A pharmacist-led, workplace-based educational intervention informed by behavioral science to influence antimicrobial prescribing in Japan Authors: Koji Masuda, Harumi Gomi, Junko Tanuma, Tamae Hiramatsu, Takamitsu Sato, Saho Takaya, Yasuyuki Kato, Tetsuya Matsumoto Published: 2026 ABSTRACT AbstractPurposeChanging antimicrobial prescribing behavior is a key challenge in antimicrobial stewardship (AMS). In postgraduate clinical education, knowledge-based teaching alone is often insufficient to achieve sustained behavior change, highlighting the need for workplace-based educational designs informed by behavioral science.MethodsWe conducted a retrospective before–after study to evaluate a pharmacist-led educational intervention designed to influence prescribing behavior of piperacillin/tazobactam (TZP) in Japan. The educational component, consisting of a pharmacist-developed video and post-test, was completed once by physicians prior to implementation, whereas a structured prescription rationale requirement was applied at each TZP prescribing event. Prescribing behavior was assessed using days of therapy (DOT) and antibiotic use density (AUD) over 6 months before and after the intervention. Educational components were treated as process indicators of engagement rather than primary outcomes.ResultsAll full-time physicians completed the instructional video and the post-test. Following the intervention, TZP use decreased by 22.3% in DOT, with consistent reductions observed in season-matched comparisons. Similar trends were observed for AUD.ConclusionWorkflow-embedded reflective prompts may be useful as a workplace-based educational strategy for influencing antimicrobial prescribing behavior.Keywords: Antimicrobial stewardship, Pharmacists, Piperacillin/tazobactam drug combination Introduction Improving prescribing behavior remains a key challenge in antimicrobial stewardship (AMS), particularly in postgraduate clinical education, where knowledge-based teaching alone is often insufficient to achieve sustained behavioral change [1]. Pharmacists have increasingly contributed to physician education in AMS by supporting prescribing decisions in routine clinical practice [2], and from a behavioral science perspective, prescribing behavior can be influenced not only through education but also through system-level design that promotes reflection and reinforces motivation [3]. Piperacillin/tazobactam (TZP) is a broad-spectrum antimicrobial frequently used for empirical therapy and a common target for stewardship interventions. We developed a pharmacist-led intervention combining brief educational components with a system-based approach to prompt reflective prescribing at the point of care. Therefore, this study aimed to evaluate a pharmacist-led intervention for physicians implemented in a real-world clinical setting, using antimicrobial consumption metrics as indicators of prescribing behavior and focusing on behavior change corresponding to Level 3 of Kirkpatrick’s model [4]. Methods 1. Ethics statement This project was conducted as an educational quality-improvement activity using aggregated, non-patient-level antimicrobial use data. Following consultation with the research support office, ethics committee review was not required. 2. Study design and setting This study employed a retrospective before–after design to evaluate a pharmacist-led multicomponent intervention to influence antimicrobial prescribing behavior at a university hospital in Japan. 3. Conceptual framework The intervention was designed based on the COM-B model, which conceptualizes behavior as the interaction of capability, opportunity, and motivation. In this framework, capability refers to the knowledge and skills required to perform a behavior, opportunity reflects environmental or system-level factors that enable or constrain behavior, and motivation encompasses reflective and automatic processes that drive decision-making. In the present intervention, physician capability was supported through a pharmacist-developed instructional video and a brief post-test addressing key principles of antimicrobial use. Opportunity was created by integrating a structured prescription rationale requirement into the prescribing workflow, thereby prompting consideration at the point of care. Motivation was reinforced by requiring explicit justification for TZP prescriptions, promoting accountability and deliberate decision-making in routine clinical practice. 4. Educational intervention A pharmacist-led educational intervention was implemented between January 18 and January 31, 2025, to influence TZP prescribing. The educational component consisted of a short, web-based instructional video (approximately 7 minutes in duration), followed by a mandatory post-test comprising three true/false questions. The video was accessible via a web-based platform, allowing physicians to complete the training regardless of location. The content focused on key principles of TZP use, including indications, local antimicrobial susceptibility patterns, and potential alternative agents based on relevant guidelines (Supplement 1). The post-test addressed clinically relevant and commonly misunderstood aspects of TZP use. Completion of the intervention was confirmed through submission of the post-test responses. A total of 109 physicians prescribed TZP during the post-intervention period. The post-test was not designed to assess knowledge acquisition but to reinforce key concepts immediately after the educational component. 5. Implementation of the structured prescription rationale requirement Following completion of the educational component, a structured prescription rationale requirement was integrated into the TZP prescribing workflow. Physicians were required to document key clinical information—including diagnosis, site of infection, suspected or confirmed pathogens, and the rationale for TZP use—primarily using structured fields, with optional free-text entries when needed. This documentation was mandatory for order completion and was intended to promote self-reflection and deliberate consideration at the time of prescribing. The structured prescription rationale requirement was embedded within the electronic medical record in general wards and implemented via a separate documentation system in the intensive care unit due to differences in prescribing platforms, ensuring consistent application across clinical settings (Table 1). No real-time audit-and-feedback process was implemented, although prescription rationale was retrospectively monitored. Table 1.Structured Prescription Rationale Requirement for Piperacillin/Tazobactam UseSectionItemsConfirmation of culture sampling (multiple choices allowed)- Two sets of blood cultures obtained (or planned)- Urine test and urine culture obtained (or planned)- Sputum culture obtained (or planned)- Other specimen obtained (please specify)Confirmation of indication (multiple choices allowed)- Neutropenia- On immunosuppressive therapy (including corticosteroids)- Other immunocompromised conditions- Recent antibiotic exposure / recent hospitalization- Nosocomial infection- Healthcare-associated infection- Suspected resistant organism infection- Suspected Pseudomonas aeruginosa infection (including anaerobic coverage)- ICU admission or shock state- Severe infection (other than ICU/shock)- Complicated infection- Other (please specify)Diagnosis (multiple choices allowed)- Pneumonia- Intra-abdominal infection (including cholecystitis, cholangitis)- Urinary tract infection- Meningitis- Skin and soft tissue infection- Under diagnosis/undiagnosed- Other (please specify)Physicians were required to complete all sections before the order could be submitted; the system did not allow optional or incomplete entries.ICU: Intensive care unit. Structured Prescription Rationale Requirement for Piperacillin/Tazobactam Use Physicians were required to complete all sections before the order could be submitted; the system did not allow optional or incomplete entries. ICU: Intensive care unit. 6. Evaluation and data collection The intervention was implemented as a multicomponent strategy, and the relative contribution of individual components was not evaluated. The primary outcome was change in antimicrobial prescribing behavior, assessed using days of therapy (DOT) and antibiotic use density (AUD; defined daily doses per 100 patient-days) [5,6], as indicators corresponding to Level 3 (behavioral change) of Kirkpatrick’s model [4]. Measures related to the intervention components, including post-test responses and completion of mandatory rationale documentation, were treated as process indicators to confirm implementation and engagement with the intervention rather than as primary educational outcomes. Outcomes were assessed by comparing antimicrobial consumption during the 6 months before (August 2024–January 2025) and 6 months after (February–July 2025) the intervention, with the intervention period (January 18–31, 2025) included in the pre-intervention phase. Consumption data for other anti-pseudomonal agents, including meropenem and cefepime, were also collected to assess potential compensatory changes. Antimicrobial consumption data were obtained from the nationwide Japan Surveillance for Infection Prevention and Healthcare Epidemiology (J-SIPHE) database. Results 1. Implementation and engagement with the intervention A total of 289 full-time physicians participated in the intervention and completed both the instructional video and post-test, resulting in a completion rate of 100% (289/289). Of these physicians, 109 prescribed TZP during the post-intervention period. 2. Changes in antimicrobial prescribing behavior Antimicrobial prescribing behavior changed following implementation of the pharmacist-led intervention. Following the intervention, TZP use decreased substantially. The DOT declined from 6.49 to 5.04 per 100 patient-days, representing a relative reduction of 22.3%, while the AUD decreased from 4.82 to 3.74 per 100 patient-days, corresponding to a 22.4% reduction. Similar season-matched reductions were observed when compared with the corresponding period in the previous year (February–July 2024) (Fig. 1). No compensatory increase in other anti-pseudomonal agents was observed. Trends in piperacillin/tazobactam use before and after a pharmacist-led educational intervention. Antimicrobial consumption of piperacillin/tazobactam is shown as antibiotic use density (AUD; defined daily doses per 100 patient-days) and days of therapy (DOT; per 100 patient-days) across three consecutive 6-month periods. The vertical dashed line indicates the timing of the pharmacist-led educational intervention, which combined a brief instructional video, a post-test, and a structured prescription rationale requirement embedded in the prescribing workflow. Both AUD and DOT demonstrated a decrease following the intervention, with similar downward trends observed in season-matched comparisons. Discussion This study demonstrated a change in antimicrobial prescribing behavior, as reflected by reduced TZP use, corresponding to Level 3 of Kirkpatrick’s model rather than evaluating prescribing appropriateness. Accordingly, this study focused on changes in antimicrobial use, as measured by DOT and AUD, rather than on clinical outcomes. Level 4 outcomes, such as patient mortality or organizational impact, were not evaluated, and patient-level effectiveness cannot be inferred. Further studies with longer follow-up and additional outcome measures are needed to assess broader clinical and system-level effects. From an educational and behavioral design perspective, this intervention aligns with principles of adult learning theory, emphasizing relevance to professional practice and problem-centered learning [7]. Rather than relying on intensive didactic education, the intervention combined brief educational input with a structured prescription rationale requirement embedded at the point of prescribing, which may have prompted more deliberate consideration. The post-test was not intended to assess knowledge acquisition but to reinforce key concepts immediately after the educational component. In contrast, consideration at each prescribing decision was primarily prompted by the structured prescription rationale requirement embedded in the prescribing workflow. This combination of components is consistent with the Behavior Change Wheel framework, which highlights the importance of opportunity and motivation, in addition to capability, in influencing professional behavior [3]. However, as this was a bundled multicomponent intervention, the relative contribution of each component could not be determined. The COM-B model provides a useful framework for interpreting the observed changes in prescribing behavior. Specifically, the educational video and post-test were designed to support capability, the structured prescription rationale requirement functioned as an opportunity within the prescribing workflow, and the requirement for explicit justification was intended to reinforce motivation. Together, these elements may explain how a low-intensity multicomponent intervention was associated with changes in prescribing behavior. In particular, the requirement for explicit justification may have promoted reflective motivation by encouraging physicians to engage in active clinical reasoning at the point of prescribing. Over time, such repeated prompts may also contribute to changes in automatic motivation by influencing habitual prescribing patterns [7]. In the Japanese clinical context, prescribing decisions may be influenced by hierarchical structures and time constraints, and pharmacist-led interventions using system-level design rather than direct interpersonal instruction may represent a practical approach to AMS. Regarding the structured prescription rationale requirement, the documentation process was designed to reduce documentation burden through a simple checkbox-based format aligned with routine prescribing workflows. Although physician burden and alert fatigue were not formally assessed, no apparent usability or workflow-related issues were observed during implementation. Several limitations should be considered. This was a single-center, before–after study, which may limit the generalizability of the findings. The 6-month observation period limits assessment of the long-term sustainability of the observed changes, and learning outcomes were not evaluated independently of prescribing behavior. In addition, clinical outcomes were not formally assessed, and the impact of reduced TZP use on patient safety therefore remains uncertain. No apparent increase in adverse clinical events was observed during the study period. In conclusion, a pharmacist-led, workplace-based educational intervention incorporating a workflow-embedded reflective prompt may be a useful strategy for influencing antimicrobial prescribing behavior. This approach may support reflective prescribing and behavior change in real-world clinical settings. Footnotes Data sharing statement Please contact the corresponding author for data availability. Acknowledgements None. Funding This work received no external funding. Conflicts of interest No potential conflict of interest relevant to this article was reported. Author contributions All authors contributed to the study design, data interpretation, and manuscript preparation, and approved the final manuscript. AI disclosure A large language model (ChatGPT, OpenAI) was used solely for language refinement. Supplementary materials Supplementary files are available from https://doi.org/10.3946/kjme.2026.181 Post-Test Items Used in the Educational Component of the Pharmacist-Led Intervention. References Associated Data Supplementary Materials Post-Test Items Used in the Educational Component of the Pharmacist-Led Intervention.