Alex Howard

Background Non-ventilator-associated hospital-acquired pneumonia (nv-HAP) diagnosis is imprecise; chest x-rays (CXR) lack detail and sputum culture delays targeted prescribing. However, the evidence base to support diagnostic improvement is sparse. We assessed the feasibility of a trial of low-dose thoracic CT and rapid molecular testing of sputum. Methods HAP-FAST was a UK NIHR funded, open-label pilot sequential multiple assignment randomised trial with qualitative and costing substudies in four UK hospitals (ClinicalTrials.gov: NCT05483309 ). Using research without prior consent (RWPC) and site-stratified web-based randomisation, clinicians allocated adults with suspected nv-HAP (excluding pregnancy and palliation), before their second antibiotic dose, 1:1 to CT or CXR; participants subsequently treated for nv-HAP with sputum available underwent another 1:1 randomisation to BIOFIRE ^® FILMARRAY ^® Pneumonia plus Panel (FAPP) with standard microbiology or standard microbiology alone . Feasibility outcomes included recruitment, consent, intervention delivery and data completeness through 90 days. Adverse events were not systematically collected. Analyses were intention-to-treat and descriptive. Results Between June 2023 and March 2024, 220 adults were randomised at stage-1 (109 CT; 111 CXR), reaching target three months early. Of 154/220 (70.0%) subsequently managed as nv-HAP, 92/154 (59.7%) could not produce sputum. 62/154 (40.3%; 28.2% of stage-1) were randomised to stage-2: 31 FAPP and 31 no-FAPP. Allocated imaging was recorded for 75/109 (68.8%) CT and 99/111 (89.2%) CXR participants; 6/31(19.4%) FAPP tests were not completed. Consent for follow-up was obtained for 142/220 (64.5%). Protocol-defined clinical cure was assessable in 1/220 participants. The day-90 resource-use questionnaire was returned by 55/110 participants (50.0%). Qualitative findings supported RWPC but identified needs for staged communication, lower-burden follow-up and improved diagnostic pathways. Excess bed days and readmissions drove costs but data capture requires refinement. Conclusions A definitive trial is plausible only with enhanced RWPC, diagnostic delivery and follow-up pathways, improved sputum procedures, stewardship-aligned outcomes and routine economic data capture.

Amoxicillin-clavulanic acid (AMX-CLV) is a widely used oral β-lactam/β-lactamase inhibitor combination against Escherichia coli . Clinical success is largely confined to urinary tract infections. The mechanistic basis for this site-specific efficacy remains unclear. Using a hollow-fibre infection model to replicate human plasma and urinary pharmacokinetics, we show that plasma-like exposures rapidly select for pre-existing resistant subpopulations; whereas, urinary exposures produce sustained bactericidal activity without resistance emergence. Genomic and transcriptomic analyses following plasma drug exposure reveal that treatment selectively enriches pre-existing resistant lineages already harbouring oxidative-stress-associated mutations that activate the SOS response and drive IS-mediated amplification of blaTEM-1 , leading to β-lactamase hyperproduction and treatment failure. In contrast, the high urinary concentrations of clavulanic acid exert direct antibacterial activity, eradicating these subpopulations. Our findings demonstrate that local pharmacokinetic environments fundamentally shape evolutionary trajectories under β-lactam/β-lactamase inhibitor therapy, explaining the restricted efficacy of AMX-CLV and revealing a dynamic interplay between stress responses, genome plasticity, and drug partitioning that governs treatment outcome.

Background: Bloodstream infections (BSIs) are a major complication during pre-engraftment neutropenia in hematopoietic stem-cell transplant (HSCT) recipients. Quinolone prophylaxis reduces BSIs but increases multidrug-resistance bacterial infections. We aimed to assess the impact of discontinuing quinolone prophylaxis in adult HSCT recipients. Methods: We performed a systematic review with meta-analysis. Five databases were searched on May 20, 2025. Studies were eligible for inclusion if they reported clinical characteristics of adult HSCT recipients and outcomes following the discontinuation of prophylaxis. Data extraction and quality assessment were performed independently by 2 reviewers. Meta-analysis was performed and pooled risk ratios with 95% CI calculated using a random-effects model. Results: Ten studies including 2363 HSCT recipients (1190 with and 1173 without prophylaxis) were included. Most were allogeneic-HSCT (53.1%). Prophylaxis was associated with a lower risk of BSIs due to any microorganism (pooled risk ratio: 0.69; 95% CI 0.54–0.87; P value = .002; I2 = 61%) and due to Gram-negative bacteria (pooled risk ratio: 0.49; 95% CI 0.33–0.74; P value = .0007; I2 = 71%). In the overall population, there was no significant difference in resistant infections. In patients with Gram-negative BSIs, prophylaxis was associated with a higher risk of quinolone-resistant (pooled risk ratio: 2.35; 95% CI 1.68–3.29; P value < .00001; I2 = 27%) and carbapenem-resistant microorganisms (pooled risk ratio: 5.32; 95% CI 1.08–26.16; P value = .04; I2 = 69%). Discontinuation was not associated with a statistically increased infection-related mortality (pooled risk ratio: 1.26; 95% CI: 0.71–2.23; P value = .43; I2 = 0%). Conclusions: Discontinuation of quinolone prophylaxis was associated with an elevated incidence of BSIs, without a concomitant reduction in the overall burden of antimicrobial resistance. Importantly, infection-related mortality did not increase to a statistically significant extent. Furthermore, Gram-negative BSIs occurring after discontinuation were less likely to exhibit resistance to quinolones or carbapenems.

Predicting antibiotic treatment outcomes could help tackle antibiotic resistance by guiding prescribing decisions. Existing approaches do not quantitatively incorporate the judgment of clinician users. Our antibiotic decision-making algorithm predicted treatment outcomes for 13 antibiotics using clinical prediction models trained on prescribing and urine culture data from 93,906 patients, then weighted outcomes using treatment decisions made by 49 clinicians in an antibiotic choice ranking exercise. In a simulation using Emergency Department data, the algorithm chose more correctly-targeted World Health Organization Access category antibiotics (75.6% of cases versus 11.9%, 95% confidence interval of difference 57.6% to 69.7%, p < 0.001) and oral antibiotics (69% versus 22.6%, 95% confidence interval of difference 39.5% to 53.4%, p < 0.001) than human prescribers, and fewer intravenous antibiotics (31.2% versus 65.8%, 95% confidence interval of difference −41.9% to −27.1%, p < 0.001). These results show that our algorithm could improve antibiotic prescribing decisions by combining human judgment with data-driven probability predictions.

Amoxicillin-clavulanic acid (AMX-CLV) is a widely used oral β-lactam/β-lactamase inhibitor combination against Escherichia coli . Clinical success is largely confined to urinary tract infections. The mechanistic basis for this site-specific efficacy remains unclear. Using a hollow-fibre infection model to replicate human plasma and urinary pharmacokinetics, we show that plasma-like exposures rapidly select for pre-existing resistant subpopulations; whereas, urinary exposures produce sustained bactericidal activity without resistance emergence. Genomic and transcriptomic analyses following plasma drug exposure revealed that treatment selectively enriches pre-existing resistant lineages already harbouring oxidative-stress-associated mutations that activate the SOS response and drive IS-mediated amplification of blaTEM-1 , leading to β-lactamase hyperproduction and treatment failure. In contrast, the high urinary concentrations of clavulanic acid exert direct antibacterial activity, eradicating these subpopulations. Our findings demonstrate that local pharmacokinetic environments fundamentally shape evolutionary trajectories under β-lactam/β-lactamase inhibitor therapy, explaining the restricted efficacy of AMX-CLV and revealing a dynamic interplay between stress responses, genome plasticity, and drug partitioning that governs treatment outcome.

Objectives: Infection diagnoses are critical to the personalized interpretation of EUCAST aminopenicillin breakpoints for Enterobacterales, but microbiology laboratories cannot predict diagnosis when specimens are received. Here, we assess whether machine learning could facilitate personalized antimicrobial susceptibility breakpoint reporting by predicting urinary tract infection (UTI) diagnoses. Methods: XGBoost models were trained using open-source electronic healthcare record data to predict complicated UTI in patients with Enterobacterales bacteriuria and to predict UTI in patients with Enterobacterales bacteraemia. These models were validated and used to provide simulated aminopenicillin dosing/regimen recommendations based on antimicrobial susceptibility results for patients with bacteriuria and bacteraemia in a holdout dataset. The main outcomes were the proportions of patients recommended appropriate aminopenicillin dosages/regimens according to EUCAST guidelines based on their diagnosis. Results: The area under the receiver operating characteristic curve was 0.62 for predicting both complicated UTI in patients with bacteriuria and UTI in patients with bacteraemia. In the simulation study, 79.3% (n = 276) and 72.7% (n = 8) of patients with ampicillin-susceptible Enterobacterales bacteriuria and bacteraemia, respectively, were recommended appropriate aminopenicillin dosages/regimens for their infection diagnosis according to EUCAST guidelines. Adjusting the probability threshold for predicting complicated UTI increased the proportion of appropriate recommendations in bacteriuria to 96.6% (n = 336). Conclusions: Using machine learning models to predict the probability of complicated UTI in patients with bacteriuria and the probability of UTI in patients with bacteraemia resulted in appropriate aminopenicillin dosages/regimens being recommended in most cases. These results provide proof-of-concept for how machine learning could facilitate the personalized implementation of EUCAST aminopenicillin breakpoints.