Evaluating the impact of artificial intelligence-assisted image analysis on the diagnostic accuracy of front-line clinicians in detecting fractures on plain X-rays (FRACT-AI): protocol for a prospective observational study.

INTRODUCTION: Missed fractures are the most frequent diagnostic error attributed to clinicians in UK emergency departments and a significant cause of patient morbidity. Recently, advances in computer vision have led to artificial intelligence (AI)-enhanced model developments, which can support clinicians in the detection of fractures. Previous research has shown these models to have promising effects on diagnostic performance, but their impact on the diagnostic accuracy of clinicians in the National Health Service (NHS) setting has not yet been fully evaluated. METHODS AND ANALYSIS: A dataset of 500 plain radiographs derived from Oxford University Hospitals (OUH) NHS Foundation Trust will be collated to include all bones except the skull, facial bones and cervical spine. The dataset will be split evenly between radiographs showing one or more fractures and those without. The reference ground truth for each image will be established through independent review by two senior musculoskeletal radiologists. A third senior radiologist will resolve disagreements between two primary radiologists. The dataset will be analysed by a commercially available AI tool, BoneView (Gleamer, Paris, France), and its accuracy for detecting fractures will be determined with reference to the ground truth diagnosis. We will undertake a multiple case multiple reader study in which clinicians interpret all images without AI support, then repeat the process with access to AI algorithm output following a 4-week washout. 18 clinicians will be recruited as readers from four hospitals in England, from six distinct clinical groups, each with three levels of seniority (early-stage, mid-stage and later-stage career). Changes in the accuracy, confidence and speed of reporting will be compared with and without AI support. Readers will use a secure web-based DICOM (Digital Imaging and Communications in Medicine) viewer (www.raiqc.com), allowing radiograph viewing and abnormality identification. Pooled analyses will be reported for overall reader performance as well as for subgroups including clinical role, level of seniority, pathological finding and difficulty of image. ETHICS AND DISSEMINATION: The study has been approved by the UK Healthcare Research Authority (IRAS 310995, approved on 13 December 2022). The use of anonymised retrospective radiographs has been authorised by OUH NHS Foundation Trust. The results will be presented at relevant conferences and published in a peer-reviewed journal. TRIAL REGISTRATION NUMBERS: This study is registered with ISRCTN (ISRCTN19562541) and ClinicalTrials.gov (NCT06130397). The paper reports the results of a substudy of STEDI2 (Simulation Training for Emergency Department Imaging Phase 2).

Evaluation of the impact of artificial intelligence-assisted image interpretation on the diagnostic performance of clinicians in identifying pneumothoraces on plain chest X-ray: a multi-case multi-reader study.

BACKGROUND: Artificial intelligence (AI)-assisted image interpretation is a fast-developing area of clinical innovation. Most research to date has focused on the performance of AI-assisted algorithms in comparison with that of radiologists rather than evaluating the algorithms' impact on the clinicians who often undertake initial image interpretation in routine clinical practice. This study assessed the impact of AI-assisted image interpretation on the diagnostic performance of frontline acute care clinicians for the detection of pneumothoraces (PTX). METHODS: A multicentre blinded multi-case multi-reader study was conducted between October 2021 and January 2022. The online study recruited 18 clinician readers from six different clinical specialties, with differing levels of seniority, across four English hospitals. The study included 395 plain CXR images, 189 positive for PTX and 206 negative. The reference standard was the consensus opinion of two thoracic radiologists with a third acting as arbitrator. General Electric Healthcare Critical Care Suite (GEHC CCS) PTX algorithm was applied to the final dataset. Readers individually interpreted the dataset without AI assistance, recording the presence or absence of a PTX and a confidence rating. Following a 'washout' period, this process was repeated including the AI output. RESULTS: Analysis of the performance of the algorithm for detecting or ruling out a PTX revealed an overall AUROC of 0.939. Overall reader sensitivity increased by 11.4% (95% CI 4.8, 18.0, p=0.002) from 66.8% (95% CI 57.3, 76.2) unaided to 78.1% aided (95% CI 72.2, 84.0, p=0.002), specificity 93.9% (95% CI 90.9, 97.0) without AI to 95.8% (95% CI 93.7, 97.9, p=0.247). The junior reader subgroup showed the largest improvement at 21.7% (95% CI 10.9, 32.6), increasing from 56.0% (95% CI 37.7, 74.3) to 77.7% (95% CI 65.8, 89.7, p<0.01). CONCLUSION: The study indicates that AI-assisted image interpretation significantly enhances the diagnostic accuracy of clinicians in detecting PTX, particularly benefiting less experienced practitioners. While overall interpretation time remained unchanged, the use of AI improved diagnostic confidence and sensitivity, especially among junior clinicians. These findings underscore the potential of AI to support less skilled clinicians in acute care settings.

Assessing the effectiveness of artificial intelligence (AI) in prioritising CT head interpretation: study protocol for a stepped-wedge cluster randomised trial (ACCEPT-AI).

INTRODUCTION: Diagnostic imaging is vital in emergency departments (EDs). Accessibility and reporting impacts ED workflow and patient care. With radiology workforce shortages, reporting capacity is limited, leading to image interpretation delays. Turnaround times for image reporting are an ED bottleneck. Artificial intelligence (AI) algorithms can improve productivity, efficiency and accuracy in diagnostic radiology, contingent on their clinical efficacy. This includes positively impacting patient care and improving clinical workflow. The ACCEPT-AI study will evaluate Qure.ai's qER software in identifying and prioritising patients with critical findings from AI analysis of non-contrast head CT (NCCT) scans. METHODS AND ANALYSIS: This is a multicentre trial, spanning four diverse sites, over 13 months. It will include all individuals above the age of 18 years who present to the ED, referred for an NCCT. The project will be divided into three consecutive phases (pre-implementation, implementation and post-implementation of the qER solution) in a stepped-wedge design to control for adoption bias and adjust for time-based changes in the background patient characteristics. Pre-implementation involves baseline data for standard care to support the primary and secondary outcomes. The implementation phase includes staff training and qER solution threshold adjustments in detecting target abnormalities adjusted, if necessary. The post-implementation phase will introduce a notification (prioritised flag) in the radiology information system. The radiologist can choose to agree with the qER findings or ignore it according to their clinical judgement before writing and signing off the report. Non-qER processed scans will be handled as per standard care. ETHICS AND DISSEMINATION: The study will be conducted in accordance with the principles of Good Clinical Practice. The protocol was approved by the Research Ethics Committee of East Midlands (Leicester Central), in May 2023 (REC (Research Ethics Committee) 23/EM/0108). Results will be published in peer-reviewed journals and disseminated in scientific findings (ClinicalTrials.gov: NCT06027411) TRIAL REGISTRATION NUMBER: NCT06027411.

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

INTRODUCTION: A 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 (AI) tools to detect abnormalities on NCCTH. These tools are intended to provide clinical decision support for clinicians, rather than stand-alone diagnostic devices. However, validation studies mostly compare AI performance against radiologists, and there is relative paucity of evidence on the impact of AI assistance on other healthcare staff who review NCCTH in their daily clinical practice. METHODS AND ANALYSIS: A retrospective data set of 150 NCCTH will be compiled, to include 60 control cases and 90 cases with intracranial haemorrhage, hypodensities suggestive of infarct, midline shift, mass effect or skull fracture. The intracranial haemorrhage cases will be subclassified into extradural, subdural, subarachnoid, intraparenchymal and intraventricular. 30 readers will be recruited across four National Health Service (NHS) trusts including 10 general radiologists, 15 emergency medicine clinicians and 5 CT radiographers of varying experience. Readers will interpret each scan first without, then with, the assistance of the qER EU 2.0 AI tool, with an intervening 2-week washout period. Using a panel of neuroradiologists as ground truth, the stand-alone performance of qER will be assessed, and its impact on the readers' performance will be analysed as change in accuracy (area under the curve), median review time per scan and self-reported diagnostic confidence. Subgroup analyses will be performed by reader professional group, reader seniority, pathological finding, and neuroradiologist-rated difficulty. ETHICS AND DISSEMINATION: The study has been approved by the UK Healthcare Research Authority (IRAS 310995, approved 13 December 2022). The use of anonymised retrospective NCCTH has been authorised by Oxford University Hospitals. The results will be presented at relevant conferences and published in a peer-reviewed journal. TRIAL REGISTRATION NUMBER: NCT06018545.

Emergency Department Resuscitative Endovascular Balloon Occlusion of the Aorta in Trauma Patients With Exsanguinating Hemorrhage: The UK-REBOA Randomized Clinical Trial.

Importance: Bleeding is the most common cause of preventable death after trauma. Objective: To determine the effectiveness of resuscitative endovascular balloon occlusion of the aorta (REBOA) when used in the emergency department along with standard care vs standard care alone on mortality in trauma patients with exsanguinating hemorrhage. Design, Setting, and Participants: Pragmatic, bayesian, randomized clinical trial conducted at 16 major trauma centers in the UK. Patients aged 16 years or older with exsanguinating hemorrhage were enrolled between October 2017 and March 2022 and followed up for 90 days. Intervention: Patients were randomly assigned (1:1 allocation) to a strategy that included REBOA and standard care (n = 46) or standard care alone (n = 44). Main Outcomes and Measures: The primary outcome was all-cause mortality at 90 days. Ten secondary outcomes included mortality at 6 months, while in the hospital, and within 24 hours, 6 hours, or 3 hours; the need for definitive hemorrhage control procedures; time to commencement of definitive hemorrhage control procedures; complications; length of stay; blood product use; and cause of death. Results: Of the 90 patients (median age, 41 years [IQR, 31-59 years]; 62 [69%] were male; and the median Injury Severity Score was 41 [IQR, 29-50]) randomized, 89 were included in the primary outcome analysis because 1 patient in the standard care alone group declined to provide consent for continued participation and data collection 4 days after enrollment. At 90 days, 25 of 46 patients (54%) had experienced all-cause mortality in the REBOA and standard care group vs 18 of 43 patients (42%) in the standard care alone group (odds ratio [OR], 1.58 [95% credible interval, 0.72-3.52]; posterior probability of an OR >1 [indicating increased odds of death with REBOA], 86.9%). Among the 10 secondary outcomes, the ORs for mortality and the posterior probabilities of an OR greater than 1 for 6-month, in-hospital, and 24-, 6-, or 3-hour mortality were all increased in the REBOA and standard care group, and the ORs were increased with earlier mortality end points. There were more deaths due to bleeding in the REBOA and standard care group (8 of 25 patients [32%]) than in standard care alone group (3 of 18 patients [17%]), and most occurred within 24 hours. Conclusions and Relevance: In trauma patients with exsanguinating hemorrhage, a strategy of REBOA and standard care in the emergency department does not reduce, and may increase, mortality compared with standard care alone. Trial Registration: isrctn.org Identifier: ISRCTN16184981.

Diagnosis of cauda equina syndrome in the emergency department.

Cauda equina syndrome (CES) is a spinal emergency that can be challenging to identify from among the many patients presenting to EDs with low back and/or radicular leg pain. This article presents a practical guide to the assessment and early management of patients with suspected CES as well as an up-to-date review of the most important studies in this area that should inform clinical practice in the ED.

Pharmacological interventions for the prevention of bleeding in people undergoing definitive fixation or joint replacement for hip, pelvic and long bone fractures.

BACKGROUND: Pelvic, hip, and long bone fractures can result in significant bleeding at the time of injury, with further blood loss if they are treated with surgical fixation. People undergoing surgery are therefore at risk of requiring a blood transfusion and may be at risk of peri-operative anaemia. Pharmacological interventions for blood conservation may reduce the risk of requiring an allogeneic blood transfusion and associated complications. OBJECTIVES: To assess the effectiveness of different pharmacological interventions for reducing blood loss in definitive surgical fixation of the hip, pelvic, and long bones. SEARCH METHODS: We used a predefined search strategy to search CENTRAL, MEDLINE, PubMed, Embase, CINAHL, Transfusion Evidence Library, ClinicalTrials.gov, and the WHO International Clinical Trials Registry Platform (ICTRP) from inception to 7 April 2022, without restrictions on language, year, or publication status. We handsearched reference lists of included trials to identify further relevant trials. We contacted authors of ongoing trials to acquire any unpublished data. SELECTION CRITERIA: We included randomised controlled trials (RCTs) of people who underwent trauma (non-elective) surgery for definitive fixation of hip, pelvic, and long bone (pelvis, tibia, femur, humerus, radius, ulna and clavicle) fractures only. There were no restrictions on gender, ethnicity, or age. We excluded planned (elective) procedures (e.g. scheduled total hip arthroplasty), and studies published since 2010 that had not been prospectively registered. Eligible interventions included: antifibrinolytics (tranexamic acid, aprotinin, epsilon-aminocaproic acid), desmopressin, factor VIIa and XIII, fibrinogen, fibrin sealants, and non-fibrin sealants. DATA COLLECTION AND ANALYSIS: Two review authors independently assessed trial eligibility and risk of bias, and extracted data. We assessed the certainty of the evidence using GRADE. We did not perform a network meta-analysis due to lack of data. MAIN RESULTS: We included 13 RCTs (929 participants), published between 2005 and 2021. Three trials did not report any of our predefined outcomes and so were not included in quantitative analyses (all were tranexamic acid versus placebo). We identified three comparisons of interest: intravenous tranexamic acid versus placebo; topical tranexamic acid versus placebo; and recombinant factor VIIa versus placebo. We rated the certainty of evidence as very low to low across all outcomes. Comparison 1. Intravenous tranexamic acid versus placebo Intravenous tranexamic acid compared to placebo may reduce the risk of requiring an allogeneic blood transfusion up to 30 days (RR 0.48, 95% CI 0.34 to 0.69; 6 RCTs, 457 participants; low-certainty evidence) and may result in little to no difference in all-cause mortality (Peto odds ratio (Peto OR) 0.38, 95% CI 0.05 to 2.77; 2 RCTs, 147 participants; low-certainty evidence).  It may result in little to no difference in risk of participants experiencing myocardial infarction (risk difference (RD) 0.00, 95% CI -0.03 to 0.03; 2 RCTs, 199 participants; low-certainty evidence), and cerebrovascular accident/stroke (RD 0.00, 95% CI -0.02 to 0.02; 3 RCTs, 324 participants; low-certainty evidence).  We are uncertain if there is a difference between groups for risk of deep vein thrombosis (Peto OR 2.15, 95% CI 0.22 to 21.35; 4 RCTs, 329 participants, very low-certainty evidence), pulmonary embolism (Peto OR 1.08, 95% CI 0.07 to 17.66; 4 RCTs, 329 participants; very low-certainty evidence), and suspected serious drug reactions (RD 0.00, 95% CI -0.03 to 0.03; 2 RCTs, 185 participants; very low-certainty evidence). No data were available for number of red blood cell units transfused, reoperation, or acute transfusion reaction. We downgraded the certainty of the evidence for imprecision (wide confidence intervals around the estimate and small sample size, particularly for rare events), and risk of bias (unclear or high risk methods of blinding and allocation concealment in the assessment of subjective measures), and upgraded the evidence for transfusion requirement for a large effect.  Comparison 2. Topical tranexamic acid versus placebo We are uncertain if there is a difference between topical tranexamic acid and placebo for risk of requiring an allogeneic blood transfusion (RR 0.31, 95% CI 0.08 to 1.22; 2 RCTs, 101 participants), all-cause mortality (RD 0.00, 95% CI -0.10 to 0.10; 1 RCT, 36 participants), risk of participants experiencing myocardial infarction (Peto OR 0.15, 95% CI 0.00 to 7.62; 1 RCT, 36 participants), cerebrovascular accident/stroke (RD 0.00, 95% CI -0.06 to 0.06; 1 RCT, 65 participants); and deep vein thrombosis (Peto OR 1.11, 95% CI 0.07 to 17.77; 2 RCTs, 101 participants).  All outcomes reported were very low-certainty evidence. No data were available for number of red blood cell units transfused, reoperation, incidence of pulmonary embolism, acute transfusion reaction, or suspected serious drug reactions. We downgraded the certainty of the evidence for imprecision (wide confidence intervals around the estimate and small sample size, particularly for rare events), inconsistency (moderate heterogeneity), and risk of bias (unclear or high risk methods of blinding and allocation concealment in the assessment of subjective measures, and high risk of attrition and reporting biases in one trial). Comparison 3. Recombinant factor VIIa versus placebo   Only one RCT of 48 participants reported data for recombinant factor VIIa versus placebo, so we have not presented the results here. AUTHORS' CONCLUSIONS: We cannot draw conclusions from the current evidence due to lack of data. Most published studies included in our analyses assessed the use of tranexamic acid (compared to placebo, or using different routes of administration).  We identified 27 prospectively registered ongoing RCTs (total target recruitment of 4177 participants by end of 2023). The ongoing trials create six new comparisons: tranexamic acid (tablet + injection) versus placebo; intravenous tranexamic acid versus oral tranexamic acid; topical tranexamic acid versus oral tranexamic acid; different intravenous tranexamic acid dosing regimes; topical tranexamic acid versus topical fibrin glue; and fibrinogen (injection) versus placebo.

Which clinical features best predict occult scaphoid fractures? A systematic review of diagnostic test accuracy studies.

BACKGROUND: Plain radiographs cannot identify all scaphoid fractures; thus ED patients with a clinical suspicion of scaphoid injury often undergo immobilisation despite normal imaging. This study determined (1) the prevalence of scaphoid fracture among patients with a clinical suspicion of scaphoid injury with normal radiographs and (2) whether clinical features can identify patients that do not require immobilisation and further imaging. METHODS: This systematic review of diagnostic test accuracy studies included all study designs that evaluated predictors of scaphoid fracture among patients with normal initial radiographs. Quality assessment was undertaken using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. Meta-analyses included all studies. RESULTS: Eight studies reported data on 1685 wrist injuries. The prevalence of scaphoid fracture despite normal radiographs was 9.0%. Most studies were at overall low risk of bias but two were at unclear risk; all eight were at low risk for applicability concerns. The most accurate clinical predictors of occult scaphoid fracture were pain when the examiner moved the wrist from a pronated to a supinated position against resistance (sensitivity 100%, specificity 97.9%, LR+ 45.0, 95% CI 6.5 to 312.5), supination strength <10% of contralateral side (sensitivity 84.6%, specificity 76.9%, LR+ 3.7, 95% CI 2.2 to 6.1), pain on ulnar deviation (sensitivity 55.2%, specificity 76.4%, LR+ 2.3, 95% CI 1.8 to 3.0) and pronation strength <10% of contralateral side (sensitivity 69.2%, specificity 64.6%, LR+ 2.0, 95% CI 1.2 to 3.2). Absence of anatomical snuffbox tenderness significantly reduced the likelihood of an occult scaphoid fracture (sensitivity 92.1%, specificity 48.4%, LR- 0.2, 95% CI 0.0 to 0.7). CONCLUSION: No single feature satisfactorily excludes an occult scaphoid fracture. Further work should explore whether a combination of clinical features, possibly in conjunction with injury characteristics (such as mechanism) and a normal initial radiograph might exclude fracture. Pain on supination against resistance would benefit from external validation. TRIAL REGISTRATION NUMBER: CRD42021290224.

Clinical predictors of fracture in patients with shoulder dislocation: systematic review of diagnostic test accuracy studies.

BACKGROUND: Prereduction radiographs are conventionally used to exclude fracture before attempts to reduce a dislocated shoulder in the ED. However, this step increases cost, exposes patients to ionising radiation and may delay closed reduction. Some studies have suggested that prereduction imaging may be omitted for a subgroup of patients with shoulder dislocations. OBJECTIVES: To determine whether clinical predictors can identify patients who may safely undergo closed reduction of a dislocated shoulder without prereduction radiographs. METHODS: A systematic review and meta-analysis of diagnostic test accuracy studies that have evaluated the ability of clinical features to identify concomitant fractures in patients with shoulder dislocation. The search was updated to 23 June 2022 and language limits were not applied. All fractures were included except for Hill-Sachs lesions. Quality assessment was undertaken using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. Data were pooled and meta-analysed by fitting univariate random effects and multilevel mixed effects logistic regression models. RESULTS: Eight studies reported data on 2087 shoulder dislocations and 343 concomitant fractures. The most important potential sources of bias were unclear blinding of those undertaking the clinical (6/8 studies) and radiographic (3/8 studies) assessment. The prevalence of concomitant fracture was 17.5%. The most accurate clinical predictors were age >40 (positive likelihood ratio (LR+) 1.8, 95% CI 1.5 to 2.1; negative likelihood ratio (LR-) 0.4, 95% CI 0.2 to 0.6), female sex (LR+ 2.0, 95% CI 1.6 to 2.4; LR- 0.7, 95% CI 0.6 to 0.8), first-time dislocation (LR+ 1.7, 95% CI 1.4 to 2.0; LR- 0.2, 95% CI 0.1 to 0.5) and presence of humeral ecchymosis (LR+ 3.0-5.7, LR- 0.8-1.1). The most important mechanisms of injury were high-energy mechanism fall (LR+ 2.0-9.8, LR- 0.4-0.8), fall >1 flight of stairs (LR+ 3.8, 95% CI 0.6 to 13.1; LR- 1.0, 95% CI 0.9 to 1.0) and motor vehicle collision (LR+ 2.3, 95% CI 0.5 to 4.0; LR- 0.9, 95% CI 0.9 to 1.0). The Quebec Rule had a sensitivity of 92.2% (95% CI 54.6% to 99.2%) and a specificity of 33.3% (95% CI 23.1% to 45.3%), but the Fresno-Quebec rule identified all clinically important fractures across two studies: sensitivity of 100% (95% CI 89% to 100%) in the derivation dataset and 100% (95% CI 90% to 100%) in the validation study. The specificity of the Fresno-Quebec rule ranged from 34% (95% CI 28% to 41%) in the derivation dataset to 24% (95% CI 16% to 33%) in the validation study. CONCLUSION: Clinical prediction rules may have a role in supporting shared decision making after shoulder dislocation, particularly in the prehospital and remote environments when delay to imaging is anticipated.

Does peripheral neuroinflammation predict chronicity following whiplash injury? Protocol for a prospective cohort study.

INTRODUCTION: Whiplash-associated disorder grade 2 (WAD2) is characterised by musculoskeletal pain/tenderness but no apparent nerve injury. However, studies have found clinical features indicative of neuropathy and neuropathic pain. These studies may indicate peripheral nerve inflammation, since preclinical neuritis models found mechanical sensitivity in inflamed, intact nociceptors. The primary aim of this study is to establish the contribution of peripheral neuroinflammation to WAD2 and its role in prognosis. Participants will be invited to participate in a sub-study investigating the contribution of cutaneous small fibre pathology to WAD2. METHODS AND ANALYSIS: 115 participants within 1 month following whiplash injury and 34 healthy control participants will be recruited and complete validated questionnaires for pain, function and psychological factors. Data collection will take place at the Universities of Sussex and Oxford, UK. Clinical examination, quantitative sensory testing and blood samples will be undertaken. MRI scans using T2-weighted and diffusion tensor images of the brachial plexus and wrist will determine nerve inflammation and nerve structural changes. Skin biopsies from a substudy will determine structural integrity of dermal and intraepidermal nerve fibres. At 6 months, we will evaluate recovery using Neck Disability Index and a self-rated global recovery question and repeat the outcome measures. Regression analysis will identify differences in MRI parameters, clinical tests and skin biopsies between participants with WAD2 and age/gender-matched controls. Linear and logistic regression analyses will assess if nerve inflammation (MRI parameters) predicts poor outcome. Mixed effects modelling will compare MRI and clinical measures between recovered and non-recovered participants over time. ETHICS AND DISSEMINATION: Ethical approval was received from London-Brighton and Sussex Research Ethics Committee (20/PR/0625) and South Central-Oxford C Ethics Committee (18/SC/0263). Written informed consent will be obtained from participants prior to participation in the study. Results will be disseminated through publications in peer-reviewed journals, presentations at national/international conferences and social media. TRIAL REGISTRATION NUMBER: NCT04940923.

Real-world evaluation of rapid and laboratory-free COVID-19 triage for emergency care: external validation and pilot deployment of artificial intelligence driven screening.

BACKGROUND: Uncertainty in patients' COVID-19 status contributes to treatment delays, nosocomial transmission, and operational pressures in hospitals. However, the typical turnaround time for laboratory PCR remains 12-24 h and lateral flow devices (LFDs) have limited sensitivity. Previously, we have shown that artificial intelligence-driven triage (CURIAL-1.0) can provide rapid COVID-19 screening using clinical data routinely available within 1 h of arrival to hospital. Here, we aimed to improve the time from arrival to the emergency department to the availability of a result, do external and prospective validation, and deploy a novel laboratory-free screening tool in a UK emergency department. METHODS: We optimised our previous model, removing less informative predictors to improve generalisability and speed, developing the CURIAL-Lab model with vital signs and readily available blood tests (full blood count [FBC]; urea, creatinine, and electrolytes; liver function tests; and C-reactive protein) and the CURIAL-Rapide model with vital signs and FBC alone. Models were validated externally for emergency admissions to University Hospitals Birmingham, Bedfordshire Hospitals, and Portsmouth Hospitals University National Health Service (NHS) trusts, and prospectively at Oxford University Hospitals, by comparison with PCR testing. Next, we compared model performance directly against LFDs and evaluated a combined pathway that triaged patients who had either a positive CURIAL model result or a positive LFD to a COVID-19-suspected clinical area. Lastly, we deployed CURIAL-Rapide alongside an approved point-of-care FBC analyser to provide laboratory-free COVID-19 screening at the John Radcliffe Hospital (Oxford, UK). Our primary improvement outcome was time-to-result, and our performance measures were sensitivity, specificity, positive and negative predictive values, and area under receiver operating characteristic curve (AUROC). FINDINGS: 72 223 patients met eligibility criteria across the four validating hospital groups, in a total validation period spanning Dec 1, 2019, to March 31, 2021. CURIAL-Lab and CURIAL-Rapide performed consistently across trusts (AUROC range 0·858-0·881, 95% CI 0·838-0·912, for CURIAL-Lab and 0·836-0·854, 0·814-0·889, for CURIAL-Rapide), achieving highest sensitivity at Portsmouth Hospitals (84·1%, Wilson's 95% CI 82·5-85·7, for CURIAL-Lab and 83·5%, 81·8-85·1, for CURIAL-Rapide) at specificities of 71·3% (70·9-71·8) for CURIAL-Lab and 63·6% (63·1-64·1) for CURIAL-Rapide. When combined with LFDs, model predictions improved triage sensitivity from 56·9% (51·7-62·0) for LFDs alone to 85·6% with CURIAL-Lab (81·6-88·9; AUROC 0·925) and 88·2% with CURIAL-Rapide (84·4-91·1; AUROC 0·919), thereby reducing missed COVID-19 cases by 65% with CURIAL-Lab and 72% with CURIAL-Rapide. For the prospective deployment of CURIAL-Rapide, 520 patients were enrolled for point-of-care FBC analysis between Feb 18 and May 10, 2021, of whom 436 received confirmatory PCR testing and ten (2·3%) tested positive. Median time from arrival to a CURIAL-Rapide result was 45 min (IQR 32-64), 16 min (26·3%) sooner than with LFDs (61 min, 37-99; log-rank p<0·0001), and 6 h 52 min (90·2%) sooner than with PCR (7 h 37 min, 6 h 5 min to 15 h 39 min; p<0·0001). Classification performance was high, with sensitivity of 87·5% (95% CI 52·9-97·8), specificity of 85·4% (81·3-88·7), and negative predictive value of 99·7% (98·2-99·9). CURIAL-Rapide correctly excluded infection for 31 (58·5%) of 53 patients who were triaged by a physician to a COVID-19-suspected area but went on to test negative by PCR. INTERPRETATION: Our findings show the generalisability, performance, and real-world operational benefits of artificial intelligence-driven screening for COVID-19 over standard-of-care in emergency departments. CURIAL-Rapide provided rapid, laboratory-free screening when used with near-patient FBC analysis, and was able to reduce the number of patients who tested negative for COVID-19 but were triaged to COVID-19-suspected areas. FUNDING: The Wellcome Trust, University of Oxford Medical and Life Sciences Translational Fund.

Changes in paediatric respiratory infections at a UK teaching hospital 2016-2021; impact of the SARS-CoV-2 pandemic.

Objective To describe the impact of the SARS-CoV-2 pandemic on the incidence of paediatric viral respiratory tract infection in Oxfordshire, UK. Methods Data on paediatric Emergency Department (ED) attendances (0-15 years inclusive), respiratory virus testing, vital signs and mortality at Oxford University Hospitals were summarised using descriptive statistics. Results Between 1-March-2016 and 30-July-2021, 155,056 ED attendances occurred and 7,195 respiratory virus PCRs were performed. Detection of all pathogens was suppressed during the first national lockdown. Rhinovirus and adenovirus rates increased when schools reopened September-December 2020, then fell, before rising in March-May 2021. The usual winter RSV peak did not occur in 2020/21, with an inter-seasonal rise (32/1,000 attendances in 0-3 yr olds) in July 2021. Influenza remained suppressed throughout. A higher paediatric early warning score (PEWS) was seen for attendees with adenovirus during the pandemic compared to pre-pandemic (p = 0.04, Mann-Witney U test), no other differences in PEWS were seen. Conclusions SARS-CoV-2 caused major changes in the incidence of paediatric respiratory viral infection in Oxfordshire, with implications for clinical service demand, testing strategies, timing of palivizumab RSV prophylaxis, and highlighting the need to understand which public health interventions are most effective for preventing respiratory virus infections.

Impact of the COVID-19 pandemic on emergency department attendances and acute medical admissions.

BACKGROUND: To better understand the impact of the COVID-19 pandemic on hospital healthcare, we studied activity in the emergency department (ED) and acute medicine department of a major UK hospital. METHODS: Electronic patient records for all adult patients attending ED (n = 243,667) or acute medicine (n = 82,899) during the pandemic (2020-2021) and prior year (2019) were analysed and compared. We studied parameters including severity, primary diagnoses, co-morbidity, admission rate, length of stay, bed occupancy, and mortality, with a focus on non-COVID-19 diseases. RESULTS: During the first wave of the pandemic, daily ED attendance fell by 37%, medical admissions by 30% and medical bed occupancy by 27%, but all returned to normal within a year. ED attendances and medical admissions fell across all age ranges; the greatest reductions were seen for younger adults in ED attendances, but in older adults for medical admissions. Compared to non-COVID-19 pandemic admissions, COVID-19 admissions were enriched for minority ethnic groups, for dementia, obesity and diabetes, but had lower rates of malignancy. Compared to the pre-pandemic period, non-COVID-19 pandemic admissions had more hypertension, cerebrovascular disease, liver disease, and obesity. There were fewer low severity ED attendances during the pandemic and fewer medical admissions across all severity categories. There were fewer ED attendances with common non-respiratory illnesses including cardiac diagnoses, but no change in cardiac arrests. COVID-19 was the commonest diagnosis amongst medical admissions during the first wave and there were fewer diagnoses of pneumonia, myocardial infarction, heart failure, cellulitis, chronic obstructive pulmonary disease, urinary tract infection and other sepsis, but not stroke. Levels had rebounded by a year later with a trend to higher levels of stroke than before the pandemic. During the pandemic first wave, 7-day mortality was increased for ED attendances, but not for non-COVID-19 medical admissions. CONCLUSIONS: Reduced ED attendances in the first wave of the pandemic suggest opportunities for reducing low severity presentations to ED in the future, but also raise the possibility of harm from delayed or missed care. Reassuringly, recent rises in attendance and admissions indicate that any deterrent effect of the pandemic on attendance is diminishing.