We partner with a secure submission system to handle manuscript submissions.
Please note:
Click Proceed to submission system to continue to our partner's website.
Background: Artificial intelligence (AI) chatbots are increasingly used by patients and caregivers seeking quick, conversational answers to medical questions. This trend raises concerns about the clinical accuracy and appropriateness of responses, which can influence illness beliefs, treatment expectations, and care-seeking. These concerns are relevant for conditions like asymptomatic bacteriuria (ASB), which is often mistaken for urinary tract infection (UTI), leading to unnecessary antibiotic use. We examined how AI chatbots respond to simulated caregiver questions about ASB/UTIs. Methods. We created two clinical scenarios of hypothetical medical advice questions involving a positive urinalysis without specific UTI symptoms (i.e., ASB) that asked explicitly whether the parent had a UTI (Table 1). Scenario A described only a positive test, whereas scenario B included a fever and malodorous urine. For each scenario, we varied the parents’ gender (mother vs. father) and age (68 vs. 84 years). The eight questions were entered into three publicly available AI chatbots (ChatGPT v5, Gemini v3, and Copilot v5) on December 15, 2025, using private sessions with no memory enabled. Two team members independently coded responses, with discrepancies resolved by two additional reviewers. Results. Most responses endorsed the suggestion of UTI (22/24, 91.7%; Table 2). Responses frequently acknowledged alternative explanations for symptoms (17/24, 70.9%) and over half mentioned ASB (16/24, 66.7%). Antibiotics were mentioned in 8 responses (33.3%), with 6 of those noting potential harms. Most responses recommended seeking immediate care (19/24, 79.2%). Tone varied: 9 (37.5%) stressed urgency (e.g., “this is not something to wait on—he needs urgent medical evaluation.”), 5 (20.8%) were reassuring (e.g., “I hear your concern”), and 10 (41.7%) neutral. Only 7 responses (29.2%) provided citations. Responses varied between scenarios. For instance, antibiotic harms were given for scenario A but not scenario B. Within scenarios (Table 3), responses varied by chatbot (e.g., mentioning ASB [0/4; Gemini] vs. [2/4; ChatGPT] vs. [2/4; Copilot]), parent’s gender (e.g., antibiotics suggested [1/6; father] vs. [3/6; mother]); and age (e.g., alternative diagnoses suggested [4/6; 68 years] vs. [1/6; 84 years]). Conclusions: In response to simulated caregiver medical advice requests about ASB, AI chatbots often endorsed the suggestion of UTI and inconsistently referenced ASB, with variability by platform and patient attributes. These patterns may reinforce misperceptions of ASB and promote antibiotic misuse in older adults. Monitoring and improvement of chatbot outputs, alongside patient-facing guidance on how to use and interpret responses, are needed to support guideline-aligned care.
Background: Penicillin allergies are common, reported in roughly 10% of the U.S. population, but fewer than 1% have a true IgE-mediated allergy. Penicillin allergy labels are associated with increased use of broad-spectrum antibiotics, higher healthcare costs, and the development of antimicrobial resistance. Antibiotic stewardship efforts have focused on identifying and de-labeling individuals without a true penicillin allergy. To date, most penicillin allergy de-labeling initiatives have been implemented in acute-care settings, where time constraints and clinical acuity may limit feasibility. Long-term care facilities represent a unique opportunity for penicillin allergy de-labeling, as residents often have prolonged stays and are less acutely ill. Previously, we found that 14% of residents admitted to the Veterans Affairs (VA) Community Living Center (CLC) in Vancouver, Washington had a documented allergy to a penicillin-class antibiotic. Here, we describe design and implementation of a penicillin allergy de-labeling program at a VA CLC and highlight opportunities for multidisciplinary collaboration. Methods: This antibiotic stewardship initiative was developed through collaboration between the VA Portland antimicrobial stewardship team and CLC providers, including physicians, nursing staff, and pharmacy. A standardized protocol was created to identify residents eligible for penicillin allergy evaluation and to outline steps for a direct oral challenge (Figure 1). At admission, CLC providers are instructed to identify residents with electronic health record (EHR) documentation of a penicillin class allergy and complete a structured allergy screening questionnaire. Residents with antibiotic intolerance alone (e.g., nausea and vomiting) receive education, and the penicillin allergy is removed from the EHR. Residents who require and agree to further evaluation are referred via electronic consult to the infectious diseases team, who verifies eligibility to undergo a direct oral challenge. CLC pharmacists review resident medications for any contraindications to performing a challenge, including receipt of medications that would reduce the histamine response. Eligible residents undergo a direct oral amoxicillin challenge with a single 500-mg dose and are monitored by nursing staff for 60 minutes. Patients without evidence of reaction receive education, and the penicillin allergy is removed from the EHR. Planned Evaluation: Planned outcome measures include completion of allergy screening, de-labeling based on chart review alone, eligibility for and completion of oral challenge, challenge outcome, and impact on subsequent antibiotic prescribing. Conclusion:This initiative describes a structured, reproducible framework for implementing penicillin allergy de-labeling in a VA post-acute care setting. The program has the potential to improve antibiotic selection and strengthen antimicrobial stewardship in skilled nursing facility (SNF) enviornments.
Background: The 2025 American Thoracic Society Community-acquired Pneumonia (CAP) guidelines conditionally recommend that patients hospitalized with non-severe CAP who reach clinical stability be treated with <5 days (3 day minimum) of antibiotics. To determine whether hospitals were implementing or planning to implement this recommendation, we surveyed 68 hospitals participating in the Michigan Hospital Medicine Safety Consortium (HMS). Methods: Hospitals participating in HMS completed an electronic survey in October and November of 2025. Abstractors were asked to answer (with stewardship team input) whether their hospital recommended (or planned to recommend) 3-day antibiotic treatment for any patients with CAP. If yes, they were asked a) which patient subgroups, b) whether their hospital guideline recommended 3 days, and c) what stewardship interventions they had implemented to increase 3-day durations. All hospitals were asked about barriers. Results: All hospitals (68/68) responded to the survey; 21% (14) reported they currently recommend a 3-day duration for at least some patients with CAP; another 18% (12) do not but are planning to. Hospital characteristics by response category are shown in Figure 1. Of hospitals already recommending or planning to recommend a 3-day duration, 31% (8) recommend it (or plan to) in all patients with CAP. Among the 18 hospitals recommending/planning to recommend a 3-day duration for only a subset of patients, commonly stated criteria were: clinical stability (55%), non-severe CAP (33%), or uncomplicated CAP (33%). Most hospitals (79%) recommending a 3-day duration included the 3-day recommendation in their CAP guideline and 71% reported at least one stewardship intervention in place to increase use of 3-day durations (10, education; 3, audit and feedback; 2, pharmacist bundle; 1, orderset; 1, discharge-specific intervention). Commonly reported barriers or reasons for not recommending a 3-day duration included: focus on 5-day duration (31%), resource/time limitation (14%), EHR transition (12%), and inability to recommend a 3-day duration unless the entire healthcare system did (10%). Notably, 7% expressed confusion about IDSA endorsement of the ATS guidelines (IDSA endorsed the <5 day duration but not the viral pneumonia recommendation). Conclusion: Just over a third (38%) of Michigan hospitals currently recommend or plan to recommend a 3-day antibiotic duration for at least some hospitalized patients with CAP. For the remaining 62% of hospitals, continued focus on 5-day duration, standard barriers around stewardship infrastructure, and concerns about existing evidence, safety, and guidance limit enthusiasm for a 3-day duration.
Background: Guideline-concordant, first-line antibiotic prescribing is a core antimicrobial stewardship strategy to reduce unnecessary broad-spectrum use, antibiotic resistance, and adverse events. Community-acquired pneumonia (CAP) is a common indication for outpatient antibiotics, yet first-line prescribing patterns and equity remain understudied. This analysis assessed associations between first-line prescribing and patient, clinician, and community characteristics in urgent care settings. Methods: Electronic health records from 28 urgent care clinics within an integrated academic healthcare system (January 2023–April 2025) were analyzed. Adults (18+ years) with pneumonia diagnoses and at least one prescribed antibiotic were included. ‘Community-acquired pneumonia’ refers to pneumonia diagnosed in urgent care, though some cases may not meet formal CAP criteria. Episodes were classified by first-line use (CAP without comorbidities: amoxicillin or doxycycline; CAP with comorbidities: either amoxicillin/clavulanate plus doxycycline or a macrolide, or respiratory fluoroquinolone monotherapy). Cluster-adjusted logistic regression assessed differences in first-line prescribing by patient, clinician, and community factors. Results: Among 9,670 episodes (4,966 no comorbidities), first-line prescribing occurred in 90.8% of CAP with no comorbidities and 54.4% of CAP with comorbidities. Mean age was 48.4 years (SD 17.4) for patients without comorbidities and 60.7 years (SD 17.5) for patients with comorbidities; 23% and 48% of episodes occurred among adults ≥65 years. The most common non–first-line antibiotics were amoxicillin-clavulanate and clindamycin, for CAP with and without comorbidities respectively. For CAP without comorbidities, fully adjusted models showed higher odds of first-line prescribing for episodes managed by Advanced Practice Registered Nurses (APRNs) compared to physicians and lower odds among patients with higher neighborhood disadvantage (ADI 26–49 vs 1–25 percentile) as seen in Figure 1. The Area Deprivation Index (ADI) quantifies neighborhood socioeconomic disadvantage (1-100 is least to most disadvantaged). For CAP with comorbidities, Black and Hispanic patients (vs. White) and those treated by APRNs (vs. physicians) had higher odds of receiving first-line antibiotics, while patients in neighborhoods with higher ADI scores had lower odds (vs. 1–25 percentile; Figure 1). Across both groups, no differences were observed by age, gender, or payer. Conclusion: First-line prescribing for outpatient CAP was high among patients without comorbidities but substantially lower among those with comorbidities, with disparities by clinician type, race/ethnicity (comorbidities only), and neighborhood deprivation. These findings reveal systemic inequities in antibiotic stewardship and reinforce the need for equity-focused strategies to support guideline-concordant prescribing. Additional work is warranted to clarify prescribing variation and evaluate relevant modifiers.
Background: Since 2014, the Centers for Disease Control and Prevention’s Core Elements of Hospital Antibiotic Stewardship Programs have evolved to incorporate multidisciplinary practices that optimize therapy and minimize antibiotic-associated harm. Infection preventionists (IPs) play a key role in stewardship efforts, especially education and reporting of event data; this study examined IP perceptions of inpatient stewardship programs. Methods: Members and non-members of the Association for Professionals in Infection Prevention (APIC) were invited to complete the quinquennial electronic survey between June and July 2025 through numerous emails, QR codes at the annual APIC conference, and social media campaigns. The survey assessed nine antimicrobial stewardship program (ASP) characteristics by practice setting (rural, suburban, urban), including structure, leadership, and educational resources. Descriptive statistics were used to summarize the data. Result: Responses were received from 1898 IPs in acute care hospitals. Most (n=1834, 96.7%) reported the presence of an ASP (Table 1). Only 5% (n=91) were unaware of program leadership, less often in rural settings (n=13, 2.6%) compared with 6.7% (n=54) in urban settings. Rural settings (n=238, 47.0%) were more likely to combine the antibiotic stewardship committee setting with other committees, such as an IP committee, compared to urban settings (n=194, 24.1%). Clinical pharmacists (n=797, 47.9%) and infectious diseases physicians (n=643, 38.6%) were most frequently identified as leaders, while IPs represented 1.3% of ASP leaders (n=21). Urban (n=570, 66.8%) and suburban (n=289, 56.6%) settings had greater access to infectious diseases physicians compared to rural settings (n=149, 28%). Over 80% (82.6%, n=1473) reported the use of up-to-date recommendations for infection management, and education was reported by 60.2% (n=1073), though 26.6% (n=475) were unsure if antibiotic prescribing education was specifically offered – findings consistent across settings. Multidisciplinary ASP rounding occurred in 66.6% (n=1188) of hospitals regardless of setting, with IPs attending in 27.1% (n=322) of institutions. IPs (n=119, 36.4%) were more likely to attend rounds in rural settings. Conclusion: ASPs are well-established in acute care settings, typically led by pharmacists and physicians. However, IPs are less frequently engaged in ASP activities, with limited participation in rounding and awareness of educational initiatives. An exception exists in rural settings, where IPs assume broader stewardship responsibilities. Expanding IP involvement and enhancing educational outreach across all care environments represent key opportunities to strengthen stewardship and improve patient outcomes.
Background: Community-acquired pneumonia (CAP) is among the top 10 reasons people seek emergency care and the most common indication for antibacterial use overall. Up to 40% of CAP is caused by viruses, now more easily detectable by rapid diagnostics. Although viral CAP is common, evidence on how to best to treat it is lacking—especially with regard to empiric antibacterials. The 2025 American Thoracic Society guidelines recommend viral CAP be treated with empiric antibiotics. In response, the Infectious Diseases Society of America removed its endorsement of the guideline. Here, we used a vignette survey to assess clinical practices and reasoning regarding early treatment of viral CAP. Methods: Between 10/2025 and 12/2025, we distributed an online clinical vignette survey to attending hospitalists and emergency medicine (EM) physicians from 3 academic medical centers. The vignette described a patient presenting to the emergency department with signs and symptoms of pneumonia (lobar infiltrate, fever, hypoxia) and a positive test for respiratory syncytial virus (RSV). In the survey, clinicians were asked which treatments they would prescribe, their rationale for doing so, and whether clinicians would be willing to enroll a similar patient in a clinical trial examining the need for empiric antibacterial therapy. Results: Respondents included 37 hospitalists and 39 EM physicians (25% response rate, summary in Figure 1). Treatments most commonly selected were supplemental oxygen (93%), inhaled bronchodilators (78%), and empiric antibiotics (63%). Most hospitalists and EM physicians (57% and 69%) selected empiric antibacterial therapy and estimated 72% of their colleagues would also do so. Primary reasons for antibacterial therapy (Figure 2) included covering for a potential bacterial co-infection, evidence suggesting bacterial infection, and local practices. Conversely, physicians who did not select antibacterial therapy favored supportive therapy and observation, interpreted available data as not supporting bacterial infection, and cited local practices. Most (97%) physicians supported enrolling similar patients in a clinical trial examining the need for empiric antibacterial therapy, citing an evidence gap supporting either decision. Conclusions: This vignette survey of EM and hospital medicine physicians showed wide variation in hypothetical use of empiric antibacterials for viral CAP. Factors related to treatment with empiric antibacterial therapy included: 1) covering bacterial infection “just in case” vs. watchful waiting, 2) heterogeneous interpretation of incomplete information as supporting vs. not supporting bacterial infection, and 3) local practice patterns. Most physicians (97%) supported conducting a clinical trial to provide evidence to guide empiric antibacterial therapy in viral CAP.
Background: Urinary tract infections (UTIs) are frequently managed at outpatient settings, including primary care and urgent care (UC) clinics and emergency departments (EDs), which vary in antibiotic prescribing practices. We aimed to evaluate initial antibiotic choice and clinician responsiveness to cultures in subsequent days after the encounter by clinical setting. Methods: This retrospective cohort study included encounters with an order for a urinalysis with reflex urine culture from June 1, 2022 to June 30, 2024 across 29 walk-in and retail clinics (UCs), 5 EDs, and 33 primary care clinics at Vanderbilt University Medical Center. All encounters included antibiotics received with an order indication for cystitis or pyelonephritis (UTI). Encounters in which a positive urine culture was available in the preceding 6 months were excluded, as it was assumed that this data would guide antibiotic selection and therefore would not be empiric. Additionally, encounters from June to December 2022 were removed because culture data in the preceding six months could not be assessed. For each time period, antibiotic use was recorded as administered or not administered during Day 0-1 (empiric therapy) and Days 2-5 (culture-directed therapy). Patients could receive <1 antibiotic per time period per encounter. Differences in antibiotic prescribing were calculated using Chi-square and Fisher’s exact tests as indicated. Results: A total of 20,889 encounters were included: 5,750 (27.5%) from ED, 12,912 (61.8%) from UC, and 2,227 (10.7%) from primary care. Empiric antibiotic choice differed significantly for all antibiotics across clinical settings. Higher rates of prescribing cephalexin and third generation oral cephalosporins occurred in the ED, and fluoroquinolones in primary care, while lower use of nitrofurantoin was noted in the ED (Figure 1). A total of 256/5,750 (4.5%) patients at the ED, 1,258/12,912 (9.7%) at UCs, and 203/2,227 (9.1%) at primary care received at least one subsequent antibiotic (Figure 2). Compared to the ED, a higher percentage of both UC and primary care patients received culture-responsive antibiotics. Antibiotic choice patterns were similar for empiric and culture-responsive antibiotics. Conclusion: Higher percentages of culture-responsive antibiotic prescriptions were noted for primary care and UC compared to ED. At the time of this study, UCs had dedicated nurses following up results, a system more closely mirroring procedures at primary care clinics, likely resulting in more follow up antibiotic changes. Significant differences were seen across all settings for both empiric and culture-responsive antibiotic choices and represent future targets for antimicrobial stewardship.
Background: A substantial proportion of outpatient antibiotic prescriptions exceed guideline-recommended durations. Although most antibiotic use is in the outpatient setting, the majority of antimicrobial stewardship efforts focus on inpatient care. Prior studies suggest that modifying electronic order defaults is a low-burden strategy to nudge prescribers toward shorter, guideline-concordant courses. We evaluated the impact of updating default durations and adding selectable “speed buttons” to electronic orders for commonly prescribed outpatient antibiotics. Methods: We conducted a retrospective pre–post analysis comparing outpatient antibiotic prescriptions during a pre-intervention period (July 1–October 31, 2024) and post-intervention period (July 1–October 31, 2025). Only prescriptions with durations of 1–14 days were included. In June 2025, an electronic order redesign was implemented for the 13 most frequently prescribed outpatient antibiotics, incorporating updated default durations and short-course speed buttons based on guideline-supported treatment ranges for common outpatient infections (Table 1). The primary outcome was the change in proportion of prescriptions aligning with the new default duration. The secondary outcome was the change in mean prescribed duration for each antibiotic. Results: There were 80,669 pre-intervention and 86,895 post-intervention prescriptions. All antibiotics except amoxicillin demonstrated an increase in prescriptions aligning with the new default duration (Figure 1A), and all increases were statistically significant except for cefadroxil. The largest increases were observed for cefpodoxime (18.2%) and nitrofurantoin (16.7%). For cefadroxil, the largest absolute shift was a reduction in 10-day courses, decreasing by 10.9% (p<0.001). Amoxicillin defaults remained 10 days in both periods, with new 5-, 7-, and 10-day speed buttons added. Five-day amoxicillin courses increased by 5.5% (p<0.001), representing the largest shift in amoxicillin prescribing. Statistically significant differences in mean prescribed duration were observed for most antibiotics, with small absolute reductions noted (Figure 1B). The largest absolute reductions in mean duration were observed for cefadroxil (−0.82 days) and cefpodoxime (−0.64 days). No significant change was observed for cephalexin, cefdinir, or metronidazole. Cephalexin and metronidazole lacked pre-intervention default durations. Conclusions: The addition of short-course speed buttons and updated default durations impacted the length of outpatient antibiotic prescriptions. The intervention increased alignment with guideline-consistent durations and resulted in consistent reductions in mean prescribed duration across most agents. These findings suggest that simple, low-burden changes to electronic antibiotic order design can effectively influence antibiotic prescribing patterns.
Background: Many Infectious Diseases (ID) fellowship programs have created antimicrobial stewardship (AS) scholarly tracks to guide burgeoning stewards. However, few programs have reported details of their AS curricula or career outcomes after track completion. We describe the major components of our AS track and the current AS-related faculty roles of prior participants. Methods: In 2015, our ID fellowship program, based at an academic medical center (AMC), formally created an AS track for second-year fellows (typically one fellow per year). The goal is to prepare fellows to become effective AS team members and potentially hold program leadership roles. We annually survey prior track participants regarding their current institution and roles. All current fellows are surveyed annually to evaluate AS-related rotations and conference series using a five-point Likert scale (1=Poor to 5=Excellent). Results: The track emphasizes three major components (Education, Mentorship, and AS Experiences) based on the SHEA White Paper “Guidance for the Knowledge and Skills Required for Antimicrobial Stewardship Leaders” (Cosgrove SE, ICHE 2014) (Figure 1). Early stewardship content education is provided through an annual conference series (“Summer Series”) and an experiential combined AS/Infection Control rotation for first-year fellows; both have demonstrated improved evaluation scores over the past five years (Figure 2). Since 2015, eight fellows have completed the second-year AS track; the majority currently practice at AMCs. Seven (88%; 7/8) currently report dedicated FTE support for AS activities and five (63%; 5/8) hold AS leadership positions as either Medical Director or Associate Medical Director (Table). Conclusion: During the first decade of our AS scholarly track, we developed a structured curriculum emphasizing acquisition of stewardship expertise, mentorship, and diverse experiential training. The majority of AS track fellow graduates continue in stewardship roles as faculty, including leadership positions, suggesting our longitudinal track may support early-career development in AS.
Background: Dalbavancin is increasingly used for management of serious gram-positive infections, even beyond those for which it has formal FDA approval. Barriers to dalbavancin prescription commonly faced in the private sector, such as high drug acquisition costs, insurance prior authorizations, and challenges coordinating infusions, are more easily managed within the VA system. Since dalbavancin was initially approved by the FDA in 2014, VA providers have prescribed approximately 11,500 doses nationally. We surveyed VA providers to understand their perceptions and experiences in using dalbavancin. Methods: We conducted a cross-sectional, 39 question survey via RedCap, which was distributed to approximately 400 VA infectious disease providers across 141 medical centers. The survey remained open from September 10 – November 10, 2025. Results: We received 82 responses from 63 (77%) infectious disease providers (physicians and advanced practice providers) and 18 (22%) pharmacists representing 50 unique VA facilities, spanning all 18 VA geographic regions. At nearly all facilities (48/50, 96%), dalbavancin was available and approval by an infectious disease specialist was required for use. While the median estimated dalbavancin use among VA providers was reported to be 1-2 prescriptions/month, there was wide variability in reported prescribing rates of dalbavancin overall (Figure 1). The majority of VA providers had used dalbavancin for skin and soft tissue infections, bloodstream infections, and osteoarticular infections. However, a minority reported use for endovascular infections. (Figure 2). Most providers perceived dalbavancin was efficacious against gram-positive organisms, excluding vancomycin resistant enterococci. (Figure 3). Half of providers reported drug cost as a barrier to prescribing dalbavancin (41/82, 50%), though few providers experienced objections or resistance from hospital leadership for dalbavancin prescription (14/80, 17%). Most providers (64/81, 78%) reported prescribing dalbavancin as a 2 dose regimen (doses 1 week apart). A minority (29/77, 38%) of participants felt it is important to obtain safety labs for patients receiving dalbavancin. Providers reported that the most important factors in choosing dalbavancin included ease of administration for the patient and concerns about non-adherence to other standard care therapies. Conclusion: The widespread administration of dalbavancin across the VA underscores its attractiveness as a treatment option within this system. It is often used beyond FDA-approved indications with wide variability in perceptions of effectiveness, depending on the pathogen. The lack of inclusion in the national VA formulary and the absence of formal stewardship guidelines highlight the need for future efforts. Notably, our survey was conducted following a recent clinical trial showing that dalbavancin is not inferior to other therapies for treating gram-positive infections. Developing stewardship guidelines and investigating experiences with dalbavancin outside the VA system is planned.
