Automatic antibiotic stewardship

One health system added a best practice alert to the electronic health record and notified clinicians when specific criteria suggesting a viral illness were met.


The antimicrobial stewardship team at Saint Luke's Health System used to manually perform prospective audit and feedback for more than 200 patients on antibiotics each day. While team members still provide active prospective audit and feedback, in 2017, they decided to let the computer take over some of that work for them.

A best practice alert was added to the electronic health record and notified clinicians when specific criteria suggesting a viral illness were met. “Therefore, it minimized the need to manually intervene on these patients, and it does it 24/7 without the need for active oversight,” said Nick Bennett, PharmD, program manager of the health system's antimicrobial stewardship program.

How it works

The alert prompts clinicians to consider antibiotic de-escalation when three criteria are met: A virus is isolated on a respiratory polymerase chain reaction (PCR) panel, a procalcitonin value is less than 0.25 ng/mL, and one or more systemic antibiotics are ordered. It says, “Antimicrobial stewardship alert: Your patient has a positive viral PCR + negative procalcitonin + one or more antibiotics ordered. These results suggest viral infection—please reassess necessity of antibiotics as indicated.”

The alert does not force clinicians to take action. “If they either were just picking up the patients and needed to decide, or maybe they weren't making the antibiotic decisions on other teams, it would just re-fire to the next physician who opened the chart,” said infectious diseases physician Sarah Boyd, MD, medical director of the health system's antimicrobial stewardship program. “It really was just a nudge to look at the values and think about the patient.”


The alert significantly reduced antibiotic use, according to a study published online in October 2019 by Clinical Infectious Diseases. In 226 patients who had the alert fired on their record, the average length of antibiotic therapy was 5.8 days, compared to 8.0 days in 161 patients who had no alert. “Interestingly enough, this matches some of the previous data that had been out there,” Dr. Bennett said. “Where the data was lacking in the past was if you really wanted to make it effective, this kind of intervention required a lot of manual effort.”

The alert was also associated with a 19% increase in antibiotics being stopped within the first 24 hours (37.8% vs. 18.6%) and fewer patients being discharged on antibiotics (20.0% vs. 47.8%). “Antibiotic exposure upon discharge was not reflected in our days of therapy, so our intervention had a subsequent effect on postdischarge plan of care,” Dr. Bennett said.

To make sure clinicians were not stopping antibiotics prematurely, the researchers looked at re-initiation of antibiotics after they had been stopped and found no increase in the alert group. “That reassured us that it wasn't firing at the wrong time or pushing people to make what they felt was not the right decision,” said Dr. Boyd. There was also no difference between groups in rates of Clostridioides difficile infection.


One challenge was developing the alert itself, which required support from the hospital's IT team, whose services are in great demand, Dr. Bennett said. There was also initial apprehension among clinicians about using a single diagnostic tool to guide treatment decisions, which the stewardship team tried to dispel with the combined alert. “The initial apprehension with the isolated values or results in and of themselves were probably one of the biggest hurdles. . . . By putting this information together for them, we helped to paint a more clear picture regarding how to use the information to make decisions,” he said.

Lessons learned

When the intervention began, clinicians were already familiar with using procalcitonin values, which made it easier to get them on board, Dr. Boyd noted. She added that the concepts of antibiotic stewardship were already in place and that the stewardship team enlisted a critical care/pulmonary physician to be a project champion. “Having that input from multiple specialties was helpful,” she said.

Next steps

The alert is still in use. Anecdotally, Dr. Bennett said it appears to be even more effective now. “We measured its effect right after implementation, and . . . if you evaluate something right after it's implemented, there is this lag phase where people buy into it,” he said. “So there's likely a greater benefit now than when we measured the effect of it right after implementation.”

Words of wisdom

“There's so many stewardship initiatives you can implement or pursue, and when the EMR can do some of the work for you, it only enhances your ability to maximize your scope of patient care intervention,” Dr. Bennett said.