, the center's chief financial officer. "There were a lot of pain points in terms of the flow of data and a lot of pain points in terms of physicians' time in charts."
The AI, only available in CGH's outpatient setting at this time, will "tee up orders through that conversation with the patient," he said. "It'll prompt them for certain things that they discuss with the patient," including tests or prescriptions, and draft notes automatically.
Early results already show a drastic reduction in clinical documentation burden and immediate gains in coding and reporting. CGH does not yet have data on documentation burdens reduced or error coding rate drops from integrated AI claims processing tools.
But with machine learning automatically writing notes and extracting diagnosis codes for ancillary and emergency department visits, it does alleviate staff and staffing burdens.
The same is true for creating data dashboards and reports.
"We have a lot more ability to do self-service reporting," Parks said. "You can basically have a conversation with the reporting module and say, 'I'm looking for XYZ,' and it can build the report for you or build the dashboard for you."
One standout success CGH was able to share is that in just the first eight weeks of being fully live with the new system, it identified 310 incidental radiology findings by pulling out discrete findings in paragraphs of notes and then prompting follow-up with patients.
Notably, launching a new EHR and new ambient listening tools helped CGH advance quickly beyond a typical new EHR learning curve.
"We were intentional about what things we wanted to turn on and how we wanted to roll them out and use them," said Parks. "I think it paid off for us to do it all at once."
Because CGH wanted to launch its new EHR with AI documentation available to everyone – including nurses with challenging workflows like end-of-shift summaries – Schaab and Parks piloted ambient documentation tools with eager providers earlier in the year.
With the migration, CGH providers were then designated certain pieces of the new workflows, "where maybe a support staff had previously been more engaged," Parks explained.
That way, "they didn't have to spend so much time learning where to document certain things" in a brand new EHR.
With AI, cardiologists, for example, can "speak through what they're doing and then get a procedure narrative note that they can just modify, and move on," she said.
CGH also tapped AI physician advocates to spark colleagues' interest, and even those superusers were shocked by some of the ways AI could enhance workflows.
"We did have some at-the-elbow support physicians that we brought in just to give a little bit of extra support to our providers," Parks said. When they saw the new AI features, "one of them commented, 'I would go back to being a hospitalist after seeing the hospital course draft note.'"
Finding their champions also meant CGH didn't need a full "all-in mandate" to start with AI – the word simply spread.
"From the feedback we're getting from [CGH] providers, I think we'd be hard-pressed to take it away from them now. They love it," Schaab added.
Embracing AI in rural healthcare
While financial return is important with new AI technologies, Teaff suggested focusing on the human element when making the case to staff: "We want to give you time back in your life."
"We did a lot of talking with the stakeholders and the providers before we went live," she said. "We made sure they knew that they had choices. This was not an all or nothing mentality."
At rural Patterson Health, doctors know that they can switch between the AI tool and traditional typing.
Parks and Schaab said that while their "Big Bang" rollout at CGH might seem intense, and while AI can be "scary" to both rural doctors and patients, the benefits are considerable. They recommend that their peers:
- Be intentional about how AI tools are rolled out
- Always emphasize that a human is in the loop
- Educate staff and patients about data security and tool accuracy
"AI is coming," said Teaff. "This is something we need to step into and start playing around with and getting accustomed to."
Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.
