Triage nurses in the emergency department at Johns Hopkins Hospital in Baltimore face a hard task. Just like in EDs across the country, they must decide in just a few minutes how critical a patient's condition is and assign them a score that will determine how quickly they are treated.
The nurses use the Emergency Severity Index to help make their decision. The ESI, a tool used widely in EDs across the U.S., is a way for caregivers to identify patients' conditions by assigning them to one of five groups, or levels. Level 1 indicates the patient needs immediate attention and is experiencing something along the lines of cardiac arrest, while Level 5 means their needs aren't urgent-a rash, for example.
That's important, obviously, because it gets the patient the right treatment more quickly. The problem is the ESI isn't always right and relies heavily on nurses' subjectivity, said Scott Levin, associate professor of emergency medicine at the Johns Hopkins University School of Medicine.
Research on the ESI shows that about 70% of patients are lumped into the medium category—Level 3—even though there can be wide variance in the severity of their symptoms and ultimate diagnoses. "The major challenge of the ESI is that it's completely subjective," Levin said. "When something is completely subjective, there can be untoward variability."
In an attempt to make the triage process more objective, Levin and his colleagues developed an electronic tool last year that is now used by triage nurses at Johns Hopkins Hospital.
The tool uses an algorithm based on data from roughly 200,000 patients treated at the six hospitals in the Johns Hopkins system to predict a patient's severity of illness. It takes into account how patients with the same symptoms were treated and what their likelihood was for dying, being admitted to the intensive-care unit or needing an emergency procedure. The tool then assigns the patient a level score using the ESI.
Nurses have been using the tool since last December and find it's helpful to guide their clinical decisionmaking.
But it took some time for staff to warm up to the tool, admits Sophia Henry, a triage nurse in the Johns Hopkins ED. Henry said she and other nurses were initially worried that the tool would take away their autonomy or that they would be "replaced by a computer."
"We were very resistant at first because for us, being trained as a triage nurse is an honor. It shows clinical excellence and that you understand clinical decisionmaking," she said.
Levin said he spent months with the triage nurses to ensure they understood that the tool was not to replace their clinical judgment but merely to support them in their work. Levin said he tells nurses they should disagree with the tool when they think it is appropriate. After all, the tool can't interact with patients the way a nurse can.
Levin, however, said he's confident in the tool's results because it's targeted at Johns Hopkins' unique patient population. He said healthcare tools that rely on algorithms aren't usually widely adopted because providers don't trust the data. But caregivers trust the accuracy of data used to develop this tool's algorithm.
The e-triage tool "is more meaningful to the people who are using it," Levin said. "Every ED is so different—the patient populations they treat, the resources they bear and the care processes they use."
The tool has been shown to work effectively. A recent study led by Levin published in the Annals of Emergency Medicine found that the tool identified 14,000 patients, or 10%, triaged to ESI Level 3 who should have been categorized as a Level 1 or 2. The tool also increased the number of patients assigned lower priority levels like Levels 4 or 5.
Identifying patients with less serious conditions sooner decreases the time they have to spend in the ED, Levin said. EDs often "fast track" patients who aren't ill enough to require prolonged care.
"If we put these patients in line with the very sick, they would never get out," he said. "The hope is to not have them wait and get out quickly."