Clinicians in the emergency department are often short on time, so researchers at Johns Hopkins are arming them with an electronic tool aimed at improving efficiency at triage.
Hopkins built the Electronic Triage System, a machine learning program that uses data from the electronic health record to grade a patient’s severity of illness and in the first year of using the tool, found a significant change in how patients were sorted at triage.
Under the traditional Emergency Severity Index (ESI), most patients are graded at a three, the exact middle of the scale, by triage nurses, said Scott Levin, Ph.D., associate professor of emergency medicine at Hopkins and the study’s lead author.
However, after using the tool, level three designations decreased by 17% and low-severity level four and five designations increased by 56%, the researchers found. Levin and his team presented the findings at the American College of Emergency Physicians Scientific Assembly this week.
The ESI “doesn’t do great in actually separating sick people from people who are lower acuity,” Levin said. The tool, meanwhile, combs the EHR for data before making a recommendation in just a few seconds to triage nurses.
RELATED: ER triage pilot reduced wait times by more than 2 hours, study finds
To date, nurses typically agree with the tool’s assessment, though they’re free to code a patient differently if they see a reason to, Levin said. Getting the nurse team on board, he said, required highlighting the downsides of using ESI.
“We were able to change the mindset of, particularly, some of the more senior nurse leadership in triage, and from there they ... don’t do ESI anymore,” Levin said.
The tool has also shorted the window from arrival to admission for emergency department patients by about an hour, as more patients are being appropriately coded with lower-severity care needs, the study found. This translates to about 14,000 saved per year.
The Electronic Triage system has been deployed at the Johns Hopkins Hospital and Howard County General Hospital, and Levin said they’re looking to expand it to additional facilities in the Hopkins system.
RELATED: Predictive analytics fuel Johns Hopkins ‘command center’
The Hopkins team is also working to make the tool available to other providers, Levin said. It is built outside of the EHR itself and instead operates in the cloud, so it could be applied at hospitals that use a different system from Hopkins.
He said Hopkins has already teamed up with other health systems to provide the tool and has garnered interest from others in addition.
Hopkins has teamed with the National Science Foundation for funding on this expansion work, Levin said. Few other emergency departments have the resources and technology background to build a tool like the Electronic Triage System themselves, he said.
The ED is a great place to launch predictive analytics tools, Levin said, as they’re a source for a wealth of diverse data these tools can learn from. Taking that data and using it “at the point of care ... is kind of what I think will be next, not just for triage, but for a lot of clinical decision making,” Levin said.
“It’s just an example of what the future may be,” he said.