Photo: Epic
A new independent study in the Journal of Critical Care Medicine found that Epic’s sepsis early warning system led to faster antibiotic administration and better patient outcomes without an increase in harmful clinical interventions, like antibiotic or IV fluid overdose.
The model, used by hospitals nationwide, detects the first risk factors of infection in patients, allowing clinicians to enact early treatment measures and save lives.
Earlier this summer, a study in JAMA Internal Medicine using retrospective data found that prediction algorithms included as part of Epic’s electronic health record may poorly predict sepsis.
But the new independent study, conducted by MetroHealth and Case Western Reserve University, shows the use of EHR sepsis warning systems flags clinicians before symptoms become visible and potentially deadly to patients. Sepsis contributes to one in three deaths in U.S. hospitals today.
As the study shows, this technology reduces the burden on clinical staff, especially those without extra resources, while giving patients better care.
Healthcare IT News interviewed Emily Barey, MSN, director of nursing at Epic, to discuss the findings of this study
Q. How does your sepsis early warning system work? How does it help achieve better patient outcomes?
A. Sepsis is a hard problem to solve, and it’s one the industry has been working for many years to address. As a nurse at the bedside, sepsis can sneak up on you because it shows up in patients in different ways.
A change in a patient’s vital signs or symptoms could be due to several different causes. Clinicians are often taking care of multiple patients and might not see a pattern developing right away. The electronic health record continuously looks for changes. It can alert the clinician in real time when it spots something significant and being embedded in the primary workspace allows the clinician to take the next step quickly.
When Epic designed our sepsis model, we looked to improve on common sepsis scoring methods, such as SIRS. We studied hundreds of potential variables to see what provided the most predictive value.
In the end, we landed on 50 variables in the model. These include things like lab test results, vital signs, medication orders, comorbidities, past ED visits and hospitalizations, and a lot more. What we found when we did our initial analysis was that our model helped identify about 10% more septic patients in the timing window, when compared with the SIRS model.
By finding 10% more septic patients compared to SIRS, that’s 10% more patients who can be treated faster and have a better chance at survival. MetroHealth found they were able to deliver antibiotics to their septic patients in the emergency department almost an hour faster and those patients had a shorter length of stay and lived longer.
Q. Physicians and nurses get plenty of warnings during the course of their day. Your EHR sepsis warning system flags clinicians before symptoms become visible and potentially deadly to patients. How do you make your alert stand out? How do you make sure not to overload clinicians with too many alerts?
A. Alert fatigue is a very real thing. Our own analysis found our model not only was more sensitive to identifying positive septic patients, but also alerted for 27% fewer patients when compared with common sepsis scoring methods like SIRS.
UCHealth in Colorado found something similar. Their implementation of the Epic model had 19% fewer alerts in comparison to the Modified Early Warning Score. This is good news for clinicians. That’s a lot less noise in sepsis alerts.
The MetroHealth study highlights the importance of validating the model first, and then trialing the model, live, in real time, in clinical practice. This two-step approach meant they could focus on all four steps of the clinical alerting process: the model to aide in monitoring and alerting the right person at the right time, then a workflow to map out a response to the alert that fit the needs of the clinicians, and finally process measures and clinical outcomes to assess if the approach helped improve care.
Effective alerting isn’t always about pop-ups or interrupting a clinician. In the study, MetroHealth improved outcomes and stated that the alerts were relatively unobtrusive along with a positive experience with the tool as a multidisciplinary pharmacy-physician team.
Q. Epic says this technology reduces the burden on clinical staff, especially those without extra resources, while giving patients better care. What does the new study say that supports this statement?
A. The study was run by MetroHealth, a safety net health system with a level 1 trauma center designation, as part of a systemwide initiative to improve sepsis care. Patients were randomized to either standard sepsis care or care augmented by the display of a sepsis early warning in the EHR.
Any improvement in sepsis care is a good improvement but being able to do it without increasing resources is another contribution this study will make to healthcare organizations considering using AI to improve their care quality.
I think it is important to underscore that this was people, process and technology working in concert. The Epic early warning system was only the first step. MetroHealth’s team-based workflow was effective enough that the trial ended early, and the flags were turned on for all patients presenting to the ED.
This is consistent with other outcomes customers have reported to us, which include a 20.6% reduction in sepsis mortality, 21% faster antibiotic administration, and increased compliance with federal quality measures.
All these results are in keeping with our commitment to provide embedded technology tools that: 1) Fit within the clinical workflow, and 2) Help save lives by providing early warning of a patient who might need intervention.
Twitter: @SiwickiHealthIT
Email the writer: [email protected]
Healthcare IT News is a HIMSS Media publication.
Source: Read Full Article