Pandemic Respiratory Infection Emergency System Triage (PRIEST) Study

The Pandemic Respiratory Infection Emergency System Triage (PRIEST) study is a National Institute for Health Research (NIHR) funded project aimed at evaluating and optimising the triage of people using the emergency care system (111 and 999 calls, ambulance conveyance, or hospital emergency department) with suspected respiratory infections during the COVID-19 pandemic.

By the 17th of June 2020 231,889 people in the UK had been confirmed to have been infected with the COVID-19 virus and over 40,000 people had died because of infection. At the peak of the pandemic NHS 111 received nearly twice the normal number of calls and over 3000 patients were admitted to hospital daily in the UK due to COVID-19 infection.

It is currently unknown how safely and effectively the emergency care system (NHS 111, the ambulance service and hospital Emergency Departments) assessed patients with suspected COVID-19 infection during the pandemic and determined whether patients needed to attend hospital or required hospital admission. There are currently no validated evidence-based risk stratification tools that can be used by clinicians in the pre-hospital and Emergency Department environments to identify patients at higher risk for deterioration with suspected COVID-19 infection who require further assessment.

In order to accurately assess whether patients with suspected COVID-19 infection who accessed the emergency care system during the pandemic had serious adverse outcomes it is necessary to link prehospital and Emergency Department cohorts with the proposed Emergency Care data Sets, Hospital Episode Statistics and Mortality data. Linking pre-hospital cohorts to GPES Data for Pandemic Planning and Research is also necessary to accurately identify patient factors associated with serious adverse outcomes.

Given the possibility of a second peak of COVID-19 infections in the UK as “lock-down” measures are relaxed the proposed research is urgently needed. Early analysis of the pre-hospital and Emergency Department cohorts will identify the characteristics of any patients advised to self-care at home who subsequently deteriorated and patients who are at high risk for serious adverse outcomes.

Study Governance

The data controller is the University of Sheffield. The study sponsor is Sheffield Teaching Hospitals NHS Foundation Trust. The study funder is the National Institute for Health Research. The study has been recognised as a nationally prioritised study in response to the COVID19 pandemic.

Background

The term triage is often used to describe a brief initial assessment in the emergency department to determine patient order of priority in the queue to be seen. In this project, the research team will use the term triage more broadly to include the full process of emergency department and pre-hospital assessment (by 111 and the ambulance service) used in decision-making regarding whether patients should attend hospital, require hospital admission or be referred for high dependency or intensive care.

On 26th March 2020, the PRIEST study began recruitment of patients with suspected COVID-19 attending Emergency Departments at participating NHS Trusts in England, Scotland, Wales and Northern Ireland. Data has been collected on over 20,000 patients. This work package (WP) is informally known as “core-PRIEST” and the data collected under this WP is known as the “core-PRIEST data”.

Between 18 March 2020 and 9 April 2020, the NHS online service completed 1,911,161 COVID-19 online assessments across England, resulting in 348,125 triages to NHS 111 or 999. Data from NHS England report an average of 95,600 calls per day to NHS 111 in March 2020, compared to an average of 46,700 a day in March 2019. Ambulance Services in England received a record number of calls per day to 999 in March 2020, possibly influenced by the COVID-19 pandemic. NHS 111 call handlers use structured questions and clinical advice to determine whether a 999 response is required. If no response is required, the patient is advised to self-care or contact their GP. If a 999 response is required, the attending ambulance personnel can use their clinical judgement to determine whether transport to hospital is needed. There has not yet been any research into the appropriateness of prehospital triage decisions with respect to patients with suspected COVID-19 and the researchers are not aware of any validated tools applicable to this situation.

Emergency department triage methods need to accurately predict an individual patient’s risk of death or severe illness. The predicted risk can then guide decision-making. Patients with a low risk may be discharged home, those with a high risk admitted to hospital, and those with a very high risk referred for high dependency or intensive care. Current risk stratification tools used in the Emergency department to help triage hospital admissions for patients with respiratory infection are based upon research conducted on patients with bacterial pneumonia and seasonal influenza. The available research indicates that no single tool performs well enough to support its sole use to inform hospital triage decisions during a pandemic and the use of these existing tools has not yet been assessed in patients with suspected COVID 19.

Research is therefore urgently needed to determine the accuracy of pre-hospital and Emergency Department triage decisions during the current COVID-19 pandemic and explore whether they could be improved.

The specific objectives during the pandemic are:

1. To report any important emerging findings regarding the performance of the emergency care triage method (or methods) used for suspected respiratory infections during a pandemic

2. To identify clinical characteristics and routine tests associated with under-triage (false negative assessment) or over-triage (false positive assessment) during a pandemic

3. To determine the discriminant value of alternative triage methods for predicting severe illness in patients presenting with suspected respiratory infection during a pandemic

4. To inform policy makers and practitioners during a pandemic of the study’s emerging findings.

The specific objectives after the first wave and, potentially for subsequent waves, of the pandemic are, for the hospital (emergency department):

1. To determine the discriminant value of emergency department triage methods for predicting severe illness in patients presenting with suspected pandemic respiratory infection

2. To determine the accuracy of presenting clinical characteristics and routine tests for predicting severe illness

3. To determine the independent predictive value of presenting clinical characteristics and routine tests for severe illness

4. To develop new triage methods based upon presenting clinical characteristics alone or presenting clinical characteristics, electrocardiogram (ECG), chest X-ray and routine blood test results, depending upon the data available and the predictive value of variables evaluated in objective 3

The specific objectives after the first wave and, potentially for subsequent waves, of the pandemic are, for prehospital services (NHS 111 and emergency ambulance services):

1. To link NHS 111 calls, identified as potentially relating to COVID19, to participating hospital and NHS Digital data, to determine whether patients calling NHS 111 were appropriately advised or provided with an ambulance response, in terms of whether they were admitted to hospital or suffered an adverse outcome.

2. To link ambulance ePR data to hospital and NHS Digital data, to determine whether patients attended by ambulance were appropriately advised to self-care at home or transported to hospital, in terms of whether they were admitted to hospital or suffered an adverse outcome.

3. To use ambulance ePR data recording patient characteristics to determine which patient characteristics, when recorded prehospital, are useful in predicting adverse outcome and determine the discriminant value of early warning scores, such as NEWS2, for predicting adverse outcome.

4. To explore the potential for data mining to provide new insights into the prediction of adverse outcome among patients contacting NHS 111 or ambulance services with suspected COVID-19.

Data is being processed under Article 6(1)(e) and Article 9(2)(j) as a task in the public interest as developing accurate risk stratification tools which are fair, robust, reproducible and allows the rapid identification of low risk patients with suspected COVID-19 infection who can safely self-care at home would help to mitigate the risk of services becoming overwhelmed and adverse patient outcomes.

Weekly descriptive analysis of the cohort of “core-PRIEST data” (patients attending participating Trust’s EDs with suspected COVID) is being undertaken including:

1. The number and geographical distribution of new cases

2. The proportion with an adverse outcome and details of adverse outcomes

3. Potential predictor variables identified in patients who were not admitted at initial presentation but had an adverse outcome

4. Triage criteria identified in patients who were admitted to hospital and had no adverse outcome.

The “Prehospital-PRIEST” work package (within the PRIEST study) will link:

(A) the core-PRIEST data from participating NHS Trusts in England only;

(B) computer aided dispatch [CAD] and electronic patient record [ePR] data from Yorkshire Ambulance Service NHS Trust (YAS), as the emergency/urgent ambulance service provider for Yorkshire, on patients who: were identified (by attending ambulance service personnel) with confirmed or suspected COVID-19; or, were the subject of a call to the ambulance service’s Emergency Operations Centre which was managed according to the Advanced Priority Medical Despatch triage card 36 (a pandemic triage process for patients with suspected COVID);

(C) NHS111 telephone triage [NHS111] data from Yorkshire Ambulance Service NHS Trust (YAS), as the NHS111 telephone service provider for Yorkshire and Humber, on patients who received a COVID-19 related final disposition;

(D) Hospital Episode Statistics: Admitted Patient Care [APC] from NHS Digital on patients identified in (A), (B) or (C);

(E) Hospital Episode Statistics: Critical Care [CC] from NHS Digital on patients identified in (A), (B) or (C);

(F) Emergency Care Dataset [ECDS] from NHS Digital on patients identified in (A), (B) or (C);

(G) Demographics [DEMO] from NHS Digital on patients identified in (A), (B) or (C);

(H) Death Registration [DR] from NHS Digital on patients identified in (A), (B) or (C);

(I) General Practice Extraction Service (GPES) Data for Pandemic Planning and Research [GDPPR] from NHS Digital on patients identified in (A), (B) or (C);

(J) Categorisation of place of residence [RESI] data (e.g. care home) from NHS England for patients identified in (A), (B) or (C).

Historic GPES Data for Pandemic Planning and Research (GDPPR) data, HES data and demographic has been requested to obtain more complete information for patients in the cohort on their premorbid status- particularly pre-existing medical conditions and routine medication use. These are important potential risk factors for deterioration in COVID 19 and other acute respiratory illnesses which may not be comprehensively collected or recorded in an Emergency treatment setting. They may, however, be extremely important in identifying high risk patients who require treatment in hospital

The University of Sheffield will supply direct patient identifiers (from (A), (B), (C)) to NHS Digital to enable NHS Digital to :

(1) establish (trace) each individuals identity by comparing with the NHS Personal Demographics Service;

(2) remove all such individuals who have registered a NHS national data (type 2) opt-out;

(3) extract from amongst the requested NHS Digital held datasets the applicable records.

The data supplied to NHS Digital by the University of Sheffield PRIEST study team will contain no health (or other special category) data other than any individual appearing within the data must have had a COVID related contact with an urgent or emergency care service between February 2020 and September 2020 (inclusive). The specific direct patient identifiers supplied to NHS Digital may include:

~ NHS Number

~ date of birth

~ postcode of residence (or postcode of incident as a proxy for postcode of resident);

~ names

~ sex

For records with both a valid NHS Number and date of birth: only NHS Number and date of birth will be provided; otherwise, records will be supplied with all available direct identifiers.

NHS Digital will then link the data and return data with the following identifiers to the University of Sheffield:

~ NHS Number

~ Postcode of residence

~ Date of birth

~ Date of Death

Identifiers will then be stripped from the data and only pseudonymised data will be analysed by the research team.

Only retrospective data is requested as the cohort has been identified by the fact that they received care on the basis of confirmed or suspected COVID-19 between March and July 2020.

The University of Sheffield PRIEST study team will supply pseudonymised (securely hashed) NHS Numbers to NHS England in order for NHS England to supply pseudonymised categorisation of place of residence (J) data to the University of Sheffield PRIEST study team.

Data from the listed data sources ((A) - (J)) will be linked using a pseudonymised identifier. De-identified, pseudonymised datasets will be provisioned on separate secure virtual environments on which analysts will conduct the statistical analyses to fulfil the objectives of the research.

The University of Sheffield (UoS) seeks full DOB (and other patient demographic data, e.g. gender) from NHS Digital for the patient cohort to populate a "single source of truth" (SSOT) table of patient characteristics. All large data studies face issues with conflicting records, a SSOT means the researchers do not have to "choose" between conflicting records nor do they need to create and document an algorithm that does the "choosing". UoS will receive DOBs for the vast majority of the cohort (from participating NHS Trusts and Yorkshire Ambulance Service) for the purposes of identifying the cohort in data held by NHS Digital. Receiving full DOB for patients in the cohort from NHS Digital will provide the practical advantage of allowing the researcher to calculate age directly from a SSOT and avoid errors in the calculation of age.

Full postcode is required so the researcher can link to a range of local area social demographic predictors.

Only patients identified in the core-PRIEST data from participating English sites and/or NHS111 / CAD / ePRD data provided by Yorkshire Ambulance Service will form the cohort on which the researcher seeks data from NHS Digital.

There will be no requirement or attempt to re-identify individuals within the data.

Data processing will only be carried out by substantive employees of the data processor/controller who have been appropriately trained in data protection and confidentiality.

The University of Sheffield anticipated around ~110,000 people will be in the cohort for this study - but this will only be confirmed once the matching algorithm and subsequent linkage is carried out by NHS Digital.

Findings from the weekly core-PRIEST data analyses are reviewed by the core research team. When appropriate, these emerging findings are summarised to inform policy makers and practitioners during the pandemic on the study website (https://www.sheffield.ac.uk/scharr/research/centres/cure/priest). Findings are discussed every month with the Study Steering Committee which includes a range of clinical experts and lay members to identify any important findings which require urgent dissemination.

The results of interim analysis of the cohort of patients attending the Emergency Department with suspected COVID 19 infection are being compiled for academic publications and are projected to contribute to at least two peer-reviewed scientific articles published in high impact clinical journals such as the British Medical Journal and Lancet. Submission is planned by the end of August 2020. One article will highlight factors found to be highly associated with adverse outcomes in patients attending the Emergency Department with suspected COVID 19 infection and the other article will summarise the characteristics of patients who attended the Emergency Department with suspected COVID-19 infection during the first peak of the pandemic.

A final peer-reviewed scientific article assessing Emergency Department triage of patients with suspected COVID 19 infection is planned to be submitted by December 2020. This will present a derived and validated risk stratification tool aimed at improving Emergency Department triage of patients with suspected COVID-19 infection. This is likely to be published in a high impact clinical journal such as the British Medical Journal or Lancet using the results of analysis of the complete cohort of patients attending Emergency Departments linked to Hospital Episode Statistics and Mortality data.

Retrospective cohorts of patients with suspected COVID 19 infection assessed by prehospital services (Yorkshire and Humber NHS 111 and emergency ambulance services) linked to Emergency Care Data Sets, Hospital Episodes Statistics, Office for National Statistics Mortality data and GPES Data for Pandemic Planning and Research are planned to be derived and available for analysis by the research team for the end of September 2020.

Interim analysis of the prehospital cohorts of patients with suspected COVID-19 infection will be conducted shortly after the linked data sets are available and will be assessed by the core research team and Study Steering Committee. Any important early findings will be summarised to inform policy makers and practitioners in the pre-hospital triage of patients with suspected COVID-19 to aid the clinical management of any subsequent peaks of the pandemic in 2020 and 2021. Important early findings will be published on the study website and compiled for academic publications in peer-reviewed scientific articles published in high impact clinical journals such as the British Medical Journal and Lancet by December 2020.

Two final peer-reviewed scientific articles assessing triage of patients with suspected COVID-19 infection by NHS 111 and emergency ambulance services are planned to be submitted by February 2021. These articles will present derived risk stratification tools to aid the prehospital triage of patients with suspected COVID-19 infection. These articles are likely to be published in high impact clinical journals such as the British Medical Journal or Lancet using the results of analysis of the finalised pre-hospital cohorts of patients with suspected COVID-19 infection linked to Emergency Care Data Sets, Hospital Episode Statistics, Office for National Statistics Mortality data and GPES Data for Pandemic Planning and Research.

Outputs will only contain aggregated data with small numbers suppressed in line with the HES analysis guide.

Dissemination of early findings is planned by December 2020 in peer reviewed journal articles, reports to the New and Emerging Respiratory Threats Advisory Group and to the Study Steering Committee which includes a range of clinical experts and lay members. This would potentially inform Royal College of Emergency Medicine COVID-19 related Safety Bulletins. These are sent to all college affiliated Emergency Care Practitioners based on emerging evidence by the Royal College of Emergency Medicine and would highlight high risk groups for deterioration to relevant clinicians.

Dissemination of such information could reduce the risk of patients with suspected COVID-19 infection being inappropriately advised to self-care and deteriorating. This would be of direct benefit to users of the emergency care system in the Health and Social care sector.

During peaks in the COVID-19 pandemic emergency care services, especially pre-hospital services, are at risk of being overwhelmed. Developing accurate risk stratification tools which are fair, robust, reproducible and allows the rapid identification of low risk patients with suspected COVID-19 infection who can safely self-care at home would help to mitigate the risk of services becoming overwhelmed and adverse patient outcomes. If successfully developed accurate risk-stratification tools for patients with suspected COVID-19 infection in the pre-hospital and Emergency Department are likely to be adopted across the UK to aid the triage of patients in any subsequent peaks of the pandemic.

Using data collected on patients attending Emergency Departments in the UK with suspected COVID-19 infection during the first peak, linked to HES and ONS mortality data, the researchers may have derived, validated and disseminated a risk-stratification tool to help triage in the Emergency Department by February 2021. If the risk stratification tool is found to improve upon current triage of suspected COVID-19 patients in the Emergency Department, it will be disseminated to the New and Emerging Respiratory Threats Advisory Group which advises the Chief Medical Officer and in publication in high impact clinical journals. A successful triage tool is likely to be incorporated by NICE guidelines for use by clinicians in the triage of patients with suspected COVID-19 and change clinical practice in the Emergency Department. This would benefit the Health and Social Care sector by reducing the risk of patients being discharged who subsequently deteriorate or unnecessarily admitting patients (using scarce health service resources during a pandemic) who are unlikely to deteriorate.

If such tools were found to improve pre-hospital triage of patients with suspected COVID-19 infection results would be disseminated to the New and Emerging Respiratory Threats Advisory Group which advises the Chief Medical Officer and in high impact clinical journals. The findings could ultimately inform national clinical guidelines, such as NICE guidelines, and change clinical practice. Pre-hospital risk stratification tools could benefit the Health and Social Care sector by reducing the risk of patients being advised inappropriately to self-care at home or unnecessarily conveying patients to hospital who are at low risk of deteriorating.

Not stated in the register.