Discussion

The primary goals of this resident-led quality improvement project were to reduce the unnecessary use of piperacillin-tazobactam in patients with CA-IAIs and to ultimately raise awareness of antibiotic resistance. Our efforts led to a reduction in piperacillin-tazobactam use, with rates decreasing from 54.1, to 36.9, to 22.3 DOT/1,000 patient-days, and from 0.3, to 0.2, to 0.1 DOT/LOS per patient, from the preintervention phase, to post PDSA1, to post PDSA2. The mean LOS decreased after the first intervention, from 7.8 days to 5.4 days, and remained stable after the second intervention at 7.0 days. This project represents an early implementation experience demonstrating the feasibility of reducing empiric antipseudomonal antibiotic use through resident-led antimicrobial stewardship interventions. Although the study period was relatively short, the observed reductions suggest that targeted educational and stewardship strategies can influence prescribing behavior in the inpatient setting. These findings align with those of other studies ^{Reference Sartelli, Catena and Abu-Zidan1–Reference Lodise, Izmailyan, Olesky and Lawrence7} and support the use of more narrow-spectrum antibiotics for specific patient populations with a low risk of complications from infection. Analyzing cost implications in future initiatives would be a valuable addition to the literature.

One driver of broad-spectrum antibiotic use in low-risk CA-IAIs is concern for Pseudomonas aeruginosa. However, this risk appears to be overestimated. In a multicenter study, P aeruginosa was identified in fewer than 5% of culture-positive low-risk cases. ^{Reference Sartelli, Catena and Ansaloni9} Consistent with this, guideline recommendations do not support empiric antipseudomonal coverage in the absence of sepsis, hemodynamic instability, or other high-risk features, as it has not been shown to improve outcomes. ^{Reference Solomkin, Mazuski and Bradley10}

These findings suggest that prescribing practices are influenced less by evidence and more by a culture of defensive or “just-in-case” medicine, in which clinicians favor broader coverage despite guideline recommendations. Prior work has similarly shown that clinicians often do not perceive broad-spectrum therapy as problematic. ^{Reference Lodise, Izmailyan, Olesky and Lawrence7} In our study, clinicians frequently believed that broader coverage was in the patient’s best interest, reinforcing this pattern.

Our findings are consistent with prior work demonstrating that clinicians may rely on “historical medicine” or anecdotal experience rather than current guidelines or local susceptibility data. ^{Reference Pinkerton, Bongu, James, Lowder and Durkin11} Together, these data support the need for multifaceted interventions-including education, guideline dissemination, and clinical decision support-to promote more evidence-based antibiotic prescribing.

Initial antibiotic selection often occurs in the emergency department prior to transfer of care to the internal medicine team. In teaching hospitals, this decision is frequently made by trainees, and subsequent de-escalation may be limited by both inexperience and hierarchical dynamics, particularly when therapy was initiated by a more senior clinician. As a result, broad-spectrum antibiotics are often continued despite limited supporting evidence. These findings highlight the importance of targeted education for both emergency department and resident physicians, as well as multidisciplinary collaboration with antimicrobial stewardship pharmacists and infectious disease specialists to support evidence-based de-escalation.

When deciding which interventions to pursue, our team made an impact effort grid to discuss the various options. After seeing a partial improvement noted after PDSA1 and further discussion with our pharmacy team stakeholders, the team decided to move forward with the inclusion of a nighttime ASP team to review evening prescribing of all piperacillin-tazobactam orders, with specific targeted interventions for de-escalation in patients meeting criteria for CA-IAI. This nighttime ASP team significantly helped with the review of prescribing practices, which previously were only conducted during the daytime hours. Interventions considered but not ultimately pursued included adding a best practice advisory (BPA) into the electronic health record and modifying the electronic order set for piperacillin-tazobactam to guide prescribing practices. The impact of these interventions was perceived to be high, but the effort was also high because changing orders in the electronic health record system requires multicommittee approval and can take a prolonged time for approval. As this manuscript was written initially, this project was selected to be presented at the Mayo Clinic Quality Committee meeting, and further discussion is being had with leadership about the addition of a BPA to create lasting changes. We also considered creating a smart phrase for physicians to use in their documentation when ordering piperacillin-tazobactam and treating CA-IAI, with reminders on the latest guidelines. These interventions are still being considered for future use and will be reevaluated after the data are monitored by our antimicrobial stewardship team.

Our team learned several valuable lessons from this quality improvement project. First, recognizing potential barriers to intervention is important. Addressing common practices in hospital culture is difficult, and changing them is not easily done without open, extensive discussions about the reasoning behind clinical decision-making. One common challenge was resistance from physicians due to their comfort with using piperacillin-tazobactam for patients. They often wanted to wait for an extended period before de-escalating, despite the patient meeting criteria for lower-risk infection. This barrier was overcome with support from infectious disease colleagues and our antimicrobial stewardship team, who assisted with regular multidisciplinary meetings, educational sessions, regular updates communicated via email, and flyers posted in commonly occupied workspaces. Another barrier was time constraints for residents and faculty in scheduling sessions that allowed maximum participation and educational benefit. Our team benefited from maintaining monthly team meetings, allocating a faculty quality improvement sponsor, and designating one resident as team leader for seamless communication.

Although the primary focus of this project was antibiotic utilization, broader clinical outcomes such as mortality, infection recurrence, adequacy of source control, and readmissions were not systematically captured as part of this quality improvement initiative. Importantly, no concerning safety signals were observed during the intervention period, and the hospital LOS remained stable. Future studies incorporating additional clinical outcomes would further clarify the impact of stewardship-driven reductions in broad-spectrum antibiotic use.

The main limitations of this study include the fact it was conducted in a single academic medical center in a relatively short time frame of 10 months with modest sample size of patients. Because the intervention periods were relatively short, it is difficult to determine whether the observed changes represents sustained practice change or a temporary shift associated with increased awareness during the study period. In addition, the Hawthorne effect may have been introduced, given that resident physicians were aware that their orders were being monitored. Future research dedicated to replicating these findings in different settings and with longer follow-up periods would be of great value in assessing the sustainability of our findings.