Article Text
PDF
Abstract
Background The diagnosis of acute aortic syndrome (AAS) is commonly delayed or missed in the ED. We describe characteristics of ED attendances with symptoms potentially associated with AAS, diagnostic performance of clinical decision tools (CDTs) and physicians and yield of CT aorta angiogram (CTA).
Methods This was a multicentre observational cohort study of adults attending 27 UK EDs between 26 September 2022 and 30 November 2022, with potential AAS symptoms: chest, back or abdominal pain, syncope or symptoms related to malperfusion. Patients were preferably identified prospectively, but retrospective recruitment was also permitted. Anonymised, routinely collected patient data including components of CDTs, was abstracted. Clinicians treating prospectively identified patients were asked to record their perceived likelihood of AAS, prior to any confirmatory testing. Reference standard was radiological or operative confirmation of AAS. 30-day electronic patient record follow-up evaluated whether a subsequent diagnosis of AAS had been made and mortality.
Results 5548 patients presented, with a median age of 55 years (IQR 37–72; n=5539). 14 (0.3%; n=5353) had confirmed AAS. 10/1046 (1.0%) patients in whom the ED clinician thought AAS was possible had AAS. 5/147 (3.4%) patients in whom AAS was considered the most likely diagnosis had AAS. 2/3319 (0.06%) patients in whom AAS was considered not possible did have AAS. 540 (10%; n=5446) patients underwent CT, of which 407 were CTA (7%). 30-day follow-up did not reveal any missed AAS diagnoses. AUROC (area under the receiver operating characteristic) curve for ED clinician AAS likelihood rating was 0.958 (95% CI 0.933 to 0.983, n=4006) and for individual CDTs were: Aortic Dissection Detection Risk Score (ADD-RS) 0.674 (95% CI 0.508 to 0.839, n=4989), AORTAs 0.689 (95% CI 0.527 to 0.852, n=5132), Canadian 0.818 (95% CI 0.686 to 0.951, n=5180) and Sheffield 0.628 (95% CI 0.467 to 0.788, n=5092).
Conclusion Only 0.3% of patients presenting with potential AAS symptoms had AAS but 7% underwent CTA. CDTs incorporating clinician gestalt appear to be most promising, but further prospective work is needed, including evaluation of the role of D-dimer.
Trial registration number NCT05582967; NCT05582967.
- emergency department
- computed tomography
- diagnostic tests
- diagnosis
Data availability statement
Data are available upon reasonable request. All study data and metadata will be available upon reasonable request to dashedstudy@gmail.com.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
ED diagnosis of acute aortic syndrome (AAS) is a substantial challenge and many patients do not receive timely diagnosis and treatment. Over-investigation with too low a threshold for CT scanning of the thoracic aorta cannot be the solution. There is little research in truly undifferentiated ED populations, or in non-North American populations with different thresholds for CT and most previous studies of AAS clinical decision tools have limited inclusion to those patients undergoing CT aorta angiogram.
WHAT THIS STUDY ADDS
In this multicentre observational study including 5548 patients with symptoms potentially attributable to AAS, 0.3% of patients presenting with potential AAS symptoms did have AAS and 10% of patients with potential AAS symptoms undergo CT. A third of proven patients with AAS still endure a diagnostic delay over 24 hours from time of arrival.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
We illustrate the diagnostic challenge of AAS and the limitations of methods for selecting patients for CT. The best decision aid to facilitate decision to CT and to outperform ED clinician gestalt is not yet clear. More research is required in truly undifferentiated ED populations such as these.
Introduction
Acute aortic syndrome (AAS) could be considered as a wolf in sheep’s clothing in our emergency departments (EDs). AAS incorporates aortic dissection (AD), intramural haematoma (IMH) and penetrating aortic ulcer (PAU) and has a ‘lethal triad’; it is rare, has high mortality, and presents in atypical ways.1 AAS affects approximately 4000 people in the UK per annum,2 many not receiving timely diagnosis and treatment, and is responsible for 43 000–47 000 deaths annually in the USA.3 Population-based studies suggest the mean annual incidence of AD ranges from 64 to 7.25 per 100 000 person-years, while a US study which included all AAS (AD, IMH and PAU) found an incidence of 7.7 per 100 000 person-years.6 The misdiagnosis rate is estimated to be 33.8%,7 with diagnostic delay of up to 24 hours for 25% of cases,2 and mortality follows a linear increase of 0.5% per hour in the first 48 hours.8
Chest pain is the most common presenting symptom of AAS (80%) although back and abdominal pain are not uncommon.9 Chest pain is responsible for 7.6 million annual visits to EDs in the USA10 and collectively, chest, back and abdominal pain accounts for over 2 million ED attendances annually in England11 and are overwhelmingly due to causes other than AAS. The estimated incidence of AAS is one in every 980 ED atraumatic chest pain attendances,12 thus creating a substantial diagnostic challenge. Too low a threshold for performing a CT aorta angiogram (CTA), the gold standard for diagnosis, would result in low diagnostic yields,13 14 significant costs and risks of ionising radiation. Clinicians therefore need to use CTA selectively, yet there is no validated clinical decision tool (CDT) for this scenario.
Several CDTs have been proposed.15–17 However, they have been tested in patients undergoing CTA. Additionally, D-dimer has been suggested as a rule-out biomarker in low pre-test probability patients (95–98% sensitivity)18 19 and has been incorporated into the Aortic Dissection Detection Risk Score (ADD-RS) CDT. No CDT has previously been studied in truly undifferentiated ED populations. It is important to test possible CDTs and biomarkers in undifferentiated ED patients, because ED clinicians are likely to apply them to all patients with possible AAS rather than just those selected for CTA. It is currently unclear whether any AAS CDTs have sufficient sensitivity to be acceptable to clinicians, which is the most accurate, and whether they are likely to lead to CTA and D-dimer over-investigation. Assessment of CTA rate versus CTA positivity has not previously been studied in the clinically relevant population.
With these challenges in mind, we aimed to describe the characteristics of ED attendances with possible AAS and to assess existing CDTs and use of CTA in an ‘all-comer’ cohort of patients.
Methods
This was a multicentre observational cohort study of ED patients with symptoms potentially attributable to AAS. The primary objective was to establish the characteristics and performance of existing clinical decision tools, including ADD-RS,15 Canadian guideline,16 AORTAs17 and Sheffield (Ben Loryman, personal communication, 30 September 2021), in this cohort of patients. Secondary objectives were to establish patient characteristics, CTA rates and patient enrolment at participating sites.
This study was conducted in 27 EDs in England, Scotland and Wales. A pragmatic approach to maximise recruitment was taken, with each ED including eligible patients for a consecutive period of between 2 and 40 days in autumn 2022. Data from each patient’s attendance was entered onto a standard Case Report Form (CRF)online supplemental appendix 1, with subsequent 30-day outcome data captured from the Electronic Patient Record (EPR).
Supplemental material
People aged 16 years or over, attending the ED with new-onset symptoms of possible AAS were eligible for inclusion. New onset was defined as starting within the past 7 days; and possible AAS symptoms included chest, back or abdominal pain, syncope or symptoms related to malperfusion. The only exclusion was the absence of any potential AAS symptoms. Patients transferred from other centres were included. Patients were either identified prospectively by the treating clinician, prospectively by the local study team reviewing real-time ED attendance data to identify presentations of chest, back or abdominal pain, syncope or symptoms related to malperfusion, or retrospectively by the local study team, where local legal and ethical consent processes allowed. Data was collected by either the treating clinicians or the local study team. Where patients were identified prospectively, either the treating clinician commenced the CRF or was approached by the study team as soon after the consultation as possible, to establish their clinical suspicion of AAS before any confirmatory testing took place. Retrospective patient identification was done from daily searches of the EPR using an ED presenting problems of chest, back or abdominal pain, syncope or symptoms related to malperfusion and radiology records of ED-requested CTAs during the study period. This enabled collection of an accurate picture of the epidemiology and management of patients attending the ED with symptoms of AAS, including at weekends and out of hours when research staffing was often reduced relative to daytime hours.
For patients identified prospectively, the treating clinician was also asked about their clinical suspicion of AAS (Yes/No), with likelihood from 0 to 10, and whether they thought AAS was the most likely diagnosis (Yes/No). If the treating clinician thought that there was a negligible likelihood of AAS, the patient was still enrolled to allow us to assess what proportion of presentations with symptoms possibly associated with AAS had no clinical concern for AAS. Clinician impression was recorded by the treating clinician at the time of reviewing the patient to ensure they were not influenced by any laboratory or radiological results. There was no change to usual clinical care and no study specific interventions for participants.
Anonymised patient data was uploaded to an electronic CRF (eCRF) (online supplemental appendix 1), sited on an online secure database (Research Electronic Data Capture (REDCap); http://www.project-redcap.org) on a University of Edinburgh server.20 21 No participant identifiable data was entered onto the eCRF, left the local hospital or was viewed outside of the clinical care team. After eCRFs were completed, hospital/study number linkage was destroyed. Data was recorded for patient demographics (age, sex), attendance date and time and for all characteristics of the clinical decision tools being evaluated. Clinical data included time of onset and features of any pain, relevant medical or family history, examination findings, results from investigations and suspected diagnosis. Diagnoses and data were not adjudicated. Data was not validated but was cleaned by the central analysis team with any data queries being addressed where possible by the site study teams.
Study endpoints
The proportion of patients in whom the ED clinician thought AAS was a possible differential diagnosis, and most likely diagnosis, who had confirmed AAS.
The proportion of patients in whom the ED clinician thought AAS was not a possible differential diagnosis but had confirmed AAS
Test characteristics of clinical acumen, ADD-RS, AORTAs, Canadian and Sheffield AAS clinical decision tools and D-dimer (separately and in combination).
CT/CTA ordering and positivity rate.
Proportion of alternative diagnoses found on CT/CTA and final hospital diagnosis.
Median time from hospital presentation to imaging diagnosis.
Median time from symptom onset to hospital presentation.
30-day outcome (including 30-day mortality in proven AAS) through EPR.
Outcomes
Each local study team kept a record on a password protected National Health Service computer linking the patient’s hospital number with their study number so that 30-day outcome data could be collected from the EPR and entered onto the eCRF noting any diagnosis of AAS, final hospital diagnosis and mortality. Reference standard was radiological or operative confirmation of AAS.
Sample size calculation
The sample size calculation was based on the expected width of 80% CIs. We estimated that approximately 5000 eligible patients would attend during the study period, of whom 125 would undergo CTA and 6 would have confirmed AAS. This would provide sufficient power to estimate key measures with an acceptable degree of precision, for example, 0.12% prevalence of confirmed AAS (80% CI 0.06% to 0.21%); 2.5% prevalence of CTA use (80% CI 2.2% to 2.8%). We used 80% CIs for the key proportions so that we were not reporting intervals that were too wide to be informative, while also enabling us to identify the most likely range of values for the true proportion with reasonable certainty. Indeed, 80% CIs are consistent with the intervals that are usually reported in pilot and feasibility studies.22 However, for the diagnostic accuracy analysis of CDTs, we used the conventional 95% CIs.
Statistical analysis
A descriptive analysis was performed. Categorical variables were summarised using frequencies and percentages and continuous variables using medians and IQRs. Exact binomial 80% CIs were constructed around key proportions such as the proportion of patients in whom the clinician considered AAS was not a possible differential.
CDTs were assessed for diagnostic accuracy, taking a confirmed AAS diagnosis as the reference standard. Reference standard results were not available to those providing data to derive the CDTs, and conversely, information on decision tools were not readily available to those making a confirmed AAS diagnosis. The following performance indices were calculated (all with exact binomial 95% CIs): sensitivity, specificity, positive predictive value and negative predictive value.
Total scores were calculated for the CDTs by using the published criteria.15–17 If more than 50% of the individual dimensions of the CDT had more than 50% missing data, then a total score was not calculated. Where scores were able to be calculated, any missing data were scored as 0.
The presence of missing data in the total scores and reference standard (confirmed AAS diagnosis) may have caused bias in the performance indices. We therefore calculated the performance indices according to three methods: (1) missing data was excluded and only valid data was analysed, (2) all missing data in the numerator was assumed to take a value of 0 (indicating a negative result); and (3) all missing data in the numerator was assumed to take a value of 1 (indicating a positive result). Thus, we were able to assess the dependence of each performance index on our assumptions about the missing data.
Receiver operating characteristic (ROC) curve analysis was performed for the clinical decision tools (and clinical acumen, ie, ED clinician rating as to likelihood of AAS before confirmatory testing) based on their raw ordinal scores, excluding any missing data. The area under the ROC (AUROC) curve was calculated for each decision tool with 95% CIs.
SPSS V.27 (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, V.27.0. Armonk, New York, USA: IBM Corp) was used to produce the graphs, and R software V.4.2.1. was used to calculate the exact 80% CIs.23 All other analyses, including descriptive analyses were performed using SAS software V.9.4 (SAS Institute, Cary, North Carolina, USA).
Results
Between 26 September 2022 and 30 November 2022, 5548 patients presented to 27 EDs during their recruitment window, with symptoms potentially attributable to AAS (online supplemental figure 1A). Figure 1 details participant recruitment and online supplemental table 1 details recruitment at each of the participating EDs. Data on ED presentations (excluding minor injuries) over 16 years of age were available for 464 of the 599 (77.5%) recruitment days (60 381 presentations; mean 130.1 presentations per day), meaning there were an estimated total at all sites of 77 949 adult major presentations during the 599 study recruitment days. The study teams identified 2037 (37%) patients prospectively and 2688 (48%) patients retrospectively through EPRs or other searches, and the method of recruitment was unknown in 823 (15%) patients. Online supplemental figure 1B details the number of patients attending hospital per hour of the day, stratified by type of recruitment). Recruitment was similarly distributed throughout the 24-hour period with a slightly greater proportion of prospective recruitment during ‘office’ hours (online supplemental figure 2).
Supplemental material
Supplemental material
DAShED participant recruitment.
The median age was 55 years (IQR 37–72; n=5539); 2591 (47%) patients were men (table 1). Table 2 details the clinical findings for the enrolled study population. Pain was described as sudden onset in 743 (15%), severe or worst ever in 1547 (32%), migrating or radiating in 1752 (34%) and 1609 (31%) had hypertension in ED, indicating that CDTs using these criteria will have high rates of positivity when using a low threshold.
Baseline characteristics of study population including history, past medical history, and physical examination findings of study population
Results of investigations and outcome of study population
Table 3 details the results of investigations and outcome of the study population. Physician gestalt was obtained in 4111 (74%) patients prior to confirmatory testing (ie, CT/D-dimer). AAS was considered a possibility by the clinician in 1082 (24%) patients but only 407 underwent CTA. Fourteen patients (0.3%) were confirmed to have AAS: five had a Stanford type A aortic dissection, three had a Stanford type B aortic dissection and six an intramural haematoma or penetrating ulcer. Median (IQR) time from ED attendance to confirmed AAS was 6 hours (3–63; n=13); one patient was diagnosed on CT scan performed prior to their ED attendance for another indication which resulted in their ED attendance and is not included in the analysis of time to diagnose AAS. A second patient had a CT scan prior to ED attendance which was misreported and was only diagnosed after ED attendance followed by further scan review by radiology and is included. Of note, this is the only patient who was diagnosed with AAS who had no high-risk condition, no high-risk pain feature, no high-risk examination features and a normal chest X-ray; CT showed metastatic cancer with an incidental mural thrombus and penetrating thoracic aortic ulcer. Four patients had a diagnostic delay over 24 hours from the time of ED arrival. This included the patient diagnosed on scan re-review.
Clinician impression of AAS
Thirty-three other patients had alternative aortic pathologies (4 ruptured thoracic aortic aneurysms, 5 ruptured abdominal aortic aneurysms, 21 non-ruptured thoracic or abdominal aortic aneurysms and 3 previously known stable aortic dissection/intramural haematoma or penetrating ulcer). Thirty-one (94%) of these were identified in the ED. This group had a 26% (9/33) 30-day mortality. No patients in our cohort were found to have been discharged with undiagnosed AAS at 30-day follow-up.
Table 4 and figure 2 detail the summary test characteristics of clinical acumen, CDTs, and D-dimer (both separately and in combination) (online supplemental table 2 also includes sensitivity analyses). Brackets denote the range of possible values of sensitivity and specificity if all data that is currently missing in the test scores had been observed. AUROC curve for ED clinician AAS likelihood rating was 0.958 (95% CI 0.933 to 0.983, n=4006) and for D-dimer was 0.658 (95% CI 0.466 to 0.850, n=644). AUROC for individual CDTs were: ADD-RS 0.674 (95% CI 0.508 to 0.839, n=4989), AORTA 0.689 (95% CI 0.527 to 0.852, n=5132), Canadian 0.818 (95% CI 0.686 to 0.951, n=5180) and Sheffield 0.628 (95% CI 0.467 to 0.788, n=5092) (online supplemental figure 3).
Summary test characteristics of clinical acumen, clinical decision tools, and d-dimer (separately and in combination).
Stacked bar chart of numbers attending hospital per hour of day (stratified by prospective/retrospective recruitment).
Discussion
This study illustrates current real-world management of AAS in EDs, and highlights the diagnostic difficulty facing ED clinicians and the limitations of methods for selecting patients for CT. Most striking is the number of patients presenting with potential AAS symptoms, who did not have AAS (99.7%). Despite the low AAS prevalence, 10% with potential AAS symptoms underwent CT and 7% underwent CTA. The median time from ED arrival to confirmed AAS diagnosis was 6 hours but ranged from just over 2 hours to almost 11 days. A third of proven patients with AAS endured a diagnostic delay over 24 hours from time of ED arrival, more than the 25% reported in the literature.2 With mortality increasing per hour of delay,8 there is room for improvement in the management of potential AAS in the ED.
Ten per cent of patients with potential AAS symptoms under went CT chest of any type (2% of scans positive for AAS). All ED-requested CT scans diagnosing AAS were CTAs (2.9% positive rate for AAS). This is comparable to previous reported figures; a North American retrospective series of patients undergoing CTA for suspected AAS, reported a prevalence rate of AAS on CTA of around 3%.24 It should be noted that 40% of all CT scans detected alternative diagnoses (201 of 503), the most common are detailed in table 2. Clinicians need to use CT selectively yet be comfortable deciding which patients presenting with AAS symptoms do not require further investigation with CTA.
It may be that CTA rate was increased (407/5446; 7%) from that anticipated in the power calculation (125/5000; 2.5%) due to the presence of this study in the ED biasing ED clinicians to over investigate for AAS. However, this increased CTA rate may also reflect a more widespread generalised change in usual care towards more ED clinicians investigating potential AAS symptoms with CTA. This study shows the potential implications of over-interpreting the recommendation that all patients presenting with potential AAS symptoms should undergo CTA. A literal interpretation could have led to 5008 further patients undergoing CT scan in our study. The Aortic Syndrome Evidence Synthesis (ASES) study, an evidence synthesis and value of information analysis is currently underway, determining what CT positivity rate would represent a cost-effective use of resources.25
While clinician gestalt appears to perform well, a sensitivity of 45% when an ED clinician rates AAS as the most likely diagnosis, suggests additional help to stratify who should undergo CT is required. While our low AAS prevalence means we must be cautious about comparing the performance of different clinical decision tools, the Sheffield score (sensitivity of 36%) is probably not suitable for clinical use, with too many missed AAS cases. The other clinical decision tools performed better, but alone did not reach sufficient sensitivity for clinical use in our cohort. Table 2 potentially explains the poor specificity of the CDTs. The ADD-RS and Canadian tools both include ‘Pain severe intensity or worst ever’ and ‘migrating or radiating pain’. These characteristics are present in 32% and 34% of people with potential AAS symptoms in our cohort, yet not all underwent CT. This suggests that clinicians are choosing not to CT all patients with these symptoms, and that CDTs need to better differentiate which patients with severe intensity, worst ever, migrating or radiating pain should undergo CT.
While the AUROC for ED clinician AAS likelihood rating was impressive compared with individual CDTs, this must be interpreted with caution. AUROC for clinician likelihood was based on an ordinal score with wider range than for all the CDTs. The clinician likelihood also had a much higher proportion of missing data (26% compared with 4–7% in the other decision tools). Finally, the study taking place in the ED with ongoing recruitment of all-comers with potential AAS symptoms may have biassed towards a higher accuracy of clinical gestalt and may have reduced the risk of ED clinicians missing AAS. Nevertheless, the finding that an ED clinician AAS likelihood of 3/10 or greater detected all AAS cases, suggests that ED clinician gestalt could be a useful addition to any AAS CDT. Currently only the Canadian clinical practice guideline includes any measure of ED clinician gestalt, and this clinical decision tool performed well in our evaluation.
In our study, around half of patients could not be recruited prospectively despite extensive advertisement, excellent site engagement and acceptance among the ED community that AAS is a top emergency medicine research priority.26 27 Conducting research in this area is challenging. Some patients with AAS are missed because the diagnosis is not considered. These patients’ care will not be improved by prospective ED research studies as they will not be included. Our data is limited by the fact that it is not possible to capture patients with AAS where the diagnosis was not considered, who subsequently died without imaging or post-mortem. If a diagnostic intervention is researched in those whom the ED clinician suspects AAS, this risks the Hawthorne effect.28
This study has limitations. Retrospective recruitment of around half of patients led to missing clinician gestalt data. D-dimer and CT scans were only available if they were ordered by the treating clinician. Generating accurate test characteristics of decision tools was therefore difficult, especially around D-dimer estimation. Here, exclusion of unknown and missing values may have led to bias in the test characteristics. However, in our analyses we carefully considered the dependence of the test characteristics on the missing data and created ranges to show the possible range of test characteristics depending on the unobserved data.
The strength of this study is the recruitment of a clinically relevant cohort. However, 90% of participants did not undergo CTA and it is possible that there were missed cases of AAS. Many patients who had D-dimer and CT were being primarily investigated for pulmonary embolism. Because of a likely higher prevalence of AAS in this group, the test characteristics of D-dimer may not reflect their performance in the entire cohort. While the CDT variables were collected prospectively where possible, the overall score was determined during analysis. Any definite study of CDT performance would need to apply the CDT at the time of ED clinician decision-making.
Previous studies of AAS clinical decision tools have included a very different population to the one we have looked at. All studies previously evaluating the accuracy of ADD-RS for identifying AAS have restricted inclusion to those patients undergoing CTA, using CTA as the reference standard.29–36 We have included everyone presenting with potential AAS symptoms, illustrating the current real-world management of AAS in EDs, the diagnostic difficulty facing ED clinicians and the limitations of methods for selecting patients for CT. Despite the difficulties conducting research in such a difficult environment at a time of the most extreme pressures on the clinical service,37 we have maximised recruitment and produced a study with maximal generalisability with the ability to shape ED research and improve patient care in this area in future. The best decision aid to facilitate decision to CT and to outperform ED clinician gestalt in AAS is not yet clear. Further research is required in truly undifferentiated ED populations such as these.
Data availability statement
Data are available upon reasonable request. All study data and metadata will be available upon reasonable request to dashedstudy@gmail.com.
Ethics statements
Patient consent for publication
Ethics approval
The study was approved by the South-Central Oxford C Research Ethics Committee on 28 June 2022 (REC reference: 22/SC/0219). A waived consent process was approved by the Ethics Committee and by HRA.
Acknowledgments
Thank you to all the sites and study teams for making this study possible. Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at the University of Edinburgh. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for data integration and interoperability with external sources.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Richard Body
Twitter @mattreed73
Collaborators On behalf of the DAShED investigators * (co-applicants, site principal investigators and study teams): DAShED Co-Applicants: Catherine Fowler (The Aortic Dissection CharitableTrust (TADCT)); Ben Loryman (Sheffield Teaching Hospitals NHS FT); Graham Cooper (The Aortic Dissection Charitable Trust (TADCT)); Rob Hirst (Trainee Emergency Research Network (TERN)); Robert Hinchliffe (University of Bristol); Steven Thomas (Sheffield Teaching Hospitals, NHS Foundation Trust). DAShED Site Principal Investigators: Benjamin Clarke (Emergency Medicine Research Group Edinburgh (EMERGE), Royal Infirmary of Edinburgh, Edinburgh, UK); Edd Carlton (North Bristol NHS Trust); Alasdair Corfield (Royal Alexandra Hospital, NNHS Greater Glasgow and Clyde); David Lowe (Queen Elizabeth University Hospital, NHS Greater Glasgow and Clyde); Angelo Giubileo (Queen Elizabeth Hospital NHS Foundation Trust, King’s Lynn); Rajendra Raman (Victoria Hospital Kirkcaldy); Oliver McNab (Manchester Royal Infirmary, Wythenshawe Hospital, Manchester University NHS FT); Amir Guirguis (James Paget University Hospital NHS Foundation Trust); Mark Mitchelson (Raigmore, Inverness); Punithan Annamalai (Harrogate hospital); Deon Louw (John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust, Horton Hospital, Oxford University Hospitals NHS Foundation Trust); Patrick Aldridge (Frimley Park Hospital, Frimley Health NHS Foundation Trust); Sarah Wilson (Wexham Park Hospital, Frimley Health NHS Foundation Trust); Salini Zaini (Royal Berkshire NHS Foundation Trust); Valentino Oriolo (Bristol Royal Infirmary); Manoj Viegas (Bedfordshire Hospitals NHS Foundation Trust; Luton &Dunstable University Hospital); Steve Goodacre (Sheffield Teaching Hospitals NHS Foundation Trust); Dr Divyansh Gulati (Milton Keynes University Hospital NHS Foundation Trust); Susie Hardwick (Addenbrookes); Mayukhmoy Maiti (Royal Glamorgan Hospital, Cwm Taf Morgannwg University health board); Dr Hamzah Azzam (Hywel Dda University Health Board); Graham Johnson (University hospitals of Derby and Burton NHS Foundation Trust); Larisa Chirila (Kettering General Hospital). DAShED Site Study Teams: Rachel McLatchie, Matt Reed, Nicola Freeman, Alicia Cowan, Benjamin Clarke, Salma Alawiye, Thomas Cox, Joe Sharpe, Harriet Jennings, Rosie Andrews, Lorenzo Zucchelli, Evie Picton, YsabelleThackray, Fiona Allan, Hanmo Zhang, Clare Moceivei, Sam Covins, Alba Garcia-Mendelez, Amina Kaddouri, Thomas Connor, Ellen Voute, Kathryn Greenslade, Adam Stanley, Eilish Murphy, Ciaran Gunn, Harriet Cole (Emergency Medicine Research Group Edinburgh (EMERGE), Royal Infirmary of Edinburgh, Edinburgh, UK); Ed Carlton, Nicola Maivel, Sanjay Patel, Emma Mullen, Lucy Rivers, James Foley, Nima Maleki, Jessica Hughes, Sophia Reynolds, Sinead Brady, Rosemarie Bird, Harry Quin, Ruth Cockburn, Karthikeyini Suresh (North Bristol NHS Trust, Bristol, UK); Alasdair Corfield, Natalie Rodden, Lynn Abel, Susan Currie, Rachel Keen, Natasha Parker (Royal Alexandra Hospital, NHS Greater Glasgow and Clyde, UK); David Lowe, Elisha Khoo, Kareem Austin, Rhys Worgan, Malcolm Gordon, Mark Wilson, Kayleigh Doran, Ornagh O'Shaughnessy, Lisa Finnen, Courtney Mcarthur (Queen Elizabeth University Hospital, NHS Greater Glasgow and Clyde, UK); Ayesha Malik, Bhaskar Sarvesh, Fatima Ismail, Jarno Holopainen, Joanna Rudnika, Mohamed Nasr, Oluwaseyi Okewole, Saranya Kathrivel, Wickramanayake Udara, Mohamed Imam, Angelo Giubileo, Tracey Fuller, Hammed Tiamiyu (Queen Elizabeth Hospital NHS Foundation Trust, King’s Lynn, Norfolk, UK); Rajendra Raman, Saqib Parwez, Keith Boath, Heather Robertson, Amy McCallum, Kyle McLachlan, Naomi Gunn, Cristina Sifiringa, Claire Stewart, Jennifer Wood, Rhys Greig (Victoria Hospital, Kirkcaldy, Fife, UK); Anila Sukumaran, Charlotte Taylor, Melanie Barker, Richard Clark, Deborah Paripoorani, Rose Jama, Oliver McNab, Agnieszka Rozanska, Melanie Barker, Dr Govind Oliver (Manchester Royal Infirmary and Wythenshawe Hospital, Manchester University NHS Foundation Trust, Manchester, UK); Amir Guirguis, Dr Magdi Mohammed, Dr Likaa Gabr, Dr Katie Landles, Dr Favour Balogun, Dr Rafiq Abdul,Helen Sutherland, Charlotte Kelly (James Paget University Hospital NHS Foundation Trust, Great Yarmouth, UK); Mark Mitchelson, Emuyemiken Egwuelu, Fiona Barrett (Raigmore, Inverness, UK); Punithan Annamalai, Caroline Bennett, Cheryl McDermott, Joyce Guy (Harrogate Hospital, North Yorkshire, UK); Deon Louw, Alex Novak, Alexis Espinosa, Gianni Tam-McMillan, Antoni Krupa, Tabitha Thorton-Swan, Sophie Roberts, Samuel Mills, Abdala Trinidad Espinosa-Morgado, Sally Beer, Elizabeth Hatton, Georgy Corby, Iris Mae Morse, Lauren Scullion, Ayan Shahid, Jessy Jindal, Aisling Curtis, Jessica Morgan, Joseph Layzell (John Radcliffe, Oxford, UK, Horton Oxford University Hospitals NHS Foundation Trust, UK); Patrick Aldridge, Sana Nadeem, Shalini Ravishankar, Hannily Cooke, Jodie Clarke, Jaskiran Dulai, Sinead Helyar, Myra Dinglasan, Durga Ponnaganti (Frimley Park Hospital, Frimley Health NHS Foundation Trust, UK); Sarah Wilson, Joana Da Rocha, Karen Chivers, Francoise Ticehurst, Louise Gardner, Vani Muthusami, Marie Broyde, Zoe Foster, Nattakarn Limphaibool, Victoria Tay, Shihana Saifudheen, Arane Prabakaran, Shahab Manouchehri, Faluade Akinroye, Rahul Shah, Joseph Mahony, Amrit Gherra, Rob Conway (Wexham Park Hospital, Frimley Health NHS Foundation Trust, UK); Liza Keating, Salini Zaini, Emma Grierson, Shauna Bartley,Nicola Jacques, Giulia Mascia, Francesca Compton, Gabriela Tomkova, Sabi Rai, Kuldeep Nijjar, Edward Neary, Jonathon Lane, Kin Yung Lam, Mohammed Albakoush, Tejas Harit, Silvia Panetta, Mae Gill, Carmen Powell, Erin Robinson, Helen Callard, Hameed Qureshi, Hira Naeem, Laura Hudson, Osama Akrama (Royal Berkshire NHS Foundation Trust, Reading, UK); Emma Redfern, Casandra Bazanlacerot, Rachel Shiel, Christina Coleman, Eva Maria Hernandez Morano, Eleanor Daniel, Kim Wright, Catherine Macleodhall, Denise Webster, Valentino Oriolo, Katie Sweet, Josie Bonnici, Irene Grossi, Carl Robinson, Omar Almousa, Jonathan Benger (Bristol Royal Infirmary, Bristol, UK); Manoj Viegas, Charlotte Whitehouse, Leigh Bissett, Fredrico Miranda, Dinkar Bisht (Bedfordshire Hospitals NHS Foundation Trust; Luton & Dunstable University Hospital, UK); Steve Goodacre, Hannah Walker, Jaden Groves, John Banner, Sayantani Mukhopadhyay, Yasmin Mahmud, Charles Wong, Shamiso Muza, Oliver Salwey (Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, UK); Divyansh Gulati, Hina Tahir, Santos Oshiotes, Devika Chaudhri, Wuraola Akande (Milton Keynes University Hospital NHS Foundation Trust, UK); Susie Hardwick, Katie Coupe (Addenbrooke’s Hospital, Cambridge, UK); Lisa Roche, Mayukhmoy Maiti, Eleanor James (Royal Glamorgan Hospital, Cwm Taf Morgannwg University Health Board, UK); Hamza Azzam, Janet Phipps, Catherine MacPhee, Michelle Edwards, Hannah Blowing, Samantha Coetzee, Lucy Hill, Jill Williams, Becky Icke, Bethan Landeg, Helen Tench, Heather McGuinness, Tanya Sims, Rebecca Wolf-Roberts, Ronda Loosley, Maddie Moores, Kimberley Walker (Hywel Dda University Health Board, (Withybush, Bronglais, Glagwilli), UK); Elisha Cousins, Graham Johnson, Alison Rockey, Alison Fletcher, Lianne Wright, Lucy-May Moulden, Tom Ward, Andrew Tabner, Sarah Allen, Hannah Phillips, Obiageli Ngwuocha (University Hospitals of Derby and Burton NHS Foundation Trust, UK); Larisa Chirila, Hannah Britton, Julimon Sebastian, John Obiakor, Nathan Hancock (Kettering General Hospital, Northamptonshire, UK).
Contributors RM, MJR, NF, RAP, SW, SG, AC, JB, BC and EC conceived the study and were involved in study design and study delivery. MJR and RAP performed data analysis. RM, MJR, NF, RAP, SW, SG, AC, JB, BC and EC were involved in data interpretation and RM, MJR, NF, RAP, SW, SG, AC and EC prepared the manuscript. RM, MJR, NF, RAP, SW, SG, AC, JB, BC and EC contributed to its revision. MJR takes responsibility for the paper as a whole and acts as guarantor, accepts full responsibility for the work and the conduct of the study, had access to the data, and controlled the decision to publish. The DAShED investigators were involved in the acquisition of data for the work.
Funding This study was supported by a Royal College of Emergency Medicine (RCEM) research grant. MJR is supported by an NHS Research Scotland Career Researcher Clinician award.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.