Time-lagged associations between nominal C. difficile exposure dose effect and colorectal cancer incidence, MM (2000–2021).

Association between nominal dose effect of CD exposure (0, 1, and >1) and colorectal cancer incidence by anatomic site (right colon, left colon, and rectum), MM (2000–2021), JHM (2016–2024).

Association between nominal dose effect of CD exposure (0, 1, and >1) and colorectal cancer incidence by sex, MM (2000–2021), JHM (2016–2024).

Baseline demographics of individuals according to CD exposure, MM (2000–2021), JHM (2016–2024).

BACKGROUND & AIMS:The low Fermentable Oligosaccharides, Disaccharides, Monosaccharides, And Polyols (FODMAP) diet (LFD) and rifaximin are effective in <50% individuals with irritable bowel syndrome (IBS), highlighting the need to identify predictors of treatment response. We therefore conducted a randomized controlled trial comparing LFD and rifaximin to identify microbial predictors of response. METHODS:Sixty-five adults with diarrhea-predominant IBS (IBS-D) were randomized to LFD or rifaximin for 5 weeks. Primary endpoints were changes in mean daily abdominal pain and bloating at week 5 vs baseline. Secondary endpoints included changes in IBS Symptom Severity Score and Bristol Stool Form Scale at week 5 vs baseline. Exploratory endpoints included responders defined as ≥30% reduction in abdominal pain or bloating. Stool samples collected at weeks 0, 2, 4, and 5 underwent 16S rRNA sequencing, and glucose breath testing (BT) was performed at weeks 0 and 5. RESULTS:Both LFD and rifaximin significantly improved abdominal pain (-0.29 with LFD vs -0.24 points/week with rifaximin); bloating (-0.29 vs -0.19 per week); and IBS Symptom Severity Score (-14.2 vs -13.3 per week) at week 5 (all P < .0001), with no significant change in Bristol Stool Form Scale. BT results were inconsistent predictors of response, with positive baseline hydrogen BT associated with lower odds of rifaximin response, and methane conversion at week 5 showed discordant associations with rifaximin response. In contrast, distinct baseline taxa were associated with treatment response. LFD responders had lower abundance of putative saccharolytic taxa (Butyricimonas, Bacteroides, Intestinibacter), whereas rifaximin responders were enriched in taxa with putative short-chain fatty acid-producing and bile acid-modifying potential (Ruminococcus, Coprococcus, Odoribacter). Nonresponders exhibited enrichment of putative proteolytic taxa (Bilophila, Alistipes, Prevotella). CONCLUSIONS:LFD and rifaximin are equally effective for IBS-D, with distinct microbial predictors of response. However, these findings require validation before informing personalized treatment approaches. CLINICALTRIALS:gov, Number: NCT03219528.

Association between any CD infection and colorectal cancer incidence by sex, MM (2000–2021), JHM (2016–2024).

Association between any CD exposure and colorectal cancer incidence by site (right colon, left colon, and rectum), MM (2000–2021), JHM (2016–2022).

Sporadic colorectal cancer remains a significant driver of worldwide morbidity and mortality. Environmental factors associated with colorectal cancer are increasingly well-described and now include generalized colonic dysbiosis and individual enteric bacteria. Clostridioides difficile is one such species, with recent mouse model work suggesting prolonged exposure to C. difficile toxin B is conducive to colonic tumorigenesis. However, there is a dearth of real-world human evidence linking C. difficile exposure and colorectal cancer. Herein, we analyzed a multicenter, longitudinal, electronic health record-based dataset to test the association between C. difficile test positivity and the risk of incident colorectal cancer utilizing unadjusted and multivariable (controlled for clinical conditions independently associated with colorectal cancer development) Cox proportional hazard modeling to compare C. difficile exposed and nonexposed cohorts. We found that individuals who tested recurrently positive for C. difficile had a significantly increased risk of incident colorectal cancer [adjusted HR (aHR) 2.05 (95% confidence interval, 1.27-3.29)] compared with those who tested positive only once [aHR 0.70 (0.45-1.10)] or never. Furthermore, we found potential trends that the effect of C. difficile test positivity on the risk of incident colorectal cancer was stronger amongst females compared with males. These findings help translate emerging mouse model work on C. difficile-influenced colorectal tumorigenesis and lay groundwork for more substantial human investigations into this connection. These findings also may begin to help guide the personalized deployment of novel fecal microbiota-based therapies designed to interrupt the life cycle of C. difficile within the gut of human hosts and, potentially, prevent long-term health sequelae of chronic C. difficile infection. SIGNIFICANCE:These findings help translate emerging mouse model work on C. difficile-influenced colorectal tumorigenesis and lay groundwork for more substantial human investigations into this connection. These findings also may begin to help guide the personalized deployment of novel fecal microbiota-based therapies designed to interrupt the life cycle of C. difficile within the gut of human hosts and, potentially, prevent long-term health sequelae of chronic C. difficile infection.

Association between nominal dose effect of CD infection (0, 1 and > 1) and colorectal incidence by race, Michigan Medicine (2000-2023), Johns Hopkins (2016-2024)

The development and use of hierarchical composite endpoints (HCEs) in pneumonia clinical trials are recommended by federal regulatory agencies, expert professional societies and pharmaceutical stakeholders alike. However, the selection and hierarchy of nonfatal outcomes in HCEs often reflect investigator opinion rather than the bedside priorities of frontline clinicians. To date, no study has systematically evaluated how clinicians prioritize nonfatal outcomes in clinical decision-making for patients with pneumonia.Table 1.Patient Outcome Features and LevelsTable 2.Demographic Characteristics of Survey RespondentsIndividual question response sizes vary per question given respondent attribution. Not all responses shown. We conducted an international web-based survey using choice-based conjoint analysis. Respondents were presented with and asked to choose a more desirable overall outcome between two hypothetical patient profiles from a ventilator-associated pneumonia (VAP) trial, each comprised of a bundle of nonfatal outcomes of varying severity (Table 1). Feature order and level selection were randomized per respondent. Bayesian hierarchical modeling was used to calculate normalized part-worth utilities and determine the relative feature importance of each nonfatal outcome. The survey targeted practicing clinicians who care for patients with VAP and was deployed via professional society listservs.Figure 1.Bar Plot of Relative Feature Importance of Nonfatal OutcomesFeature importance values reflect the measure of influence of each nonfatal outcome on the desirability of the overall patient outcome and were were derived by dividing the range of part-worth utilities across a given feature's levels by the sum of all attribute part-worth utility score ranges. 95% confidence intervals derived via bootstrapping and display in error bars.Figure 2.Lollipop Plot of Part-Worth Utility Scores of Nonfatal Outcomes.Relative utility values (normalized part-worth utility scores) represent the weighted value respondents assigned to each nonfatal outcome level. A higher relative utility value implies a greater preference for a given feature level and a greater weight the presence of said feature level has on respondent choice, whereas a lower/negative relative utility value implies a lower preference for a given feature level and a lower weight the presence of said feature level has on respondent choice. The survey had 506 respondents and a 93% completion rate; demographics are shown in Table 2. Clinicians prioritized antibiotic-related adverse events (median feature importance 26.4% [95% CI 25.7-26.8%]), duration of invasive mechanical ventilation (24.9% [95% CI 24.4-25.9%]), and 30-day discharge disposition (24.2% [95% CI 23.6-25.1%]), over time to symptoms resolution (13.8% [95% CI 13.4-14.2%]), and pneumonia recurrence (9.2% [95% CI 8.6-9.6%]) (Figure 1). Part-worth utility scores for individual feature levels are shown in Figure 2. In this international conjoint analysis survey, clinicians prioritized outcomes often excluded from current pneumonia HCEs, such as discharge disposition and antibiotic-related adverse events. In contrast, they placed less value on pneumonia recurrence and time to clinical cure, despite their frequent use in existing pneumonia regulatory/comparative effectiveness trials. These findings suggest that pneumonia trial HCEs may be misaligned with the priorities of frontline clinicians. Krishna Rao, MD, MS, Merck & Co, Inc.: Grant/Research Support|Rebiotix Inc.: Advisor/Consultant|Seres Therapeutics: Advisor/Consultant|SUmmit Therapeutics Inc: Advisor/Consultant Keith S. Kaye, MD, MPH, AbbVie: Advisor/Consultant|GSK: Advisor/Consultant|Merck: Advisor/Consultant|Shionogi: Advisor/Consultant Owen Albin, MD, Biomerieux: Advisor/Consultant|Biomerieux: Grant/Research Support|Charles River Laboratories: Advisor/Consultant

Introduction Current guideline-recommended antibiotic treatment durations for ventilator-associated pneumonia (VAP) are largely standardised, with limited consideration of individual patient characteristics, pathogens or clinical context. This one-size-fits-all approach risks both overtreatment—promoting antimicrobial resistance and adverse drug events—as well as undertreatment, increasing the likelihood of pneumonia recurrence and sepsis-related complications. There is a critical need for VAP-specific biomarkers to enable individualised treatment strategies. The Ventilator-associated pneumonia Biomarker Evaluation (VIBE) study aims to identify a dynamic alveolar biomarker signature associated with treatment response, with the goal of informing personalised antibiotic duration in future clinical trials.Methods and analysis VIBE is a prospective, observational, case-cohort study of 125 adult patients with VAP in Michigan Medicine University Hospital intensive care units. Study subjects will undergo non-bronchoscopic bronchoalveolar lavage on the day of VAP diagnosis (Day 1) and then on Days 3 and 5. Alveolar biomarkers (quantitative respiratory culture bioburden, alveolar neutrophil percentage and pathogen genomic load assessed via BioFire FilmArray polymerase chain reaction) will be assessed. An expert panel of intensivists, blinded to biomarker data, will adjudicate each patient’s Day 10 outcome as VAP clinical cure (control) or treatment failure (case). Absolute biomarker levels and mean-fold changes in biomarker levels will be compared between groups. Data will be used to derive a composite temporal alveolar biomarker signature predictive of VAP treatment failure.Ethics and dissemination Ethical approval was obtained from the University of Michigan Institutional Review Board (IRB #HUM00251780). Informed consent will be obtained from all study participants or their legally authorised representatives. Findings will be disseminated through peer-reviewed publications, conferences and feedback into clinical guidelines committees.

Clostridioides difficile infection (CDI) and recurrent CDI (rCDI) are significant causes of morbidity and mortality. The microbiome plays a significant role in the body's defense against CDI and rCDI. Antibiotics can cause significant injury to the microbiome which leads to an increased risk of CDI and rCDI. Ongoing perturbations of the microbiome perpetuate this risk. Antibiotic treatments for CDI can kill C difficile but also can impact the microbiome. Microbiome therapeutics are effective in restoring the function of the gut microbiota and re-establishing colonization resistance. The field of microbiome therapeutics is evolving with newer, more refined, modalities in development.

Abstract Background Infections with Clostridioides difficile are associated with prolonged hospital stays, higher costs, and significant morbidity. Artificial intelligence (AI) tools can accurately predict which hospitalized patients are most likely to acquire C. difficile infection (CDI). However, to date, such tools have not been used in clinical practice. We investigated how AI tools for CDI risk stratification could be integrated into clinical workflows to promote targeted infection prevention efforts. Details of the infection prevention bundle (a) Screenshot of the BPA for enhanced handwashing precautions. This BPA instructs the receiving provider to place an order for putting up the “Enhanced Handwashing Precautions” sign, depicted in Figure 2. (b) Screenshot of the BPA for antimicrobial stewardship. This BPA is educational and provides a list of recommendations for reducing risk of CDI, including discontinuing unnecessary acid suppressants, minimizing unnecessary antibiotics, consulting the beta-lactam allergy evaluation service, and encouraging patient to eat yogurt if appropriate. Methods A previously validated AI model for predicting CDI risk from routinely collected data in electronic health records was used to generate daily risk scores for adult inpatients presenting to Michigan Medicine between January 1, 2023 and December 31, 2023. These scores were used to focus infection prevention efforts on high-risk patients in 10 selected hospital units with the greatest concentration of CDI cases. The infection prevention bundle, aimed at reducing both susceptibility and exposure, included provider-facing best practice alerts (BPAs) for enhanced handwashing precautions and antimicrobial stewardship (Figure 1). Using retrospective data, we determined a risk threshold that targets 5 alerts/unit/week on average. Clinical staff on selected units were educated about the AI tool by the study team. Picture of the “Enhanced Handwashing Precautions” sign This sign is placed on the door of the rooms for high-risk patients in selected hospital units and instructs all persons to wash their hands with soap and water upon room entry. Results During the study, 12,983 hospitalizations corresponding to 10,815 patients were assessed daily by the model, totaling 109,068 CDI risk scores. Among this population, 2,151 (16.6%) high-risk hospitalizations exceeded the risk threshold and triggered BPAs (an average of 4.1 alerts/unit/week). Among the high-risk population, 1,647 (76.6%) and 117 (5.4%) hospitalizations received an order for enhanced handwashing precautions and an order for a β-lactam allergy evaluation consultation, respectively. Field observations and interviews with clinical staff revealed challenges associated with behavior changes such as compliance with handwashing using soap and water to remove spores. Conclusion AI tools can be integrated into clinical workflows to promote targeted infection prevention efforts. However, continuous monitoring of how such tools interact with existing workflows and education on novel infection prevention strategies are key to success. Disclosures Krishna Rao, MD, MS, Merck and Company, Inc.: Grant/Research Support|Rebiotix Inc.: Advisor/Consultant|Seres Therapeutics: Advisor/Consultant|Summit pharmaceuticals: Advisor/Consultant

Importance:Increasingly, artificial intelligence (AI) is being used to develop models that can identify patients at high risk for adverse outcomes. However, the clinical impact of these models remains largely unrealized. Objective:To evaluate the association of an AI-guided infection prevention bundle with Clostridioides difficile infection (CDI) incidence in a hospital setting. Design, Setting, and Participants:This prospective, single-center quality improvement study evaluated adult inpatient hospitalizations before (September 1, 2021, to August 31, 2022) and after (January 1, 2023, to December 31, 2023) AI implementation. Data analysis was performed from January to August 2024. Intervention:A previously validated institution-specific AI model for CDI risk prediction was integrated into clinical workflows at the study site. The model was used to guide infection prevention practices for reducing pathogen exposure through enhanced hand hygiene and reducing host susceptibility through antimicrobial stewardship. Main Outcomes and Measures:The primary outcome was CDI incidence rate. Secondary outcomes included antimicrobial use and qualitative assessments of bundle implementation. Results:Pre-AI and post-AI samples included 39 046 (21 645 [55.4%] female; median [IQR] age, 58 [36-70] years) and 40 515 (22 575 [55.7%] female; median [IQR] age, 58 [37-70] years) hospitalizations, respectively. After adjusting for differences in clinical characteristics, there was no significant reduction in CDI incidence (pre-AI period: 5.76 per 10 000 patient-days vs post-AI period: 5.65 per 10 000 patient-days; absolute difference, -0.11; 95% CI, -1.43 to 1.18; P = .85). Relative reductions greater than 10% in normalized antimicrobial days were seen for piperacillin-tazobactam (-9.64; 95% CI, -12.93 to -6.28; P < .001) and clindamycin (-1.04; 95% CI, -1.60 to -0.47; P = .03), especially for high-risk patients alerted by AI (relative reduction for piperacillin-tazobactam, 16.8%; 95% CI, 8.0%-24.6%). On the basis of qualitative assessments via semistructured interviews and field observations, the study found that health care staff's experiences with AI-guided workflows varied. In particular, the enhanced hand hygiene protocols were met with poor adherence, whereas pharmacists consistently engaged with the alerts. Conclusions and Relevance:In this quality improvement study, the implementation of an AI-guided infection prevention bundle was not associated with a significant reduction in the already low CDI incidence rate at the study site, but it was associated with reduced CDI-associated antimicrobial use. The results highlight the potential of AI in supporting antimicrobial stewardship. Barriers to implementation, including infrastructure, staff knowledge, and workflow integration, need to be addressed in future applications.

Despite 2 decades of effort, there is a lack of clinically deployed models for predicting incident, severe, or recurrent Clostridioides difficile infection (CDI). This review outlines the promise of machine learning and biomarker-augmented models for targeted prevention and treatment, but also emphasizes the challenges of real-world deployment-namely integration into clinical workflows and governance. Moving forward, progress will depend on translational biomarker development, pragmatic modeling pipelines, and continuous monitoring. With these elements in place, CDI prediction tools can become a template for precision prevention of healthcare-associated infections.