Steadily increasing antibiotic use, the lack of new antimicrobials coming down the pipeline and the growth of multidrug-resistant organisms have together forced some real creativity among those who head up hospitals' efforts to control drug costs and combat overuse.

In 2007, spurred by the significant uptick in MDROs such as methicillin-resistant Staphylococcus aureus, or MRSA, and vancomycin-resistant enterococci, or VRE, the Infectious Disease Society of America and the Society for Healthcare Epidemiology of America jointly released a set of guidelines for antimicrobial stewardship, an activity they defined as including “appropriate selection, dosing, route and duration of antimicrobial therapy.”

And as those who lead antimicrobial stewardship programs search for ways to bolster appropriate use, they are turning to strategies that leverage and build on their existing health information technology systems.

“IT plays a tremendous role in antimicrobial stewardship because it makes it easy to identify which patients have been given which antibiotics, and it can also show us what microorganism might have been involved,” says Dr. Harold Standiford, medical director for infection control and antimicrobial effectiveness at 702-bed University of Maryland Medical Center, Baltimore.

In a 2006 article in the Journal of the American Medical Informatics Association, co-authored by Standiford, researchers found that the University of Maryland Medical Center's use of a Web-based computerized clinical decision support system with specialized alerts resulted in $84,000 in cost savings over three months in the intervention group. In addition, they also found that the hospital's antimicrobial management team was able to intervene on nearly twice as many patients' therapies in less time because of increased efficiency.

“The system allowed us to monitor more patients on a daily basis, it decreased the time needed for clinical pharmacists to determine who to monitor and it saved the hospital a considerable amount of money,” Standiford says.

In spite of those successes, the hospital ceased its antimicrobial stewardship program two years ago in favor of an approach that ramped up the number of infectious-disease consultations. The hospital's rationale for the switch, according to Standiford, was that increased participation of infectious-disease physicians would lead to more careful selection of antibiotics.

Unfortunately, costs have risen considerably since the discontinuation of the antimicrobial stewardship program, a result that suggests “many more antibiotics are being used than necessary,” he says. In response, the hospital is in the process of purchasing a new monitoring tool and is transitioning back to the old system

Ever since NorthShore University HealthSystem, Evanston, Ill., went live with an enterprise-wide electronic health record in 2003, clinicians and IT staff have been looking for ways to exploit the new data source to ease workflow and improve patient care. For Dr. Ari Robicsek, an epidemiologist and associate chief medical information officer for the two-hospital system, the target was antibiotic overuse.

Robicsek and his informatics colleagues have used the data collected in the health system's EHR to spearhead several programs aimed at improving antimicrobial stewardship, including a newly launched project focused on postoperative fever.

Some temperature elevation following surgery is expected, Robicsek says. But until recently, physicians had little to no idea just how much temperature elevation was normal and how much indicated a fever and possible infection.

To clear up that confusion and reduce excessive use of antibiotics, Robicsek and his team created the Wunderlich Project, named for a 19th-century German physician who conducted pioneering research on human body temperatures. They used NorthShore's EHR to identify every patient—15,000 in total—who had undergone any one of about 20 surgical procedures. Then they analyzed half a million temperature readings taken from those patients.

“What we found is that different types of procedures had very different postoperative temperature ranges,” Robicsek says.

Normal postoperative temperature ranges also varied widely based on patient characteristics such as age and gender. In other words, one temperature could indicate a fever in a particular patient who had undergone a particular procedure, such as knee replacement, but that same temperature reading could fall well within the normal range for another patient.

“There is no one magic number,” Robicsek says.

Together with the IT department, Robicsek designed a website that asks physicians what type of surgery was performed, the number of postoperative days and a few other procedure-specific and demographic questions. The system then produces a graph with a range of normal temperatures for that patient.

The model's accuracy continually improves, he adds, because new patient data is added each day.

In its first version, the system is a stand-alone website, meaning physicians must click away from the EHR to get to it. But the next version will be built into the record and eventually, Robicsek says, the data will automatically populate and the system will provide a personalized range of temperatures, reducing guesswork and facilitating appropriate antibiotic use.

Robicsek also is involved in several other projects that use EHR data to guide antibiotic administration. A project titled “What's Going Around?” provides point-of-care data to physicians about illnesses that are currently prevalent in their region. The hope, Robicsek says, is that when physicians become aware that particular viral illnesses are “going around” in their area, they will be less inclined to prescribe unnecessary antibiotics. And the results so far have shown that to be true, he says.

“We found that when physicians were highly aware that the flu was going around, they did the right thing and did not prescribe antibiotics,” Robicsek says.

To ensure patients at eight-hospital Memorial Hermann Healthcare System in Houston were receiving the right antibiotics, Thanh Dao, the health system's lead clinical informatics project manager, relied on a cross-matching approach she had crafted while working at her previous position in infection control at 719-bed St. Luke's Episcopal Hospital, also in Houston.

The system uses data-mining technology to match lab results with physicians' medication orders. For instance, if a patient is on the antibiotic Cipro, the system cross-matches culture data from the hospital lab, which the clinical pharmacist can use to see whether the patient's infection is susceptible to that drug.

Dao joined Memorial Hermann in 2009 and began to create an adapted version of the program soon after. Working alongside an information systems programmer, a clinical pharmacist and an infectious disease specialist, she compiled a list of priorities.

Over the course of the next year, Dao led a five-phase rollout and validation of the system. It works like this: Each afternoon, Dao and her team pull pharmacy orders and cultures and they conduct the cross-matching. They give the results to the clinical pharmacy managers in the form of once-daily reports. If there is a discrepancy—for instance, if a patient is on a particular antibiotic and the lab culture shows a resistance to that drug—the clinical pharmacist calls the physician and recommends a change.

“Normally when you speak with data, the physician says yes,” says Dao, who recently led a session at the annual meeting of the Healthcare Information and Management Systems Society titled “Clinical Informatics and Antimicrobial Stewardship: You Can Do It.” Still, she points out, the physician always makes the final decision regarding any treatment change.

The reports also alert clinical pharmacists to opportunities for de-escalation. For example, if a patient is on vancomycin—a powerful antibiotic associated with increased antimicrobial resistance—but the lab culture shows susceptibility to ampicillin, the pharmacist will receive a notification to contact the physician and recommend that the patient's therapy be changed.

“The logic seems very simple, but the program is very complex,” Dao says. “It's not a one-size-fits-all approach. There are different lab processes at each facility, and it takes time and creativity to get the queries right. I thought it would be a piece of cake to implement this system after my work at St. Luke's, but it was really different.”

Since the launch of the system in April 2010, 22% of antibiotics have been changed during patients' hospital stays. The time it takes to change a patient's antibiotic has been lowered from one or two days before to a same-day change after the system was put in place. And the health system has seen “significant” cost savings, according to Dao, although exact figures are not yet available.

“Before, we used a manual exhaustive review process,” Dao says. “Now the computer does it for us and the time to review each patient is only four minutes. That's a valuable difference.”