National Health Service Trusts Research Articles

Implementation issues and barriers for assessing oral health in dependent patients after stroke: A qualitative study

ObjectivesTo explore implementation issues and potential barriers for assessing oral health in dependent post-stroke patients. MethodsSemi-structured interviews were conducted with a purposively identified sample of healthcare service providers who work in two National Health Service (NHS) Trusts in the north of England. Interviews were conducted until data saturation was achieved (n = 30). Data were analysed using the constant comparative method. ResultsSix themes were drawn out in this study, which described potential barriers to assessing oral health in post-stroke patients, aspects of oral health that need assessment, streamlining the oral health assessment, input methods for oral health assessment, characteristics of assessors, and how oral care should be planned. ConclusionsAssessment of oral health for post-stroke patients has been viewed as a complex task because of several identified barriers. Several suggestions have been proposed to overcome these barriers, aiming to enable more feasible and effective oral health assessments for post-stroke patients. Clinical SignificanceThe findings from this study have the potential to contribute to developing oral health measurement instruments that might be more successfully implemented and guide oral care planning for dependent patients after stroke.

Efficiency Assessment of Breast Clinics for Patients Under 35: A Comparative Analysis of Targeted Models in a University Hospital.

Despite the higher incidence of breast cancer in older age groups, it remains pertinent not to overlook breast cancer occurrence in those aged 35 years and below. Recent transitions towardtargeted under-35 clinics in Englandaim to enhance efficiency and meet referral standards. Three models were planned, and we assessed the efficiency of each model. This study, conducted for five months within a single National Health Service (NHS) trust, analyzed data from the following clinics: the General One-Stop Clinic, the Under 35 One-Stop Clinic with ultrasound services (USS), and the Under 35 Clinic without USS services. Of the 300 patients recruited (100 consecutive patients from each clinic), 94.3% were female. The average age at presentation was 27.53 years. The most frequently encountered age group was between 26 and 30 years, and the majorityof patients had palpable lumps (78, 51.6%). Out of 300 patients who attended the clinics, 151 had USS, and of these, 15 biopsies were performed. Fibroadenomas (32, 21.2%) and cysts (22, 14.6%) were the most common radiological findings.We found that more breast imaging was being undertaken for under-35 patients who attended the general one-stop clinics compared to the specific under-35 clinics. Targeted breast clinics for individuals 35 years and below offer an effective approach in terms of resource allocation and meeting cancer targets.

Health outcomes after myocardial infarction: A population study of 56 million people in England.

The occurrence of a range of health outcomes following myocardial infarction (MI) is unknown. Therefore, this study aimed to determine the long-term risk of major health outcomes following MI and generate sociodemographic stratified risk charts in order to inform care recommendations in the post-MI period and underpin shared decision making. This nationwide cohort study includes all individuals aged ≥18 years admitted to one of 229 National Health Service (NHS) Trusts in England between 1 January 2008 and 31 January 2017 (final follow-up 27 March 2017). We analysed 11 non-fatal health outcomes (subsequent MI and first hospitalisation for heart failure, atrial fibrillation, cerebrovascular disease, peripheral arterial disease, severe bleeding, renal failure, diabetes mellitus, dementia, depression, and cancer) and all-cause mortality. Of the 55,619,430 population of England, 34,116,257 individuals contributing to 145,912,852 hospitalisations were included (mean age 41.7 years (standard deviation [SD 26.1]); n = 14,747,198 (44.2%) male). There were 433,361 individuals with MI (mean age 67.4 years [SD 14.4)]; n = 283,742 (65.5%) male). Following MI, all-cause mortality was the most frequent event (adjusted cumulative incidence at 9 years 37.8% (95% confidence interval [CI] [37.6,37.9]), followed by heart failure (29.6%; 95% CI [29.4,29.7]), renal failure (27.2%; 95% CI [27.0,27.4]), atrial fibrillation (22.3%; 95% CI [22.2,22.5]), severe bleeding (19.0%; 95% CI [18.8,19.1]), diabetes (17.0%; 95% CI [16.9,17.1]), cancer (13.5%; 95% CI [13.3,13.6]), cerebrovascular disease (12.5%; 95% CI [12.4,12.7]), depression (8.9%; 95% CI [8.7,9.0]), dementia (7.8%; 95% CI [7.7,7.9]), subsequent MI (7.1%; 95% CI [7.0,7.2]), and peripheral arterial disease (6.5%; 95% CI [6.4,6.6]). Compared with a risk-set matched population of 2,001,310 individuals, first hospitalisation of all non-fatal health outcomes were increased after MI, except for dementia (adjusted hazard ratio [aHR] 1.01; 95% CI [0.99,1.02];p = 0.468) and cancer (aHR 0.56; 95% CI [0.56,0.57];p < 0.001). The study includes data from secondary care only-as such diagnoses made outside of secondary care may have been missed leading to the potential underestimation of the total burden of disease following MI. In this study, up to a third of patients with MI developed heart failure or renal failure, 7% had another MI, and 38% died within 9 years (compared with 35% deaths among matched individuals). The incidence of all health outcomes, except dementia and cancer, was higher than expected during the normal life course without MI following adjustment for age, sex, year, and socioeconomic deprivation. Efforts targeted to prevent or limit the accrual of chronic, multisystem disease states following MI are needed and should be guided by the demographic-specific risk charts derived in this study.

AI assisted reader evaluation in acute CT head interpretation (AI-REACT): protocol for a multireader multicase study

IntroductionA non-contrast CT head scan (NCCTH) is the most common cross-sectional imaging investigation requested in the emergency department. Advances in computer vision have led to development of several artificial intelligence...

Are Clinical Prediction Rules Used in Spinal Cord Injury Care? A Survey of Practice.

Accurate outcome prediction is desirable post spinal cord injury (SCI), reducing uncertainty for patients and supporting personalized treatments. Numerous attempts have been made to create clinical prediction rules that identify patients who are likely to recover function. It is unknown to what extent these rules are routinely used in clinical practice. To better understand knowledge of, and attitudes toward, clinical prediction rules amongst SCI clinicians in the United Kingdom. An online survey was distributed via mailing lists of clinical special interest groups and relevant National Health Service Trusts. Respondents answered questions about their knowledge of existing clinical prediction rules and their general attitudes to using them. They also provided information about their level of experience with SCI patients. One hundred SCI clinicians completed the survey. The majority (71%) were unaware of clinical prediction rules for SCI; only 8% reported using them in clinical practice. Less experienced clinicians were less likely to be aware. Lack of familiarity with prediction rules was reported as being a barrier to their use. The importance of clinical expertise when making prognostic decisions was emphasized. All respondents reported interest in using clinical prediction rules in the future. The results show widespread lack of awareness of clinical prediction rules amongst SCI clinicians in the United Kingdom. However, clinicians were positive about the potential for clinical prediction rules to support decision-making. More focus should be directed toward refining current rules and improving dissemination within the SCI community.

Models of care in secure services for people with intellectual and developmental disability: Implementing the Walkway to Wellness.

Changes to policy around inpatient services for people with intellectual and developmental disability (IDD) who offend, have led to a need for services to reconsider their models of care. This has led to calls for more tailored, patient-centred care models, with less reliance solely on offence-related treatment programmes which can be unsuitable for a growing proportion of patients with more complex cognitive and behavioural difficulties. In response, the Walkway to Wellness (W2W) was developed at one National Health Service Trust providing secure services to people with IDD, with the intention of delivering a more collaborative, co-produced and goal-oriented care model that was better understood by staff and patient stakeholders. To evaluate the implementation of the W2W using Normalisation Process Theory (NPT), an evidence-based theoretical approach is used across a number of health settings. Staff were invited to complete a short questionnaire, using the NPT informed Normalisation Measure Development questionnaire, at two time points along the implementation process. Patients were invited to complete a simplified questionnaire. Both groups were asked for their views on the W2W and the process of its implementation. Although the W2W was more familiar to staff at the second time point, scores on the four NPT constructs showed a trend for it being less embedded in practice, with significant results concerning the ongoing appraisal of the new model. Patient views were mixed; some saw the benefit of more goal-oriented processes, but others considered it an additional chore hindering their own perceived goals. Early involvement of all stakeholders is required to enhance the understanding of changes to models of care. Live feedback should be used to refine and revise the model to meet the needs of patients, carers and staff members.

Estimating demand for potential disease-modifying therapies for Alzheimer's disease in the UK.

Phase three trials of the monoclonal antibodies lecanemab and donanemab, which target brain amyloid, have reported statistically significant differences in clinical end-points in early Alzheimer's disease. These drugs are already in use in some countries and are going through the regulatory approval process for use in the UK. Concerns have been raised about the ability of healthcare systems, including those in the UK, to deliver these treatments, considering the resources required for their administration and monitoring. To estimate the scale of real-world demand for monoclonal antibodies for Alzheimer's disease in the UK. We used anonymised patient record databases from two National Health Service trusts for the year 2019 to collect clinical, demographic, cognitive and neuroimaging data for these cohorts. Eligibility for treatment was assessed using the inclusion criteria from the clinical trials of donanemab and lecanemab, with consideration given to diagnosis, cognitive performance, cerebrovascular disease and willingness to receive treatment. We examined the records of 82 386 people referred to services covering around 2.2 million people. After applying the trial criteria, we estimate that a maximum of 906 people per year would start treatment with monoclonal antibodies in the two services, equating to 30 200 people if extrapolated nationally. Monoclonal antibody treatments for Alzheimer's disease are likely to present a significant challenge for healthcare services to deliver in terms of the neuroimaging and treatment delivery. The data provided here allows health services to understand the potential demand and plan accordingly.

Retrospective cohort study of the association between socioeconomic deprivation and incidence of gestational diabetes and perinatal outcomes

IntroductionSocioeconomic disparities have been shown to correlate with perinatal mortality and the incidence of type 2 diabetes. Few studies have explored the relationship between deprivation and the incidence of gestational diabetes (GDM). We aimed to identify the relationship between deprivation and incidence of GDM, after adjusting for age, BMI, and ethnicity. We also examined for relationships between deprivation and perinatal outcomes.MethodsA retrospective cohort analysis of 23,490 pregnancies from a major National Health Service Trust in Northwest London was conducted. The 2019 English Indices of Multiple Deprivation was used to identify the deprivation rank and decile for each postcode. Birthweight centile was calculated from absolute birthweight after adjusting for ethnicity, maternal height, maternal weight, parity, sex and outcome (live birth/stillbirth). Logistic regression and Kendall’s Tau were used to identify relationships between variables.ResultsAfter controlling for age, BMI & ethnicity, Index of Multiple Deprivation postcode decile was not associated with an increased risk of developing gestational diabetes. Each increase in decile of deprivation was associated with an increase in birthweight centile by 0.471 (p < 0.001). After adjusting for confounders, age was associated with a 7.1% increased GDM risk (OR: 1.076, p < 0.001); BMI increased risk by 5.81% (OR: 1.059, p < 0.001). There was no significant correlation between Index of Multiple Deprivation rank and perinatal outcomes.DiscussionOur analysis demonstrates that socioeconomic deprivation was not associated with incidence of GDM or adverse perinatal outcomes. Factors such as genetic predisposition and lifestyle habits may likely play a larger role in the development of GDM compared to socioeconomic deprivation alone.

“Quite simply they don’t communicate”: a case study of a National Health Service response to staff suicide

Workplace suicide can have significant knock-on effects within an organisation, yet research has shown within the healthcare profession, not all staff receive suicide prevention training, and few employers take the...

Women’s experiences of the consenting process for pregnancy remains disposal following early miscarriage

Background and methodologyUK clinical practices around managing pregnancy remains after pregnancy loss involve a process of documenting consent. Women are typically offered options for disposal, which may include cremation, burial,...

N-desethyl isotonitazene detected in polydrug users admitted to hospital in Birmingham, United Kingdom

Introduction Nitazenes are potent synthetic opioids and N-desethyl isotonitazene, a metabolite of isotonitazene, has emerged as a drug in its own right. Methods This is an observational case series of patients with suspected or declared substance use who were admitted to hospitals in the Sandwell and West Birmingham National Health Service Trust between July and October 2023. All patients were found on toxicological screening to have been exposed to N-desethyl isotonitazene. Results Twenty presentations involving 19 patients who tested positive for N-desethyl isotonitazene were included in the study. In 19 presentations, multiple substances were detected on toxicological screening. The number of patients testing positive for other substances were: 19 for cocaine and its main metabolite benzoylecgonine, 13 for morphine, 11 for the heroin-specific metabolite 6-monoacetylmorphine, ten for xylazine, eight for gabapentinoids (pregabalin and/or gabapentin), seven for methadone and/or the metabolite, 2-ethylidene-1, 5-dimethyl-3, 3-diphenylpyrrolidine, six for benzodiazepines and five for the synthetic cannabinoid MDMB-4en-PINACA. Only one patient had no other substances detected apart from N-desethyl isotonitazene. This patient presented with coma, miosis, bradypnoea and hypercapnia and responded to naloxone. In this cohort, the median concentration of N-desethyl isotonitazene was 1.53 µg/L (n = 14; range 0.59–5.48) in whole blood and 27.75 µg/L (n = 16; range 0.51–91.53) in urine. Discussion The majority of the patients in this cohort presented with features typical of an opioid overdose, which is unsurprising as they were all experienced users of diamorphine. Although these features are also consistent with the known effects of N-desethyl isotonitazene, in only one case is it possible to attribute the patient’s features to N-desethyl isotonitazene toxicity alone. Conclusions This case series highlights the need for toxicovigilance in the illicit drug market as patterns of substance misuse evolve and novel psychoactive substances continue to emerge.

Establishing the size and configuration of the imaging support workforce: a census of national workforce data in England.

The imaging support workforce is a key enabler in unlocking imaging capacity and capability, yet no evidence exists of the workforce size and configuration. This research provides the first comprehensive analysis of workforce data to explore the deployment of the support workforce within National Health Service (NHS) imaging services in England. Using a census methodology, an anonymized electronic staff record (ESR) data set extracted in December 2022 was analysed to identify support workers and their employment bandings at NHS Trust, regional and national (England) level. Support workforce proportions, median values, and Spearman's rank correlations were calculated. Analysis of 137 NHS Trusts, comprising 100% of acute trusts (n = 124) and specialist trusts with imaging services (n = 13), identified that the support workforce (pay bands 2-4) constitutes 23.6% of the imaging staff base. Ranking trusts into 3 categories based on the proportion of support workers in their imaging establishment, median values ranged from 30.7% (high) to 22.2% (medium) and 10.5% (low). Two opposing deployment models of band 2 and band 3 support workers were identified. Comprising almost one-quarter of the imaging establishment, models of deployment at bands 2 and 3 are highly variable. Assistant practitioners (band 4) are under-utilised, providing an opportunity to introduce innovations to address workforce demands. This census is the first to provide evidence of the size and structure of the support workforce, the first step in enabling effective workforce transformation. Further research is required to explain the two opposing deployment models.

NeuroRehabilitation OnLine: Description of a regional multidisciplinary group telerehabilitation innovation for stroke and neurological conditions using the Template for Intervention Description and Replication checklist.

Providing recommended amounts of rehabilitation for stroke and neurological patients is challenging. Telerehabilitation is viable for delivering rehabilitation and an acceptable adjunct to in-person therapy. NeuroRehabilitation OnLine (NROL) was developed as a pilot and subsequently operationalised as a regional innovation embedded across four National Health Service (NHS) Trusts. To describe the NROL innovation to assist future implementation and replication efforts. The Template for Intervention Description and Replication (TIDieR) checklist, with guidance from the TIDieR-Telehealth extension, was used to describe NROL. The description was developed collaboratively by clinical academics, therapists, managers and researchers. Updated Consolidated Framework for Implementation Research domains were used to describe the context in which the innovation was delivered. NROL delivers online group-based real-time neurorehabilitation with technology assistance. It incorporates multidisciplinary targeted therapy and peer support to complement existing therapy. Procedures, materials and structure are detailed to demonstrate how NROL is embedded within a healthcare system. NROL uses existing NHS therapy workforce alongside dedicated NROL roles, including an essential technology support role. Selection of NROL groups is dependent on patient needs. The NROL innovation is tailored over time in response to feedback. NROL described here is successfully integrated within a regional stroke and neurorehabilitation network, aligns with local and national strategies and capitalises on an existing clinical-academic partnership. This comprehensive description of a regional NROL innovation, and clarification of core components, should facilitate other healthcare settings to adapt and implement NROL for their context. Continuous evaluation alongside implementation will ensure maximal impact for neurorehabilitation.

Real-time monitoring and feedback to improve shared decision-making for surgery (the ALPACA Study): protocol for a mixed-methods study to inform co-development of an inclusive intervention

IntroductionHigh-quality shared decision-making (SDM) is a priority of health services, but only achieved in a minority of surgical consultations. Improving SDM for surgical patients may lead to more effective care...

Effects of First Wave of COVID-19 on Colon Cancer Multi-disciplinary Team Performance: A Two-Year Analysis.

We investigated the effects of COVID-19-related delay on two-year outcomes of colon cancer treatment during the first wave of the pandemic. Ninety-two patients were referred for bowel cancer at our National Health Service (NHS) trust between March and July 2020, and 41 patients were treated for colon cancer and followed up (a two-year) through a multidisciplinary team (MDT). Treatment delays and overall survival (OS) were also assessed. Treatment delays were observed in 48% of patients. The average delay was 31 days beyond the 62-day mark (P0.001). Logistic and binary logistic regression models showed that a comorbid diagnosis of respiratory disease had a significant effect on delays in management and two-year outcomes (P=0.04), but without the likelihood of upstaging or a poorer outcome (P=0.942). The overall survival rate was 81.5%. Eight percent of bowel cancer surgeries could have been avoided if endoscopic visualization and biopsy were available, and 8% more surgeries could have been performed laparoscopically without fear of surrounding aerosols. The findings showed that oncologic care provided minimal disruption to trust during the COVID-19 pandemic owing to a quick association between the NHS site and a green non-NHS site, resulting in acceptable two-year outcomes for colon cancer patients.

Length of stay and economic sustainability of virtual ward care in a medium-sized hospital of the UK: a retrospective longitudinal study

ObjectiveTo evaluate the length of stay difference and its economic implications between hospital patients and virtual ward patients.DesignRetrospective longitudinal study.SettingWrightington, Wigan and Leigh (WWL) Teaching Hospitals, National Health Service (NHS)...

The important factors nurses consider when choosing shift patterns: A cross-sectional study.

To gain a deeper understanding of what is important to nurses when thinking about shift patterns and the organisation of working time. A cross-sectional survey of nursing staff working across the UK and Ireland collected quantitative and qualitative responses. We recruited from two National Health Service Trusts and through an open call via trade union membership, online/print nursing profession magazines and social media. Worked versus preferred shift length/pattern, satisfaction and choice over shift patterns and nurses' views on aspects related to work and life (when working short, long, rotating shifts) were analysed with comparisons of proportions of agreement and crosstabulation. Qualitative responses on important factors related to shift preferences were analysed with inductive thematic analysis. Eight hundred and seventy-three survey responses were collected. When nurses worked long shifts and rotating shifts, lower proportions reported being satisfied with their shifts and working their preferred shift length and pattern. Limited advantages were realised when comparing different shift types; however, respondents more frequently associated 'low travel costs' and 'better ability to do paid overtime' with long shifts and 'healthy diet/exercise' with short shifts; aspects related to rotating shifts often had the lowest proportions of agreement. In the qualitative analysis, three themes were developed: 'When I want to work', 'Impacts to my life outside work' and 'Improving my work environment'. Reasons for nurses' shift preferences were frequently related to nurses' priorities outside of work, highlighting the importance of organising schedules that support a good work-life balance. General scheduling practices like adhering to existing shift work guidelines, using consistent and predictable shift patterns and facilitating flexibility over working time were identified by nurses as enablers for their preferences and priorities. These practices warrant meaningful consideration when establishing safe and efficient nurse rosters. This survey was developed and tested with a diverse group of stakeholders, including nursing staff, patients, union leads and ward managers. The Strengthening the Reporting of Observational Studies (STROBE) checklist for cross-sectional studies was used to guide reporting.

Deploying artificial intelligence software in an NHS trust: a how-to guide for clinicians.

Over the past 10 years, artificial intelligence (AI) has become one of the fastest-growing sectors in healthcare. There are now numerous new technologies designed to cut costs and improve diagnoses and treatment pathways. However, there is significant scepticism amongst National Health Service (NHS) staff regarding the usefulness of AI and it's cost to the NHS. This has likely resulted in underuse and slow adoption of software that may revolutionize our healthcare system and ensure its continued survival and effectiveness. Several governing bodies have put forward guidance on the safe and effective adoption of AI tools, but this rarely covers the reality of selecting and deploying new software. This article set out clear guidance on the practicalities and pitfalls of deploying digital solutions in healthcare, using the example of a deep learning algorithm designed to improve the accuracy of chest X-ray (CXR) interpretation in the emergency department.

Accuracy of an artificial intelligence-enabled diagnostic assistance device in recognizing normal chest radiographs: a service evaluation.

Artificial intelligence (AI) enabled devices may be able to optimize radiologists' productivity by identifying normal and abnormal chest X-rays (CXRs) for triaging. In this service evaluation, we investigated the accuracy of one such AI device (qXR). A randomly sampled subset of general practice and outpatient-referred frontal CXRs from a National Health Service Trust was collected retrospectively from examinations conducted during November 2022 to January 2023. Ground truth was established by consensus between 2 radiologists. The main objective was to estimate negative predictive value (NPV) of AI. A total of 522 CXRs (458 [87.74%] normal CXRs) from 522 patients (median age, 64 years [IQR, 49-77]; 305 [58.43%] female) were analysed. AI predicted 348 CXRs as normal, of which 346 were truly normal (NPV: 99.43% [95% CI, 97.94-99.93]). The sensitivity, specificity, positive predictive value, and area under the ROC curve of AI were found to be 96.88% (95% CI, 89.16-99.62), 75.55% (95% CI, 71.34-79.42), 35.63% (95% CI, 28.53-43.23), and 91.92% (95% CI, 89.38-94.45), respectively. A sensitivity analysis was conducted to estimate NPV by varying assumptions of the prevalence of normal CXRs. The NPV ranged from 88.96% to 99.54% as prevalence increased. The AI device recognized normal CXRs with high NPV and has the potential to increase radiologists' productivity. There is a need for more evidence on the utility of AI-enabled devices in identifying normal CXRs. This work adds to such limited evidence and enables researchers to plan studies to further evaluate the impact of such devices.

Development and validation of open-source deep neural networks for comprehensive chest x-ray reading: a retrospective, multicentre study

Artificial intelligence (AI) systems for automated chest x-ray interpretation hold promise for standardising reporting and reducing delays in health systems with shortages of trained radiologists. Yet, there are few freely accessible AI systems trained on large datasets for practitioners to use with their own data with a view to accelerating clinical deployment of AI systems in radiology. We aimed to contribute an AI system for comprehensive chest x-ray abnormality detection. In this retrospective cohort study, we developed open-source neural networks, X-Raydar and X-Raydar-NLP, for classifying common chest x-ray findings from images and their free-text reports. Our networks were developed using data from six UK hospitals from three National Health Service (NHS) Trusts (University Hospitals Coventry and Warwickshire NHS Trust, University Hospitals Birmingham NHS Foundation Trust, and University Hospitals Leicester NHS Trust) collectively contributing 2 513 546 chest x-ray studies taken from a 13-year period (2006-19), which yielded 1 940 508 usable free-text radiological reports written by the contemporary assessing radiologist (collectively referred to as the "historic reporters") and 1 896 034 frontal images. Chest x-rays were labelled using a taxonomy of 37 findings by a custom-trained natural language processing (NLP) algorithm, X-Raydar-NLP, from the original free-text reports. X-Raydar-NLP was trained on 23 230 manually annotated reports and tested on 4551 reports from all hospitals. 1 694 921 labelled images from the training set and 89 238 from the validation set were then used to train a multi-label image classifier. Our algorithms were evaluated on three retrospective datasets: a set of exams sampled randomly from the full NHS dataset reported during clinical practice and annotated using NLP (n=103 328); a consensus set sampled from all six hospitals annotated by three expert radiologists (two independent annotators for each image and a third consultant to facilitate disagreement resolution) under research conditions (n=1427); and an independent dataset, MIMIC-CXR, consisting of NLP-annotated exams (n=252 374). X-Raydar achieved a mean AUC of 0·919 (SD 0·039) on the auto-labelled set, 0·864 (0·102) on the consensus set, and 0·842 (0·074) on the MIMIC-CXR test, demonstrating similar performance to the historic clinical radiologist reporters, as assessed on the consensus set, for multiple clinically important findings, including pneumothorax, parenchymal opacification, and parenchymal mass or nodules. On the consensus set, X-Raydar outperformed historical reporter balanced accuracy with significance on 27 of 37 findings, was non-inferior on nine, and inferior on one finding, resulting in an average improvement of 13·3% (SD 13·1) to 0·763 (0·110), including a mean 5·6% (13·2) improvement in critical findings to 0·826 (0·119). Our study shows that automated classification of chest x-rays under a comprehensive taxonomy can achieve performance levels similar to those of historical reporters and exhibit robust generalisation to external data. The open-sourced neural networks can serve as foundation models for further research and are freely available to the research community. Wellcome Trust.