At Penn Medicine, Leaders Leverage Predictive Analytics to Improve Antibiotic Use
Healthcare Innovation-Leaders at Penn Medicine initiated a project to leverage health IT to improve antibiotic prescribing, with the aim of delivering decision support within clinicians’ workflow, ensuring the right patient gets the right antibiotic at the right time.
More than half of patients in U.S. hospitals receive at least one antibiotic during their hospital stay; however, one-third of antibiotic prescriptions in hospitals involve potential prescribing problems, such as prescribing an antibiotic when it is not needed or giving an antibiotic for too long, according to the Centers for Disease Control and Prevention (CDC).
And, in hospitals, misuse of antibiotics can lead to the development of antibiotic resistance, which adversely affects morbidity, mortality, length of stay, and cost, the CDC reports.
A study, recently published in the Joint Commission Journal on Quality and Patient Safety, found that antimicrobial-resistant organisms account for more than two million infections and 23,000 deaths annually in the United States. Studies from diverse settings estimate that between 25 percent and 50 percent of antibiotic use in hospitals is sub-optimal or unnecessary. Hospital antimicrobial stewardship programs can reduce inappropriate antimicrobial use, length of stay, Clostridium difficile infection, rates of resistant infections and cost, the researchers concluded.
To combat the threat of antimicrobial resistance, The Joint Commission and the Centers for Medicare & Medicaid Services (CMS) have initiated or proposed requirements for hospitals to have antimicrobial stewardship programs, but implementation remains challenging, according to the above study, led by Shashi Kapadia, M.D., instructor of medicine, division of infectious diseases at Weill Cornell Medicine in New York City. As of 2014, only 39 percent of hospitals in the United States reported having a program that met all recommended elements of stewardship programs, and only 55 percent had any antimicrobial stewardship program infrastructure.
That study also examined top antimicrobial stewardship programs in U.S. hospitals and found that innovative programs are integrating IT systems to enable real-time interventions to optimize antimicrobial therapy and patient management.
At Penn Medicine, based in Philadelphia, clinical leaders have developed a robust IT system for stewardship to facilitate clinical decision support and identify opportunities for intervention. Penn Medicine, the University of Pennsylvania Health System, operates six hospitals in eastern Pennsylvania. The Hospital of the University of Pennsylvania is home to one of the oldest antibiotic stewardship programs in the country, as it was formed in 1992 by associate chief medical officer Neil Fishman, M.D., according to Keith Hamilton, M.D., associate healthcare epidemiologist and director of antimicrobial stewardship at the Hospital of the University of Pennsylvania.
Working with health technology company ILUM Health Solutions, the hospital’s antimicrobial stewardship team, led by Hamilton, initiated a project to leverage health IT to improve antimicrobial prescribing, with the aim of delivering decision support within clinicians’ workflow to help ensure the right patient gets the right antibiotic at the right time.
“If we were to choose an antibiotic that is too narrow for a given patient, too narrow spectrum, then we may not be treating their infection adequately, especially those with more severe infections, like sepsis. But, if we’re giving an antibiotic that is too broad spectrum, then we may be exposing the patient to unnecessary toxicity and side effects. Ultimately, our goal is to give an antibiotic that is effective but is not more than what a patient needs,” Hamilton says.
While the hospital had an established stewardship program infrastructure, Hamilton notes that there was room for improvement in antibiotic prescribing practices. “The existing solutions to antibiotic stewardship didn’t have an efficient approach to identifying the patients who may benefit from an improvement in their antibiotic treatments. Tracking antibiotic use on a healthcare system-level is also fairly challenging to do with current solutions,” he says.
ILUM Health Solutions, a Merck subsidiary, provides enterprise-wide disease management tools and services as well as assists with antimicrobial stewardship programs. Last fall, ILUM acquired Teqqa LLC, which provides precision analytics to help physicians assess what antibiotic to prescribe to patients. The Hospital at the University of Pennsylvania has been working with Teqqa since 2014, and Hamilton and his team saw an opportunity to leverage ILUM and Teqqa’s technologies to bring relevant data and lab results right to the point of care, within the clinicians’ workflow.
When the hospital first began working with Teqqa back in 2014, the antimicrobial stewardship team focused on creating software to track antibiotic use and resistance to help clinicians determine the resistance patterns in their given settings, Hamilton says. The team then focused on moving more toward a “precision medicine-type approach” using predictive models, and that software was implemented a year ago, he notes.
The platform integrates real-time electronic health record (EHR), lab and pharmacy data from across disparate systems, and can pull in vital signs, drug data, microbiology and other laboratory findings. And, ILUM has developed a series of alerts identifying patients that may need a change in antibiotics, Hamilton says. “These alerts can alert both us [antibiotic stewardship teams] as well as individual clinicians where a change may be warranted and we can make those interventions,” he says.
What’s more, Hamilton says, the technology platform provides a user-friendly, interactive platform that allows the antimicrobial stewardship team to track antibiotic use in a hospital, on an ongoing basis, to better identify overall areas where improvement may be needed. And, the platform allows the user to focus in on patient-level data to better identify educational interventions or changes in guidelines that may be warranted, he says.
“The platform allows us to create a real-time, continually updated table of susceptible and resistance data for our institution, which, in turn, informs the guidelines that we set up for antibiotic use on a health-system level,” he says, adding, “But where even that falls short is that a guideline indicates what is appropriate for at least a majority of the patient population, but every patient is different. There are some patients who may not fit into guidelines, and we know based on prior research which risk factors may identify a patient as not being one that actually fits into those guidelines and may need an antibiotic that is broader spectrum in order to effectively treat their infection.”
To this end, ILUM has developed a predictive model that clinicians can use in real-time to better identify what antibiotics are going to work for a given patient. “Those predictions also can alert our stewardship team before we even get susceptibility data for a given patient’s infection so we can get them the antibiotic they need much sooner,” he says.
The predictive model is based on risk factors associated with antibiotic resistance, and integrates patient data from the EHR. Risk factors include prior antibiotic exposure, infections with other resistant microorganisms, age and prolonged hospitalizations in the intensive care unit (ICU). Clinicians can then use the software to actually predict the chances of susceptibility of a particular antibiotic given to a patient based on those risk factors. The software automatically pulls those risks factors from the health record, and comes up with predictions on what percentage likelihood a given antibiotic would be effective.
One significant way prescriptions are managed across the health system is through prior approval, meaning prescribers need authorization to prescribe certain antibiotics. To improve this process, Hamilton and his team worked with ILUM to implement a platform, available via smartphone and through a web version, enabling clinicians to request approval for antibiotics in a more efficient way.
As a result of implementing the health IT tools, clinicians can easily access the information they need to make an antibiotic decision. “The platform pulls all the biology data, recent antibiotic exposure that the patient has had and it’s very easy to have that data in front of you in the palm of your hand,” Hamilton says. “Clinicians use it as an informational tool as they are making decisions. They can also run a predictive model on their patient, just by looking up their patient and pressing go. They can access our institutional treatment guidelines through the software program as well, and the clinicians use it to communicate with our stewardship team both for asking permission to use antibiotics as well as asking general questions of the stewardship team.”
The platform also provides an interactive, dynamic, Web-based antibiogram to replace the program’s traditional static antibiogram (antibiograms are overall profiles that reveal the antimicrobial susceptibility testing results when a specific microorganism is subjected to a battery of antimicrobial drugs). Since implementing the new interactive antibiogram, the website has had about 3,000 hits per month, compared to only 30 hits per month with the previous table antibiogram, Hamilton says. “That’s definitely a tribute to how well the software was designed and how much the clinicians find the software useful. Right now, they are going outside their workflow, because the software is not in our EHR, so even despite that, they are still seeking out the information,” he says.
For the antimicrobial stewardship teams, the software provides an assessment of the hospitals’ current antimicrobial usage, the effectiveness of these programs, and helps providers re-focus their program as needed. “It basically forms a command center where we can get alerts and then reach out to the prescribers to make suggestions on how to optimize those antibiotics. It also allows us to more easily respond to requests for approval and questions and then also allows the team to more easily track antibiotic use on a health system-level and on a unit level,” he says.
He continues, “That has allowed us to start to provide feedback to individual patient care units on their antibiotic use and develop action plans on those units, and to further optimize their antibiotic prescribing practices. It also has allowed each provider to become an antibiotic steward themselves, and it empowers unit-based staff to make sure that their patients are receiving the right antibiotics.”
Improving the efficiency of antimicrobial stewardship programs and leveraging IT to enable real-time interventions can have a notable impact on patient care. “Our primary goal is to improve patient outcomes; that’s really the driving force of antibiotic stewardship. But, as an added benefit, antimicrobial stewardship programs have consistently shown to decrease cost, both cost of antibiotics as well as hospital costs,” Hamilton says, adding, ‘If you are effectively treating patients for the infections that they have, their length of stay in the hospital is going to be shorter as well, so that leads to dramatic decreases in healthcare costs related to antimicrobial stewardship as well as decreases in adverse events, such as antibiotic resistant infections and C. difficle infections as well.”