Obtaining participant consent Patients were invited to read a study information sheet and encouraged to ask questions about the research, after which written informed consent was sought from patients for: a) completion of a baseline health questionnaire; b) provision of a blood sample for long-term storage and analysis of DNA and biomarkers; c) extraction of patient data from their medical records; and d) linkage of patient data with national electronic health record data sources (e.g. coded hospitalisations and procedures, emergency admissions and cause-specific mortality).
Sample size For sufficient power to detect relative risk differences of 1.2 at a 5% significance level a minimum of 3000 patients was required.
Baseline data collection and measures Three unique sources of data contributed to the participant baseline assessment, which took place during patients' scheduled hospital appointments. Research measures (biological specimens and information not routinely collected during patients' clinical assessments) included 1) blood samples for DNA and biomarker analysis and a 2) short health questionnaire completed by participants (usual completion time 5-10 minutes). Clinical measures were 3) detailed clinical information collected as part of usual care, extracted from the hospital electronic health record up to six months after the baseline assessment to allow time for procedure and test results to be uploaded to hospital electronic systems.
A. Collection of research measures (i) Blood resource Venous blood sampling and processing of samples Non-fasting blood samples were obtained from consenting participants, including those in whom a diagnosis of angina was subsequently ruled out, by staff trained in venepuncture during the hospital appointment. Twenty millilitres of blood was drawn to enable a wide range of research measures to be obtained. Blood was drawn into five vacutainers in the following order: 2x 4.0 mL ethylenediaminetetraacetic acid tube (EDTA), 1x 4.5 mL gel-separator lithium heparin tube (plasma separation tube (PST)), 1x 5.0 mL gel serum separator tube (SST), 2x 4.0 mL, and 1x 2.5 mL PAXgene ribonucleic acid (RNA) blood tube (PreAnalytiX, Franklin Lakes, NJ). Vacutainers were gently inverted and reverted up to five times to allow additives to mix with the blood, labelled and temporarily stored at approximately 5°C until processing. Date and time of collection was recorded. Different blood fractions were separated by centrifugation at 2500 g for 10 minutes. EDTA and PST samples were centrifuged within four hours after sampling and the SST sample was left to clot at room temperature for 25-30 minutes prior to centrifugation.
EDTA plasma, heparinised plasma, serum and white cells (buffy coat) were aliquoted into a total of nineteen 0.5 mL 2D QR-coded cryovial tubes suitable for long-term cryopreservation. Each cryovial was individually logged in two un-linked databases using a single-tube QR-code reader, placed in a 96-position rack and stored locally at -20°C. PAXgene RNA tubes were stored as whole blood at -20°C.
Buccal swab sample collection Consenting patients who did not provide a venous blood sample were invited to provide a buccal swab for subsequent DNA extraction. Samples were collected by rubbing a sterile swab firmly against the inside of the participant's cheek or between their lips and the gum line for one minute. The swab was stored in a sterile container with a silica gel capsule (Isohelix Dri-Capsules, Cell Projects Ltd) to stabilise the sample at room temperature for up to two years.
Samples were transported in temperature-controlled shipping boxes for long-term storage at a biobank at -80°C.
Extraction of DNA and single-nucleotide polymorphism (SNP) genotyping DNA was extracted from 2556 individual participant samples and genotyping of 51 CHD-specific SNPs identified in Coronary ARtery DIsease Genome wide Replication and Meta-analysis (CARDIoGRAM PlusC4D) was completed, according to specification, by a company with specialised services for extraction of DNA and genotyping.
Biomarkers and metabolomic profiling of patients The investigators propose to measure routinely collected and novel markers that appear most promising for clinical predictive value, including N-terminal brain natriuretic peptide (NT-BNP), high-sensitivity C-reactive protein (hs-CRP), apolipoprotein A1 (apoA1), apolipoprotein B (apoB), lipoprotein A (Lp(a)), and cystatin C.
(ii) Baseline health questionnaire All participants were invited to complete a questionnaire to ascertain or validate participants' sex, level of education, ethnicity and health. General health functioning was assessed using the EQ-5D, (http://www.euroqol.org/) a 5-item standardised measure of health outcome. Functional chest pain was determined using the Rose angina questionnaire and angina severity using the Canadian Cardiovascular Society (CCS) classification. Symptoms of depression were assessed using the Patient Health Questionnaire-9 (PHQ-9), which scores each of the nine Diagnostic and Statistical Manual of Mental Disorders (DSM-IV 4th Edition, American Psychiatric Association) criteria for depression in the PRIME-MD (Primary Care Evaluation of Mental Disorders screening questionnaire for depressive symptoms) diagnostic instrument for common mental disorders. Anxiety was assessed using the 7-item Generalized Anxiety Disorder (GAD-7) scale, a self-reported measure for screening and quantifying severity of generalized anxiety disorder.
(iii) Extended physical examination and measures obtained in sub-study Additional measures obtained from a subset of participants included a sixty-minute physical examination (anthropometric measures, lung function, aortic blood pressure, pulse wave \[augmentation\] index, pulse wave velocity and ankle brachial index), 31-hour ambulatory ECG, tri-axial accelerometry and sleep measures. Sub-study participants also completed an extended health questionnaire which included questions on diet and sleep.
B. Extraction of clinical information from the electronic health record Trained extractors reviewed hospital databases and recorded detailed clinical information onto standardised case report forms. Depending on the clinic or hospital, different clinical systems were searched to manually extract data that could not be downloaded or exported as a data file. Information included reason(s) for referral, history of chest pain, physician-recorded risk factors for coronary artery disease, previously diagnosed medical history, current medication and test results.
All participants underwent a resting 12-lead electrocardiogram (ECG) as part of their routine clinical assessment. An exercise tolerance test (ETT) was undertaken by an experienced cardiac electrophysiologist in hospitals that performed ETTs for routine assessment of chest pain. The ETT followed the modified Bruce protocol (Bruce, RA, 1963) and was conducted with concurrent 12-lead electrocardiography.
Data were manually entered onto a research database at the study coordinating centre, cleaned and coded ready for analysis.
Follow-up Postal questionnaire A postal follow-up questionnaire was sent to study participants after recruitment (mean time to follow-up 469 days) to assess general health status (EQ-5D), functional status (Rose angina questionnaire, Seattle angina questionnaire), symptoms of depression (PHQ-2) and anxiety (GAD-7) and use of nitrate medication.
Electronic health record linkages With their consent, the investigators linked participants to data from Hospital Episode Statistics, a national data warehouse of administrative data containing ICD-10-coded hospital diagnoses and OPCS4-coded procedures. The investigators additionally linked participants with ICD-10-coded cause-specific mortality data from the Office for National Statistics. The linkages were conducted via the NHS Health and Social Care Information Centre (formerly known as the NHS Information Centre) and patients were identified using their NHS number (a unique ten digit numeric identifier for the NHS), gender, date of birth and postcode.
Data collection and management The design, conduct, analysis and reporting of the study follow the methodological standards set out in the REporting recommendations for tumour MARKer prognostic studies (REMARK) guidelines. To maintain confidentiality, identifiers were removed from all sources of data and replaced with a unique participant study number assigned at the point of enrolment. Researchers were trained to follow standardised operating procedures to ensure high quality, consistent collection of data. Data were checked and validated, and inconsistencies in clinical data were resolved by revisiting hospital data at source.
