Health system uses Epic EHR, communications tech to reduce sepsis mortality rate by 20%
University of Utah Health is the Mountain West’s only academic health system, combining care, research and teaching. The health system frequently is ranked among U.S. News & World Report’s Best Hospitals.
In the U.S., sepsis is the most common cause of death among critically ill patients in noncoronary intensive care units – and the most expensive condition treated at U.S. hospitals, costing $24 billion annually. Consequently, identifying and responding to sepsis is a priority for virtually every organization.
Sepsis is a challenge for many hospitals, and it is the most expensive inpatient condition to treat in part because decompensation can happen quickly and it can be difficult to detect, said Dr. Devin Horton, a hospitalist and assistant professor in the division of internal medicine at the University of Utah Health.
“The primary problem we had was that we had the information that showed decompensation – that is, worsening vital signs – but sometimes those became stuck in the black hole of our EHR, from Epic,” he said. “The information was not getting to a person who could act on it in real time.”
Sometimes staff would see Epic importing the entered vital signs into the daily note, and there is objective data the patient is not well, but there is no mention of abnormalities from the resident, he added.
"We erred on the side of fewer alerts at first, then slowly increased."
Dr. Devin Horton, University of Utah Health
“We thought, ‘If only we could clearly push this information to someone who knew what to do with it,’“ he explained. “How can we get the technology to say this is a red flag for the patient, and then create a forced function for an immediate response?”
So the University of Utah launched an initiative, led by Dr. Horton and Dr. Kencee Graves, to improve its sepsis response in 2016. Staff leveraged the Epic EHR and the Spok Care Connect clinical communication platform from vendor Spok to hardwire sepsis alerts, automating steps within the recognition and communication processes in the sepsis response workflow.
Staff used a best practice alert within Epic to trigger a MEWS score alert, which then automatically is sent by Spok to the rapid response team for better patient outcomes. MEWS stands for Modified Early Warning System.
“We implemented a calculator into Epic that calculates the vital signs using the Modified Early Warning Score,” Horton said. “What this does is accumulate all of the vital sign abnormalities and then aggregate them into one single score to simplify the information. It boils down all of the vital signs into a single digit that represents the severity of illness of that patient, without even talking to a human.”
When a patient’s MEWS score exceeds the threshold the institution has set for response to illness, Epic then automatically sends a page through Spok. That detection by the EHR activates a chain of events with forcing functions to nudge a provider to evaluate the patient. Epic and Spok automatically send an alert for elevated MEWS scores to the mobile device of the charge nurse (tier one), or to a rapid response team (tier two).
“When the clinicians arrive at the bedside and open the patient chart in the EHR, they’re prompted with an automatic order set populated with the gold standard of treatment for sepsis – check lactate, then administer antibiotics and IV fluids,” Horton explained. “We also added a real-time display of all MEWS scores of all patients, all of the time, within Epic.”
There is a wide variety of electronic health records systems vendors on the market, including Allscripts, athenahealth, Cerner, eClinicalWorks, Greenway Health, Meditech and NextGen Healthcare. On the clinical communications technology front, vendors include Avaya, Halo, HipLink Software, Mobile Heartbeat, PatientSafe Solutions, PerfectServe, Telmediq and Vocera.
MEETING THE CHALLENGE
University of Utah Health originally was not sure it had the existing technology to pull off this sepsis effort. But staff knew they first wanted to look at the technology the whole hospital used, in this case the EHR and the communication system.
“Our former CMIO, Dr. Michael Strong, was a champion for the effort, and Dave Roach, one of our IT experts, was key to implementation on the technology side,” Horton said. “It was an exploring adventure of what the systems could do together. I did not take no for an answer.”
Since implementing its sepsis program, University of Utah Health has reduced the sepsis mortality rate for patients with elevated MEWS scores (7-11) by 20 percent, and is saving about 40 patients a year. It also has reduced length of stay and total direct costs for patients with sepsis by at least 10 percent, and is on track to approach 20 percent.
“The technology undoubtedly helped by building automation into the workflow and proactively delivering meaningful information to the right people who can act on it,” Horton remarked. “But it was also a big effort to change the culture.”
Now, Horton said he walks through the hospital and hears clinicians talking about MEWS, and they have no idea who he is or know that this program did not exist five years ago. How the health system treats patients with worsening vital signs has totally changed, Horton added.
ADVICE FOR OTHERS
“My advice would be that this has to be a big stakeholder buy-in situation,” Horton said. “You can have the coolest technology in the world, but if you’re making half the hospital upset, it’s not a good solution. You have to ensure that the technology matches what the providers want and bring in clinician opinion throughout the process.”
IT and clinical staff must hold many meetings, negotiate thresholds, talk about downstream effects and be flexible, he added. It is a lot of human interaction and dealing with emotions and opinions before one can introduce a new way to use technology and change the workflow, he said.
“Then after buy-in, it requires a lot of practice sessions in a testing environment, followed by pilots in individual units,” he advised. “We wanted to find and address any issues within a small population in one unit before we rolled out hospital-wide. We also erred on the side of fewer alerts at first, then slowly increased. This helped us improve patient safety and outcomes while avoiding alert fatigue.”