Speaker
Shruti K. Gohil, MD, MPH
Assistant Professor, Infectious Diseases
Associate Medical Director, Epidemiology & Infection Prevention
University of California, Irvine School of Medicine
Keywords
Antibiotics; Antibiotic Resistance; Infection; Prompts; Clinical Decision-Making
Key Points
- Multidrug-resistant organisms are a major public health threat, with more than 2.8 million antibiotic-resistant infections occurring annually in the U.S. This leads to increased readmissions, healthcare costs, and thousands of attributable deaths. Overprescribing is a major contributor to antibiotic resistance.
- The culture of empiric antibiotic selection currently reflects a “start broad, narrow later” approach: an effort to avoid missing resistant infections at the cost of unnecessary broad spectrum antibiotics. The INSPIRE trials aim to encourage a new practice: “start narrow, broaden if needed,” in which physicians use data on resistance risk to spare broad spectrum antibiotic use. This delay in broadening is not high-risk for the vast majority of patients.
- The 4 most commonly treated infections in U.S. hospitals (pneumonia, urinary tract infections (UTI), abdominal infections, and skin/soft tissue infections) do not reflect the multi-drug resistant organisms that physicians most worry about, but comprise the bulk of what physicians prescribe extended spectrum antibiotics for.
- Risk factor research typically reports relative risk, AKA odds. This can exaggerate physician perception of risk. The INSPIRE trials sought to reframe risk perceptions by using absolute risk, based on a series of models for every extended-spectrum antibiotic and multidrug-resistant organism pair.
- The research team developed a computerized provider order entry (CPOE) smart prompt, activated when a physician in a non-ICU location selected extended-spectrum antibiotics to treat abdominal or skin/soft tissue infections. These real-time prompts provided the absolute risk of infection due to multidrug-resistant organisms, in a manner that was patient-specific, infection syndrome-specific, and pathogen-specific.
- The INSPIRE Abdominal and Skin/Soft Tissue Trials consisted of 2 cluster-randomized, 92-hospital trials. Arm 1 received routine care, and Arm 2 received the INSPIRE CPOE bundle intervention, in which physicians received the CPOE smart prompts recommending appropriate antibiotic choice.
- The INSPIRE trials demonstrated the effectiveness of real-time, individualized smart prompts. In both the abdominal infection trial and the skin and soft tissue infection trial, the group that received the CPOE prompts saw a 35% and 28% reduction in extended-spectrum days of therapy, respectively. There was no difference in safety outcomes for either trial.
- The intervention was rolled out as a bundle; this makes it difficult to know the impact of the individual components. One limitation of this intervention is that activating it requires CPOE capability and a live link to the EHR.
- The research team took some proactive steps, including actively extracting and cleaning data starting in the CPOE prompt development stage. That allowed them to identify issues early in the process, adjust accordingly, and ultimately speed the analysis.
- Precision medicine trials for antibiotic stewardship can work; physicians are willing to make different choices when they’re given the right kind of information. If you thoughtfully construct a prompt-based intervention and populate it with useful and applicable information, these interventions could help lead to new best practices for empiric treatment in the U.S.
Discussion Themes
The trial defined the threshold for acceptable risk – i.e., differentiated between high and low risk – at 10%. This was based on precedence set by the Infectious Diseases Society of America (IDSA) guidelines and vetted by the INSPIRE steering committee and national experts. Each patient had their own risk profile – comprised of a series of risk factors, pulled from the EHR – that the algorithm drew upon to place the patient on either side of the risk threshold.
Patient rates of ICU transfer and length of stay were equal between the intervention and control groups. This addresses a significant concern amongst physicians when it comes to reducing broad-spectrum antibiotic prescription, as many worry about incurring any unnecessary risk for their patients.
The prompts were intensely vetted. One important consideration making sure they appeared only for low-risk patients. This was an important step in getting local physicians on board with the prompts as a clinical decision support tool. Stewardship teams played an important role in this provider education.
Nurse practitioners, physician assistants, and residents were another key audience, as they often play a significant role in ordering antibiotics.