Data linkage request for FAMOUS-NSTEMI study

Ischaemic heart disease (IHD) persists as the leading global cause of death and lost life years in adults. In the UK, there are more than 2 million men and women living with angina (form of IHD, a condition marked by severe pain in the chest that often spreads to the shoulders, arms, and neck, leading to an inadequate blood supply to the heart) in the UK. Overall, IHD due to coronary artery disease (CAD) remains a worldwide public health problem of unmet need. The national audit of myocardial infarction (MINAP) in England and Wales indicates around 80,000 patients experience a non-ST segment elevation myocardial infarction (NSTEMI - a type of heart attack that happens when a part of your heart is not getting enough oxygen) each year; please see here - https://my.clevelandclinic.org/health/diseases/22233-nstemi-heart-attack. NSTEMI is the more common type of myocardial infarction (MI, also known as ‘heart attack’, comprising approximately 70% of MI events). The standard care for acute MI is ‘invasive management’ with a view to identifying a culprit blockage, or blockages, that could be treated with a stent or bypass surgery. Around 100,000 patients undergo a stent procedure in the NHS each year and approximately 60% have a diagnosis of NSTEMI.

NSTEMI is a condition associated with considerable diversity– patients may present sub-acutely (delayed presentation; >3.0 days), may have multiple health problems, and multiple coronary artery (arteries providing oxygen to the heart muscles) narrowings. These complexities create challenges for clinicians, leading to variations in treatment decisions that sometimes may underpin adverse outcomes. The appropriate adoption of novel medical technologies, such as Fractional Flow Reserve (FFR) tests, can lead to improvements in medical decisions and patient outcomes. FFR is a pressure-derived ratio of whether a coronary artery narrowing is limiting blood flow in a coronary artery and reflects its severity. In patients with stable angina, an FFR value of 0.80 is a prognostically validated treatment threshold. In other words, clinical trials have found that patients with an FFR value of ≤ 0.80 have a lower risk of adverse cardiac events (e.g. cardiac death and MI) when treated with stents than patients treated with medication (tablets) only without stents. Equally, patients with stable angina and an FFR > 0.80 have a good prognosis when treated with medication alone.

For patients with MI and multiple narrowed coronary arteries, clinicians and their patients lack information on whether all narrowings should be treated or just the ‘culprit’. The FAMOUS-NSTEMI trial was the first to address this question in a registry-based, randomised, controlled trial. The FAMOUS-NSTEMI trial focused on the diagnostic and clinical value of using FFR to guide treatment decisions in patients with a recent heart attack. The FFR threshold for a flow-limiting coronary narrowing is less certain in patients with MI as compared to patients with stable angina symptoms. In the FAMOUS-NSTEMI trial, FFR was measured in NSTEMI patients at the time of the initial baseline angiogram (blood vessel health check), and the treatment plan was randomly assigned (FFR-guided management vs. standard care without FFR-guided management).

FAMOUS-NSTEMI was a registry-based clinical trial. Participants were enrolled in 6 NHS hospitals, two in Scotland and four in England, between October 2011 – May 2013. In patients with visually obvious, obstructive coronary artery narrowings, an angiogram is sufficient for clinicians to make a definitive treatment plan for use of stents or bypass surgery. All participants who gave consent but were not eligible to continue in the study based on the exclusion criteria at angiography were enrolled into the registry only. All participants who gave informed consent and who were eligible for randomisation continued into the randomised trial and were also enrolled into the registry. The main reason for not progressing from enrolment to randomisation was disclosure of severe coronary artery disease. The registry patients who were enrolled following informed consent, but not randomised, will provide a control group to contextualise the trial results e.g. with respect to selection bias, or not. Combining registry and randomised together, in total, 580 participants were enrolled between the Scottish hospitals, and 167 between the English hospitals. The University of Glasgow have section 251 approval form the Health Research Authority (HRA) Confidentiality Advisory Group (CAG) to address the common law duty of confidentiality and to provide a lawful basis for the patient identifiers to flow. Section 251 support was sought because the patient information sheet and consent form did not explicitly state that data would be held and analysed at the University of Glasgow. Section 251 approval was granted in April 2021 and gave approval for the University of Glasgow to transfer patient identifiers (Name, NHS Number, date of birth and postcode) to NHS Digital and for the flow of HES and mortality data back to University of Glasgow.

At the time of designing the FAMOUS-NSTEMI trial (2010), the clinical evidence to support the use of FFR-guided management in patients with recent MI was not well established. Remarkably, this remains the case in 2021. Despite this, the FFR test is widely used in the NHS, including in MI patients. Since the publication of the original trial results in 2014, there is still no other publication of long-term outcomes following FFR-guided management in patients with acute MI. However, in recent years, the validity of FFR-guided management has been called into question. Notable publications include

- limited reproducibility of FFR in patients with recent MI (van der Hoeven NW et al JAMA Cardiol. 2019 Aug 1;4(8):736-744; PMID: 31268466),

- a lack of overall benefit in the large, multicentre FLOWER-AMI trial in France (Puymirat N Engl J Med. 2021 Jul 22;385(4):297-308. PMID: 33999545) and

- the as yet unpublished FUTURE trial (including patients with stable symptoms or acute MI) in France which reported an increase in mortality in patients with FFR-guided management (https://www.tctmd.com/news/future-results-confirm-more-deaths-ffr-guided-treatment-choice).

These new findings are very concerning. They highlight the possibility of adverse outcomes in patients with myocardial infarction undergoing FFR-guided management. The FAMOUS-NSTEMI trial is the only trial with the potential to provide data on longer term prognosis. These results are critically important and potentially practice-changing information to inform health policy in this patient group and best practice in the NHS.

FAMOUS-NSTEMI was originally published in the European Heart Journal in 2014 using health outcome data to 1 year. The study was funded by a 3-year Project Grant from the British Heart Foundation. Follow-up involving contact with study participants was truncated to 12 months since the BHF grant duration was 3-years. Since only 39 patients experienced a major adverse cardiac event (MACE), the study lacked statistical power to assess the relationships between FFR and health outcomes.

In designing this trial, use of electronic record linkage to permit an analysis of the effect of the intervention (FFR-guided management) on longer term prognosis was predefined. The cohort being submitted to NHS Digital will comprise the 167 patients recruited in English hospitals. A linkage has already been carried out for the Scottish patients by e-DRIS (NHS Digital’s equivalent organisation in Scotland). The size of the Scottish cohort ended up at 553 (there were 576 patients sent through to eDRIS, but there were some data issues/no matches). The criteria for preparing the cohort to be provided to NHS Digital will be as follows:

- Relevant sites only (for NHS Digital this will be English sites)

- Consented randomised participants - data requested from Consent Date onwards

- Consented registry participants - data requested from Consent Date onwards

- Participants who have 'withdrawn consent and use of data' (as noted on the withdrawal screen of the eCRF) will be removed from the datasets

- Participants with missing key identifier fields will be flagged and removed from the dataset.

Inclusion of data from the 167 participants recruited in NHS England is of the utmost importance in order that the full study population can be included in the main analysis, hence the results can be reported as a complete representation of the longer term, pre-specified analysis.

Given the duration of follow-up, taking into account data from eDRIS for sites in NHS Scotland and NHS Digital for sites in NHS England, the University of Glasgow anticipate that around 100 MACE events will have occurred by 2020/2021. This number of events is expected to confer adequate power, and a preliminary analysis of the data provided by eDRIS confirms this will be the case.

The University of Glasgow now wish to obtain longer-term follow-up data from NHS Digital for health outcomes through case record linkage. The maximum duration of follow-up will be from the date of the first patient enrolled in NHS Southampton (April 2012) to the most recent date of linkage (2021/2022). The longer follow-up period will result in more events being included therefore the statistical power of the analysis will be increased implying enhanced clinical relevance of the final results. The event rate of predefined adverse events including all-cause death, cardiovascular death and hospitalisation for MI, heart failure and stroke was 10% in the first year and in line with published literature on longer term outcomes post-MI, the annual event rate is anticipated to be ≥5% in subsequent years. Accordingly, taking the full study population of participants enrolled in NHS Scotland and NHS England, this analysis will be adequately powered to identify differences in clinically important outcomes between the randomised groups, and in a position to report reliable results that will be directly relevant to clinical practice. Inclusion of NHS Digital data is essential for the delivery of the study’s predefined analysis.

The datasets being requested are pseudonymised Hospital Episode Statistics (HES) Admitted Patient Care (APC) and Civil Registrations (Deaths). These individual-level data are essential since they will enable an analysis of whether, compared with standard care, FFR-guided management is associated with fewer or more deaths and hospitalisations for a cardiovascular cause in patients undergoing invasive coronary angiography for a NSTEMI. This question has not been previously addressed in the medical literature, hence patient care in the NHS is not well informed. The University of Glasgow have focused on clinically important outcomes e.g. death, hospitalisation for myocardial infarction, hence the results of the predefined analysis should be impactful on patient care in the NHS and clinical guidelines. If the study analysis identifies a signal of harm with FFR-guided management (as has been suggested in recent studies), then the analysis will have practice-changing implications.

The data requested will relate only to patients in England within a clinical trial. Data requested is restricted to the period from the date of informed consent to present day which will allow the University of Glasgow to evaluate the within-trial and post-trial events. There are no other alternative, less intrusive ways to obtain this information. Without the data from NHS Digital the study cannot be completed.

The health outcome events will then be linked with the initial FFR values to assess for a prognostic relationship between initial FFR and future clinical outcomes. This analysis will provide unique information on whether FFR is valid in this patient group and whether FFR-guided treatment decisions might improve treatment and patient outcomes in the longer term.

The data controller is the University of Glasgow, and the data will be processed in the Robertson Centre for Biostatistics, University of Glasgow. Statistical analysis using NHS Digital data will be conducted by statisticians in the Robertson Centre for Biostatistics. The results of these analyses will be reported in scientific manuscripts/publications. This is the scope of work and there is no other programme involved.

Longer term studies of patient survival and wellbeing are crucial to inform medical decision making. In the case of this analysis, there are no data on the long term implications of using FFR-guided management in patients with a heart attack. The data from NHS Digital will very much enhance this research and provide clinically important information that could not be obtained in any other way. The University of Glasgow’s lawful bases for processing are GDPR Article 6 (1) (e) (task in the public interest) and Article 9 (2) (j) (archiving in the public interest, scientific or historical research or statistical purposes).

The study sponsor is The Golden Jubilee National Hospital (health-board in Glasgow). For the record linkage data, the sponsor does not determine the purpose and the means of processing the data from NHS Digital or process any NHS Digital data. The University of Glasgow is the sole data controller for record linkage data.

The long-term follow-up study is supported by a peer-reviewed grant application to the British Heart Foundation (BHF, Project Grant no. PG/18/52/33892). BHF do not determine the purpose or the manner in which the data will be processed.

The plans for record linkage have been described in the Patient and Public Involvement meetings

• A lecture followed by discussion at the Annual Conference of the West of Scotland Research Ethics Service (a patient and public involvement day that took place on 16 November 2012 in the Central Hotel, Glasgow),

• A lecture followed by discussion to the Kings Fund on 10 October 2012 in London

• A lecture followed by discussion at the British Heart Foundation, the Annual Meeting of the British Heart Foundation (23 October 2013, Edinburgh, attended by patients and community nurses)

• NHS Glasgow R&D Patients Day – A lecture followed by discussion at the annual patient’s day, 14 March 2017, in the Queen Elizabeth University Hospital, Glasgow.

• Meetings of the guideline committee of the European Society of Cardiology (17 June 2022 and 1 July 2022, – this group includes a patient advocacy representative who herself experienced an acute myocardial infarction. The meeting in June occurred on line as part of a recurring planning meeting of the group. The meeting in July occurred in person. The plan for long term follow-up was highlighted in these meetings.

• The current long term follow-up study has been funded by a second, peer-reviewed grant from the British Heart Foundation. The Project Grant committee includes patient / lay representatives and non-medical BHF staff. The study is highlighted in a dedicated, unrestricted webpage - Towards less-invasive monitoring after a heart attack (bhf.org.uk)

• https://www.bhf.org.uk/research-projects/the-bhf-famousnstemi-longterm-followup-study

• More generally, the FAMOUS-NSTEMI trial received widespread media coverage at the time of public presentation of the main results in 2014. The results were covered by the BBC News. It would be reasonable to believe that new findings of a mortality signal with FFR-guided management would be of public interest and again covered in the media.

The NHS Glasgow / University of Glasgow PPI group meets regularly (at least quarterly) to discuss studies. This group was updated during July 2021. Final results of this analysis will be discussed with this group when they become available and as early as possible during 2023. The Site PIs (n=6 across the UK, including 5 outwith Glasgow) will also be encouraged to share the final results with their local hospital PPI groups.

All organisations party to this agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined with the Data Sharing Framework Contract i.e.: employees, agents and contractors of the Data Recipient who may have access to that data). Individual patient data will not be published or shared with a third party.

The Robertson Centre for Biostatistics (RCB) at the University of Glasgow will securely transfer a file of the following identifiers to NHS Digital for the trial participants recruited across the four English hospital sites.:

1. NHS number,

2. Full name,

3. Date of birth

4. Postcode

5. Unique study ID

NHS Digital will return the linked Hospital Episode Statistics (HES) Admitted Patient Care (APC) and Mortality data including the unique study ID and no other identifiers.

The RCB at the University of Glasgow will store the data on a server in the RCB. Researchers can remotely access the pseudonymised data via Virtual Private Network (VPN) having gone through an internal approval process to gain VPN access. Removal of any data is technically prevented. Data will be controlled by using password encoded files and computers. The RCB has ISO 27001 standard data management processes. Data will only be held and processed by the University of Glasgow. Data will only be accessed by individuals who have authorisation to access the data for the purpose described, all of whom are substantive employees of the University of Glasgow and have been trained in data protection and confidentiality. Any data shared outside of the University of Glasgow will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

The pseudonymised data is stored in a separate location to the participant identifiers. All personal details are kept in a separate more restricted database and only visible to appropriately approved study team members. The two datasets will not be re-linked and the data will remain pseudonymised. Access to patient identifiers is limited to a small number of named individuals to allow the cohort file to be sent to NHS Digital. Staff with access to patient identifiers are not involved in the analysis of the data. The data linkage file preparation is documented by the RCB staff members carrying out the process and this work is carried out in restricted file store and database areas.

The data will be accessed via the Robertson Centre Safe Haven at the University of Glasgow, within a secure network. Summary reports will be prepared and any additional information from the study database will be included i.e., adjacent events already reported via the study web portal. Clinically qualified study staff will review the reports, via the Robertson Centre analytical environment. Only statisticians will see the data and since the data are deidentified there will be no possibility of identifying any of the participants. Staff members within RCB will have access to the patient identifiable data to perform the data linkage preparation work but this access can be revoked once the work is finalised.

The returned record linkage datasets will be able to be analysed alongside the other study data by joining on the unique study ID (the pseudonymised ID). Outputs from these joins will not be presented at record-level outside of the permitted staff group within the University of Glasgow. Record-level data will only be accessed by the University of Glasgow, and will not be shared with any organisation, including funding organisations. All of this work will be conducted in the UK.

Outputs will contain only aggregate level data with small numbers suppressed in line with HES analysis guide. The outputs will be in the form of research reports and analytical outputs. The raw data (NHS Digital output) will be retained within the Robertson Centre for Biostatistics.

The University of Glasgow intend to produce manuscript drafts within 3 months of NHS Digital providing the data. It is anticipated that the manuscripts will be publicly available 6 – 9 months from the date provided by NHS Digital. Clinicians, the public, and academics will have access to the outputs in the form of Plain English summaries and peer-reviewed scientific papers submitted to reputable journals. The reports will be made ‘Open Access’ so freely available at no cost.

The publication of the results of this analysis are intended to be made available to the National Institute for Clinical Excellence (NICE). Since NICE provide clinical guidelines and updates for the management of patients with acute myocardial infarction, the University of Glasgow believe the results will be relevant to NICE and the NHS. It is intended that the results will also be highlighted to guideline committees of international societies e.g. European Society of Cardiology.

It is anticipated the results will be published in a high impact cardiology journal e.g. European Heart Journal, impact factor 23. There is also intended to be a paper published in a top health economics journal e.g. Value In Health, impact factor 5.9. The aim is to produce these outputs as soon as possible following data receipt.

There will be a report back to the funder (British Heart Foundation) (this will be the final report at the end of the project grant), but ownership and control of the outputs rests with the University of Glasgow.

Inclusion of the clinical event data from NHS Digital is essential to deliver the primary aim of the longer-term analysis of the BHF FAMOUS-NSTEMI trial. This study included 6 NHS sites, 2 in NHS Scotland and 4 in NHS England, hence the majority of the enrolling sites were located in NHS England. Of 350 randomised patients with recent non-ST segment elevation myocardial infarction (NSTEMI), 100 (29%) were enrolled in NHS England, and an even larger number of patients were enrolled into the follow-up registry. The trial results hope to directly inform whether the clinical management of heart attack patients should be guided by a diagnostic test (fractional flow reserve) during coronary angiography. The angiogram is the standard of care test, but it is widely recognised to have limitations. FFR-guided management is evidence based for patients with stable angina, but it is not evidence-based in patients with a heart attack, which is the gap this study aims to address.

Approximately 100,000 heart attack patients undergo invasive coronary angiography in the NHS each year. FAMOUS-NSTEMI is the only study (anywhere in the world) that has assessed whether FFR-guided management might improve clinical outcomes in heart attack patients.

A prognostic analysis will be undertaken using data on health outcomes (death, myocardial infarction, etc) accrued over the years. Including information obtained from e-DRIS in NHS Scotland, it is anticipated that the number of events overall will be >100. Given that the clinical trial involved two randomised groups, it is believed that there will be sufficient statistical power to discriminate a difference in health outcomes between the randomised groups, should such a difference exist. This information hopes to be directly relevant to the Health and Social Care sector since the diagnostic test (FFR) is widely available but its use is variable. This analysis hopes to be informative in the sense that if there is a prognostic advantage conferred by FFR-guided management, then the results would support an increase in adoption of FFR in the NHS. On the other hand, if the results do not favour a benefit to patients then the future publication would serve as an appropriate disincentive to use this technology in the NHS. In this case, the analysis would serve to reduce inappropriate use of this technology (currently £300 per test). Currently, there is a knowledge gap on whether or not FFR-guided management should be adopted for patients with a recent heart attack in the NHS. This gap can be addressed through this analysis, which represents justification for future data sharing. Whatever the results may be from this analysis, it hopes to be relevant to the doctors and managers in the NHS. Results will be submitted for publication in peer-reviewed journals. Following publication, the results will become known to authors of clinical practice guidelines (such as NICE), and the conclusions can be expected to inform practice guideline recommendations for patient care in the NHS.

Regardless of the nature of the results, a long term health outcome analysis enabled by NHS Digital information will add value to medical knowledge and the literature.

Yielded Benefits is not a requirement for new applications.