ARRISA-UK study request for secondary care data for asthma patients at participating GP practices in England and Wales

The University of East Anglia requires pseudonymised secondary care data (hospital admissions, Accident and Emergency (A&E), outpatient and critical care data, and mortality data) for a selected cohort of previously identified patients for the At-Risk Registers Integrated into primary care to Stop Asthma crises in the UK (ARRISA-UK) study.

Approximately 74,000 people with asthma are admitted to hospital and 1,150 die unnecessarily in the UK every year. Excellent drugs are available for asthma and clear advice on prescribing them that should allow asthma to be controlled in most patients. It is known that certain asthma patients are at greater risk of being admitted or dying than others and that targeting intensive support and care to these patients improves their health.

Development of the ARRISA approach began with a single-practice pilot study in 2002, which was followed by a regional, cluster-randomised trial in 30 GP practices in Norfolk (2006-2009) funded by Asthma UK (Project no. 06/047). The current ARRISA-UK cluster-randomised trial of 275 practices in England, Scotland and Wales will be a definitive study which, if it has a positive outcome, it would be expected to change UK primary care policy and practice.

In the regional study, ‘at-risk’ patients were identified within GP practices and computer-based systems were used to create pop-up alerts when these patients contacted the practice. Practice staff were trained on what to do when they see the alert. This didn’t reduce the total number of attacks but reduced the hospital admissions as more patients appeared to receive appropriate treatment for their asthma (J. R. Smith et al, Thorax, 2012, 12, 1052). Based on these promising findings the study is now undertaking a nationwide study to confirm that it is possible to improve the care of these patients in a way that is cost-effective and doesn’t affect the care of other asthma patients within GP practices. (https://www.uea.ac.uk/arrisa-uk/about).

The ARRISA-UK study is sponsored and led by the data controller University of East Anglia (UEA), with analysis to be completed at UEA who is also acting as data processor. The purpose of the study is to determine whether flagging the electronic health records of people identified as being at risk of asthma attacks and training staff on the action to take when seeing the flag reduces asthma related crisis events (defined as hospital admissions, A&E attendances and deaths) over a 12 month period. The primary outcome is the difference in the proportion of at-risk patients (as identified by a prior database search) who have an asthma-related crisis event in the 12 months from the date that alert flags go live on the computer system in the intervention practices compared to the control group practices. This is a clinically meaningful, patient centred outcome that represents the severest form of asthma attacks that are the most costly to patients and society.

The study will also find out how many people have well controlled asthma, what medications are prescribed for asthma, how often patients attend appointments and if they stop smoking. The study will calculate how much this costs and whether it improves (or interferes with) the care of other patients with asthma in the practice.

So far, the study has collected pseudonymised primary care data from participating GP practices in England, Scotland and Wales. However, this data lacks sufficient detail to answer the primary outcome (asthma-related crisis event) as this requires secondary care data. The data requested here will build a more precise picture of asthma patient outcomes during the study.

Public and patient involvement has been fundamental to all aspects of the study and follows on from PPI involvement in the previous study, a smaller regional study of 30 GP practices in Norfolk. This includes review of the end of study report. PPI contributors - including a local representative who was involved in the previous study and a National PPI representative supported by the Asthma UK Centre for Applied Research - were co-applicants on the grant submission and have helped with the study design and the patient-facing documents in particular. They are also members of the Trial Management Group and will have a crucial role in the dissemination of findings.

The plans and processes for extracting data from GP practices and linking it with secondary care data from NHS England have been reviewed by the Asthma UK Centre for Applied Research (AUKCAR) including their public and patient group, lay members of the Trial Management Group and Independent Trial Steering Committee, and members of a patient discussion forum around the proposed use of patient data without consent. Feedback from PPI representatives was used to improve fair processing materials (poster and leaflet) produced by the research team.

The ARRISA-UK study is publicly funded research being undertaken in the public interest to inform knowledge on how best to improve care for patients with asthma who are at high risk of exacerbations of their disease. The study is funded by the National Institute for Health Research's Health Technology Assessment Programme (NIHR HTA) under the Department of Health UK and has undergone rigorous peer review. The data requested is to be used in the primary analysis of the ARRISA-UK study and will be reported as aggregated results in the NIHR journal monograph series and presented in peer-reviewed publications. The results of the research may inform public guidance and decisions about funding, resources and technologies available in GP practices, and as such may help improve care for asthma patients in the future.

Processing is therefore lawful under Article 6 (1)(e) of the UK General Data Protection Regulation (UK GDPR) (processing is necessary for the performance of a task carried out in the public interest) and article 9(2)(j) (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes…).

The justification for processing the data by University of East Anglia is Article 6(1)€ of the UK GDPR: (processing is necessary for the performance of a task in the public interest or in the exercise of official authority vested in the controller).

The justification for the processing of the special category data (health data) by University of East Anglia is Article 9(2)(j) of the UK GDPR: (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

The data will not be used for commercial purposes, will not be provided in record level to any third party, and will not be used for direct marketing. All analysis will be conducted at the University of East Anglia, Norwich, UK.

Data from GP practice records is captured remotely in a manner that is completely non-identifiable to the researchers, so individual patients are not asked to give informed consent. Instead, the NHS Research Ethics Committee agreed that the Caldicott Guardian or Information Governance lead for each GP practice (who is responsible for the use of their patients’ records) could approve the conduct of the study at their practice as they had been made aware of the purpose of the study and how their patients’ data would be used. There is no anticipated harm to the public by the dissemination of the data requested. Patients who have opted-out for use of their data in research are excluded from both the primary and secondary care data requests.

With around 160,000 asthma patients across 222 participating GP practices in England, it would have been too great a burden on research and GP practice staff, and impossible in a reasonable timescale, to individually consent patients to provide this information. Instead, this study had section 251 approval in place which allowed the study to collect secondary care data from NHS England. This is coupled with data available routinely from the GP practices and organisations.

Patient identifiers (NHS number, date of birth and sex) were provided to NHS England for matching, the data disseminated by NHS England was pseudonymised. Disseminated data is required to be patient level rather than aggregated, as it will then be linked by UEA to pseudonymised primary care data via the Study ID, before being fully anonymised for analysis. Analysis of the primary and secondary outcome measures requires a patient level dataset, as the effectiveness of the ARRISA approach cannot be obtained from practice or region-level aggregated data. It is unfeasible to undertake a study with sufficient statistical power to understand the effectiveness of the study teams' intervention which is based on practice level data in the UK.

It is essential for the purposes of this study to collect primary/secondary and mortality data for the population included in the study. While the intervention is randomised at practice level, the flagging is at patient level, and study analysis will need to distinguish the outcome of patients who have been identified as at-risk vs those not at-risk. NHS England required patient identifiers to match records at a patient level. It is not possible to use an alternative study design to answer our question as there will not be sufficient statistical power and it is not possible to obtain written informed consent from patients. There are no alternative ways of collecting this data which are less intrusive. The study team have collected pseudonymised data from GP practice records, but secondary care service use is not adequately recorded in GP practice records to enable the primary outcome to be answered. By using the linkage service provided by NHS England with the proposed data flows, no identifiable data needed to be shared with the research team.

The study closed recruitment in April 2018 with 275 participating GP practices in total across England, Wales and Scotland; 35 of those are in Wales and 18 in Scotland. Each GP practice was randomised to either the intervention or control arm of the study. The study will receive GP practice record data for all asthma patients at both intervention and control practices, and the request for secondary care data covers this same group. Therefore, this data request to NHS England covered approximately 160,000 patients with a current diagnosis of asthma from 228 participating GP practices from England and border Wales.

The geographical spread of the data requested reflects the locations of the GP practices in the trial. There are participating practices in 14 of the 15 Clinical Research Network regions of England, and the request also covers six practices in Welsh local health boards near the border with England. Separate requests for secondary care data for Scotland and Wales have been made to the relevant data custodians (eData Research and Innovation Service and NHS Wales Informatics Service). This request is for secondary care data in England, and includes all of the GP practices in England participating in the study. However, participating GP practices have advised that patients registered at Welsh practices near the border with England may attend hospital in England. Therefore, this request includes six GP practices in Wales. In Scotland, GP practices participating in ARRISA-UK are not close to the border with England, and all asthma patients of interest are therefore expected to routinely be using secondary care services in Scotland only.

All patients in the study will have had asthma, according to the study definition based on primary care data, at some point in the study. Automated electronic searches will be used to identify patients with asthma: those with asthma-related codes in their electronic medical record. Some may not have had the diagnosis at the start of the study period, but may qualify for the diagnosis and inclusion in the study during the observation period. Thus the data request includes both people with current asthma at the beginning of recruitment, as well as those diagnosed with asthma throughout the recruitment period. Patients in the cohort may include those who died or left the practice during the follow-up period.

In earlier work, an algorithm was developed and validated for aiding GP practice staff in recognising those patients most at risk of being admitted to hospital or dying from an asthma exacerbation. This is based on known risk factors such as previous exacerbation history, coding for anxiety or depression, smoking history and prescribing data and laboratory results all of which are associated with more severe asthma and asthma exacerbations. This candidate at-risk list is generated by each GP practice and reviewed by their clinical asthma lead. The practice’s respiratory clinical lead can add new at-risk patients who join the practice to their register or remove patients for whom a flag on their health record would not be appropriate (e.g. if they are receiving palliative care). To meet the intended study outcomes, including the impact of the ARRISA intervention on all asthma patients at a GP practice, data from NHS England was requested for all patients with asthma, not just those deemed to be at higher risk.

Because asthma is a variable condition, the date of initial presentation with the condition, or the date(s) of presentation within any specified study dates is not relevant for the diagnosis in our study protocol. Their inclusion in the population with active asthma does depend on having a prescription issue within a year period, but no single date or date range for an asthma diagnosis or event is critical across the whole study population.

The data required from NHS England is individual level data on patient outcomes for ARRISA patients of interest (asthma patients in participating GP practices). The study require data on admitted patient care, A&E attendance, outpatient visits, critical care, and mortality for the primary and secondary outcome measures of the study. The primary outcome is the difference in the proportion of at-risk asthma patients who have an asthma-related crisis event (A&E attendance, hospitalisation or death) in the 12 months from the date the pop-up alerts go live on the computer system in the intervention practices compared to the control group practices.

The secondary outcomes include time to first crisis event, asthma control, asthma medications, attendance at appointments, medication adherence, smoking status, all cause admission and death, health care costs in all and “at-risk” asthma patients. Data contained in the Hospital Episode Statistics and mortality datasets listed above, linked with data collected in primary care, will provide the information needed for the analysis of these outcome measures.

Obtaining the requested data from NHS England is vital to the study analysis, as primary care data alone does not contain the level of detail on hospital visits and deaths needed to give an accurate picture of the effectiveness of the intervention.

By requesting data regarding all deaths and hospital episodes for the cohort, rather than just those recorded as ‘due to asthma’, the data regarding secondary care resource utilisation will be as complete as possible, and additional resources used due to poorly controlled asthma as a co-morbidity will be included. The intention is to assess all episodes for any asthma-related elements as part of the study analysis at Norwich CTU.

By including data from HES Critical Care and Outpatients, in addition to HES APC and A&E, the data regarding secondary care resource utilisation will be as complete as possible, benefiting the assessment of the effectiveness of the ARRISA study intervention. In addition, out-patient attendance data will be used to measure a key secondary outcome as part of a health economics (cost-effectiveness) analysis.

Data is requested only for the participants in the study (asthma patients at GP practices participating in ARRISA-UK). Data from all patients with asthma in a research practice is required to determine if there are any consequences of our intervention by diverting care towards an at-risk group.

Data is requested only for the relevant time period (April 2015 – September 2019). This period facilitates three years of data for all patients of interest at GP practices recruited into the study, within which the recruitment period ran from April 2016 – April 2018; practices started study activities at different times. The overall time period of the data request includes, for all practices:

- A baseline year before the practice was randomised to the intervention or control arm of the study

- A variable time period of less than one year for staff training, at-risk register creation etc

- A one year follow-up period of flags live for intervention practices or normal care for control practices.

Each data item requested has been carefully assessed and is considered to be essential for the completion of the study objectives. The study are requesting the minimum amount of data that will answer our research questions: does our intervention reduce asthma crisis events, does it improve other aspects of patient care, is it cost effective and does it adversely affect the care of asthma patients who are NOT at a high risk of an attack?

The datasets and fields requested will allow the research team to assess whether a hospital event or death is asthma-related. For example, hospital admission and A&E data includes the diagnosis and procedure data, as well as dates of admission and discharge and Healthcare Resource Group (HRG)/administrative data. These data items will facilitate analysis of the reasons for admission and procedures carried out and the frequency and lengths of admissions, as well as the health economics data to support the study objectives. The reason for hospitalisation and A&E attendance is not reliably available from primary care electronic health records. Mortality data includes the month and year of death, as well as the cause of death, which will facilitate analysis of the study endpoint of death.

The day-to-day running of the study is managed by Norwich Clinical Trials Unit at UEA, and all statistical and health economic analysis of all primary and secondary care clinical and mortality data (including all of the data to be provided by NHSD) will be controlled and carried out at UEA by the researchers at UEA. The sole data controller for the request is therefore UEA, and only UEA will be processing the data provided. UEA received pseudonymised data from both NHS England and Optimum Patient Care, containing the Study ID, and the secondary/mortality and primary care data respectively.

The study is supported by a group of collaborators from around the UK including University of Southampton, University of Exeter, University of Manchester, University of Oxford, Queen Mary University of London, University of Aberdeen, and University of Edinburgh. The initiative is carried out with the support of the Asthma UK Centre for Applied Research, and Asthma UK. The roles and responsibilities of the collaborating research institutions are laid out in the Collaboration Agreement; full control over analysis and data processing remains with UEA.

Only aggregate data is shared with the research collaborators, or any other third party. Additionally, small numbers are suppressed (i.e. values of five or fewer people per outcome variable are not made available).

Other organisations involved in the wider project but not processing data requested:

Name: Optimum Patient Care Ltd

Role:

1) To facilitate extraction of pseudonymised primary care data from participating GP practices.

2) To obtain pseudonymised secondary care data for GP practices in Scotland and Wales only.

3) To provide the research team at UEA with the ARRISA pseudonymised primary care dataset for England, and the linked pseudonymised primary/secondary care dataset for Scotland and Wales.

Study data held: Pseudonymised patient-level data from participating GP practices in England. Pseudonymised patient-level linked primary/secondary care data for participating GP practices in Scotland and Wales. Optimum Patient Care Ltd will not have access to or process any data received from NHS England.

Name: Wellbeing Software Group Ltd (Apollo)

Role: Sub-contractor of Optimum Patient Care Ltd. Facilitating extraction of primary care data from approx. 70% of participating GP practices.

Study data held: None

Name: Harvey Walsh Ltd

Role: Sub-contractor of Optimum Patient Care Ltd. To host a secure online portal to which participating GP practices upload asthma patient study ID and key identifiers (NHS number, date of birth and gender), for the purpose of transferring this data to NHS England for matching with secondary care/mortality data.

Study data held: Patient Study ID and key identifiers (NHS number, date of birth, and gender only) of asthma patients at ARRISA GP practices, hosted on a secure server, for the sole purpose of transferring this data to NHS England for matching. No clinical data will be obtained or held.

The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme funded the research but do not make any decisions regarding the data under this Agreement and therefore are not considered a data controller for the study.

The results of the trial will be reported first to the Trial Management Group (i.e. University of East Anglia, co-applicants from the collaborating universities and Asthma UK, and patient representatives). The main report will be drafted by members of the trial team for submission to funder in January 2024 in draft form. The final version will be agreed by the Trial Steering Committee before anticipated submission for publication in Spring 2024. This will be presented in the HTA monograph series published on NIHR journals. The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Further publications may be submitted during 2024/5, after the main report, where the study team explore specific aspects of the work in more detail.

Findings from the study analysis will be reported and disseminated through peer reviewed scientific journals, internal reports, conference presentations, published on the study website, and through the infrastructure of the Asthma UK Centre for Applied Research to support innovative approaches to dissemination (e.g. via social media, Science Festivals, etc). PPI contributors will be involved in the dissemination planning to ensure that the results are widely available and accessible. The result of the trial will be disseminated regardless of the direction of effect. The results will also be made available to the wider community via the websites of Asthma UK, HTA and academic, patient care, and research organisations such as the Asthma UK Centre for Applied Research and the Primary Care Respiratory Society. The participating GP practices will be given and encouraged to display links to the research. Press releases and website links to summaries of the research will be publicised on the websites of prominent campaigning and charitable organisations such as Asthma-UK.

Summary results will be publicised via social media such as the Twitter accounts of the ARRISA study (@arrisauk), Norwich Clinical Trials Unit (@norwichctu), Asthma UK (@asthmauk) and Asthma UK Centre for Applied Research (@aukcar).

This will be a definitive study and therefore it is expected that, if the study has a positive outcome, then guideline writers and commissioners throughout the world will incorporate similar methods of identifying and managing patients with at-risk asthma in management algorithms within primary care or family practice. British asthma guidelines are updated annually and would expect a definitive study to lead to new recommendations within a year of publication. The study has a strong representation within the BTS/SIGN guideline committee and within the Primary Care Respiratory Society UK. Thus the study could change UK practice within a year of publication.

The data tables and statistics included in all outputs will contain only aggregate data. No patient level data will be included, and small numbers, if they arise will be suppressed in line with the HES Analysis Guide (i.e. values of five or fewer people per outcome variable are not made available). All research outputs are prepared and reviewed for consistency with UEA Guidelines on Good Practice in Research, specifically the UEA Research Data Management Policy and UEA Research Data Management Procedures and Guidance v1.3.

As of January 2025, receipt of data from NHS England has permitted UEA to begin analysis of the primary and secondary outcomes of the trial. This work is close to completion, and full benefits will be achieved following the outcome of analysis.

The University of East Anglia requires pseudonymised secondary care data (hospital admissions, Accident and Emergency (A&E), outpatient and critical care data, and mortality data) for a selected cohort of previously identified patients for the At-Risk Registers Integrated into primary care to Stop Asthma crises in the UK (ARRISA-UK) study.

Approximately 74,000 people with asthma are admitted to hospital and 1,150 die unnecessarily in the UK every year. Excellent drugs are available for asthma and clear advice on prescribing them that should allow asthma to be controlled in most patients. It is known that certain asthma patients are at greater risk of being admitted or dying than others and that targeting intensive support and care to these patients improves their health.

Development of the ARRISA approach began with a single-practice pilot study in 2002, which was followed by a regional, cluster-randomised trial in 30 GP practices in Norfolk (2006-2009) funded by Asthma UK (Project no. 06/047). The current ARRISA-UK cluster-randomised trial of 275 practices in England, Scotland and Wales will be a definitive study which, if it has a positive outcome, it would be expected to change UK primary care policy and practice.

In the regional study, ‘at-risk’ patients were identified within GP practices and computer-based systems were used to create pop-up alerts when these patients contacted the practice. Practice staff were trained on what to do when they see the alert. This didn’t reduce the total number of attacks but reduced the hospital admissions as more patients appeared to receive appropriate treatment for their asthma (J. R. Smith et al, Thorax, 2012, 12, 1052). Based on these promising findings the study is now undertaking a nationwide study to confirm that it is possible to improve the care of these patients in a way that is cost-effective and doesn’t affect the care of other asthma patients within GP practices. (https://www.uea.ac.uk/arrisa-uk/about).

The ARRISA-UK study is sponsored and led by the data controller University of East Anglia (UEA), with analysis to be completed at UEA who is also acting as data processor. The purpose of the study is to determine whether flagging the electronic health records of people identified as being at risk of asthma attacks and training staff on the action to take when seeing the flag reduces asthma related crisis events (defined as hospital admissions, A&E attendances and deaths) over a 12 month period. The primary outcome is the difference in the proportion of at-risk patients (as identified by a prior database search) who have an asthma-related crisis event in the 12 months from the date that alert flags go live on the computer system in the intervention practices compared to the control group practices. This is a clinically meaningful, patient centred outcome that represents the severest form of asthma attacks that are the most costly to patients and society.

The study will also find out how many people have well controlled asthma, what medications are prescribed for asthma, how often patients attend appointments and if they stop smoking. The study will calculate how much this costs and whether it improves (or interferes with) the care of other patients with asthma in the practice.

So far, the study has collected pseudonymised primary care data from participating GP practices in England, Scotland and Wales. However, this data lacks sufficient detail to answer the primary outcome (asthma-related crisis event) as this requires secondary care data. The data requested here will build a more precise picture of asthma patient outcomes during the study.

Public and patient involvement has been fundamental to all aspects of the study and follows on from PPI involvement in the previous study, a smaller regional study of 30 GP practices in Norfolk. This includes review of the end of study report. PPI contributors - including a local representative who was involved in the previous study and a National PPI representative supported by the Asthma UK Centre for Applied Research - were co-applicants on the grant submission and have helped with the study design and the patient-facing documents in particular. They are also members of the Trial Management Group and will have a crucial role in the dissemination of findings.

The plans and processes for extracting data from GP practices and linking it with secondary care data from NHS England have been reviewed by the Asthma UK Centre for Applied Research (AUKCAR) including their public and patient group, lay members of the Trial Management Group and Independent Trial Steering Committee, and members of a patient discussion forum around the proposed use of patient data without consent. Feedback from PPI representatives was used to improve fair processing materials (poster and leaflet) produced by the research team.

The ARRISA-UK study is publicly funded research being undertaken in the public interest to inform knowledge on how best to improve care for patients with asthma who are at high risk of exacerbations of their disease. The study is funded by the National Institute for Health Research's Health Technology Assessment Programme (NIHR HTA) under the Department of Health UK and has undergone rigorous peer review. The data requested is to be used in the primary analysis of the ARRISA-UK study and will be reported as aggregated results in the NIHR journal monograph series and presented in peer-reviewed publications. The results of the research may inform public guidance and decisions about funding, resources and technologies available in GP practices, and as such may help improve care for asthma patients in the future.

Processing is therefore lawful under Article 6 (1)(e) of the UK General Data Protection Regulation (UK GDPR) (processing is necessary for the performance of a task carried out in the public interest) and article 9(2)(j) (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes…).

The justification for processing the data by University of East Anglia is Article 6(1)€ of the UK GDPR: (processing is necessary for the performance of a task in the public interest or in the exercise of official authority vested in the controller).

The justification for the processing of the special category data (health data) by University of East Anglia is Article 9(2)(j) of the UK GDPR: (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

The data will not be used for commercial purposes, will not be provided in record level to any third party, and will not be used for direct marketing. All analysis will be conducted at the University of East Anglia, Norwich, UK.

Data from GP practice records is captured remotely in a manner that is completely non-identifiable to the researchers, so individual patients are not asked to give informed consent. Instead, the NHS Research Ethics Committee agreed that the Caldicott Guardian or Information Governance lead for each GP practice (who is responsible for the use of their patients’ records) could approve the conduct of the study at their practice as they had been made aware of the purpose of the study and how their patients’ data would be used. There is no anticipated harm to the public by the dissemination of the data requested. Patients who have opted-out for use of their data in research are excluded from both the primary and secondary care data requests.

With around 160,000 asthma patients across 222 participating GP practices in England, it would have been too great a burden on research and GP practice staff, and impossible in a reasonable timescale, to individually consent patients to provide this information. Instead, this study had section 251 approval in place which allowed the study to collect secondary care data from NHS England. This is coupled with data available routinely from the GP practices and organisations.

Patient identifiers (NHS number, date of birth and sex) were provided to NHS England for matching, the data disseminated by NHS England was pseudonymised. Disseminated data is required to be patient level rather than aggregated, as it will then be linked by UEA to pseudonymised primary care data via the Study ID, before being fully anonymised for analysis. Analysis of the primary and secondary outcome measures requires a patient level dataset, as the effectiveness of the ARRISA approach cannot be obtained from practice or region-level aggregated data. It is unfeasible to undertake a study with sufficient statistical power to understand the effectiveness of the study teams' intervention which is based on practice level data in the UK.

It is essential for the purposes of this study to collect primary/secondary and mortality data for the population included in the study. While the intervention is randomised at practice level, the flagging is at patient level, and study analysis will need to distinguish the outcome of patients who have been identified as at-risk vs those not at-risk. NHS England required patient identifiers to match records at a patient level. It is not possible to use an alternative study design to answer our question as there will not be sufficient statistical power and it is not possible to obtain written informed consent from patients. There are no alternative ways of collecting this data which are less intrusive. The study team have collected pseudonymised data from GP practice records, but secondary care service use is not adequately recorded in GP practice records to enable the primary outcome to be answered. By using the linkage service provided by NHS England with the proposed data flows, no identifiable data needed to be shared with the research team.

The study closed recruitment in April 2018 with 275 participating GP practices in total across England, Wales and Scotland; 35 of those are in Wales and 18 in Scotland. Each GP practice was randomised to either the intervention or control arm of the study. The study will receive GP practice record data for all asthma patients at both intervention and control practices, and the request for secondary care data covers this same group. Therefore, this data request to NHS England covered approximately 160,000 patients with a current diagnosis of asthma from 228 participating GP practices from England and border Wales.

The geographical spread of the data requested reflects the locations of the GP practices in the trial. There are participating practices in 14 of the 15 Clinical Research Network regions of England, and the request also covers six practices in Welsh local health boards near the border with England. Separate requests for secondary care data for Scotland and Wales have been made to the relevant data custodians (eData Research and Innovation Service and NHS Wales Informatics Service). This request is for secondary care data in England, and includes all of the GP practices in England participating in the study. However, participating GP practices have advised that patients registered at Welsh practices near the border with England may attend hospital in England. Therefore, this request includes six GP practices in Wales. In Scotland, GP practices participating in ARRISA-UK are not close to the border with England, and all asthma patients of interest are therefore expected to routinely be using secondary care services in Scotland only.

All patients in the study will have had asthma, according to the study definition based on primary care data, at some point in the study. Automated electronic searches will be used to identify patients with asthma: those with asthma-related codes in their electronic medical record. Some may not have had the diagnosis at the start of the study period, but may qualify for the diagnosis and inclusion in the study during the observation period. Thus the data request includes both people with current asthma at the beginning of recruitment, as well as those diagnosed with asthma throughout the recruitment period. Patients in the cohort may include those who died or left the practice during the follow-up period.

In earlier work, an algorithm was developed and validated for aiding GP practice staff in recognising those patients most at risk of being admitted to hospital or dying from an asthma exacerbation. This is based on known risk factors such as previous exacerbation history, coding for anxiety or depression, smoking history and prescribing data and laboratory results all of which are associated with more severe asthma and asthma exacerbations. This candidate at-risk list is generated by each GP practice and reviewed by their clinical asthma lead. The practice’s respiratory clinical lead can add new at-risk patients who join the practice to their register or remove patients for whom a flag on their health record would not be appropriate (e.g. if they are receiving palliative care). To meet the intended study outcomes, including the impact of the ARRISA intervention on all asthma patients at a GP practice, data from NHS England was requested for all patients with asthma, not just those deemed to be at higher risk.

Because asthma is a variable condition, the date of initial presentation with the condition, or the date(s) of presentation within any specified study dates is not relevant for the diagnosis in our study protocol. Their inclusion in the population with active asthma does depend on having a prescription issue within a year period, but no single date or date range for an asthma diagnosis or event is critical across the whole study population.

The data required from NHS England is individual level data on patient outcomes for ARRISA patients of interest (asthma patients in participating GP practices). The study require data on admitted patient care, A&E attendance, outpatient visits, critical care, and mortality for the primary and secondary outcome measures of the study. The primary outcome is the difference in the proportion of at-risk asthma patients who have an asthma-related crisis event (A&E attendance, hospitalisation or death) in the 12 months from the date the pop-up alerts go live on the computer system in the intervention practices compared to the control group practices.

The secondary outcomes include time to first crisis event, asthma control, asthma medications, attendance at appointments, medication adherence, smoking status, all cause admission and death, health care costs in all and “at-risk” asthma patients. Data contained in the Hospital Episode Statistics and mortality datasets listed above, linked with data collected in primary care, will provide the information needed for the analysis of these outcome measures.

Obtaining the requested data from NHS England is vital to the study analysis, as primary care data alone does not contain the level of detail on hospital visits and deaths needed to give an accurate picture of the effectiveness of the intervention.

By requesting data regarding all deaths and hospital episodes for the cohort, rather than just those recorded as ‘due to asthma’, the data regarding secondary care resource utilisation will be as complete as possible, and additional resources used due to poorly controlled asthma as a co-morbidity will be included. The intention is to assess all episodes for any asthma-related elements as part of the study analysis at Norwich CTU.

By including data from HES Critical Care and Outpatients, in addition to HES APC and A&E, the data regarding secondary care resource utilisation will be as complete as possible, benefiting the assessment of the effectiveness of the ARRISA study intervention. In addition, out-patient attendance data will be used to measure a key secondary outcome as part of a health economics (cost-effectiveness) analysis.

Data is requested only for the participants in the study (asthma patients at GP practices participating in ARRISA-UK). Data from all patients with asthma in a research practice is required to determine if there are any consequences of our intervention by diverting care towards an at-risk group.

Data is requested only for the relevant time period (April 2015 – September 2019). This period facilitates three years of data for all patients of interest at GP practices recruited into the study, within which the recruitment period ran from April 2016 – April 2018; practices started study activities at different times. The overall time period of the data request includes, for all practices:

- A baseline year before the practice was randomised to the intervention or control arm of the study

- A variable time period of less than one year for staff training, at-risk register creation etc

- A one year follow-up period of flags live for intervention practices or normal care for control practices.

Each data item requested has been carefully assessed and is considered to be essential for the completion of the study objectives. The study are requesting the minimum amount of data that will answer our research questions: does our intervention reduce asthma crisis events, does it improve other aspects of patient care, is it cost effective and does it adversely affect the care of asthma patients who are NOT at a high risk of an attack?

The datasets and fields requested will allow the research team to assess whether a hospital event or death is asthma-related. For example, hospital admission and A&E data includes the diagnosis and procedure data, as well as dates of admission and discharge and Healthcare Resource Group (HRG)/administrative data. These data items will facilitate analysis of the reasons for admission and procedures carried out and the frequency and lengths of admissions, as well as the health economics data to support the study objectives. The reason for hospitalisation and A&E attendance is not reliably available from primary care electronic health records. Mortality data includes the month and year of death, as well as the cause of death, which will facilitate analysis of the study endpoint of death.

The day-to-day running of the study is managed by Norwich Clinical Trials Unit at UEA, and all statistical and health economic analysis of all primary and secondary care clinical and mortality data (including all of the data to be provided by NHSD) will be controlled and carried out at UEA by the researchers at UEA. The sole data controller for the request is therefore UEA, and only UEA will be processing the data provided. UEA received pseudonymised data from both NHS England and Optimum Patient Care, containing the Study ID, and the secondary/mortality and primary care data respectively.

The study is supported by a group of collaborators from around the UK including University of Southampton, University of Exeter, University of Manchester, University of Oxford, Queen Mary University of London, University of Aberdeen, and University of Edinburgh. The initiative is carried out with the support of the Asthma UK Centre for Applied Research, and Asthma UK. The roles and responsibilities of the collaborating research institutions are laid out in the Collaboration Agreement; full control over analysis and data processing remains with UEA.

Only aggregate data is shared with the research collaborators, or any other third party. Additionally, small numbers are suppressed (i.e. values of five or fewer people per outcome variable are not made available).

Other organisations involved in the wider project but not processing data requested:

Name: Optimum Patient Care Ltd

Role:

1) To facilitate extraction of pseudonymised primary care data from participating GP practices.

2) To obtain pseudonymised secondary care data for GP practices in Scotland and Wales only.

3) To provide the research team at UEA with the ARRISA pseudonymised primary care dataset for England, and the linked pseudonymised primary/secondary care dataset for Scotland and Wales.

Study data held: Pseudonymised patient-level data from participating GP practices in England. Pseudonymised patient-level linked primary/secondary care data for participating GP practices in Scotland and Wales. Optimum Patient Care Ltd will not have access to or process any data received from NHS England.

Name: Wellbeing Software Group Ltd (Apollo)

Role: Sub-contractor of Optimum Patient Care Ltd. Facilitating extraction of primary care data from approx. 70% of participating GP practices.

Study data held: None

Name: Harvey Walsh Ltd

Role: Sub-contractor of Optimum Patient Care Ltd. To host a secure online portal to which participating GP practices upload asthma patient study ID and key identifiers (NHS number, date of birth and gender), for the purpose of transferring this data to NHS England for matching with secondary care/mortality data.

Study data held: Patient Study ID and key identifiers (NHS number, date of birth, and gender only) of asthma patients at ARRISA GP practices, hosted on a secure server, for the sole purpose of transferring this data to NHS England for matching. No clinical data will be obtained or held.

The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme funded the research but do not make any decisions regarding the data under this Agreement and therefore are not considered a data controller for the study.

The results of the trial will be reported first to the Trial Management Group (i.e. University of East Anglia, co-applicants from the collaborating universities and Asthma UK, and patient representatives). The main report will be drafted by members of the trial team for submission to funder in January 2024 in draft form. The final version will be agreed by the Trial Steering Committee before anticipated submission for publication in Spring 2024. This will be presented in the HTA monograph series published on NIHR journals. The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Further publications may be submitted during 2024/5, after the main report, where the study team explore specific aspects of the work in more detail.

Findings from the study analysis will be reported and disseminated through peer reviewed scientific journals, internal reports, conference presentations, published on the study website, and through the infrastructure of the Asthma UK Centre for Applied Research to support innovative approaches to dissemination (e.g. via social media, Science Festivals, etc). PPI contributors will be involved in the dissemination planning to ensure that the results are widely available and accessible. The result of the trial will be disseminated regardless of the direction of effect. The results will also be made available to the wider community via the websites of Asthma UK, HTA and academic, patient care, and research organisations such as the Asthma UK Centre for Applied Research and the Primary Care Respiratory Society. The participating GP practices will be given and encouraged to display links to the research. Press releases and website links to summaries of the research will be publicised on the websites of prominent campaigning and charitable organisations such as Asthma-UK.

Summary results will be publicised via social media such as the Twitter accounts of the ARRISA study (@arrisauk), Norwich Clinical Trials Unit (@norwichctu), Asthma UK (@asthmauk) and Asthma UK Centre for Applied Research (@aukcar).

This will be a definitive study and therefore it is expected that, if the study has a positive outcome, then guideline writers and commissioners throughout the world will incorporate similar methods of identifying and managing patients with at-risk asthma in management algorithms within primary care or family practice. British asthma guidelines are updated annually and would expect a definitive study to lead to new recommendations within a year of publication. The study has a strong representation within the BTS/SIGN guideline committee and within the Primary Care Respiratory Society UK. Thus the study could change UK practice within a year of publication.

The data tables and statistics included in all outputs will contain only aggregate data. No patient level data will be included, and small numbers, if they arise will be suppressed in line with the HES Analysis Guide (i.e. values of five or fewer people per outcome variable are not made available). All research outputs are prepared and reviewed for consistency with UEA Guidelines on Good Practice in Research, specifically the UEA Research Data Management Policy and UEA Research Data Management Procedures and Guidance v1.3.

As of October 2023, receipt of data from NHS England has permitted UEA to begin analysis of the primary and secondary outcomes of the trial. This work is ongoing, and full benefits will be achieved following the completion of analysis - December 2023.

The University of East Anglia requires pseudonymised secondary care data (hospital admissions, Accident and Emergency (A&E), outpatient and critical care data, and mortality data) for a selected cohort of previously identified patients for the At-Risk Registers Integrated into primary care to Stop Asthma crises in the UK (ARRISA-UK) study.

Approximately 74,000 people with asthma are admitted to hospital and 1,150 die unnecessarily in the UK every year. Excellent drugs are available for asthma and clear advice on prescribing them that should allow asthma to be controlled in most patients. It is known that certain asthma patients are at greater risk of being admitted or dying than others and that targeting intensive support and care to these patients improves their health.

Development of the ARRISA approach began with a single-practice pilot study in 2002, which was followed by a regional, cluster-randomised trial in 30 GP practices in Norfolk (2006-2009) funded by Asthma UK (Project no. 06/047). The current ARRISA-UK cluster-randomised trial of 275 practices in England, Scotland and Wales will be a definitive study which, if it has a positive outcome, it would be expected to change UK primary care policy and practice.

In the regional study, ‘at-risk’ patients were identified within GP practices and computer-based systems were used to create pop-up alerts when these patients contacted the practice. Practice staff were trained on what to do when they see the alert. This didn’t reduce the total number of attacks but reduced the hospital admissions as more patients appeared to receive appropriate treatment for their asthma (J. R. Smith et al, Thorax, 2012, 12, 1052). Based on these promising findings the study is now undertaking a nationwide study to confirm that it is possible to improve the care of these patients in a way that is cost-effective and doesn’t affect the care of other asthma patients within GP practices. (https://www.uea.ac.uk/arrisa-uk/about).

The ARRISA-UK study is sponsored and led by the data controller University of East Anglia (UEA), with analysis to be completed at UEA who is also acting as data processor. The purpose of the study is to determine whether flagging the electronic health records of people identified as being at risk of asthma attacks and training staff on the action to take when seeing the flag reduces asthma related crisis events (defined as hospital admissions, A&E attendances and deaths) over a 12 month period. The primary outcome is the difference in the proportion of at-risk patients (as identified by a prior database search) who have an asthma-related crisis event in the 12 months from the date that alert flags go live on the computer system in the intervention practices compared to the control group practices. This is a clinically meaningful, patient centred outcome that represents the severest form of asthma attacks that are the most costly to patients and society.

The study will also find out how many people have well controlled asthma, what medications are prescribed for asthma, how often patients attend appointments and if they stop smoking. The study will calculate how much this costs and whether it improves (or interferes with) the care of other patients with asthma in the practice.

So far, the study has collected pseudonymised primary care data from participating GP practices in England, Scotland and Wales. However, this data lacks sufficient detail to answer the primary outcome (asthma-related crisis event) as this requires secondary care data. The data requested here will build a more precise picture of asthma patient outcomes during the study.

Public and patient involvement has been fundamental to all aspects of the study and follows on from PPI involvement in the previous study, a smaller regional study of 30 GP practices in Norfolk. This includes review of the end of study report. PPI contributors - including a local representative who was involved in the previous study and a National PPI representative supported by the Asthma UK Centre for Applied Research - were co-applicants on the grant submission and have helped with the study design and the patient-facing documents in particular. They are also members of the Trial Management Group and will have a crucial role in the dissemination of findings.

The plans and processes for extracting data from GP practices and linking it with secondary care data from NHS Digital have been reviewed by the Asthma UK Centre for Applied Research (AUKCAR) including their public and patient group, lay members of the Trial Management Group and Independent Trial Steering Committee, and members of a patient discussion forum around the proposed use of patient data without consent. Feedback from PPI representatives was used to improve fair processing materials (poster and leaflet) produced by the research team.

The ARRISA-UK study is publicly funded research being undertaken in the public interest to inform knowledge on how best to improve care for patients with asthma who are at high risk of exacerbations of their disease. The study is funded by the National Institute for Health Research's Health Technology Assessment Programme (NIHR HTA) under the Department of Health UK and has undergone rigorous peer review. The data requested is to be used in the primary analysis of the ARRISA-UK study and will be reported as aggregated results in the NIHR journal monograph series and presented in peer-reviewed publications. The results of the research may inform public guidance and decisions about funding, resources and technologies available in GP practices, and as such may help improve care for asthma patients in the future.

Processing is therefore lawful under Article 6 1(e) of the General Data Protection Regulation (GDPR) (processing is necessary for the performance of a task carried out in the public interest) and article 9(2)(j) (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes…).

The justification for processing the data by University of East Anglia is Article 6(1)(e) of the GDPR: (processing is necessary for the performance of a task in the public interest or in the exercise of official authority vested in the controller).

The justification for the processing of the special category data (health data) by University of East Anglia is Article 9(2)(j) of the GDPR: (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

The data will not be used for commercial purposes, will not be provided in record level to any third party, and will not be used for direct marketing. All analysis will be conducted at the University of East Anglia, Norwich, UK.

Data from GP practice records is captured remotely in a manner that is completely non-identifiable to the researchers, so individual patients are not asked to give informed consent. Instead, the NHS Research Ethics Committee agreed that the Caldicott Guardian or Information Governance lead for each GP practice (who is responsible for the use of their patients’ records) could approve the conduct of the study at their practice as they had been made aware of the purpose of the study and how their patients’ data would be used. There is no anticipated harm to the public by the dissemination of the data requested. Patients who have opted-out for use of their data in research are excluded from both the primary and secondary care data requests.

With around 160,000 asthma patients across 222 participating GP practices in England, it would have been too great a burden on research and GP practice staff, and impossible in a reasonable timescale, to individually consent patients to provide this information. Instead, this study has section 251 approval in place which allows the study to collect secondary care data from NHS Digital. This is coupled with data available routinely from the GP practices and organisations.

Patient identifiers (NHS number, date of birth and sex) will be provided to NHS Digital for matching, but the data to be disseminated by NHS Digital will be pseudonymised. Disseminated data is required to be patient level rather than aggregated, as it will then be linked by UEA to pseudonymised primary care data via the Study ID, before being fully anonymised for analysis. Analysis of the primary and secondary outcome measures requires a patient level dataset, as the effectiveness of the ARRISA approach cannot be obtained from practice or region-level aggregated data. It is unfeasible to undertake a study with sufficient statistical power to understand the effectiveness of the study teams' intervention which is based on practice level data in the UK.

It is essential for the purposes of this study to collect primary/secondary and mortality data for the population included in the study. While the intervention is randomised at practice level, the flagging is at patient level, and study analysis will need to distinguish the outcome of patients who have been identified as at-risk vs those not at-risk. NHS Digital requires patient identifiers to match records at a patient level. It is not possible to use an alternative study design to answer our question as there will not be sufficient statistical power and it is not possible to obtain written informed consent from patients. There are no alternative ways of collecting this data which are less intrusive. The study team have collected pseudonymised data from GP practice records, but secondary care service use is not adequately recorded in GP practice records to enable the primary outcome to be answered. By using the linkage service provided by NHS Digital with the proposed data flows, no identifiable data needs to be shared with the research team.

The study closed recruitment in April 2018 with 275 participating GP practices in total across England, Wales and Scotland; 35 of those are in Wales and 18 in Scotland. Each GP practice was randomised to either the intervention or control arm of the study. The study will receive GP practice record data for all asthma patients at both intervention and control practices, and the request for secondary care data covers this same group. Therefore, this data request to NHS Digital covers approximately 160,000 patients with a current diagnosis of asthma from 228 participating GP practices from England and border Wales.

At this stage it is not possible to give an exact cohort size, as some GP practices may choose not to take part in the secondary care data collection activity. They will be approached with a variation of the site agreement if IGARD approval is granted.

The geographical spread of the data requested reflects the locations of the GP practices in the trial. There are participating practices in 14 of the 15 Clinical Research Network regions of England, and the request also covers six practices in Welsh local health boards near the border with England. Separate requests for secondary care data for Scotland and Wales have been made to the relevant data custodians (eData Research and Innovation Service and NHS Wales Informatics Service). This request is for secondary care data in England, and includes all of the GP practices in England participating in the study. However, participating GP practices have advised that patients registered at Welsh practices near the border with England may attend hospital in England. Therefore, this request includes six GP practices in Wales. In Scotland, GP practices participating in ARRISA-UK are not close to the border with England, and all asthma patients of interest are therefore expected to routinely be using secondary care services in Scotland only.

All patients in the study will have had asthma, according to the study definition based on primary care data, at some point in the study. Automated electronic searches will be used to identify patients with asthma: those with asthma-related codes in their electronic medical record. Some may not have had the diagnosis at the start of the study period, but may qualify for the diagnosis and inclusion in the study during the observation period. Thus the data request includes both people with current asthma at the beginning of recruitment, as well as those diagnosed with asthma throughout the recruitment period. Patients in the cohort may include those who died or left the practice during the follow-up period.

In earlier work, an algorithm was developed and validated for aiding GP practice staff in recognising those patients most at risk of being admitted to hospital or dying from an asthma exacerbation. This is based on known risk factors such as previous exacerbation history, coding for anxiety or depression, smoking history and prescribing data and laboratory results all of which are associated with more severe asthma and asthma exacerbations. This candidate at-risk list is generated by each GP practice and reviewed by their clinical asthma lead. The practice’s respiratory clinical lead can add new at-risk patients who join the practice to their register or remove patients for whom a flag on their health record would not be appropriate (e.g. if they are receiving palliative care). To meet the intended study outcomes, including the impact of the ARRISA intervention on all asthma patients at a GP practice, data from NHS Digital is being requested for all patients with asthma, not just those deemed to be at higher risk.

Because asthma is a variable condition, the date of initial presentation with the condition, or the date(s) of presentation within any specified study dates is not relevant for the diagnosis in our study protocol. Their inclusion in the population with active asthma does depend on having a prescription issue within a year period, but no single date or date range for an asthma diagnosis or event is critical across the whole study population.

The data required from NHS Digital is individual level data on patient outcomes for ARRISA patients of interest (asthma patients in participating GP practices). The study require data on admitted patient care, A&E attendance, outpatient visits, critical care, and mortality for the primary and secondary outcome measures of the study. The primary outcome is the difference in the proportion of at-risk asthma patients who have an asthma-related crisis event (A&E attendance, hospitalisation or death) in the 12 months from the date the pop-up alerts go live on the computer system in the intervention practices compared to the control group practices.

The secondary outcomes include time to first crisis event, asthma control, asthma medications, attendance at appointments, medication adherence, smoking status, all cause admission and death, health care costs in all and “at-risk” asthma patients. Data contained in the Hospital Episode Statistics and mortality datasets listed above, linked with data collected in primary care, will provide the information needed for the analysis of these outcome measures.

Obtaining the requested data from NHS Digital is vital to the study analysis, as primary care data alone does not contain the level of detail on hospital visits and deaths needed to give an accurate picture of the effectiveness of the intervention.

By requesting data regarding all deaths and hospital episodes for the cohort, rather than just those recorded as ‘due to asthma’, the data regarding secondary care resource utilisation will be as complete as possible, and additional resources used due to poorly controlled asthma as a co-morbidity will be included. The intention is to assess all episodes for any asthma-related elements as part of the study analysis at Norwich CTU.

By including data from HES Critical Care and Outpatients, in addition to HES APC and A&E, the data regarding secondary care resource utilisation will be as complete as possible, benefiting the assessment of the effectiveness of the ARRISA study intervention. In addition, out-patient attendance data will be used to measure a key secondary outcome as part of a health economics (cost-effectiveness) analysis.

Data is requested only for the participants in the study (asthma patients at GP practices participating in ARRISA-UK). Data from all patients with asthma in a research practice is required to determine if there are any consequences of our intervention by diverting care towards an at-risk group.

Data is requested only for the relevant time period (April 2015 – September 2019). This period facilitates three years of data for all patients of interest at GP practices recruited into the study, within which the recruitment period ran from April 2016 – April 2018; practices started study activities at different times. The overall time period of the data request includes, for all practices:

- A baseline year before the practice was randomised to the intervention or control arm of the study

- A variable time period of less than one year for staff training, at-risk register creation etc

- A one year follow-up period of flags live for intervention practices or normal care for control practices.

Each data item requested has been carefully assessed and is considered to be essential for the completion of the study objectives. The study are requesting the minimum amount of data that will answer our research questions: does our intervention reduce asthma crisis events, does it improve other aspects of patient care, is it cost effective and does it adversely affect the care of asthma patients who are NOT at a high risk of an attack?

The datasets and fields requested will allow the research team to assess whether a hospital event or death is asthma-related. For example, hospital admission and A&E data includes the diagnosis and procedure data, as well as dates of admission and discharge and Healthcare Resource Group (HRG)/administrative data. These data items will facilitate analysis of the reasons for admission and procedures carried out and the frequency and lengths of admissions, as well as the health economics data to support the study objectives. The reason for hospitalisation and A&E attendance is not reliably available from primary care electronic health records. Mortality data includes the month and year of death, as well as the cause of death, which will facilitate analysis of the study endpoint of death.

The day-to-day running of the study is managed by Norwich Clinical Trials Unit at UEA, and all statistical and health economic analysis of all primary and secondary care clinical and mortality data (including all of the data to be provided by NHSD) will be controlled and carried out at UEA by the researchers at UEA. The sole data controller for the request is therefore UEA, and only UEA will be processing the data provided by this request. UEA will receive pseudonymised data from both NHS Digital and Optimum Patient Care, containing the Study ID, and the secondary/mortality and primary care data respectively.

The study is supported by a group of collaborators from around the UK including University of Southampton, University of Exeter, University of Manchester, University of Oxford, Queen Mary University of London, University of Aberdeen, and University of Edinburgh. The initiative is carried out with the support of the Asthma UK Centre for Applied Research, and Asthma UK. The roles and responsibilities of the collaborating research institutions are laid out in the Collaboration Agreement; full control over analysis and data processing remains with UEA.

Only aggregate data is shared with the research collaborators, or any other third party. Additionally, small numbers are suppressed (i.e. values of five or fewer people per outcome variable are not made available).

Other organisations involved in the wider project but not processing data requested:

Name: Optimum Patient Care Ltd

Role:

1) To facilitate extraction of pseudonymised primary care data from participating GP practices.

2) To obtain pseudonymised secondary care data for GP practices in Scotland and Wales only.

3) To provide the research team at UEA with the ARRISA pseudonymised primary care dataset for England, and the linked pseudonymised primary/secondary care dataset for Scotland and Wales.

Study data held: Pseudonymised patient-level data from participating GP practices in England. Pseudonymised patient-level linked primary/secondary care data for participating GP practices in Scotland and Wales. Optimum Patient Care Ltd will not have access to or process any data received from NHS Digital.

Name: Wellbeing Software Group Ltd (Apollo)

Role: Sub-contractor of Optimum Patient Care Ltd. Facilitating extraction of primary care data from approx. 70% of participating GP practices.

Study data held: None

Name: Harvey Walsh Ltd

Role: Sub-contractor of Optimum Patient Care Ltd. To host a secure online portal to which participating GP practices upload asthma patient study ID and key identifiers (NHS number, date of birth and gender), for the purpose of transferring this data to NHS Digital for matching with secondary care/mortality data.

Study data held: Patient Study ID and key identifiers (NHS number, date of birth, and gender only) of asthma patients at ARRISA GP practices, hosted on a secure server, for the sole purpose of transferring this data to NHS Digital for matching. No clinical data will be obtained or held.

The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme funded the research but do not make any decisions regarding the data under this Agreement and therefore are not considered a data controller for the study.

The results of the trial will be reported first to the Trial Management Group (i.e. University of East Anglia, co-applicants from the collaborating universities and Asthma UK, and patient representatives). The main report will be drafted by members of the trial team for submission to funder in October 2023 in draft form. The final version will be agreed by the Trial Steering Committee before anticipated submission for publication in early 2024. This will be presented in the HTA monograph series published on NIHR journals. The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Further publications may be submitted during 2024, after the main report, where the study team explore specific aspects of the work in more detail.

Findings from the study analysis will be reported and disseminated through peer reviewed scientific journals, internal reports, conference presentations, published on the study website, and through the infrastructure of the Asthma UK Centre for Applied Research to support innovative approaches to dissemination (e.g. via social media, Science Festivals, etc). PPI contributors will be involved in the dissemination planning to ensure that the results are widely available and accessible. The result of the trial will be disseminated regardless of the direction of effect. The results will also be made available to the wider community via the websites of Asthma UK, HTA and academic, patient care, and research organisations such as the Asthma UK Centre for Applied Research and the Primary Care Respiratory Society. The participating GP practices will be given and encouraged to display links to the research. Press releases and website links to summaries of the research will be publicised on the websites of prominent campaigning and charitable organisations such as Asthma-UK.

Summary results will be publicised via social media such as the Twitter accounts of the ARRISA study (@arrisauk), Norwich Clinical Trials Unit (@norwichctu), Asthma UK (@asthmauk) and Asthma UK Centre for Applied Research (@aukcar).

This will be a definitive study and therefore it is expected that, if the study has a positive outcome, then guideline writers and commissioners throughout the world will incorporate similar methods of identifying and managing patients with at-risk asthma in management algorithms within primary care or family practice. British asthma guidelines are updated annually and would expect a definitive study to lead to new recommendations within a year of publication. The study has a strong representation within the BTS/SIGN guideline committee and within the Primary Care Respiratory Society UK. Thus the study could change UK practice within a year of publication.

The data tables and statistics included in all outputs will contain only aggregate data. No patient level data will be included, and small numbers, if they arise will be suppressed in line with the HES Analysis Guide (i.e. values of five or fewer people per outcome variable are not made available). All research outputs are prepared and reviewed for consistency with UEA Guidelines on Good Practice in Research, specifically the UEA Research Data Management Policy and UEA Research Data Management Procedures and Guidance v1.3.

As of November 2022, receipt of data from NHS Digital has permitted UEA to begin data processing and planning for analysis of the primary and secondary outcomes of the trial. This work is ongoing, and full benefits will be achieved on completion of analysis, which is planned for October 2023.

The University of East Anglia requires pseudonymised secondary care data (hospital admissions, Accident and Emergency (A&E), outpatient and critical care data, and mortality data) for a selected cohort of previously identified patients for the At-Risk Registers Integrated into primary care to Stop Asthma crises in the UK (ARRISA-UK) study.

Approximately 74,000 people with asthma are admitted to hospital and 1,150 die unnecessarily in the UK every year. Excellent drugs are available for asthma and clear advice on prescribing them that should allow asthma to be controlled in most patients. It is known that certain asthma patients are at greater risk of being admitted or dying than others and that targeting intensive support and care to these patients improves their health.

Development of the ARRISA approach began with a single-practice pilot study in 2002, which was followed by a regional, cluster-randomised trial in 30 GP practices in Norfolk (2006-2009) funded by Asthma UK (Project no. 06/047). The current ARRISA-UK cluster-randomised trial of 275 practices in England, Scotland and Wales will be a definitive study which, if it has a positive outcome, it would be expected to change UK primary care policy and practice.

In the regional study, ‘at-risk’ patients were identified within GP practices and computer-based systems were used to create pop-up alerts when these patients contacted the practice. Practice staff were trained on what to do when they see the alert. This didn’t reduce the total number of attacks but reduced the hospital admissions as more patients appeared to receive appropriate treatment for their asthma (J. R. Smith et al, Thorax, 2012, 12, 1052). Based on these promising findings the study is now undertaking a nationwide study to confirm that it is possible to improve the care of these patients in a way that is cost-effective and doesn’t affect the care of other asthma patients within GP practices. (https://www.uea.ac.uk/arrisa-uk/about).

The ARRISA-UK study is sponsored and led by the data controller University of East Anglia (UEA), with analysis to be completed at UEA who is also acting as data processor. The purpose of the study is to determine whether flagging the electronic health records of people identified as being at risk of asthma attacks and training staff on the action to take when seeing the flag reduces asthma related crisis events (defined as hospital admissions, A&E attendances and deaths) over a 12 month period. The primary outcome is the difference in the proportion of at-risk patients (as identified by a prior database search) who have an asthma-related crisis event in the 12 months from the date that alert flags go live on the computer system in the intervention practices compared to the control group practices. This is a clinically meaningful, patient centred outcome that represents the severest form of asthma attacks that are the most costly to patients and society.

The study will also find out how many people have well controlled asthma, what medications are prescribed for asthma, how often patients attend appointments and if they stop smoking. The study will calculate how much this costs and whether it improves (or interferes with) the care of other patients with asthma in the practice.

So far, the study has collected pseudonymised primary care data from participating GP practices in England, Scotland and Wales. However, this data lacks sufficient detail to answer the primary outcome (asthma-related crisis event) as this requires secondary care data. The data requested here will build a more precise picture of asthma patient outcomes during the study.

Public and patient involvement has been fundamental to all aspects of the study and follows on from PPI involvement in the previous study, a smaller regional study of 30 GP practices in Norfolk. This includes review of the end of study report. PPI contributors - including a local representative who was involved in the previous study and a National PPI representative supported by the Asthma UK Centre for Applied Research - were co-applicants on the grant submission and have helped with the study design and the patient-facing documents in particular. They are also members of the Trial Management Group and will have a crucial role in the dissemination of findings.

The plans and processes for extracting data from GP practices and linking it with secondary care data from NHS Digital have been reviewed by the Asthma UK Centre for Applied Research (AUKCAR) including their public and patient group, lay members of the Trial Management Group and Independent Trial Steering Committee, and members of a patient discussion forum around the proposed use of patient data without consent. Feedback from PPI representatives was used to improve fair processing materials (poster and leaflet) produced by the research team.

The ARRISA-UK study is publicly funded research being undertaken in the public interest to inform knowledge on how best to improve care for patients with asthma who are at high risk of exacerbations of their disease. The study is funded by the National Institute for Health Research's Health Technology Assessment Programme (NIHR HTA) under the Department of Health UK and has undergone rigorous peer review. The data requested is to be used in the primary analysis of the ARRISA-UK study and will be reported as aggregated results in the NIHR journal monograph series and presented in peer-reviewed publications. The results of the research may inform public guidance and decisions about funding, resources and technologies available in GP practices, and as such may help improve care for asthma patients in the future.

Processing is therefore lawful under Article 6 1(e) of the General Data Protection Regulation (GPDR) (processing is necessary for the performance of a task carried out in the public interest) and article 9(2)(j) (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes…).

The data will not be used for commercial purposes, will not be provided in record level to any third party, and will not be used for direct marketing. All analysis will be conducted at the University of East Anglia, Norwich, UK.

Data from GP practice records is captured remotely in a manner that is completely non-identifiable to the researchers, so individual patients are not asked to give informed consent. Instead, the NHS Research Ethics Committee agreed that the Caldicott Guardian or Information Governance lead for each GP practice (who is responsible for the use of their patients’ records) could approve the conduct of the study at their practice as they had been made aware of the purpose of the study and how their patients’ data would be used. There is no anticipated harm to the public by the dissemination of the data requested. Patients who have opted-out for use of their data in research are excluded from both the primary and secondary care data requests.

With around 160,000 asthma patients across 222 participating GP practices in England, it would have been too great a burden on research and GP practice staff, and impossible in a reasonable timescale, to individually consent patients to provide this information. Instead, this study has section 251 approval in place which allows the study to collect secondary care data from NHS Digital. This is coupled with data available routinely from the GP practices and organisations.

Patient identifiers (NHS number, date of birth and sex) will be provided to NHS Digital for matching, but the data to be disseminated by NHS Digital will be pseudonymised. Disseminated data is required to be patient level rather than aggregated, as it will then be linked by UEA to pseudonymised primary care data via the Study ID, before being fully anonymised for analysis. Analysis of the primary and secondary outcome measures requires a patient level dataset, as the effectiveness of the ARRISA approach cannot be obtained from practice or region-level aggregated data. It is unfeasible to undertake a study with sufficient statistical power to understand the effectiveness of the study teams' intervention which is based on practice level data in the UK.

It is essential for the purposes of this study to collect primary/secondary and mortality data for the population included in the study. While the intervention is randomised at practice level, the flagging is at patient level, and study analysis will need to distinguish the outcome of patients who have been identified as at-risk vs those not at-risk. NHS Digital requires patient identifiers to match records at a patient level. It is not possible to use an alternative study design to answer our question as there will not be sufficient statistical power and it is not possible to obtain written informed consent from patients. There are no alternative ways of collecting this data which are less intrusive. The study team have collected pseudonymised data from GP practice records, but secondary care service use is not adequately recorded in GP practice records to enable the primary outcome to be answered. By using the linkage service provided by NHS Digital, with the data flows proposed in this application, no identifiable data needs to be shared with the research team.

The study closed recruitment in April 2018 with 275 participating GP practices in total across England, Wales and Scotland; 35 of those are in Wales and 18 in Scotland. Each GP practice was randomised to either the intervention or control arm of the study. The study will receive GP practice record data for all asthma patients at both intervention and control practices, and the request for secondary care data covers this same group. Therefore, this data request to NHS Digital covers approximately 160,000 patients with a current diagnosis of asthma from 228 participating GP practices from England and border Wales.

At this stage it is not possible to give an exact cohort size, as some GP practices may choose not to take part in the secondary care data collection activity. They will be approached with a variation of the site agreement if IGARD approval is granted.

The geographical spread of the data requested reflects the locations of the GP practices in the trial. There are participating practices in 14 of the 15 Clinical Research Network regions of England, and the request also covers six practices in Welsh local health boards near the border with England. Separate requests for secondary care data for Scotland and Wales have been made to the relevant data custodians (eData Research and Innovation Service and NHS Wales Informatics Service). This request is for secondary care data in England, and includes all of the GP practices in England participating in the study. However, participating GP practices have advised that patients registered at Welsh practices near the border with England may attend hospital in England. Therefore, this request includes six GP practices in Wales. In Scotland, GP practices participating in ARRISA-UK are not close to the border with England, and all asthma patients of interest are therefore expected to routinely be using secondary care services in Scotland only.

All patients in the study will have had asthma, according to the study definition based on primary care data, at some point in the study. Automated electronic searches will be used to identify patients with asthma: those with asthma-related codes in their electronic medical record. Some may not have had the diagnosis at the start of the study period, but may qualify for the diagnosis and inclusion in the study during the observation period. Thus the data request includes both people with current asthma at the beginning of recruitment, as well as those diagnosed with asthma throughout the recruitment period. Patients in the cohort may include those who died or left the practice during the follow-up period.

In earlier work, an algorithm was developed and validated for aiding GP practice staff in recognising those patients most at risk of being admitted to hospital or dying from an asthma exacerbation. This is based on known risk factors such as previous exacerbation history, coding for anxiety or depression, smoking history and prescribing data and laboratory results all of which are associated with more severe asthma and asthma excerbations. This candidate at-risk list is generated by each GP practice and reviewed by their clinical asthma lead. The practice’s respiratory clinical lead can add new at-risk patients who join the practice to their register or remove patients for whom a flag on their health record would not be appropriate (e.g. if they are receiving palliative care). To meet the intended study outcomes, including the impact of the ARRISA intervention on all asthma patients at a GP practice, data from NHS Digital is being requested for all patients with asthma, not just those deemed to be at higher risk.

Because asthma is a variable condition, the date of initial presentation with the condition, or the date(s) of presentation within any specified study dates is not relevant for the diagnosis in our study protocol. Their inclusion in the population with active asthma does depend on having a prescription issue within a year period, but no single date or date range for an asthma diagnosis or event is critical across the whole study population.

The data required from NHS Digital is individual level data on patient outcomes for ARRISA patients of interest (asthma patients in participating GP practices). The study require data on admitted patient care, A&E attendance, outpatient visits, critical care, and mortality for the primary and secondary outcome measures of the study. The primary outcome is the difference in the proportion of at-risk asthma patients who have an asthma-related crisis event (A&E attendance, hospitalisation or death) in the 12 months from the date the pop-up alerts go live on the computer system in the intervention practices compared to the control group practices.

The secondary outcomes include time to first crisis event, asthma control, asthma medications, attendance at appointments, medication adherence, smoking status, all cause admission and death, health care costs in all and “at-risk” asthma patients. Data contained in the Hospital Episode Statistics and mortality datasets listed above, linked with data collected in primary care, will provide the information needed for the analysis of these outcome measures.

Obtaining the requested data from NHS Digital is vital to the study analysis, as primary care data alone does not contain the level of detail on hospital visits and deaths needed to give an accurate picture of the effectiveness of the intervention.

By requesting data regarding all deaths and hospital episodes for the cohort, rather than just those recorded as ‘due to asthma’, the data regarding secondary care resource utilisation will be as complete as possible, and additional resources used due to poorly controlled asthma as a co-morbidity will be included. The intention is to assess all episodes for any asthma-related elements as part of the study analysis at Norwich CTU.

By including data from HES Critical Care and Outpatients, in addition to HES APC and A&E, the data regarding secondary care resource utilisation will be as complete as possible, benefiting the assessment of the effectiveness of the ARRISA study intervention. In addition, out-patient attendance data will be used to measure a key secondary outcome as part of a health economics (cost-effectiveness) analysis.

Data is requested only for the participants in the study (asthma patients at GP practices participating in ARRISA-UK). Data from all patients with asthma in a research practice is required to determine if there are any consequences of our intervention by diverting care towards an at-risk group.

Data is requested only for the relevant time period (April 2015 – September 2019). This period facilitates three years of data for all patients of interest at GP practices recruited into the study, within which the recruitment period ran from April 2016 – April 2018; practices started study activities at different times. The overall time period of the data request includes, for all practices:

- A baseline year before the practice was randomised to the intervention or control arm of the study

- A variable time period of less than one year for staff training, at-risk register creation etc

- A one year follow-up period of flags live for intervention practices or normal care for control practices.

Each data item requested has been carefully assessed and is considered to be essential for the completion of the study objectives. The study are requesting the minimum amount of data that will answer our research questions: does our intervention reduce asthma crisis events, does it improve other aspects of patient care, is it cost effective and does it adversely affect the care of asthma patients who are NOT at a high risk of an attack?

The datasets and fields requested will allow the research team to assess whether a hospital event or death is asthma-related. For example, hospital admission and A&E data includes the diagnosis and procedure data, as well as dates of admission and discharge and Healthcare Resource Group (HRG)/administrative data. These data items will facilitate analysis of the reasons for admission and procedures carried out and the frequency and lengths of admissions, as well as the health economics data to support the study objectives. The reason for hospitalisation and A&E attendance is not reliably available from primary care electronic health records. Mortality data includes the month and year of death, as well as the cause of death, which will facilitate analysis of the study endpoint of death.

The day-to-day running of the study is managed by Norwich Clinical Trials Unit at UEA, and all statistical and health economic analysis of all primary and secondary care clinical and mortality data (including all of the data to be provided by NHSD) will be controlled and carried out at UEA by the researchers at UEA. The sole data controller for the request is therefore UEA, and only UEA will be processing the data provided by this request. UEA will receive pseudonymised data from both NHS Digital and Optimum Patient Care, containing the Study ID, and the secondary/mortality and primary care data respectively.

The study is supported by a group of collaborators from around the UK including University of Southampton, University of Exeter, University of Manchester, University of Oxford, Queen Mary University of London, University of Aberdeen, and University of Edinburgh. The initiative is carried out with the support of the Asthma UK Centre for Applied Research, and Asthma UK. The roles and responsibilities of the collaborating research institutions are laid out in the Collaboration Agreement; full control over analysis and data processing remains with UEA.

Only aggregate data is shared with the research collaborators, or any other third party. Additionally, small numbers are suppressed (i.e. values of five or fewer people per outcome variable are not made available).

Other organisations involved in the wider project but not processing data requested in this application:

Name: Optimum Patient Care Ltd

Role:

1) To facilitate extraction of pseudonymised primary care data from participating GP practices.

2) To obtain pseudonymised secondary care data for GP practices in Scotland and Wales only.

3) To provide the research team at UEA with the ARRISA pseudonymised primary care dataset for England, and the linked pseudonymised primary/secondary care dataset for Scotland and Wales.

Study data held: Pseudonymised patient-level data from participating GP practices in England. Pseudonymised patient-level linked primary/secondary care data for participating GP practices in Scotland and Wales. Optimum Patient Care Ltd will not have access to or process any data received from NHS Digital.

Name: Wellbeing Software Group Ltd (Apollo)

Role: Sub-contractor of Optimum Patient Care Ltd. Facilitating extraction of primary care data from approx. 70% of participating GP practices.

Study data held: None

Name: Harvey Walsh Ltd

Role: Sub-contractor of Optimum Patient Care Ltd. To host a secure online portal to which participating GP practices upload asthma patient study ID and key identifiers (NHS number, date of birth and gender), for the purpose of transferring this data to NHS Digital for matching with secondary care/mortality data.

Study data held: Patient Study ID and key identifiers (NHS number, date of birth, and gender only) of asthma patients at ARRISA GP practices, hosted on a secure server, for the sole purpose of transferring this data to NHS Digital for matching. No clinical data will be obtained or held.

The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme funded the research but do not make any decisions regarding the data under this Agreement and therefore are not considered a data controller for the study.

The results of the trial will be reported first to the Trial Management Group (i.e. University of East Anglia, co-applicants from the collaborating universities and Asthma UK, and patient representatives). The main report will be drafted by members of the trial team for submission in draft form two weeks after the end of study, and the final version will be agreed by the Trial Steering Committee before anticipated submission for publication in Spring 2022. This will be presented in the HTA monograph series published on NIHR journals. The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Further publications may be submitted during 2022, after the main report, where the study team explore specific aspects of the work in more detail.

Findings from the study analysis will be reported and disseminated through peer reviewed scientific journals, internal reports, conference presentations, published on the study website, and through the infrastructure of the Asthma UK Centre for Applied Research to support innovative approaches to dissemination (e.g. via social media, Science Festivals, etc). PPI contributors will be involved in the dissemination planning to ensure that the results are widely available and accessible. The result of the trial will be disseminated regardless of the direction of effect. The results will also be made available to the wider community via the websites of Asthma UK, HTA and academic, patient care, and research organisations such as the Asthma UK Centre for Applied Research and the Primary Care Respiratory Society. The participating GP practices will be given and encouraged to display links to the research. Press releases and website links to summaries of the research will be publicised on the websites of prominent campaigning and charitable organisations such as Asthma-UK.

Summary results will be publicised via social media such as the Twitter accounts of the ARRISA study (@arrisauk), Norwich Clinical Trials Unit (@norwichctu), Asthma UK (@asthmauk) and Asthma UK Centre for Applied Research (@aukcar).

This will be a definitive study and therefore it is expected that, if the study has a positive outcome, then guideline writers and commissioners throughout the world will incorporate similar methods of identifying and managing patients with at-risk asthma in management algorithms within primary care or family practice. British asthma guidelines are updated annually and would expect a definitive study to lead to new recommendations within a year of publication. The study has a strong representation within the BTS/SIGN guideline committee and within the Primary Care Respiratory Society UK. Thus the study could change UK practice within a year of publication.

The data tables and statistics included in all outputs will contain only aggregate data. No patient level data will be included, and small numbers, if they arise will be suppressed in line with the HES Analysis Guide (i.e. values of five or fewer people per outcome variable are not made available). All research outputs are prepared and reviewed for consistency with UEA Guidelines on Good Practice in Research, specifically the UEA Research Data Management Policy and UEA Research Data Management Procedures and Guidance v1.3.

Under a subsequent version of this Agreement, details of yielded benefits will be provided.

1) Introduction

The purpose of this application is to request pseudonymised secondary care data (hospital admissions, Accident and Emergency (A&E), outpatient and critical care data, and mortality data) for a selected cohort of previously identified patients for the At-Risk Registers Integrated into primary care to Stop Asthma crises in the UK (ARRISA-UK) study.

Approximately 74,000 people with asthma are admitted to hospital and 1,150 die unnecessarily in the UK every year. Excellent drugs are available for asthma and clear advice on prescribing them that should allow asthma to be controlled in most patients. It is known that certain asthma patients are at greater risk of being admitted or dying than others and that targeting intensive support and care to these patients improves their health.

Development of the ARRISA approach began with a single-practice pilot study in 2002, which was followed by a regional, cluster-randomised trial in 30 GP practices in Norfolk (2006-2009) funded by Asthma UK (Project no. 06/047). The current ARRISA-UK cluster-randomised trial of 275 practices in England, Scotland and Wales will be a definitive study which, if it has a positive outcome, it would be expected to change UK primary care policy and practice.

In the regional study, ‘at-risk’ patients were identified within GP practices and computer-based systems were used to create pop-up alerts when these patients contacted the practice. Practice staff were trained on what to do when they see the alert. This didn’t reduce the total number of attacks but reduced the hospital admissions as more patients appeared to receive appropriate treatment for their asthma (J. R. Smith et al, Thorax, 2012, 12, 1052). Based on these promising findings the study is now undertaking a nationwide study to confirm that it is possible to improve the care of these patients in a way that is cost-effective and doesn’t affect the care of other asthma patients within GP practices. (https://www.uea.ac.uk/arrisa-uk/about).

The ARRISA-UK study is sponsored and led by the data controller University of East Anglia (UEA), with analysis to be completed at UEA who is also acting as data processor. The purpose of the study is to determine whether flagging the electronic health records of people identified as being at risk of asthma attacks and training staff on the action to take when seeing the flag reduces asthma related crisis events (defined as hospital admissions, A&E attendances and deaths) over a 12 month period. The primary outcome is the difference in the proportion of at-risk patients (as identified by a prior database search) who have an asthma-related crisis event in the 12 months from the date that alert flags go live on the computer system in the intervention practices compared to the control group practices. This is a clinically meaningful, patient centred outcome that represents the severest form of asthma attacks that are the most costly to patients and society.

The study will also find out how many people have well controlled asthma, what medications are prescribed for asthma, how often patients attend appointments and if they stop smoking. The study will calculate how much this costs and whether it improves (or interferes with) the care of other patients with asthma in the practice.

So far, the study has collected pseudonymised primary care data from participating GP practices in England, Scotland and Wales. However, this data lacks sufficient detail to answer the primary outcome (asthma-related crisis event) as this requires secondary care data. The data requested here will build a more precise picture of asthma patient outcomes during the study.

Public and patient involvement has been fundamental to all aspects of the study and follows on from PPI involvement in the previous study, a smaller regional study of 30 GP practices in Norfolk. This includes review of the end of study report. PPI contributors - including a local representative who was involved in the previous study and a National PPI representative supported by the Asthma UK Centre for Applied Research - were co-applicants on the grant submission and have helped with the study design and the patient-facing documents in particular. They are also members of the Trial Management Group and will have a crucial role in the dissemination of findings.

The plans and processes for extracting data from GP practices and linking it with secondary care data from NHS Digital have been reviewed by the Asthma UK Centre for Applied Research (AUKCAR) including their public and patient group, lay members of the Trial Management Group and Independent Trial Steering Committee, and members of a patient discussion forum around the proposed use of patient data without consent. Feedback from PPI representatives was used to improve fair processing materials (poster and leaflet) produced by the research team.

2) Legal Basis

The ARRISA-UK study is publicly funded research being undertaken in the public interest to inform knowledge on how best to improve care for patients with asthma who are at high risk of exacerbations of their disease. The study is funded by the National Institute for Health Research's Health Technology Assessment Programme (NIHR HTA) under the Department of Health UK and has undergone rigorous peer review. The data requested is to be used in the primary analysis of the ARRISA-UK study and will be reported as aggregated results in the NIHR journal monograph series and presented in peer-reviewed publications. The results of the research may inform public guidance and decisions about funding, resources and technologies available in GP practices, and as such may help improve care for asthma patients in the future.

Processing is therefore lawful under Article 6 1(e) of the General Data Protection Regulation (GPDR) (processing is necessary for the performance of a task carried out in the public interest) and article 9(2)(j) (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes…).

The data will not be used for commercial purposes, will not be provided in record level to any third party, and will not be used for direct marketing. All analysis will be conducted at the University of East Anglia, Norwich, UK.

Data from GP practice records is captured remotely in a manner that is completely non-identifiable to the researchers, so individual patients are not asked to give informed consent. Instead, the NHS Research Ethics Committee agreed that the Caldicott Guardian or Information Governance lead for each GP practice (who is responsible for the use of their patients’ records) could approve the conduct of the study at their practice as they had been made aware of the purpose of the study and how their patients’ data would be used. There is no anticipated harm to the public by the dissemination of the data requested. Patients with have opted-out for use of their data in research are excluded from both the primary and secondary care data requests.

With around 160,000 asthma patients across 222 participating GP practices in England, it would have been too great a burden on research and GP practice staff, and impossible in a reasonable timescale, to individually consent patients to provide this information. Instead, this study has section 251 approval in place which allows the study to collect secondary care data from NHS Digital. This is coupled with data available routinely from the GP practices and organisations.

Patient identifiers (NHS number, date of birth and sex) will be provided to NHS Digital for matching, but the data to be disseminated by NHS Digital will be pseudonymised. Disseminated data is required to be patient level rather than aggregated, as it will then be linked by UEA to pseudonymised primary care data via the Study ID, before being fully anonymised for analysis. Analysis of the primary and secondary outcome measures requires a patient level dataset, as the effectiveness of the ARRISA approach cannot be obtained from practice or region-level aggregated data. It is unfeasible to undertake a study with sufficient statistical power to understand the effectiveness of our intervention which is based on practice level data in the UK.

It is essential for the purposes of this study to collect primary/secondary and mortality data for the population included in the study. While the intervention is randomised at practice level, the flagging is at patient level, and study analysis will need to distinguish the outcome of patients who have been identified as at-risk vs those not at-risk. NHS Digital requires patient identifiers to match records at a patient level. It is not possible to use an alternative study design to answer our question as there will not be sufficient statistical power and it is not possible to obtain written informed consent from patients. There are no alternative ways of collecting this data which are less intrusive. The study team have collected pseudonymised data from GP practice records, but secondary care service use is not adequately recorded in GP practice records to enable the primary outcome to be answered. By using the linkage service provided by NHS Digital, with the data flows proposed in this application, no identifiable data needs to be shared with the research team.

3) The ARRISA Cohort

The study closed recruitment in April 2018 with 275 participating GP practices in total across England, Wales and Scotland; 35 of those are in Wales and 18 in Scotland. Each GP practice was randomised to either the intervention or control arm of the study. The study will receive GP practice record data for all asthma patients at both intervention and control practices, and the request for secondary care data covers this same group. Therefore, this data request to NHS Digital covers approximately 160,000 patients with a current diagnosis of asthma from 228 participating GP practices from England and border Wales.

At this stage it is not possible to give an exact cohort size, as some GP practices may choose not to take part in the secondary care data collection activity. They will be approached with a variation of the site agreement if IGARD approval is granted.

The geographical spread of the data requested reflects the locations of the GP practices in the trial. There are participating practices in 14 of the 15 Clinical Research Network regions of England, and the request also covers six practices in Welsh local health boards near the border with England. Separate requests for secondary care data for Scotland and Wales have been made to the relevant data custodians (eData Research and Innovation Service and NHS Wales Informatics Service). This request is for secondary care data in England, and includes all of the GP practices in England participating in the study. However, participating GP practices have advised that patients registered at Welsh practices near the border with England may attend hospital in England. Therefore, this request includes six GP practices in Wales. In Scotland, GP practices participating in ARRISA-UK are not close to the border with England, and all asthma patients of interest are therefore expected to routinely be using secondary care services in Scotland only.

All patients in the study will have had asthma, according to our study definition based on primary care data, at some point in the study. Automated electronic searches will be used to identify patients with asthma: those with asthma-related codes in their electronic medical record. Some may not have had the diagnosis at the start of the study period, but may qualify for the diagnosis and inclusion in the study during the observation period. Thus the data request includes both people with current asthma at the beginning of recruitment, as well as those diagnosed with asthma throughout the recruitment period. Patients in the cohort may include those who died or left the practice during the follow-up period.

In earlier work, an algorithm was developed and validated for aiding GP practice staff in recognising those patients most at risk of being admitted to hospital or dying from an asthma exacerbation. This is based on known risk factors such as previous exacerbation history, coding for anxiety or depression, smoking history and prescribing data and laboratory results all of which are associated with more severe asthma and asthma excerbations. This candidate at-risk list is generated by each GP practice and reviewed by their clinical asthma lead. The practice’s respiratory clinical lead can add new at-risk patients who join the practice to their register or remove patients for whom a flag on their health record would not be appropriate (e.g. if they are receiving palliative care). To meet the intended study outcomes, including the impact of the ARRISA intervention on all asthma patients at a GP practice, data from NHS Digital is being requested for all patients with asthma, not just those deemed to be at higher risk.

Because asthma is a variable condition, the date of initial presentation with the condition, or the date(s) of presentation within any specified study dates is not relevant for the diagnosis in our study protocol. Their inclusion in the population with active asthma does depend on having a prescription issue within a year period, but no single date or date range for an asthma diagnosis or event is critical across the whole study population.

4) Data Requested

The data required from NHS Digital is individual level data on patient outcomes for ARRISA patients of interest (asthma patients in participating GP practices). The study require data on admitted patient care, A&E attendance, outpatient visits, critical care, and mortality for the primary and secondary outcome measures of the study. The primary outcome is the difference in the proportion of at-risk asthma patients who have an asthma-related crisis event (A&E attendance, hospitalisation or death) in the 12 months from the date the pop-up alerts go live on the computer system in the intervention practices compared to the control group practices.

The secondary outcomes include time to first crisis event, asthma control, asthma medications, attendance at appointments, medication adherence, smoking status, all cause admission and death, health care costs in all and “at-risk” asthma patients. Data contained in the Hospital Episode Statistics and mortality datasets listed above, linked with data collected in primary care, will provide the information needed for the analysis of these outcome measures.

Obtaining the requested data from NHS Digital is vital to the study analysis, as primary care data alone does not contain the level of detail on hospital visits and deaths needed to give an accurate picture of the effectiveness of the intervention.

By requesting data regarding all deaths and hospital episodes for the cohort, rather than just those recorded as ‘due to asthma’, the data regarding secondary care resource utilisation will be as complete as possible, and additional resources used due to poorly controlled asthma as a co-morbidity will be included. The intention is to assess all episodes for any asthma-related elements as part of the study analysis at Norwich CTU.

By including data from HES Critical Care and Outpatients, in addition to HES APC and A&E, the data regarding secondary care resource utilisation will be as complete as possible, benefiting the assessment of the effectiveness of the ARRISA study intervention. In addition, out-patient attendance data will be used to measure a key secondary outcome as part of a health economics (cost-effectiveness) analysis.

5) Data Minimisation

Data is requested only for the participants in the study (asthma patients at GP practices participating in ARRISA-UK). Data from all patients with asthma in a research practice is required to determine if there are any consequences of our intervention by diverting care towards an at-risk group.

Data is requested only for the relevant time period (April 2015 – September 2019). This period facilitates three years of data for all patients of interest at GP practices recruited into the study, within which the recruitment period ran from April 2016 – April 2018; practices started study activities at different times. The overall time period of the data request includes, for all practices:

- A baseline year before the practice was randomised to the intervention or control arm of the study

- A variable time period of less than one year for staff training, at-risk register creation etc

- A one year follow-up period of flags live for intervention practices or normal care for control practices.

Each data item requested has been carefully assessed and is considered to be essential for the completion of the study objectives. The study are requesting the minimum amount of data that will answer our research questions: does our intervention reduce asthma crisis events, does it improve other aspects of patient care, is it cost effective and does it adversely affect the care of asthma patients who are NOT at a high risk of an attack?

The datasets and fields requested will allow the research team to assess whether a hospital event or death is asthma-related. For example, hospital admission and A&E data includes the diagnosis and procedure data, as well as dates of admission and discharge and Healthcare Resource Group (HRG)/administrative data. These data items will facilitate analysis of the reasons for admission and procedures carried out and the frequency and lengths of admissions, as well as the health economics data to support the study objectives. The reason for hospitalisation and A&E attendance is not reliably available from primary care electronic health records. Mortality data includes the month and year of death, as well as the cause of death, which will facilitate analysis of the study endpoint of death.

6) Organisational structure of ARRISA-UK

The day-to-day running of the study is managed by Norwich Clinical Trials Unit at UEA, and all statistical and health economic analysis of all primary and secondary care clinical and mortality data (including all of the data to be provided by NHSD) will be controlled and carried out at UEA by the researchers at UEA. The sole data controller for the request is therefore UEA, and only UEA will be processing the data provided by this request. UEA will receive pseudonymised data from both NHS Digital and Optimum Patient Care, containing the Study ID, and the secondary/mortality and primary care data respectively.

The study is supported by a group of collaborators from around the UK including University of Southampton, University of Exeter, University of Manchester, University of Oxford, Queen Mary University of London, University of Aberdeen, and University of Edinburgh. The initiative is carried out with the support of the Asthma UK Centre for Applied Research, and Asthma UK. The roles and responsibilities of the collaborating research institutions are laid out in the Collaboration Agreement; full control over analysis and data processing remains with UEA.

Only aggregate data is shared with the research collaborators, or any other third party. Additionally, small numbers are suppressed (i.e. values of five or fewer people per outcome variable are not made available).

Other organisations involved in the wider project but not processing data requested in this application:

Name: Optimum Patient Care Ltd

Role:

1) To facilitate extraction of pseudonymised primary care data from participating GP practices.

2) To obtain pseudonymised secondary care data for GP practices in Scotland and Wales only.

3) To provide the research team at UEA with the ARRISA pseudonymised primary care dataset for England, and the linked pseudonymised primary/secondary care dataset for Scotland and Wales.

Study data held: Pseudonymised patient-level data from participating GP practices in England. Pseudonymised patient-level linked primary/secondary care data for participating GP practices in Scotland and Wales. Optimum Patient Care Ltd will not have access to or process any HES data for this study.

Name: Wellbeing Software Group Ltd (Apollo)

Role: Sub-contractor of Optimum Patient Care Ltd. Facilitating extraction of primary care data from approx. 70% of participating GP practices.

Study data held: None

Name: Harvey Walsh Ltd

Role: Sub-contractor of Optimum Patient Care Ltd. To host a secure online portal to which participating GP practices upload asthma patient study ID and key identifiers (NHS number, date of birth and gender), for the purpose of transferring this data to NHS Digital for matching with secondary care/mortality data.

Study data held: Patient Study ID and key identifiers (NHS number, date of birth, and gender only) of asthma patients at ARRISA GP practices, hosted on a secure server, for the sole purpose of transferring this data to NHS Digital for matching. No clinical data will be obtained or held.

The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme funded the research but play no further role.

The results of the trial will be reported first to the Trial Management Group (i.e. University of East Anglia, co-applicants from the collaborating universities and Asthma UK, and patient representatives). The main report will be drafted by members of the trial team for submission in draft form two weeks after the end of study, and the final version will be agreed by the Trial Steering Committee before anticipated submission for publication on the 15th of May 2021. This will be presented in the HTA monograph series published on NIHR journals. The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Further publications may be submitted from May 2021 onwards, after the main report, where they explore specific aspects of the work in more detail.

Findings from the study analysis will be reported and disseminated through peer reviewed scientific journals, internal reports, conference presentations, published on the study website, and through the infrastructure of the Asthma UK Centre for Applied Research to support innovative approaches to dissemination (e.g. via social media, Science Festivals, etc). PPI contributors will be involved in the dissemination planning to ensure that the results are widely available and accessible. The result of the trial will be disseminated regardless of the direction of effect. The results will also be made available to the wider community via the websites of Asthma UK, HTA and academic, patient care, and research organisations such as the Asthma UK Centre for Applied Research and the Primary Care Respiratory Society. The participating GP practices will be given and encouraged to display links to the research. Press releases and website links to summaries of the research will be publicised on the websites of prominent campaigning and charitable organisations such as Asthma-UK.

Summary results will be publicised via social media such as the Twitter accounts of the ARRISA study (@arrisauk), Norwich Clinical Trials Unit (@norwichctu), Asthma UK (@asthmauk) and Asthma UK Centre for Applied Research (@aukcar).

This will be a definitive study and therefore it is expected that, if the study has a positive outcome, then guideline writers and commissioners throughout the world will incorporate similar methods of identifying and managing patients with at-risk asthma in management algorithms within primary care or family practice. British asthma guidelines are updated annually and would expect a definitive study to lead to new recommendations within a year of publication. The study has a strong representation within the BTS/SIGN guideline committee and within the Primary Care Respiratory Society UK. Thus the study could change UK practice within a year of publication.

The data tables and statistics included in all outputs will contain only aggregate data. No patient level data will be included, and small numbers, if they arise will be suppressed in line with the HES Analysis Guide (i.e. values of five or fewer people per outcome variable are not made available). All research outputs are prepared and reviewed for consistency with UEA Guidelines on Good Practice in Research, specifically the UEA Research Data Management Policy and UEA Research Data Management Procedures and Guidance v1.3.

Yielded Benefits is not a requirement for new applications.