Active Monitoring for AtriaL Fibrillation - AMALFI trial

The University of Oxford requires linked healthcare data for consented participants in the Active Monitoring for AtriaL Fibrillation (AMALFI) study for baseline characterisation of participants and follow-up purposes. This includes centrally-collected data on primary care records, medications dispensed in the community, mortality, and Hospital Episode Statistics (HES) datasets.

Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia worldwide, and is estimated to affect over 1 million people in the UK. In AF, the atria (upper chambers of the heart) beat in an uncoordinated way, which disturbs the normal blood flow and can lead to the formation of blood clots inside the heart. These can travel through the bloodstream and create a blockage in the arteries supplying the brain, causing a stroke. Patients with AF are at a 5-fold increased risk of stroke, but this risk can be effectively reduced by up to two thirds with anticoagulation (blood-thinners). However, AF can occur only in short and infrequent episodes that make it hard to capture and start treatment, and it may also not cause any symptoms; as a result, some patients might have undetected AF until the time when they have a stroke.

A potential solution for this problem is to actively look for silent AF through screening in patients who are considered at risk. This is already routinely done in primary care when patients over the age of 65 years present for some other reason (for example for a routine appointment or a flu jab), in which their doctor might assess their pulse or do an electrocardiogram (ECG). However, this assessment is very short and is unlikely to detect short and infrequent AF episodes. New technology such as extended cardiac monitors and digital wearables offer the potential for increased and more frequent monitoring periods, and easier access to patients - but they are also more expensive and create additional workload for the healthcare system. At the moment, there is no conclusive evidence that screening for silent AF improves either clinical outcomes or patient quality-of-life, or if employing new AF screening strategies is a cost-effective strategy (and if so, how exactly it should be performed).

The AMALFI study is a randomised clinical trial of screening for subclinical (undiagnosed) AF in elderly patients with no previous AF, who are at increased risk of both AF and a subsequent stroke – this selection is based on a CHA2DS2VASc score of 3 or higher in men and 4 or higher in women (the CHA2DS2VASc score is a standard tool used by clinicians to assess stroke risk in patients with AF and help them decide who should receive anticoagulation). AMALFI is comparing a two-week remote continuous cardiac monitoring period with a ZioPatch to usual care alone: consenting participants will be assigned to one of the two groups by chance, in a similar way to tossing a coin, called “randomization”. The study has recruited 5,043 people through primary care practices in England. Recruitment started in 2019 and finished on February 28th 2022. The cohort is composed of 5,043 people, minus those that have withdrawn consent (6 thus far). This number of unconsented individuals may increase within the proposed duration of this Agreement, in which case, the University of Oxford will send NHS England an updated cohort. The main outcome of the study is the proportion of participants with newly-detected AF in both arms 2.5 years after randomisation (as shown in primary care records), with additional analyses planned in subgroups of age and sex, and at 5 years. The study may then continue collecting follow-up data for a period of up to 20 years after the initial 5 years. This long follow-up period was envisaged so that further data may be collected for longer-term outcomes (i.e. beyond 5 years), if a longer timeframe for assessment is considered important following the initial results.

If shown to be effective (and cost-effective), this approach to screening could form the basis for a potential future nationwide screening program, which could prevent stroke, disability, and premature death in this patient population. A second objective of the study is to develop streamlined procedures for running clinical trials in primary care, namely by comparing the outputs if using data collected directly from primary care practices or centrally-held data such as the datasets managed by NHS England.

AMALFI is sponsored by the University of Oxford and funded by the National Institute for Healthcare Research (NIHR) through the NIHR Oxford Biomedical Research Centre, with additional support from the NIHR Thames Valley and South Midlands Clinical Research Network (CRN) in the form of logistical support with recruitment only. The protocol and study procedures have been developed by a team of investigators spanning the Nuffield Department of Population Health (NDPH), the Radcliffe Department of Medicine, and the Nuffield Department of Primary Healthcare Sciences (all of which are departments of the University of Oxford) and is being run by the Clinical Trial Service Unit (CTSU) at NDPH.

Trial procedures are simple and remote, with no physical sites or dedicated staff apart from a small team at the CTSU, and no study visits. Eligible patients are being identified from primary care records in participating practices associated with the CRN. The practice runs a search of their records and generates a list of eligible patients, which is then uploaded to DocMail (a standard NHS mailing service). DocMail processes the mailing of the study documentation and, if interested, patients will return a completed questionnaire and consent form to the study team at CTSU – this includes their personal details such as name, address, date-of-birth, and NHS number. After this, participants are randomised and mailed a letter with or without a patch to self-administer (depending on the treatment allocation). Patients in the active/screening group are asked to wear their patch for 14 days and then return it using an enclosed box to iRhythm, the patch manufacturer, which analyses the data collected via the patch. iRhythm have not had and will not have any involvement in the design, conduct, analysis, or reporting of this study, nor will they have any access to NHS England data. A monitoring report is provided to the study team and the relevant findings shared with the patient’s GP, who manages subsequent treatment if needed. There are no study visits and follow-up data is only currently being collected from electronic record extractions at the individual practices taking part in the study. Information on quality-of-life will be collected through remote completion of mailed EQ5D (EuroQoL-5 Dimensions) questionnaires at two occasions (and for all participants at the same time): once the overall study recruitment has finished, and again approximately 2.5 years after that (i.e., coinciding with the timing of the primary outcome assessment). These questionnaires are a standard tool used in both research and clinical care settings within the NHS to assess several domains of quality-of-life (QoL; mobility, self-care, usual activities, pain/discomfort, anxiety/depression).

The University of Oxford will be the sole data controller who also process data for the study. The University of Oxford is the Data Controller as they determine the purposes for which and the manner in which the personal data disseminated under this Agreement are to be processed. The data will be stored and processed at the University of Oxford in secure servers and under strict access restrictions, and it will not be released to any other organisations or used for other purposes other than this study.

Altogether, the study requires the following datasets:

1. Medicines Dispensed in the Community (NHSBSA data)

2. Civil Registrations (mortality data)

3. HES Critical Care

4. HES Admitted Patient Care

5. Emergency Care Data Set

6. HES Outpatients

The study has also identified the need for General Practice (GP) Data, but at the current time, this data is not available. A national collection of primary care data would be a valuable resource for this study to draw upon for the research.

The data requested will be used to assess the study outcomes (namely the detection of AF) as well as other important events, such as the initiation of medication for AF, the need for additional appointments and tests, as well as hospital admissions and deaths. Altogether, these data will help the University of Oxford understand if AF screening can improve AF detection rates, what are the associated costs of doing so, and if there are particular safety signals of concern. It may also provide some initial indication of whether this procedure can improve clinical outcomes, although this is not the main purpose of the study. In particular, the NHSBSA data will be used to calculate the number of people starting anticoagulation in the active and control groups, which in turn will help the University of Oxford understand what are the likely benefits and risks of anticoagulation for subclinical AF in reducing strokes versus bleeding events.

AMALFI is collecting medications data via two routes: primary records extracted from each GP practice taking part in the study, and the centralised, nationwide NHSBSA dataset. Both sources will be used to assess rates of initiation of anticoagulation and other medications in the study cohort, and their impact on effectiveness (stroke, heart failure) and safety (bleeding) events resulting from new diagnoses of atrial fibrillation.

Given the novelty of the NHSBSA data, it is not yet known how well it overlaps with other pre-existent data sources (primary care records, in this case), and whether the NHSBSA dataset can be used on its own to retrieve medication exposure. If that was the case this would potentially improve clinical trial efficiency significantly via simplified data collection, both in AMALFI and beyond.

This work is to be undertaken in line with the NHSBSA Direction of assessing effectiveness and safety of medicines as the team is developing the methodological work required to guide future research using these data.

Finally, the conjunction of clinical events (detailed in the Civil Registrations, Medicines, HES Admitted Patient Care and HES Accident & Emergency datasets) with information on healthcare resource use (complemented by the HES Outpatients data) will provide a rich resource to assess the relative cost implications of screening from the perspective of the healthcare system, which will be crucial to inform future discussions on implementation. An additional aim of this work will be to undertake trial methodology research, specifically comparing the NHSBSA medicines data to medication data that is already being collected by the study from local GP practices. This exercise will focus on comparing the numbers of people identifying as taking particular drugs in each data source (rather than trying to audit the data collected in the NHSBSA dataset), and is hoped to help pave the way for broader use of the NSHBSA dataset for clinical trials.

Data minimisation will be pursued by only requesting datasets directly related to the outcomes of interest to the trial (clinical outcomes using primary care, admissions, and mortality data; medication initiation using primary care and dispensed medicines data; and health-care resource use using the remaining HES datasets), and requesting only the fields needed in each dataset; these have been reviewed by a clinician and a health-economist to ensure that only the necessary data is requested. Data is also only being requested for as far back as 5 years before the start of recruitment, i.e., from 2014/15 to latest available.

The objective of undertaking the AMALFI research study aiming to improve future patients’ health is the justification for processing under GDPR Article 6(1)(e) and Article 9(2)(j). In particular, AMALFI aims to provide high-grade evidence on the potential efficacy and cost-effectiveness of atrial fibrillation screening in the NHS, which it is hoped will inform future considerations regarding whether and how to roll-out a nationwide screening program, much like the existing programs for breast, cervical, and colorectal cancer. All participants in AMALFI have provided written informed consent to take part in the study, including specific consent for access to data held by NHS England to be provided to the University of Oxford for the purpose of this study. The study has received favourable Research Ethics Committee approval from the London - Bromley Research Ethics Committee (REC reference 19/LO/0220) and is registered in the International Standard Randomised Controlled Trials registry (reference ISRCTN15544176) and the NIHR portfolio; the Integrated Research Application System (IRAS) reference number is 234837.

Recruitment under each version of the consent:

Consent Form (v1.1) 06-FEB-19; (284 consented; in use May 2019 - August 2019) - this version only had HES/ONS wording.

Consent Form (v2.0) 15-AUG-19; (1000 consented; in use February 2020 - December 2020) - this version only had HES/ONS wording - edit not related to linkage.

Consent Form (v2.1) 27-APR-20; (2045 consented; in use February 2021 – July 2021) - this version only had HES/ONS - edit not related to linkage.

Consent Form (v3.0) 01-MAY-21; (812 consented; in use September 2021 - October 2021) - this version had GP/Medicines wording.

Consent Form (v3.1) 25-OCT-21; (902 consented; in use November 2021 - January 2022) - this version had GP/Medicines wording - edit not related to linkage.

The University of Oxford established the Clinical Trial Service Unit (CTSU), now within the Nuffield Department of Population Health, in the 1980s to conduct large trials such as the International Study of Infarct Survival (ISIS) trials. Since then CTSU has successfully completed a number of landmark studies including the 20,000 participant Heart Protection Study, the 9500 participant SHARP study, the 26,000 participant HPS-2/THRIVE study and the 30,000 participant HPS-3/REVEAL study. More recently, CTSU has coordinated the 40,000+ RECOVERY trial of treatments for COVID-19, which showed that dexamethasone, tocilizumab, and the REGN-COVID2 monoclonal antibodies were effective in reducing all-cause mortality in this population.

AMALFI is sponsored by the University of Oxford and is adopted onto the NIHR Portfolio as an academic study. iRhythm are the manufacturers of ZioPatch, the device which is being used in the study, and are providing it free of charge. iRhythm have not had and will not have any involvement in the design, conduct, analysis, or reporting of this study.

Although iRhythm could have an indirect commercial benefit if the study was to be positive (i.e., expanded use of the Zio Patch device in UK practice), this would only occur as a consequence of the study providing evidence of a benefit for public health (and pending a detailed assessment by policymakers such as the National Institute for Health and Care Excellence and others, whom would only recommend screening with the Zio Patch if any commercial benefit gained by iRhythm was outweighed by the benefits to health and social care). Moreover, the study will include cost-effectiveness analyses that will factor in not only the benefits but also any costs from the proposed intervention, producing important data for health economics analyses to be conducted in the future by policymakers, such as the National Institute for Health and Care Excellence.

The flows of data:

1. The University of Oxford will provide NHS England with personal identifiers (sex, date-of-birth, and NHS number) plus a study ID for a cohort of consenting participants in the AMALFI study, via Secure Electronic File Transfer (SEFT). The University of Oxford will resend the cohort to NHS England before each data dissemination to ensure that any potential new withdrawals of consent are removed.

2. NHS England will link the cohort to the datasets requested and send the data back to University of Oxford via SEFT. The data will contain

- identifiable data of date-of-birth and NHS number.

- for the Civil Registration Deaths dataset, Cause of Death will also be flowed which is classed as identifiable as it is a free text field and so could potentially contain identifying information.

- requested data from the datasets.

University of Oxford are requesting that date-of-birth and NHS number (which is provided in step 1 above) are flowed back to University of Oxford for the purpose of linkage validation by the research team. The research team would like to ensure that the NHS England data is correctly linked given a specific cohort is being studied and many of the study outcomes will rely on the provision of correctly linked data.

The study ID is used to link the information received from NHS England to the cohort. However, this does not allow the University of Oxford to confirm that the data actually belongs to the correct person – therefore it is standard practice to cross-check identifiers provided by NHS England with those that the University of Oxford already hold to confirm correct linkage (note that NHS number and date of birth are already known to the University of Oxford, therefore NHS England will not be providing data that is not already held).

If a participant withdraws their consent to participation in the study (including for data linkage with NHS England), their wish will be recorded and they will not be included in the cohort sent to NHS England for linkage. The linkage cohort will be updated before each data dissemination to accommodate any potential new withdrawals of consent. Data from withdrawn participants is stored up until the point of withdrawal, with no further data collection taking place after that point.

The record-level identifiable data received from NHS England will be stored in a secure location within England and Wales and will only be accessed by individuals within the Clinical Trial Service Unit who have authorisation to access the data for the purpose(s) described, all of whom are substantive employees of the University of Oxford and have been appropriately trained in data protection and confidentiality. The raw data will be securely held at the CTSU within The University of Oxford in a restricted database (with access limited to a very small number of individuals), and it will not be shared outside the University of Oxford.

Data will be stored in an encrypted study database in a pseudonymised form, with identifiable fields such as NHS number or names encrypted in situ using state-of-the-art encryption methods (in line with the NHS Data Security and Protection toolkit). Access to this database is restricted to the study team and on a need to know basis, protected with username and password, and can only be accessed from internal networks. Information stored in the database can be accessed either via an internal desktop program (used to log information and contacts with participants), or directly within the database server. For the desktop program, only the study team individuals who have direct contact with participants can see identifiable fields. Within the database server, access to the identifiable fields is locked with a decryption key held by the two database managers, who are qualified programmers and substantial employees of the University of Oxford. The data will not be used by individuals outside the AMALFI study team and therefore there are no perceived risks of re-identification at this stage (outside the group of people who already have access to identifiable information as a consequence of their roles in managing the trial and contacting with participants).

Besides the data linkage described in this Agreement, AMALFI is also collecting electronic primary care records directly from participating GP practices via local data linkage; besides this, no other linkage is currently being pursued. The linked data received from NHS England will be stored separately from the primary care data collected via the participating GP practices until the data analysis stage (where it may be combined to produce the study outcomes). The data to be provided will not be matched to publicly available data.

The data collected via this Agreement will be used to determine the presence of atrial fibrillation diagnoses and associated symptoms (primary care and HES APC), quantify and characterise reasons for hospital admission (HES APC) and death (Civil Registrations), assess medication use (primary care and Dispensing data). In addition to this, the University of Oxford will further estimate healthcare resource use (hospital admissions, A&E attendances, outpatient appointments, need for diagnostic tests) for health economics analyses (using HES APC, HES critical care, Emergency Care Data Set, and HES outpatients).

The main results of AMALFI are expected in mid-late 2024, with long-term results in 2026. Further long-term results may be published if a longer timeframe for assessment is considered important following these results.

Dissemination of the results will be aimed at three different audiences: 1) clinical and research community; 2) patients and charities; and 3) policymakers.

The results will be disseminated widely within the clinical and research community, including presentation at relevant conferences (such as the European Society of Cardiology congress or the European Heart Rhythm Association congress) and publication in a high-impact medical journal such as Circulation, Journal of the American College of Cardiology, or European Heart Journal. Further academic papers (including results of cost-effectiveness analysis and papers about the trial methods) will be published in high impact, peer-reviewed journals (possible outlets include Trials, Clinical Trials, EuroPace, European Heart Journal Quality of Care and Clinical Outcomes) and on the trial website.

For patients and charities, a non-technical summary of the main study findings will be sent to participants and relevant charities, such as the British Heart Foundation and Arrhythmia Alliance (as well as NIHR who has funded the trial up to this stage) and published on the study website. Patient and Public Involvement will be sought in the design and sharing of outputs from the Public Advisory Panel at the Nuffield Department of Population Health (https://www.ndph.ox.ac.uk/research/participant-panel).

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

The medicines data is not deemed disclosive and information on a GP level is available in the public domain. However, should the study team consider that published information poses a risk of re-identification, the following suppression methodology should be applied:

· Zeros should be shown.

· 1-7 to be rounded to 5.

· Any other numbers rounded to nearest 5.

· Rounding unnecessary for averages etc.

· Percentages calculated from rounded values.

· If zeros need to be suppressed, round to 5.

1. Provision of the data requested may help AMALFI provide important results that may affect standard clinical practice in the UK and beyond, by generating reliable evidence on the efficacy and cost-effectiveness of remote screening for subclinical AF using a self-applied patch. In particular, the datasets requested in this data dissemination are expected to form the basis for the assessment of the impacts of AF screening on anticoagulation rates, hospital admissions, outpatient appointments and A&E attendances, and deaths. While several monitoring devices are available, there is scarce randomised evidence of the added value of using such devices on top of usual care, particularly in a longer time frame (which may lead to AF cases being detected regardless of the use of the device, diluting its benefit and therefore cost-effectiveness). It is therefore expected that results from AMALFI might be incorporated in future National Institute for Health and Care Excellence (NICE), the Scottish Intercollegiate Guidelines Network (SIGN), and UK National Screening Committee guidelines; in particular, and depending on what the results of efficacy and cost-effectiveness show, AMALFI might provide the basis for a recommendation of targeted screening in elderly patients with additional comorbidities using a wearable patch in the UK, most likely in the primary care setting (and potentially managed via a central coordinating system replicating the methodology used in AMALFI). These results also have the potential to affect the care of millions of people worldwide, with potentially significant benefits in terms of reduction of fatal or disabling stroke and reduced healthcare costs (by means of reduced number of hospitalisations or A&E attendances) – the extent of which will be assessed by the detailed efficacy and cost-effectiveness analyses planned.

2. AMALFI is a streamlined and remote study with minimal data collection undertaken by participants, and none by their GPs. Participants are only asked to complete a short one-page questionnaire at enrolment (with remote EQ5D questionnaires planned at 2.5 and 5 years after inclusion), while GPs will perform a standard data extraction from their records at approximately 1, 2.5, and 5 years after randomisation. While these simple procedures make it easier for both participants and GPs to take part, they are limited in their capacity to provide detailed data that is of interest to the study (such as hospitalizations, secondary care appointments, mortality, and others). Therefore, access to data held by NHS England for participants in the AMALFI study will be crucial for detailed assessments of the impact of AF screening on a range of outcomes that would be hard to capture through more bespoke methods. Moreover, the provision of data on primary care records and medications as held by NHS England will allow the study team to potentially drop the need for GPs to run any data extractions for follow-up, further reducing the burden of taking part in research and making clinical trials more efficient, cost-effective, and attractive to busy clinicians. The comparative assessments of data collected directly from GPs or similar data held by NHS England may fuel the use of centrally held data by other researchers. The methods developed as part of this study may then be shared with the scientific community to improve the development of streamlined trials in cardiovascular disease and other fields.

3. The results should inform future discussions on the potential implementation of a nationwide screening program for subclinical atrial fibrillation and will be developed and disseminated to the wider public with the assistance of the Public Advisory Panel at the Nuffield Department of Population Health. If such a program were to be implemented based on these results, it could potentially involve several thousands of eligible patients in the UK each year, while paving the way for similar programs in other countries around the world.

4. Finally, the University of Oxford will aim to reach policymakers such as the National Institute for Health and Care Excellence (NICE), the Scottish Intercollegiate Guidelines Network (SIGN), and UK National Screening Committee (UKNSC). Although the University of Oxford does not expect an immediate policy change following the results of this study, the data produced may be used by both bodies to guide future recommendations for subclinical AF screening in UK practice. The University of Oxford will do this by registering as a stakeholder with these institutions and commenting on their AF management and screening guidance documents based on the findings.

Data received so far was used initial exploratory work, which informed the development of the statistical analysis plan and baseline characteristics and outcome derivation procedures. It also helped identify deaths of study participants so that communication with them could be stopped.

Data already received during the initial agreement period will be used for production of the final results (commencing Q1 2025 and due Q3 2025), as specified in the Data Sharing Agreement.

Data to be collected as part of the agreement renewal requested for the period 2025-2027 will complement the data already received for use in the main analyses scheduled in 2025, and support the long-term analyses scheduled for 2027.