Augusta Health builds its own sepsis warning system, cuts cases

Augusta Health has created a custom warning system that uses vital signs and other data from its electronic health records system to provide advance warnings of potential sepsis.

The result has been significant improvements in care. Virginia’s mortality rate for sepsis is 12.7 percent, but at Augusta Health, an independent community hospital with 255 beds and a medical staff of 280, the rate is less than 4.8 percent. Since April 2017, 282 lives have been saved from sepsis, the organization estimates.

The hospital recently won a Health Quality Innovator Award from HQI, which serves as the Medicare Quality Innovative Network that assists health organizations in Maryland, Virginia, West Virginia and the District of Columbia.

Augusta Health’s warning system looks at vital signs in the Meditech electronic health record and a data warehouse, as well as other types of data such as respiratory, heart rate, shock index, high pulse and other factors.

Cooper-Penny-CROP.jpg
Penny Cooper

If factors for a specific patient get above a certain threshold that totals a value of 3.0, an automated alert is sent to the assigned nurse, says Penny Cooper, a data scientist who helped build the warning system.

Augusta’s system works by automatically identifying common warning signs and sending alerts through the Vocera communication badge that clinicians and staff wear to talk with each other.

Particularly important warning signs include a body temperature over 38 degrees Celsius, a heart rate of about 90 beats per minute, a respiratory rate exceeding 20 and abnormal white blood count.

Sepsis is a difficult condition to treat, and it requires the eyes of everyone caring for patients to recognize the signs early, and it needs top support from hospital executives and clinical leadership, Cooper advises. But to put the program together isn’t that difficult, she adds.

“We did this with off-the-shelf products with no consulting firms, and we’re a small community hospital with a small data team,” she explains. “You can do this; it’s a simple algorithm that parses the data and looks to see who’s at risk. The bottom line is we collect all this data on patients, and we need to use it. We have so much data and we have the responsibility to use it and get the patient home.”

For reprint and licensing requests for this article, click here.