MR1291: Clinical Cohorts in Coronary Disease Collaboration (4C)

Mortality data was supplied to University College London (UCL) by ONS and subsequently the Health and Social Care Information Centre (which has since become NHS Digital) for the purpose of a research study referred to as MR1291, Clinical Cohorts in Coronary Disease Collaboration (4C).

This Data Sharing Agreement permits the retention of the data for an interim period but no other processing of the data is permitted.

Permission to retain the data for the interim period is a practical step to enable the study to comply with the necessary NHS Digital standards, legal and ethical requirements. If, for any reason, it is not possible for the study to meet the necessary requirements, this Agreement will be terminated and destruction of the data will be required.

The following information provides background information on the purpose of the original study:

Background:

The Clinical Cohorts in Coronary disease Collaboration (4C), led by Prof Harry Hemingway and Dr Riyaz Patel at University College London, was established in 2009 as part of a large NIHR programme grant for applied research (Ref: RP-PG-0407-10314) to evaluate important new opportunities for improving quality of care and outcomes for cardiovascular disease (CVD) patients, integrated across the patient journey, and at different levels of care. Preventing the onset or progression of CVD and improving outcomes remains an important national health priority. We have previously shown that chest pain is a more frequent initial presentation of coronary artery disease than unheralded acute myocardial infarction (AMI) (Herrett E et al. Eur Heart J Acute Cardiovasc Care 2013; 2(3):235-245) thus new-onset chest pain may represent a potential intervention point that could facilitate the identification of those at risk of acute cardiac events. Assessing risk of future health outcomes is high on research, clinical healthcare and policy agendas, and consented cohorts recruited in clinical settings make important contributions towards addressing translational research questions in prognosis research. High quality prognosis research in clinical populations can improve understanding of how biomarkers or genes interact with environmental factors associated with disease by linking imaging, -omic and bespoke phenotypic data with rich clinical information held in patients electronic health records (EHR). Despite their potential advantages, large, prospective clinical cohorts are lacking across all disease areas, and few exist for coronary artery disease.

4C consists of 3345 patients who underwent investigation for new-onset, stable chest pain recruited from rapid access chest pain clinics, cardiac catheter laboratories and coronary angiogram pre-assessment clinics at four NHS hospitals in London and Bristol between 2009 and 2014. Patients consented to providing blood samples for biomarker and DNA analysis and completed a health questionnaire to assess functional status, quality of life, mood and social position at baseline. 4C forms part of the Global GENIUS-CHD (GENetIcs of sUbSequent Coronary Heart Disease) Consortium (http://www.genius-chd.com/), an international consortium of 63 clinical international cohorts with the aim of studying the genetic and non-genetic determinants of future events in patients with heart disease. The consortium was established in 2014 and contains data from cohorts from 17 countries, including over 270,000 patients - the largest effort of its kind studying determinants of risk for subsequent coronary heart disease events.

Patients involved in the 4C study have consented to the collection of:

• Self-reported health questionnaire data including information about the patients’ chest pain and general health and one-year follow up health questionnaire

• A blood sample for research (circulating biomarkers and DNA)

• Access to their GP and hospital medical record

• Access to information relating to their past, present and future hospital admissions (HES – received under NIC-152452-N5J8N) and mortality, embarkation and registration data (Medical Research (ONS))

Detailed clinical information was manually extracted from hospital electronic databases by trained study staff. -Omics (genomics and metabolomics) data are linked to the detailed phenotypic data obtained from the EHR and participant questionnaire data. Data collection ceased in 2012. Genetic and metabolomic analyses of samples were carried out between 2014-15. Data were analysed and a paper prepared for publication in 2016. Proteomic analyses were subject to funding and are planned for Q4 2019.

Data Summary:

University College London (UCL) currently holds record-level, identifiable Medical Research (then ONS) data against the cohort of patients. Under the previous version of this agreement, UCL received the following MRIS reports to receive mortality and demographic data:

- MRIS Cause of Death report

- MRIS Cohort Event Notification report

- MRIS Flagging Current Status report

- MRIS Members and Postings report

- MRIS Personal Demographics Service

The data was minimised to the cohort and requested on a Quarterly basis to allow continued follow-up on patients. Using the above Medical Research data, linked with HES (under NIC-152452) and bespoke study data, UCL the applicants aimed to study the long-term (10-years, from the close of recruitment in 2014) causes and consequences of coronary disease among patients with stable chest pain by following-up patients for the primary pre-specified composite endpoint of all cardiovascular deaths, non-fatal acute myocardial infarction, stroke and peripheral arterial disease.

UCL also received HES data for part of their cohort (833 patients) under enquiry reference NIC-152452-N5J8N. UCL intended to receive HES data for the full cohort of patients, however this was not disseminated under NIC-152452-N5J8N. UCL intend to amend this agreement, once a Short-Term Extension has been approved, to received further HES data on their participants. The delay in receiving further HES data under NIC-152452-N5J8N has effected the production of outputs during the validity period of version 0.1 of this agreement, details of which are listed below.

This application forms part of a long standing agreement with NHS Digital, with the previous data sharing agreement running from 2012 to 2017. The purpose of this application is to request a short-term extension to grant us permission to retain data received up until 2017 when this data sharing agreement expired. The legal basis for processing personal data within this application under General Data Protection Regulation, is Article 6(1)(e) “processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller”. For processing of special categories of personal data, the legal basis is Article 9(2)(j) “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 sole data controller is UCL, Gower Street, London, WC1E 6BT. The data processor is UCL Institute of Health Informatics (IHI), 222 Euston Road, London NW1 2DA. No other organisations are involved in the wider project and no funders/commissioners are involved.

Under this agreement, UCL previously received demographic and mortality (date and causes of death), registration & embarkation data for all patients recruited to 4C since date of enrolment to the study in order to ensure complete follow-up data, including outcomes and confounding variables, for all patients in the study.

Patients additionally consented to access and linkage to their past, present and future medical and other health-related records for research purposes.

Data was processed according to the following steps:

1. UCL sent patient NHS Numbers and demographic details (date of birth, sex and postcode) of the established cohort of participants to NHS Digital for tracing.

2. NHS Digital produced reports for mortality and registration data using participant data.

3. NHS Digital sent the resulting Medical Research reports to UCL to be received by the study data manager.

4. The study data manager removed patient identifiers and stored the data, pseudonymised at Study ID, in the UCL Data Safe Haven secure study database.

5. UCL IHI approved analysts processed the data using the statistical analysis package STATA.

6. Baseline characteristics of the cohort and DNA analyses were published in a funder final report using bespoke study data (see Outputs below). UCL intends to produce data reporting 10-year outcomes of patients using MRIS requested under this agreement and HES data to be requested under an amendment, in conjunction with forthcoming proteomic analyses. Outputs will be produced in aggregated form with small number suppressed and published in an open access journal and included on the news section of the UCL IHI website (https://www.ucl.ac.uk/health-informatics/) which is made available for UCL researchers.

Medical Research (MRIS) data was downloaded from NHS Digital by the study data manager within UCL and stored in a secure database (project share) in the UCL Data Safe Haven (http://www.ucl.ac.uk/isd/itforslms/services/handling-sens-data/tech-soln) to which only approved study staff have access. The server is held on-site at UCL and access is restricted to named individuals according to UCL’s security policy. The Data Safe Haven is governed by a strict information governance framework that is used throughout the UCL School of Life and Medical Sciences of which the Senior Information Risk Owner (SIRO) is the Dean of the Faculty of Population Health Sciences, Professor Graham Hart.

DNA and biomarker samples, questionnaire and clinical data extracted from EHRs were linked to Medical Research report (formerly Office for National Statistics [ONS]) mortality and demographic data (DARS-NIC-152228-DL5MK) and Hospital Episode Statistics (HES) inpatient, outpatient and A&E data (NIC-152452-N5J8N - HES data were sent on 833 patients) using patients NHS number, date of birth, sex and postcode.

Remote access to the project, shared by researchers analysing the data, is permitted but only via secure token for approved researchers (so processing is carried out on site), and with local printing and access to internet disabled. Only study staff who have passed NHS Digital information governance training (https://www.ucl.ac.uk/isd/it-for-slms/research-ig/information-governance-training-awareness-service) and who are authorised as a member of the study team were able to access NHS Digital data for this study. No record-level data will be exported outside the Data Safe Haven or shared with any third party organisation or user. No aggregate data with small numbers un-suppressed will be exported from the Data Safe Haven.

Study team members who will be responsible for processing study data will hold a substantive contract with UCL. Researchers responsible for analysing the data will only have access to de-identified data. The data requested will only be used for the purposes described in this application. The full HES-mortality data will be accessible only by the PI or researchers responsible for managing (downloading and extracting data) the resource, curating and cleaning the data.

Researchers cannot export results from their own analyses from the UCL Data Safe Haven. Rights to export will be restricted to only authorised staff within the study team. Aggregate data outputs for export can only be exported through the established disclosure control procedure. The UCL Data Safe Haven requires specific authorisation of staff who can export outputs. Staff authorised to export aggregate outputs will control outputs by scrutinising aggregate tables and figures to assess whether the outputs meet the following requirements:

• the HES analysis guide

• the ICO Anonymisation Code of Practice

• the Anonymisation Standard for Publishing Health and Social Care Data Specification (ISB1523)

• the ONS Disclosure control guidance for birth and death statistics

• UCL SLMS Health Informatics requirements

• Pseudonymisation ISO/TS 25237:2008 Overview for audit purposes

A joiners and leavers standard operating procedure will apply to identify training requirements for new staff and rescind

access to the Data Safe Haven on completion of the study or when staff leave/staff appointment status changes.

Data analysis is carried out using the statistical analysis package STATA. Baseline data have been analysed and published (see 5c. below). HES-mortality data will be linked to bespoke study data (bloods, questionnaire and clinical data extracted from EHRs) by the study data manager at UCL IHI using the unique participant study identifier for patients enrolled in the study. Following approval of this application for a short-term extension we will submit a DARS amendment to request full renewal of the data sharing agreement with NHS Digital and request follow-up HES and ONS data for study patients from date of expiry of the data sharing agreement (2017) to date so we may carry out analyses on 10-year outcomes (CVD events and all-cause mortality) for study patients.

Direct outputs from the study were produced in aggregated form with small numbers suppressed in the form of an open access journal available to all in line with UCL’s open access publication policy. Access to this data is via secure token permissions for approve researchers. To date, outputs produced in this form have utilised bespoke study data from the sources below:

• Self-reported health questionnaire data including information about the patients’ chest pain and general health and one-year follow up health questionnaire

• A blood sample for research (circulating biomarkers and DNA)

• Access to their GP and hospital medical record

Outputs to date have currently excluded HES linked to MRIS report data due to the UCL only holding hospital records for a portion of their cohort.

Applicant researchers using these aggregate reports have a strong track record in publishing research in high impact journals. Examples of the use of 4C data include the protocol, baseline characteristics of the cohort and patient’s genotyping data being used within projects aimed at improving patient care for patients suffering from angina and heart attacks (https://www.journalslibrary.nihr.ac.uk/programmes/pgfar/RP-PG-0407-10314/#/) and 4C data contributing to the GENIUS-CHD Consortium (Patel R et al. Circ Genom Precis Med. 2019 Mar 21 DOI: 10.1161/CIRCGEN.119.002470; DOI: 10.1161/CIRCGEN.119.002471) have been published. This research has been cited as a resource for CVD research (https://academic.oup.com/eurheartj/article/39/16/1481/4096831) and is registered on ClinicalTrials.gov (Identifier: NCT02402478).

Planned outputs:

The HES-mortality data which forms the subject of the planned application will lead to submission of at least two research papers for publication in leading peer-reviewed journals. It is currently envisaged that tThe first paper will describe 10-year follow-up of patients for CVD events and all-cause mortality. The second paper will focus on multi-omic (genomic, metabolomic and proteomic) associations with CVD events and mortality at 10-year follow up among this cohort. Any publication arising from analysis of study data will include only aggregated data summarised in tables and figures. We will submit research papers within 12-18 months of approval of our application for full renewal of the data sharing agreement. In accordance with UCL’s policy on open access publishing, we will submit all publications to fully open access journals.

4C forms part of GENIUS-CHD (http://www.genius-chd.com/), an international consortium of 63 clinical cohorts containing coronary heart disease patients and DNA to associate genetic variants with subsequent events. The consortium was established in 2014 and contains data from cohorts from 17 countries, including over 270,000 patients - the largest effort of its kind studying determinants of risk for subsequent CHD events. 4C's contribution has already seen benefits being realised in the development of the consortium, as detailed above. Further benefits will be gained through 4C's role in GENIUS utilising further outputs from the study which will incorporate the use of HES & MRIS report data to better understand patient pathways.

Collaborative research based on individual patient data should be encouraged to strengthen prognostic model development and external validation of models. This is timely, with new standards to improve the quality of prognosis research and new opportunities to link clinical health-care data with other sources of electronic information and bespoke data at scale. To provide a new resource for prognosis research, we established 4C, a novel, contemporary clinical cohort which includes a DNA and biomarker resource linked to hospital EHR, questionnaire data and health outcomes. A combination of -omic, clinical and phenotypic data, 4C will support research on the causes of stable coronary disease and its main complication, acute coronary syndrome, and the complex interactions between genetic and environmental determinants of post-acute myocardial infarction outcomes.

Use of MRIS report (ONS) data provided as part of this application will support investigation of long-term outcomes for patients with stable coronary disease. Further use of this data will benefit the study by working to achieve the aim of follow-up on patient journey over a 10 year period (2014-2024). Use of HES & MRIS data within outputs made available to researchers are expected to broaden the scope of impact. Research findings will be published in a high impact, open access scientific journal. Study findings can be used to inform policy around the management of patients with stable coronary disease.

Due to the lower than expected frequency of outputs produced from the study and HES linked to MRIS report data being excluded form outputs due to only receiving HES data on a sub-cohort of participants from the total cohort, there is a lack of benefits being realised as a result of previous outputs. Future benefits are expected as follows:

1) As a resource

Despite their value in translational research, few large prospective clinical cohorts have been established in stable coronary artery disease. In the report to the National Institute of Health Research (Hemingway H et al. Programme Grants Appl Res 2017 (Feb);5(4)) the researcher highlighted a need for embedding genetic information in hospital EHR at scale in order to carry out a wide range of translational research, from discovery efforts, through drug repurposing and embedded pharmacogenetics for patient safety. 4C was established as a resource for research into cardiovascular disease and will fill this important gap.

2) Translational benefits

Data are contributing to the GENIUS Consortium, (http://www.genius-chd.com/) an international collaboration of investigators seeking to collectively better understand the genetic and non-genetic drivers of subsequent or recurrent events in those who have established CHD.

Clinical record data are highly effective in distinguishing risk groups, for diverse diseases and in diverse settings and higher risk patients usually have more absolute benefit than those in lower risk groups (i.e. without biologic interaction). Clinical risk prediction algorithms and decision support are rapidly proliferating in CVD and many tools can be envisaged in the management of a single patient, spanning benefits and harms at different time points. Clinical data can outperform the Framingham risk score and can flexibly model start point populations and endpoints and be easily updated in the light of new imaging, genetic information, and implemented in clinical practice. Predictions may be improved by incorporating clinical trajectories. For example patients in whom blood pressure declines over time, without diagnosed heart failure, have a worse survival than those whose blood pressure remains stable.