MMMehrdad A. Mizani

Mehrdad A. Mizani

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Mehrdad A. Mizani has a total of 625 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Predicting COVID-19 hospitalisation and common disease risk from comorbid diagnoses in 13 million individuals

    by Hongjiao Liu, Mehrdad A. Mizani, Yujie Zhao, Angela Wood, Michael Inouye, Alkes L. Price, Xilin Jiang

    Predicting disease risk from prior diagnoses is fundamental to clinical decision-making, particularly during health emergencies such as the COVID-19 pandemic, when individuals with long-term conditions may be disproportionately vulnerable to adverse outcomes. Despite intense interest in developing models to predict disease risk from prior diagnoses 1–3 , most prediction models do not estimate effects of each prior diagnosis on disease risk conditional on other diagnoses, limiting interpretability and clinical utility. We developed the Comorbidity Risk Score (CRS), trained on 13 million individuals (age 40-69) from linked electronic health record (EHR) datasets of the entire population of England, to predict COVID-19 hospitalisation and 87 other disease outcomes. CRS was trained at close to saturated sample size and precisely estimated the effects of 212 prior diagnoses on the 88 disease outcomes, conditional on all other prior diagnoses. Correlations of CRS effect sizes across outcomes (e.g. 0.76 for myocardial infarction vs. hyperlipidaemia) matched the corresponding genetic correlations (e.g. 0.79 for myocardial infarction vs. hyperlipidaemia), confirming that comorbidity architectures capture disease aetiology. On average, CRS identified 5% of the population with 3.4-fold higher disease risk, including myocardial infarction (4.4-fold), lung cancer (6.5-fold), and COVID-19 hospitalisation (6.3-fold). Using prior diagnoses alone, CRS outperformed state-of-the-art clinical COVID-19 models 4 . Furthermore, CRS ( N =13 million) substantially outperformed state-of-the-art AI 1 ( N =0.5 million) and linear 3 ( N =0.5 million) models in predicting disease risk, suggesting that training sample size outweighs model complexity. CRS attained near-perfect transferability across self-reported ethnicities (e.g., Black vs. White: AUROC ratio = 97.3%). Finally, CRS distinguished independently predictive comorbidities from indirect associations, e.g., lipid metabolism disorder was a strong predictor of myocardial infarction risk but not ischaemic stroke, after conditioning on other prior diagnoses. In conclusion, CRS provides a comprehensive resource for understanding the impact of comorbidities on COVID-19 and other future diseases, revealing disease aetiology while enabling powerful prediction of disease risk.

  • Open Access

    Healthcare utilisation and cost among individuals with Long COVID in England and Wales: potential value of specialised post-COVID services

    by Ashkan Dashtban, Mehrdad Mizani, Yi Mu, Mamas Mamas, Evan Kontopantelis, Spiros Denaxas, Qi Huang, Kamlesh Khunti, Melissa Heightman, Rouven Priedon, et al.

    Objective: To compare healthcare utilisation and costs for Long COVID (LC) in England and Wales. Design: Case–control cohort analysis with multiple age-, sex-, ethnicity-, deprivation-, region- and comorbidity-matched control groups: (1) COVID-only, no LC; (2) pre-pandemic; (3) contemporary non-COVID and (4) pre-LC (self-controlled, pre-COVID pandemic). Setting: National, population-based, linked UK electronic health records (British Heart Foundation/NHS England Secure Data Environment). Participants: Adults aged ⩾18 years with LC between January 2020 and December 2023. Main outcome measures: Healthcare utilisation (number of consultations/visits per person): primary care (general practitioner (GP)), secondary care (outpatient, inpatient and emergency department, investigations and procedures) and inflation-adjusted cost (£) for LC and control populations. Results: In England ( n = 295,180) and Wales ( n = 7925), LC was associated with higher utilisation and costs across primary and secondary care. Mean annual total cost for LC versus contemporary controls was £6681.6 versus £2039.9 in Wales (3.27-fold) and £3378.3 versus £1355.8 in England (2.49-fold). GP consultations averaged 26.76 (SD 22.07) per person-year in the Welsh LC cohort versus 14.63 (SD 14.34) in England (absolute difference 12.13 visits/year). Cross-national multiplicative difference-in-differences estimates indicated a 25% greater relative LC-related cost increase in Wales compared with England (ratio 1.25, 95% CI 1.09–1.42), corresponding to an absolute excess of £2501 per person-year (95% CI £1707–£3270). Conclusions: LC is associated with sustained increases in healthcare utilisation and costs in England and Wales. Our findings suggest that system-level service design may influence healthcare use and costs following COVID-19, warranting further evaluation of post-viral care models.


  • Open Access

    Incidence and prevalence of Takayasu arteritis in England before and after the COVID-19 pandemic: results from RECORDER

    by Stephanie J Lax, Mehrdad A Mizani, Megan Rutter, David Groves, Peter Lanyon, Matthew J Grainge, Sean McPhail, Peter Stilwell, Jeanette Aston, Mary Bythell, et al.

    Objectives: Takayasu arteritis (TAK) is a rare large vessel vasculitis associated with substantial morbidity and mortality. Evidence on its epidemiology, including the COVID-19 pandemic’s impacts, is incomplete. We aimed to describe the incidence and prevalence of TAK in England using national routinely collected healthcare data. Methods: Using Hospital Episode Statistics for the whole of England (Admitted Patient Care and Outpatients), we identified all people with the International Classification of Diseases, 10th revision code for TAK. We validated our case ascertainment and diagnosis date refinement strategies in five National Health Service Trusts and using the unique Imperial College TAK cohort of 123 confirmed cases. We calculated the incidence of TAK in England from 2006 to 2023 (cohort design), particularly focusing on the COVID-19 pandemic, and point prevalence for 1 July 2022 (cross-sectional design). Results were reported by key demographic variables. Results: The incidence of TAK in England was 1.3 (95% CI 1.3, 1.4) per million per year and was stable except for a 30% decrease in 2020–2021 (COVID-19 pandemic). The prevalence of TAK in mid-2022 was 22.5 per million (95% CI 21.3, 23.8). The median age at diagnosis was 53 years. Incidence and prevalence were both higher in women and in Asian/Asian British and Other ethnic groups. There was variation between different geographic areas, but socio-economic status did not have a marked effect. Conclusion: Nearly 1300 people have TAK in England, with >70 new cases each year apart from during the COVID-19 pandemic. These estimates, based on a 10-fold larger cohort than previous European studies, will be essential in planning and commissioning pandemic-prepared specialist services.

  • Open Access

    An empirical assessment of differential privacy in real-world observational data: a case-control study of asthma exacerbation in UK Biobank linked with electronic health records

    by Mehrdad A Mizani, Aziz Sheikh, Amitava Banerjee

    Objectives: Electronic health records (EHRs) provide substantial resources for observational studies, yet present significant challenges in safeguarding patient privacy while maintaining research quality. Differential privacy (DP) offers a quantifiable privacy guarantee; however, its impact on observational studies remains underexplored. We empirically evaluated the effects of DP across varying values of its privacy parameter, epsilon, on case-control analysis outcomes using EHR data. This study aims to inform DP parameter selection and examines the influence of study characteristics on differentially private observational studies. Materials and Methods: We assessed the effects of DP on a case-control study of 1-year asthma exacerbations, including 22 165 participants with a history of asthma from UK Biobank linked to EHR data. Odds ratios (ORs) for sociodemographic factors and comorbidities were analyzed using adjusted and propensity score-matched models across epsilon values. Results: DP influenced the magnitude, direction, and statistical significance of ORs, occasionally resembling patterns of misclassification, residual confounding, and false-positive bias. Rare and imbalanced covariates showed greater OR variability, especially in matched studies. Epsilons smaller than ln(2) led to noticeable OR fluctuations. Discussion: The impact of DP on ORs and selection of an optimal epsilon depends on sample size, covariate prevalence, confounders, case-to-control ratios in propensity score matching, mitigation of random seed p-hacking, and trust models. Conclusion: The effects of DP on ORs are highly context-dependent. In this study, epsilon values below ln(2) led to unstable ORs across random seeds. Averaging results or using predetermined seeds may help reduce variability and mitigate p-hacking.

  • Open Access

    Healthcare utilisation of 282,080 individuals with long COVID over two years: a multiple matched control, longitudinal cohort analysis

    by Yi Mu, Ashkan Dashtban, Mehrdad A Mizani, Chris Tomlinson, Mohamed Mohamed, Mark Ashworth, Mamas Mamas, Rouven Priedon, Steffen Petersen, Evan Kontopantelis, et al.

    Objectives: To investigate healthcare utilisation and cost in individuals with long COVID (LC) at population level. Design: Case–control cohort analysis with multiple age-, sex-, ethnicity-, deprivation-, region- and comorbidity-matched control groups: (1) COVID only, no LC; (2) pre-pandemic; (3) contemporary non-COVID; and (4) pre-LC (self-controlled, pre-COVID pandemic). Setting: National, population-based, linked UK electronic health records (British Heart Foundation/NHS England Secure Data Environment). Participants: Adults aged ≥18 years with LC between January 2020 and January 2023. Main outcome measures: Healthcare utilisation (number of consultations/visits per person: primary care (general practitioner [GP]), secondary care (outpatient [OP], inpatient [IP] and emergency department [ED], investigations and procedures) and inflation-adjusted cost (£) for LC and control populations per month, calendar year and pandemic year for each category. Results: A total of 282,080 individuals with LC were included between January 2020 and January 2023. The control groups were COVID only, no LC ( n = 1,112,370), pre-pandemic ( n = 1,031,285), contemporary non-COVID ( n = 1,118,360) and pre-LC ( n = 282,080). Healthcare utilisation per person (per month/year) was higher in LC than controls across GP, OP and ED. For IP, LC had higher healthcare utilisation than pre-LC and contemporary non-COVID (all p < 0.0001). Healthcare utilisation of the LC group increased progressively between 2020 and 2023, compared with controls. Median cost per patient/year was also higher in individuals with LC than all control groups. Conclusions: LC has been associated with substantial, persistent healthcare utilisation and cost over the last three years. Future funding, resources and staff for LC prevention, treatment and research must be prioritised to reduce sustained primary and secondary healthcare utilisation and costs.


  • Open Access

    Routine measurement of cardiometabolic disease risk factors in primary care in England before, during, and after the COVID-19 pandemic: A population-based cohort study

    by Frederick K. Ho, Caroline Dale, Mehrdad A. Mizani, Thomas Bolton, Ewan R. Pearson, Jonathan Valabhji, Christian Delles, Paul Welsh, Shinya Nakada, Daniel Mackay, et al.

    Background: This study estimated to what extent the number of measurements of cardiometabolic risk factors (e.g., blood pressure, cholesterol, glycated haemoglobin) were impacted by the COVID-19 pandemic and whether these have recovered to expected levels. Methods and findings: A cohort of individuals aged ≥18 years in England with records in the primary care—COVID-19 General Practice Extraction Service Data for Pandemic Planning and Research (GDPPR) were identified. Their records of 12 risk factor measurements were extracted between November 2018 and March 2024. Number of measurements per 1,000 individuals were calculated by age group, sex, ethnicity, and area deprivation quintile. The observed number of measurements were compared to a composite expectation band, derived as the union of the 95% confidence intervals of 2 estimates: (1) a projected trend based on data prior to the COVID-19 pandemic; and (2) an assumed stable trend from before pandemic. Point estimates were calculated as the mid-point of the expectation band. A cohort of 49,303,410 individuals aged ≥18 years were included. There was sharp drop in all measurements in March 2020 to February 2022, but overall recovered to the expected levels during March 2022 to February 2023 except for blood pressure, which had prolonged recovery. In March 2023 to March 2024, blood pressure measurements were below expectation by 16% (−19 per 1,000) overall, in people aged 18 to 39 (−23%; −18 per 1,000), 60 to 79 (−17%; −27 per 1,000), and ≥80 (−31%; −57 per 1,000). There was suggestion that recovery in blood pressure measurements was socioeconomically patterned. The second most deprived quintile had the highest deviation (−20%; −23 per 1,000) from expectation compared to least deprived quintile (−13%; −15 per 1,000). Conclusions: There was a substantial reduction in routine measurements of cardiometabolic risk factors following the COVID-19 pandemic, with variable recovery. The implications for missed diagnoses, worse prognosis, and health inequality are a concern.


  • Open Access

    Identifying subtypes of type 2 diabetes mellitus with machine learning: development, internal validation, prognostic validation and medication burden in linked electronic health records in 420 448 individuals

    by Mehrdad A Mizani, Ashkan Dashtban, Laura Pasea, Qingjia Zeng, Kamlesh Khunti, Jonathan Valabhji, Jil Billy Mamza, He Gao, Tamsin Morris, Amitava Banerjee

    Introduction: None of the studies of type 2 diabetes (T2D) subtyping to date have used linked population-level data for incident and prevalent T2D, incorporating a diverse set of variables, explainable methods for cluster characterization, or adhered to an established framework. We aimed to develop and validate machine learning (ML)-informed subtypes for type 2 diabetes mellitus (T2D) using nationally representative data. Research design and methods: In population-based electronic health records (2006–2020; Clinical Practice Research Datalink) in individuals ≥18 years with incident T2D (n=420 448), we included factors (n=3787), including demography, history, examination, biomarkers and medications. Using a published framework, we identified subtypes through nine unsupervised ML methods (K-means, K-means++, K-mode, K-prototype, mini-batch, agglomerative hierarchical clustering, Birch, Gaussian mixture models, and consensus clustering). We characterized clusters using intracluster distributions and explainable artificial intelligence (AI) techniques. We evaluated subtypes for (1) internal validity (within dataset; across methods); (2) prognostic validity (prediction for 5-year all-cause mortality, hospitalization and new chronic diseases); and (3) medication burden. Results: Development: We identified four T2D subtypes: metabolic, early onset, late onset and cardiometabolic. Internal validity: Subtypes were predicted with high accuracy (F1 score >0.98). Prognostic validity: 5-year all-cause mortality, hospitalization, new chronic disease incidence and medication burden differed across T2D subtypes. Compared with the metabolic subtype, 5-year risks of mortality and hospitalization in incident T2D were highest in late-onset subtype (HR 1.95, 1.85–2.05 and 1.66, 1.58–1.75) and lowest in early-onset subtype (1.18, 1.11–1.27 and 0.85, 0.80–0.90). Incidence of chronic diseases was highest in late-onset subtype and lowest in early-onset subtype. Medications: Compared with the metabolic subtype, after adjusting for age, sex, and pre-T2D medications, late-onset subtype (1.31, 1.28–1.35) and early-onset subtype (0.83, 0.81–0.85) were most and least likely, respectively, to be prescribed medications within 5 years following T2D onset. Conclusions: In the largest study using ML to date in incident T2D, we identified four distinct subtypes, with potential future implications for etiology, therapeutics, and risk prediction.


  • Open Access

    Identifying Subtypes of Chronic Kidney Disease with Machine Learning: Development, Internal Validation and Prognostic Validation Using Electronic Health Records in 350067 Individuals

    by Ashkan Dashtban, Mehrdad A. Mizani, Laura Pasea, Spiros Denaxas, Richard Corbett, Jil Billy Mamza, He Gao, Tamsin Morris, Harry Hemingway, Amitava Banerjee

    Background: Although chronic kidney disease (CKD) is associated with high multimorbidity, polypharmacy, morbidity and mortality, existing classification systems (mild to severe, usually based on estimated glomerular filtration rate, proteinuria or urine albumin-creatinine ratio) and risk prediction models largely ignore the complexity of CKD, its risk factors and its outcomes. Improved subtype definition could improve prediction of outcomes and inform effective interventions. Methods: We analysed individuals ≥18 years with incident and prevalent CKD (n=350067 and 195422 respectively) from a population-based electronic health record resource (2006-2020; Clinical Practice Research Datalink, CPRD). We included factors (n=264 with 2670 derived variables), e.g. demography, history, examination, blood laboratory values and medications. Using a published framework, we identified subtypes through seven unsupervised machine learning (ML) methods (K-means, Diana, HC, Fanny, PAM, Clara, Model-based) with 66 (of 2670) variables in each dataset. We evaluated subtypes for: (i) internal validity (within dataset, across methods); (ii) prognostic validity (predictive accuracy for 5-year all-cause mortality and admissions); and (iii) medications (new and existing by British National Formulary chapter). Findings: After identifying five clusters across seven approaches, we labelled CKD subtypes: 1.Early-onset, 2.Late-onset, 3. Cancer, 4. Metabolic, and 5.Cardiometabolic. Internal validity: We trained a high performing model (using XGBoost) that could predict disease subtypes with 95% accuracy for incident and prevalent CKD (Sensitivity: 0.81-0.98, F1 score:0.84-0.97). Prognostic validity: 5-year all-cause mortality, hospital admissions, and incidence of new chronic diseases differed across CKD subtypes. The 5-year risk of mortality and admissions in the overall incident CKD population were highest in cardiometabolic subtype: 43.3% (42.3-42.8%) and 29.5% (29.1-30.0%), respectively, and lowest in the early-onset subtype: 5.7% (5.5-5.9%) and 18.7% (18.4-19.1%). Medications: Across CKD subtypes, the distribution of prescription medication classes at baseline varied, with highest medication burden in cardiometabolic and metabolic subtypes, and higher burden in prevalent than incident CKD. Interpretation: In the largest CKD study using ML, to-date, we identified five distinct subtypes in individuals with incident and prevalent CKD. These subtypes have relevance to study of aetiology, therapeutics and risk prediction. Funding Information: AstraZeneca, Health Data Research UK. Declaration of Interests: JBM, HG and TM are employed by AstraZeneca UK Ltd, a biopharmaceutical company. AB is supported by research funding from the National Institute for Health Research (NIHR), British Medical Association, AstraZeneca, and UK Research and Innovation, and Trustee of the South Asian Health Foundation. HH is a National Institute for Health Research (NIHR) Senior Investigator. HH is funded by the National Institute for Health Research University College London Hospitals Biomedical Research Centre, Health Data Research UK (grant No. LOND1, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care [England], Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division [Welsh Government], Public Health Agency [Northern Ireland], British Heart Foundation and Wellcome Trust). AB and HH are part of the BigData@Heart Consortium, funded by the Innovative Medicines Initiative-2 Joint Undertaking under grant agreement No. 116074. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation programme and EFPIA; it is chaired, by DE Grobbee and SD Anker, partnering with 20 academic and industry partners and ESC. Other authors report no conflicts of interest. Ethics Approval Statement: Study approvals were by: (i) MHRA Independent Scientific Advisory Committee [18_217R]: Section 251 (NHS Social Care Act 2006), (ii) Scientific Review Committee [17THIN038-A1] and (iii) UKB 15422: Patient informed consent was not required or provided.

  • Open Access

    Understanding COVID-19 trajectories from a nationwide linked electronic health record cohort of 56 million people: phenotypes, severity, waves & vaccination

    by Johan H Thygesen, Christopher Tomlinson, Sam Hollings, Mehrdad Mizani, Alex Handy, Ashley Akbari, Amitava Banerjee, Jennifer Cooper, Alvina Lai, Ken Li, et al.

    Background Updatable understanding of the onset and progression of individuals COVID-19 trajectories underpins pandemic mitigation efforts. In order to identify and characterize individual trajectories, we defined and validated ten COVID-19 phenotypes from linked electronic health records (EHR) on a nationwide scale using an extensible framework. Methods Cohort study of 56.6 million people in England alive on 23/01/2020, followed until 31/05/2021, using eight linked national datasets spanning COVID-19 testing, vaccination, primary & secondary care and death registrations data. We defined ten COVID-19 phenotypes reflecting clinically relevant stages of disease severity using a combination of international clinical terminologies (e.g. SNOMED-CT, ICD-10) and bespoke data fields; positive test, primary care diagnosis, hospitalisation, critical care (four phenotypes), and death (three phenotypes). Using these phenotypes, we constructed patient trajectories illustrating the transition frequency and duration between phenotypes. Analyses were stratified by pandemic waves and vaccination status. Findings We identified 3,469,528 infected individuals (6.1%) with 8,825,738 recorded COVID-19 phenotypes. Of these, 364,260 (11%) were hospitalised and 140,908 (4%) died. Of those hospitalised, 38,072 (10%) were admitted to intensive care (ICU), 54,026 (15%) received non-invasive ventilation and 21,404 (6%) invasive ventilation. Amongst hospitalised patients, first wave mortality (30%) was higher than the second (23%) in non-ICU settings, but remained unchanged for ICU patients. The highest mortality was for patients receiving critical care outside of ICU in wave 1 (51%). 13,083 (9%) COVID-19 related deaths occurred without diagnoses on the death certificate, but within 30 days of a positive test while 10,403 (7%) of cases were identified from mortality data alone with no prior phenotypes recorded. We observed longer patient trajectories in the second pandemic wave compared to the first. Interpretation Our analyses illustrate the wide spectrum of severity that COVID-19 displays and significant differences in incidence, survival and pathways across pandemic waves. We provide an adaptable framework to answer questions of clinical and policy relevance; new variant impact, booster dose efficacy and a way of maximising existing data to understand individuals progression through disease states. Evidence before the study We searched PubMed on October 14, 2021, for publications with the terms “COVID-19” or “SARS-CoV-2”, “severity”, and “electronic health records” or “EHR” without date or language restrictions. Multiple studies explore factors associated with severity of COVID-19 infection, and model predictions of outcome for hospitalised patients. However, most work to date focused on isolated facets of the healthcare system, such as primary or secondary care only, was conducted in subpopulations (e.g. hospitalised patients) of limited sample size, and often utilized dichotomised outcomes (e.g. mortality or hospitalisation) ignoring the full spectrum of disease. We identified no studies which comprehensively detailed severity of infections while describing disease severity across pandemic waves, vaccination status, and patient trajectories. Added value of this study To our knowledge, this is the first study providing a comprehensive view of COVID-19 across pandemic waves using national data and focusing on severity, vaccination, and patient trajectories. Drawing on linked electronic health record (EHR) data on a national scale (56.6 million people alive and registered with GP in England), we describe key demographic factors, frequency of comorbidities, impact of the two main waves in England, and effect of full vaccination on COVID-19 severities. Additionally, we identify and describe patient trajectory networks which illustrate the main transition pathways of COVID-19 patients in the healthcare system. Finally, we provide reproducible COVID-19 phenotyping algorithms reflecting clinically relevant stages of disease severity i.e. positive tests, primary care diagnoses, hospitalisation, critical care treatments (e.g. ventilatory support) and mortality. Implications of all the available evidence The COVID-19 phenotypes and trajectory analysis framework outlined produce a reproducible, extensible and repurposable means to generate national-scale data to support critical policy decision making. By modelling patient trajectories as a series of interactions with healthcare systems, and linking these to demographic and outcome data, we provide a means to identify and prioritise care pathways associated with adverse outcomes and highlight healthcare system ‘touch points’ which may act as tangible targets for intervention.


  • Open Access

    Evaluation of antithrombotic use and COVID-19 outcomes in a nationwide atrial fibrillation cohort

    by Alex Handy, Amitava Banerjee, Angela Wood, Caroline E Dale, Cathie Sudlow, Christopher Tomlinson, Daniel Bean, Johan H Thygesen, Mehrdad A Mizani, Michail Katsoulis, et al.

    Objective: Evaluate antithrombotic (AT) use in individuals with atrial fibrillation (AF) and high stroke risk (CHA 2 DS 2 -VASc score>=2) and investigate whether pre-existing AT use may improve COVID-19 outcomes. Methods: Individuals with AF and a CHA 2 DS 2 -VASc score>=2 on January 1 st 2020 were identified using pseudonymised, linked electronic health records for 56 million people in England and followed-up until May 1 st 2021. Factors associated with pre-existing AT use were analysed using logistic regression. Differences in COVID-19 related hospitalisation and death were analysed using logistic and Cox regression for individuals exposed to pre-existing AT use vs no AT use, anticoagulants (AC) vs antiplatelets (AP) and direct oral anticoagulants (DOACs) vs warfarin. Results: From 972,971 individuals with AF and a CHA 2 DS 2 -VASc score>=2, 88.0% (n=856,336) had pre-existing AT use, 3.8% (n=37,418) had a COVID-19 related hospitalisation and 2.2% (n=21,116) died. Factors associated with no AT use included comorbidities that may contraindicate AT use (liver disease and history of falls) and demographics (socioeconomic status and ethnicity). Pre-existing AT use was associated with lower odds of death (OR=0.92 [0 . 87-0 . 96 at 95% CI] ), but higher odds of hospitalisation OR=1.20 [1 . 15-1 . 26 at 95% CI] ). The same pattern was observed for AC vs AP (death (OR=0.93 [0.87-0.98]), hospitalisation (OR=1.17 [1.11-1.24])) but not for DOACs vs warfarin (death (OR=1.00 [0.95-1.05]), hospitalisation (OR=0.86 [0.82-0.89]). Conclusions: Pre-existing AT use may offer marginal protection against COVID-19 death, with AC offering more protection than AP. Although this association may not be causal, it provides further incentive to improve AT coverage for eligible individuals with AF. KEY QUESTIONS: What is already known about this subject?: Anticoagulants (AC), a sub-class of antithrombotics (AT), reduce the risk of stroke and are recommended for individuals with atrial fibrillation (AF) and at high risk of stroke (CHA 2 DS 2 -VASc score>=2, National Institute for Health and Care Excellence threshold). However, previous evaluations suggest that up to one third of these individuals may not be taking AC. Over estimation of bleeding and fall risk in elderly patients have been identified as potential factors in this under medicating. In response to the COVID-19 pandemic, several observational studies have observed correlations between pre-existing AT use, particularly anticoagulants (AC), and lower risk of severe COVID-19 outcomes such as hospitalisation and death. However, these correlations are inconsistent across studies and have not compared all major sub-types of AT in one study. What does this study add?: This study uses datasets covering primary care, secondary care, pharmacy dispensing, death registrations, multiple COVID-19 diagnoses routes and vaccination records for 56 million people in England and is the largest scale evaluation of AT use to date. This provides the statistical power to robustly analyse targeted sub-types of AT and control for a wide range of potential confounders. All code developed for the study is opensource and an updated nationwide evaluation can be rapidly created for future time points. In 972,971 individuals with AF and a CHA 2 DS 2 -VASc score>=2, we observed 88.0% (n=856,336) with pre-existing AT use which was associated with marginal protection against COVID-19 death (OR=0.92 [0 . 87-0 . 96 at 95% CI] ). How might this impact on clinical practice?: These findings can help shape global AT medication policy and provide population-scale, observational analysis results alongside gold-standard randomised control trials to help assess whether a potential beneficial effect of pre-existing AT use on COVID-19 death alters risk to benefit assessments in AT prescribing decisions.