Creating a next-generation phenotype library: the health data research UK Phenotype Library.

Objective: To enable reproducible research at scale by creating a platform that enables health data users to find, access, curate, and re-use electronic health record phenotyping algorithms. Materials and Methods: We undertook a structured approach to identifying requirements for a phenotype algorithm platform by engaging with key stakeholders. User experience analysis was used to inform the design, which we implemented as a web application featuring a novel metadata standard for defining phenotyping algorithms, access via Application Programming Interface (API), support for computable data flows, and version control. The application has creation and editing functionality, enabling researchers to submit phenotypes directly. Results: We created and launched the Phenotype Library in October 2021. The platform currently hosts 1049 phenotype definitions defined against 40 health data sources and >200K terms across 16 medical ontologies. We present several case studies demonstrating its utility for supporting and enabling research: the library hosts curated phenotype collections for the BREATHE respiratory health research hub and the Adolescent Mental Health Data Platform, and it is supporting the development of an informatics tool to generate clinical evidence for clinical guideline development groups. Discussion: This platform makes an impact by being open to all health data users and accepting all appropriate content, as well as implementing key features that have not been widely available, including managing structured metadata, access via an API, and support for computable phenotypes. Conclusions: We have created the first openly available, programmatically accessible resource enabling the global health research community to store and manage phenotyping algorithms. Removing barriers to describing, sharing, and computing phenotypes will help unleash the potential benefit of health data for patients and the public.

Prospective study design and data analysis in UK Biobank.

Population-based prospective studies, such as UK Biobank, are valuable for generating and testing hypotheses about the potential causes of human disease. We describe how UK Biobank's study design, data access policies, and approaches to statistical analysis can help to minimize error and improve the interpretability of research findings, with implications for other population-based prospective studies being established worldwide.

Understanding Views Around the Creation of a Consented, Donated Databank of Clinical Free Text to Develop and Train Natural Language Processing Models for Research: Focus Group Interviews With Stakeholders.

BACKGROUND: Information stored within electronic health records is often recorded as unstructured text. Special computerized natural language processing (NLP) tools are needed to process this text; however, complex governance arrangements make such data in the National Health Service hard to access, and therefore, it is difficult to use for research in improving NLP methods. The creation of a donated databank of clinical free text could provide an important opportunity for researchers to develop NLP methods and tools and may circumvent delays in accessing the data needed to train the models. However, to date, there has been little or no engagement with stakeholders on the acceptability and design considerations of establishing a free-text databank for this purpose. OBJECTIVE: This study aimed to ascertain stakeholder views around the creation of a consented, donated databank of clinical free text to help create, train, and evaluate NLP for clinical research and to inform the potential next steps for adopting a partner-led approach to establish a national, funded databank of free text for use by the research community. METHODS: Web-based in-depth focus group interviews were conducted with 4 stakeholder groups (patients and members of the public, clinicians, information governance leads and research ethics members, and NLP researchers). RESULTS: All stakeholder groups were strongly in favor of the databank and saw great value in creating an environment where NLP tools can be tested and trained to improve their accuracy. Participants highlighted a range of complex issues for consideration as the databank is developed, including communicating the intended purpose, the approach to access and safeguarding the data, who should have access, and how to fund the databank. Participants recommended that a small-scale, gradual approach be adopted to start to gather donations and encouraged further engagement with stakeholders to develop a road map and set of standards for the databank. CONCLUSIONS: These findings provide a clear mandate to begin developing the databank and a framework for stakeholder expectations, which we would aim to meet with the databank delivery.

Incidence, morbidity, mortality and disparities in dementia: A population linked electronic health records study of 4.3 million individuals.

INTRODUCTION: We report dementia incidence, comorbidities, reasons for health-care visits, mortality, causes of death, and examined dementia patterns by relative deprivation in the UK. METHOD: A longitudinal cohort analysis of linked electronic health records from 4.3 million people in the UK was conducted to investigate dementia incidence and mortality. Reasons for hospitalization and causes of death were compared in individuals with and without dementia. RESULTS: From 1998 to 2016 we observed 145,319 (3.1%) individuals with incident dementia. Repeated hospitalizations among senior adults for infection, unknown morbidity, and multiple primary care visits for chronic pain were observed prior to dementia diagnosis. Multiple long-term conditions are present in half of the individuals at the time of diagnosis. Individuals living in high deprivation areas had higher dementia incidence and high fatality. DISCUSSION: There is a considerable disparity of dementia that informs priorities of prevention and provision of patient care.

A survey on clinical natural language processing in the United Kingdom from 2007 to 2022.

Much of the knowledge and information needed for enabling high-quality clinical research is stored in free-text format. Natural language processing (NLP) has been used to extract information from these sources at scale for several decades. This paper aims to present a comprehensive review of clinical NLP for the past 15 years in the UK to identify the community, depict its evolution, analyse methodologies and applications, and identify the main barriers. We collect a dataset of clinical NLP projects (n = 94; £ = 41.97 m) funded by UK funders or the European Union's funding programmes. Additionally, we extract details on 9 funders, 137 organisations, 139 persons and 431 research papers. Networks are created from timestamped data interlinking all entities, and network analysis is subsequently applied to generate insights. 431 publications are identified as part of a literature review, of which 107 are eligible for final analysis. Results show, not surprisingly, clinical NLP in the UK has increased substantially in the last 15 years: the total budget in the period of 2019-2022 was 80 times that of 2007-2010. However, the effort is required to deepen areas such as disease (sub-)phenotyping and broaden application domains. There is also a need to improve links between academia and industry and enable deployments in real-world settings for the realisation of clinical NLP's great potential in care delivery. The major barriers include research and development access to hospital data, lack of capable computational resources in the right places, the scarcity of labelled data and barriers to sharing of pretrained models.

'What is the risk to me from COVID-19?': Public involvement in providing mortality risk information for people with 'high-risk' conditions for COVID-19 (OurRisk.CoV).

Patients and public have sought mortality risk information throughout the pandemic, but their needs may not be served by current risk prediction tools. Our mixed methods study involved: (1) systematic review of published risk tools for prognosis, (2) provision and patient testing of new mortality risk estimates for people with high-risk conditions and (3) iterative patient and public involvement and engagement with qualitative analysis. Only one of 53 (2%) previously published risk tools involved patients or the public, while 11/53 (21%) had publicly accessible portals, but all for use by clinicians and researchers.Among people with a wide range of underlying conditions, there has been sustained interest and engagement in accessible and tailored, pre- and postpandemic mortality information. Informed by patient feedback, we provide such information in 'five clicks' (https://covid19-phenomics.org/OurRiskCoV.html), as context for decision making and discussions with health professionals and family members. Further development requires curation and regular updating of NHS data and wider patient and public engagement.

Estimated impact of the COVID-19 pandemic on cancer services and excess 1-year mortality in people with cancer and multimorbidity: near real-time data on cancer care, cancer deaths and a population-based cohort study.

OBJECTIVES: To estimate the impact of the COVID-19 pandemic on cancer care services and overall (direct and indirect) excess deaths in people with cancer. METHODS: We employed near real-time weekly data on cancer care to determine the adverse effect of the pandemic on cancer services. We also used these data, together with national death registrations until June 2020 to model deaths, in excess of background (pre-COVID-19) mortality, in people with cancer. Background mortality risks for 24 cancers with and without COVID-19-relevant comorbidities were obtained from population-based primary care cohort (Clinical Practice Research Datalink) on 3 862 012 adults in England. RESULTS: Declines in urgent referrals (median=-70.4%) and chemotherapy attendances (median=-41.5%) to a nadir (lowest point) in the pandemic were observed. By 31 May, these declines have only partially recovered; urgent referrals (median=-44.5%) and chemotherapy attendances (median=-31.2%). There were short-term excess death registrations for cancer (without COVID-19), with peak relative risk (RR) of 1.17 at week ending on 3 April. The peak RR for all-cause deaths was 2.1 from week ending on 17 April. Based on these findings and recent literature, we modelled 40% and 80% of cancer patients being affected by the pandemic in the long-term. At 40% affected, we estimated 1-year total (direct and indirect) excess deaths in people with cancer as between 7165 and 17 910, using RRs of 1.2 and 1.5, respectively, where 78% of excess deaths occured in patients with ≥1 comorbidity. CONCLUSIONS: Dramatic reductions were detected in the demand for, and supply of, cancer services which have not fully recovered with lockdown easing. These may contribute, over a 1-year time horizon, to substantial excess mortality among people with cancer and multimorbidity. It is urgent to understand how the recovery of general practitioner, oncology and other hospital services might best mitigate these long-term excess mortality risks.

A semi-supervised approach for rapidly creating clinical biomarker phenotypes in the UK Biobank using different primary care EHR and clinical terminology systems.

OBJECTIVES: The UK Biobank (UKB) is making primary care electronic health records (EHRs) for 500 000 participants available for COVID-19-related research. Data are extracted from four sources, recorded using five clinical terminologies and stored in different schemas. The aims of our research were to: (a) develop a semi-supervised approach for bootstrapping EHR phenotyping algorithms in UKB EHR, and (b) to evaluate our approach by implementing and evaluating phenotypes for 31 common biomarkers. MATERIALS AND METHODS: We describe an algorithmic approach to phenotyping biomarkers in primary care EHR involving (a) bootstrapping definitions using existing phenotypes, (b) excluding generic, rare, or semantically distant terms, (c) forward-mapping terminology terms, (d) expert review, and (e) data extraction. We evaluated the phenotypes by assessing the ability to reproduce known epidemiological associations with all-cause mortality using Cox proportional hazards models. RESULTS: We created and evaluated phenotyping algorithms for 31 biomarkers many of which are directly related to COVID-19 complications, for example diabetes, cardiovascular disease, respiratory disease. Our algorithm identified 1651 Read v2 and Clinical Terms Version 3 terms and automatically excluded 1228 terms. Clinical review excluded 103 terms and included 44 terms, resulting in 364 terms for data extraction (sensitivity 0.89, specificity 0.92). We extracted 38 190 682 events and identified 220 978 participants with at least one biomarker measured. DISCUSSION AND CONCLUSION: Bootstrapping phenotyping algorithms from similar EHR can potentially address pre-existing methodological concerns that undermine the outputs of biomarker discovery pipelines and provide research-quality phenotyping algorithms.

UK phenomics platform for developing and validating electronic health record phenotypes: CALIBER.

OBJECTIVE: Electronic health records (EHRs) are a rich source of information on human diseases, but the information is variably structured, fragmented, curated using different coding systems, and collected for purposes other than medical research. We describe an approach for developing, validating, and sharing reproducible phenotypes from national structured EHR in the United Kingdom with applications for translational research. MATERIALS AND METHODS: We implemented a rule-based phenotyping framework, with up to 6 approaches of validation. We applied our framework to a sample of 15 million individuals in a national EHR data source (population-based primary care, all ages) linked to hospitalization and death records in England. Data comprised continuous measurements (for example, blood pressure; medication information; coded diagnoses, symptoms, procedures, and referrals), recorded using 5 controlled clinical terminologies: (1) read (primary care, subset of SNOMED-CT [Systematized Nomenclature of Medicine Clinical Terms]), (2) International Classification of Diseases-Ninth Revision and Tenth Revision (secondary care diagnoses and cause of mortality), (3) Office of Population Censuses and Surveys Classification of Surgical Operations and Procedures, Fourth Revision (hospital surgical procedures), and (4) DM+D prescription codes. RESULTS: Using the CALIBER phenotyping framework, we created algorithms for 51 diseases, syndromes, biomarkers, and lifestyle risk factors and provide up to 6 validation approaches. The EHR phenotypes are curated in the open-access CALIBER Portal (https://www.caliberresearch.org/portal) and have been used by 40 national and international research groups in 60 peer-reviewed publications. CONCLUSIONS: We describe a UK EHR phenomics approach within the CALIBER EHR data platform with initial evidence of validity and use, as an important step toward international use of UK EHR data for health research.

Phenotyping UK Electronic Health Records from 15 Million Individuals for Precision Medicine: The CALIBER Resource.

Electronic health records (EHR) are increasingly being used for observational research at scale. In the UK, we have established the CALIBER research resource which utilizes national primary and hospital EHR data sources and enables researchers to create and validate longitudinal disease phenotypes at scale. In this work, we will describe the core components of the resource and provide results from three exemplar research studies on high-resolution epidemiology, disease risk prediction and subtype discovery which demonstrate both the opportunities and challenges of using EHR for research.

Between Scylla and Charybdis: Charting the Wicked Problem of Reusing Health Data for Clinical Research Informatics.

OBJECTIVES: Recognising dilemmas posed by the sharing and reuse of health data as a classic wicked problem and uncover some current key challenges to clinical research informatics. METHODS: A modified thematic review process including identification of agreed critical research questions, appropriate query terms and search strategy, identification of relevant papers in accordance with inclusion criteria, and authors' co-review of full text papers. RESULTS: Queries returned 4,779 papers published between January 2014 and November 2017. A shortlist of 197 abstracts was analysed and 18 papers were finally selected for review. Thematic assessment of findings revealed four key challenges: (1) uncertain reliability of consent as a cornerstone of trust due to the limits to understanding and awareness of data sharing; (2) ethical challenges around equity and autonomy; (3) ambitious overly theoretical governance frameworks lacking practical validity; and (4) a clear desire for further public and individual engagement to achieve clearer and more nuanced knowledge dissemination around data sharing practice and governance frameworks. CONCLUSIONS: Understanding the wicked problem of reusing clinically acquired health data for research purposes is essential if clinical research is to benefit from informatics advances. A lack of understanding around the context of data acquisition and sharing undermines the foundations of patient-professional trust. Efforts to protect privacy, where tailoring to specific contexts is a key driver, should support the development of solutions which more adequately honour privacy needs, justify access, and protect equity and autonomy.

Serotonin reuptake inhibitors and mortality in epilepsy: A linked primary-care cohort study.

OBJECTIVE: Preliminary evidence suggests that serotonin reuptake inhibitor (SRI) use may increase postictal respiratory drive and prevent death. We sought to determine whether SRIs are associated with improved all-cause and possible seizure-specific mortality in patients with epilepsy. METHODS: Patients with epilepsy and a random 10:1 sample without epilepsy were extracted from The ClinicAl research using LInked Bespoke studies and Electronic health Records (CALIBER) resource. The hazard ratio (HR) of all-cause and possible seizure-specific mortality, treating SRI use as a time-varying covariate, was determined using the date of a second SRI prescription as exposure and in discrete 6-month periods over the entire duration of follow-up. We used Cox regression and competing risk models with Firth correction to calculate the HR. We controlled for age, sex, depression, comorbidity, (Charlson comorbidity index) and socioeconomic status (Index of Multiple Deprivation). RESULTS: We identified 2,718,952 eligible patients in CALIBER, of whom 16,379 (0.60%) had epilepsy. Median age and follow-up were 44 (interquartile range [IQR] 29-61]) and 6.4 years (IQR 2.4-10.4 years), respectively, and 53% were female. A total of 2,178 patients (13%) had at least two SRI prescriptions. Hazard of all-cause mortality was significantly elevated following a second prescription for an SRI (HR 1.64 95% confidence interval [95% CI] 1.44-1.86; p < 0.001). The HR was similar in 163,778 age, sex, and general practitioner (GP) practice-matched controls without epilepsy. Exposure to an SRI was not associated with seizure-related death (HR 1.08, 95% CI 0.59-1.97; 0.796). SIGNIFICANCE: There is no evidence in this large population-based cohort that SRIs protect against all-cause mortality or seizure-specific mortality. Rather, SRI use was associated with increased mortality, irrespective of epilepsy, which is probably due to various factors associated with the use of antidepressants. Larger studies with systematically collected clinical data are needed to shed further light on these findings.

Evaluation of Machine Learning Methods to Predict Coronary Artery Disease Using Metabolomic Data.

Metabolomic data can potentially enable accurate, non-invasive and low-cost prediction of coronary artery disease. Regression-based analytical approaches however might fail to fully account for interactions between metabolites, rely on a priori selected input features and thus might suffer from poorer accuracy. Supervised machine learning methods can potentially be used in order to fully exploit the dimensionality and richness of the data. In this paper, we systematically implement and evaluate a set of supervised learning methods (L1 regression, random forest classifier) and compare them to traditional regression-based approaches for disease prediction using metabolomic data.

Use of electronic health records to ascertain, validate and phenotype acute myocardial infarction: A systematic review and recommendations.

Electronic health records (EHRs) offer the opportunity to ascertain clinical outcomes at large scale and low cost, thus facilitating cohort studies, quality of care research and clinical trials. For acute myocardial infarction (AMI) the extent to which different EHR sources are accessible and accurate remains uncertain. Using MEDLINE and EMBASE we identified thirty three studies, reporting a total of 128658 patients, published between January 2000 and July 2014 that permitted assessment of the validity of AMI diagnosis drawn from EHR sources against a reference such as manual chart review. In contrast to clinical practice, only one study used EHR-derived markers of myocardial necrosis to identify possible AMI cases, none used electrocardiogram findings and one used symptoms in the form of free text combined with coded diagnosis. The remaining studies relied mostly on coded diagnosis. Thirty one studies reported positive predictive value (PPV)≥ 70% between AMI diagnosis from both secondary care and primary care EHRs and the reference. Among fifteen studies reporting EHR-derived AMI phenotypes, three cross-referenced ST-segment elevation AMI diagnosis (PPV range 71-100%), two non-ST-segment elevation AMI (PPV 91.0, 92.1%), three non-fatal AMI (PPV range 82-92.2%) and six fatal AMI (PPV range 64-91.7%). Clinical coding of EHR-derived AMI diagnosis in primary care and secondary care was found to be accurate in different clinical settings and for different phenotypes. However, markers of myocardial necrosis, ECG and symptoms, the cornerstones of a clinical diagnosis, are underutilised and remain a challenge to retrieve from EHRs.

Translational phases of evidence in a prognostic biomarker: a systematic review and meta-analysis of natriuretic peptides and the prognosis of stable coronary disease.

CONTEXT: Translational phases of study are important in evaluating whether a prognostic biomarker is likely to have impact on clinical practice but systematic evaluations of such evidence are lacking. OBJECTIVE: To systematically evaluate the clinical usefulness of the published literature on the association of natriuretic peptides (NP) and prognosis in stable coronary disease. DATA SOURCES: MEDLINE and EMBASE until the end of July 2009, without restrictions. STUDY SELECTION: Prospective studies measuring NP in people with stable coronary disease who were followed-up for all cause mortality, coronary or cardiovascular events. DATA EXTRACTION: Two independent reviewers categorised studies according to the American Heart Association phase of study, and extracted data according to the study reporting guidelines from the American Heart Association and REMARK. RESULTS: Systematic review of 19 studies found 17 which were phase 2, reporting an association between NP and events, two phase 3 studies, statistically examining the incremental prognostic value of NP, but no studies assessing whether NP predicted risk sufficiently to change management (phase 4), improve clinical outcomes (phase 5) or cost effectiveness (phase 6). No study referred to a statistical analytic protocol. Meta-analysis of 14 studies, reporting 18,841 patients and 1655 outcome events, found an RR for events of 3.28 (95% CI 2.45 to 4.38) comparing top versus bottom third of NP. This effect was 26% lower among the five studies which adjusted for a priori confounders (age, sex, renal function and left ventricular function) and 38% lower when adjusting for publication bias (Egger's p=0.001). CONCLUSION: The unbiased strength of association of NP with prognosis in stable coronary disease is unclear, and there is a lack of reports of clinically useful measures of prediction and discrimination or studies relating NP levels to clinical decision making. The available literature is confined to early phases and is of limited clinical usefulness.

Evaluating the quality of research into a single prognostic biomarker: a systematic review and meta-analysis of 83 studies of C-reactive protein in stable coronary artery disease.

BACKGROUND: Systematic evaluations of the quality of research on a single prognostic biomarker are rare. We sought to evaluate the quality of prognostic research evidence for the association of C-reactive protein (CRP) with fatal and nonfatal events among patients with stable coronary disease. METHODS AND FINDINGS: We searched MEDLINE (1966 to 2009) and EMBASE (1980 to 2009) and selected prospective studies of patients with stable coronary disease, reporting a relative risk for the association of CRP with death and nonfatal cardiovascular events. We included 83 studies, reporting 61,684 patients and 6,485 outcome events. No study reported a prespecified statistical analysis protocol; only two studies reported the time elapsed (in months or years) between initial presentation of symptomatic coronary disease and inclusion in the study. Studies reported a median of seven items (of 17) from the REMARK reporting guidelines, with no evidence of change over time. The pooled relative risk for the top versus bottom third of CRP distribution was 1.97 (95% confidence interval [CI] 1.78-2.17), with substantial heterogeneity (I(2) = 79.5). Only 13 studies adjusted for conventional risk factors (age, sex, smoking, obesity, diabetes, and low-density lipoprotein [LDL] cholesterol) and these had a relative risk of 1.65 (95% CI 1.39-1.96), I(2) = 33.7. Studies reported ten different ways of comparing CRP values, with weaker relative risks for those based on continuous measures. Adjusting for publication bias (for which there was strong evidence, Egger's p<0.001) using a validated method reduced the relative risk to 1.19 (95% CI 1.13-1.25). Only two studies reported a measure of discrimination (c-statistic). In 20 studies the detection rate for subsequent events could be calculated and was 31% for a 10% false positive rate, and the calculated pooled c-statistic was 0.61 (0.57-0.66). CONCLUSION: Multiple types of reporting bias, and publication bias, make the magnitude of any independent association between CRP and prognosis among patients with stable coronary disease sufficiently uncertain that no clinical practice recommendations can be made. Publication of prespecified statistical analytic protocols and prospective registration of studies, among other measures, might help improve the quality of prognostic biomarker research.

Assessing the cost effectiveness of using prognostic biomarkers with decision models: case study in prioritising patients waiting for coronary artery surgery.

OBJECTIVE: To determine the effectiveness and cost effectiveness of using information from circulating biomarkers to inform the prioritisation process of patients with stable angina awaiting coronary artery bypass graft surgery. DESIGN: Decision analytical model comparing four prioritisation strategies without biomarkers (no formal prioritisation, two urgency scores, and a risk score) and three strategies based on a risk score using biomarkers: a routinely assessed biomarker (estimated glomerular filtration rate), a novel biomarker (C reactive protein), or both. The order in which to perform coronary artery bypass grafting in a cohort of patients was determined by each prioritisation strategy, and mean lifetime costs and quality adjusted life years (QALYs) were compared. DATA SOURCES: Swedish Coronary Angiography and Angioplasty Registry (9935 patients with stable angina awaiting coronary artery bypass grafting and then followed up for cardiovascular events after the procedure for 3.8 years), and meta-analyses of prognostic effects (relative risks) of biomarkers. RESULTS: The observed risk of cardiovascular events while on the waiting list for coronary artery bypass grafting was 3 per 10,000 patients per day within the first 90 days (184 events in 9935 patients). Using a cost effectiveness threshold of pound20,000- pound30,000 (euro22,000-euro33,000; $32,000-$48,000) per additional QALY, a prioritisation strategy using a risk score with estimated glomerular filtration rate was the most cost effective strategy (cost per additional QALY was < pound410 compared with the Ontario urgency score). The impact on population health of implementing this strategy was 800 QALYs per 100,000 patients at an additional cost of pound 245,000 to the National Health Service. The prioritisation strategy using a risk score with C reactive protein was associated with lower QALYs and higher costs compared with a risk score using estimated glomerular filtration rate. CONCLUSION: Evaluating the cost effectiveness of prognostic biomarkers is important even when effects at an individual level are small. Formal prioritisation of patients awaiting coronary artery bypass grafting using a routinely assessed biomarker (estimated glomerular filtration rate) along with simple, routinely collected clinical information was cost effective. Prioritisation strategies based on the prognostic information conferred by C reactive protein, which is not currently measured in this context, or a combination of C reactive protein and estimated glomerular filtration rate, is unlikely to be cost effective. The widespread practice of using only implicit or informal means of clinically ordering the waiting list may be harmful and should be replaced with formal prioritisation approaches.

Ethnic differences in long-term improvement of angina following revascularization or medical management: a comparison between south Asians and white Europeans.

BACKGROUND: It is not known whether there are disparities in morbidity outcomes between south Asians and whites with established coronary disease. METHODS: Six-year prospective cohort study to determine whether improvement of angina symptoms differs between 196 south Asians and 1508 whites following revascularization or medical management. RESULTS: 43.9% of south Asians reported improvement in angina at 6 years compared with 60.3% of whites (age-adjusted OR 0.56, 95% CI 0.41-0.76, adjusted for diabetes, hypertension, smoking, number of diseased vessels, left ventricular function and social class OR 0.59, 95% CI 0.41-0.85). Similar proportions of whites and south Asians underwent percutaneous coronary intervention (PCI) (19.6% versus 19.9%) and coronary artery bypass surgery (CABG) (32.8% versus 30.1%). South Asians were less likely to report improved angina after PCI (OR 0.19, 95% CI 0.06-0.56) or CABG (OR 0.36, 95% CI 0.17-0.74). There was less evidence of ethnic differences in angina improvement when treatment was medical (OR 0.87, 95% CI 0.48-1.57). CONCLUSION: South Asians were less likely to experience long-term improvements in angina than whites after receipt of revascularization. Further research is needed to identify why these ethnic groups differ in symptomatic prognosis following revascularization for coronary disease and how these differences may be mitigated.