An Electronic Health Record-Wide Association Study to identify populations at increased risk of E. coli bloodstream infections.

Objectives: Escherichia coli bacteraemias have been under mandatory surveillance in the UK for fifteen years, but cases continue to rise. Systematic searches of all features present within electronic healthcare records (EHRs), described here as an EHR-wide association study (EHR-WAS), could potentially identify under-appreciated factors that could be targeted to reduce infections.

Methods: We used data from Oxfordshire, UK, and an EHR-WAS method developed for use with large-scale COVID-19 data to estimate associations between E.

Interplay between C-reactive protein responses and antibiotic prescribing in people with suspected infection.

Background: Serial measurements of C-reactive protein (CRP) are often taken in hospitals to assess recovery from infection, but their utility remains debated. Previous studies, including our development of CRP centile reference charts for suspected bloodstream infections (BSI), suggest variability in CRP responses across infection types. Here we investigated the association between serial CRP percentile changes, antibiotic prescribing patterns, and patient outcomes in a large cohort with suspected infection, acknowledging that CRP is one of multiple factors in clinical decision-making.

Community-acquired pneumonia identification from electronic health records in the absence of a gold standard: A Bayesian latent class analysis.

Community-acquired pneumonia (CAP) is common and a significant cause of mortality. However, CAP surveillance commonly relies on diagnostic codes from electronic health records (EHRs), with imperfect accuracy. We used Bayesian latent class models with multiple imputation to assess the accuracy of CAP diagnostic codes in the absence of a gold standard and to explore the contribution of various EHR data sources in improving CAP identification.

Epidemiology and excess mortality of antimicrobial resistance in bacteraemias among cancer patients: a cohort study using routinely collected health data from regional hospital trusts in Oxford and Oslo, 2008-2018.

Objectives: We investigated the epidemiology and impact on mortality of antimicrobial resistance (AMR) in cancer patients with bacteraemia at Oxford University Hospitals (OxUH), UK, and Oslo University Hospital (OsUH), Norway, during 2008-2018.

Design: Historical cohort study.

Setting: Regional hospital trusts with multiple sites in OxUH and OsUH.

Utilizing large and diverse bacterial genome datasets to improve the detection and identification of via PCR-based diagnostics.

The accurate identification of (pneumococcus) is crucial for diagnostics and surveillance but is complicated by the use of molecular assays that may also detect non-pneumococcal (NPS) species. Therefore, the aim of this study was to use a combination of and analyses to evaluate PCR assays for the molecular detection and identification of pneumococci. A diverse dataset of over 9,300 pneumococcal and NPS genomes was investigated to determine the sensitivity and specificity of assays for seven recommended gene targets: , , , , Spn9802, SP2020 and Xisco.

Transformers and large language models are efficient feature extractors for electronic health record studies.

Background: Free-text data is abundant in electronic health records, but challenges in accurate and scalable information extraction mean less specific clinical codes are often used instead.

Methods: We evaluated the efficacy of feature extraction using modern natural language processing methods (NLP) and large language models (LLMs) on 938,150 hospital antibiotic prescriptions from Oxfordshire, UK. Specifically, we investigated inferring the type(s) of infection from a free-text "indication" field, where clinicians state the reason for prescribing antibiotics.

Molecular clock complexities of Clostridioides difficile.

Objectives: Reconstruct the phylogenetic status of a collection of historical Clostridioides difficile isolates and evaluate the congruence of their evolutionary trajectories with established molecular clock models.

Methods: Phylogenetic analysis was performed on Illumina sequence reads from previously analysed historic C. difficile isolates (1980-86; n = 75) demonstrating multiple antimicrobial resistances.

The impact of antimicrobial stewardship ward rounds on antimicrobial use and predictors of advice, uptake, and outcomes.

Objective: To identify the impact of introducing antimicrobial stewardship (AMS) ward rounds.

Methods: We used an interrupted time-series approach to investigate the impact of implementing AMS ward rounds with in-person feedback from a multidisciplinary team in Hospital-1, also comparing to Hospital-2 in the same city where AMS ward rounds were not yet implemented. Regression models were used to identify predictors of advice given and of whether advice was followed, and associations between advice uptake and length of stay.

Developing a framework for tracking antimicrobial resistance gene movement in a persistent environmental reservoir.

Mobile genetic elements are key to the global emergence of antibiotic resistance. We successfully reconstructed the complete bacterial genome and plasmid assemblies of isolates sharing the same carbapenemase gene to understand evolution over time in six confined hospital drains over five years. From 82 isolates we identified 14 unique strains from 10 species with 113 carrying plasmids across 16 distinct replicon types.

Machine learning and clinician predictions of antibiotic resistance in Enterobacterales bloodstream infections.

Background: Patients with Gram-negative bloodstream infections are at risk of serious adverse outcomes without active treatment, but identifying who has antimicrobial resistance (AMR) to target empirical treatment is challenging.

Methods: We used XGBoost machine learning models to predict antimicrobial resistance to seven antibiotics in patients with Enterobacterales bloodstream infection. Models were trained using hospital and community data from Oxfordshire, UK, for patients with positive blood cultures between 01-January-2017 and 31-December-2021.

Addition of Macrolide Antibiotics for Hospital Treatment of Community-Acquired Pneumonia.

Background: Guidelines recommend combining macrolides with β-lactam antibiotics for moderate-to-high severity community-acquired pneumonia (CAP); however, macrolides pose risks of adverse events and anti-microbial resistance.

Methods: We analyzed electronic health data from 8872 adults hospitalized with CAP in Oxfordshire, UK (2016-2024), initially treated with amoxicillin or co-amoxiclav. Using inverse probability treatment weighting, we examined the effects of adjunctive macrolides on 30-day all-cause mortality, time to hospital discharge, and changes in Sequential Organ Failure Assessment (SOFA) score.

Predicting individual patient and hospital-level discharge using machine learning.

Background: Accurately predicting hospital discharge events could help improve patient flow and the efficiency of healthcare delivery. However, using machine learning and diverse electronic health record (EHR) data for this task remains incompletely explored.

Methods: We used EHR data from February-2017 to January-2020 from Oxfordshire, UK to predict hospital discharges in the next 24 h.

Genomic epidemiology describes introduction and outbreaks of antifungal drug-resistant .

is a globally emerged fungal pathogen causing nosocomial invasive infections. Here, we use cutting-edge genomic approaches to elucidate the temporal and geographic epidemiology of drug-resistant within the UK. We analysed a representative sample of over 200 isolates from multiple UK hospitals to assess the number and timings of introductions and infer subsequent patterns of inter- and intra-hospital transmission of azole drug-resistant isolates.

Retrospective analysis of hospital electronic health records reveals unseen cases of acute hepatitis with unknown aetiology in adults in Oxfordshire.

Background: An outbreak of acute severe hepatitis of unknown aetiology (AS-Hep-UA) in children during 2022 was subsequently linked to infections with adenovirus-associated virus 2 and other 'helper viruses', including human adenovirus. It is possible that evidence of such an outbreak could be identified at a population level based on routine data captured by electronic health records (EHR).

Methods: We used anonymised EHR to collate retrospective data for all emergency presentations to Oxford University Hospitals NHS Foundation Trust in the UK, between 2016-2022, for all ages from 18 months and older.

Immunogenicity of COVID-19 vaccines in patients with follicular lymphoma receiving frontline chemoimmunotherapy.

Immune responses to primary COVID-19 vaccination were investigated in 58 patients with follicular lymphoma (FL) as part of the PETReA trial of frontline therapy (EudraCT 2016-004010-10). COVID-19 vaccines (BNT162b2 or ChAdOx1) were administered before, during or after cytoreductive treatment comprising rituximab (depletes B cells) and either bendamustine (depletes CD4 T cells) or cyclophosphamide-based chemotherapy. Blood samples obtained after vaccine doses 1 and 2 (V1, V2) were analysed for antibodies and T cells reactive to the SARS-CoV-2 spike protein using the Abbott Architect and interferon-gamma ELISpot assays respectively.

Changes in the investigation and management of suspected myocardial infarction and injury during COVID-19: a multi-centre study using routinely collected healthcare data.

Objective: The COVID-19 pandemic was associated with a reduction in the incidence of myocardial infarction (MI) diagnosis, in part because patients were less likely to present to hospital. Whether changes in clinical decision making with respect to the investigation and management of patients with suspected MI also contributed to this phenomenon is unknown.

Methods: Multicentre retrospective cohort study in three UK centres contributing data to the National Institute for Health Research Health Informatics Collaborative.

Detecting changes in population trends in infection surveillance using community SARS-CoV-2 prevalence as an exemplar.

Detecting and quantifying changes in the growth rates of infectious diseases is vital to informing public health strategy and can inform policymakers' rationale for implementing or continuing interventions aimed at reducing their impact. Substantial changes in SARS-CoV-2 prevalence with the emergence of variants have provided an opportunity to investigate different methods for doing this. We collected polymerase chain reaction (PCR) results from all participants in the United Kingdom's COVID-19 Infection Survey between August 1, 2020, and June 30, 2022.

Nanopore sequencing of influenza A and B in Oxfordshire and the United Kingdom, 2022-23.

Objectives: We evaluated Nanopore sequencing for influenza surveillance.

Methods: Influenza A and B PCR-positive samples from hospital patients in Oxfordshire, UK, and a UK-wide population survey from winter 2022-23 underwent Nanopore sequencing following targeted rt-PCR amplification.

Results: From 941 infections, successful sequencing was achieved in 292/388 (75 %) available Oxfordshire samples: 231 (79 %) A/H3N2, 53 (18 %) A/H1N1, and 8 (3 %) B/Victoria and in 53/113 (47 %) UK-wide samples.