How Halifax Health is harnessing AI to boost clinical documentation
Update: HIMSS20 has been canceled due to the coronavirus. Read more here.
Halifax Health CIO Tom Stafford says its hospitals' clinical documentation improvement efforts used to be arduous for the CDI specialists themselves – and often aggravating for the physicians who had to spruce up their charts when issues were detected.
"Our documentation specialists used to look for a needle in a haystack," Stafford explained. "They would literally go to the floor and review every chart – room one to room 36."
And for busy acute care physicians – who often consider the encounter to be "case closed" once patients are discharged from the hospital – to be compelled to reopen the EHR to revisit a problematic chart that may be days or weeks old "really causes a lot of frustration," he said.
At HIMSS20, Stafford will show AI-powered documentation technologies from Nuance have helped the health system to better prioritize CDI for those patient encounters that have the greatest opportunity for improvement.
"Two years before we put in this solution that had the AI component to it, two different groups of people were doing things in a wasteful way," he said.
"First and foremost, the physicians were facing clinician burnout. There are too many clicks and they have to spend all this time in the record. And they have to be good tech writers. All of their care has to be documented contextually and accurately, or it's considered not done. And if it's not done that affects their Healthgrades score, and it can affect the institution they're working for because it affects case mix index."
And when it came to clinical documentation specialists, they would "look at our records after the fact and see if the physician needed to be more specific when they were doing their assessments and documenting the care for the patient," Stafford explained.
"And if they found something, they would create a document – paper, this was years ago – and give it to the physician: 'Here's your chart, these are the areas that need more specificity.'"
Understandably, those aren't welcome words for a harried and time-pressed ER doc.
"Think about a physician and the work they do," said Stafford. "And this is especially in the acute environment, rather than the ambulatory space where you're caring for someone for years and years. In the hospital, you're treating someone and then they get discharged to go back to their primary care physician. So when a physician does the documentation in the hospital and closes it out, they're kind of done with that case.
"If two weeks later a clinical documentation specialist comes to them and says, 'Hey, I need you to review this chart and look at it. I think these areas aren't specific enough. Can you please correct?' It's hard to do, because now they have to go back and open up the chart again: They care about the documentation, so they're not just going to make the edit that's suggested they're going to go in there and read the chart and make sure that the suggested edits are accurate. They have added work to do, and that adds a lot of stress."
"Now [CDI specialists] have a really good directional tool that focuses their day and makes them more efficient. The case load they work against is more meaningful than in the past, where there could have been a lot of busywork just reviewing the chart."
Tom Stafford, CIO, Halifax Health
At HIMSS20, Stafford will show how Halifax Health is using cloud-based artificial intelligence to provide documentation guidance to help its physicians and documentation gurus achieve that specificity more efficiently.
He'll explain how closed-loop workflows between clinicians and CDI teams are improving both clinical and financial outcomes at Halifax. For instance, the health system has improved case mix index and fine-tuned for more accurate reimbursements. It was able to lower the average number of severity queries by 63% and increase total case review rates by 20%.
Rather than an arduous, paper-based back and forth between physicians and documentation specialists – oftentimes on marginal charting improvements that may or may not translate into more valuable reimbursement, "today, we pull data out of our EHR and it flows into the engine," said Stafford.
"The engine has natural language processing and we get to set rules around the engine," he explained. "It ranks it as a complex chart that needs to be addressed, or a chart that is just fine."
That means that documentation specialists, rather than looking for a needle in a haystack, "can open up this triage screen and it tells them what they should be working on – the most complex cases that are in the hospital," said Stafford.
And physicians can avoid the obligation to review and improve EHR documentation for patients they may have treated days ago.
"We can even set an alert where, if someone's coming out of surgery, they can review the notes right away to make sure it's accurate," he said. "So now they have a really good directional tool that focuses their day and makes them more efficient. The case load they work against is more meaningful than in the past, where there could have been a lot of busywork just reviewing the chart."
Stafford admits that it "seems kind of simple: Why wouldn't you just focus on what you need to focus on? But the way they had access to the data before didn't allow it. This AI engine has changed that. We're able to do more, and more meaningful, work in the same amount of time than we ever were in the past."
Tom Stafford will offer much more detail during his HIMSS20 session, "Using AI to Connect and Improve Physician Documentation and CDI Performance." It's scheduled for Thursday, March 12, from 1-2 p.m. in room W304A.
Full coverage of HIMSS20
An inside look at the innovation, education, technology, networking and key events at the HIMSS20 global conference in Orlando.