Dynamic Updating of Clinical Survival Prediction Models in a Rapidly Changing Environment (preprint)

Over time, the performance of clinical prediction models may deteriorate due to changes in clinical management, data quality, disease risk and/or patient mix. Such prediction models must be updated in order to remain useful. Here, we investigate methods for discrete and dynamic model updating of clinical survival prediction models based on refitting, recalibration and Bayesian updating. In contrast to discrete or one-time updating, dynamic updating refers to a process in which a prediction model is repeatedly updated with new data. Motivated by infectious disease settings, our focus was on model performance in rapidly changing environments. We first compared the methods using a simulation study. We simulated scenarios with changing survival rates, the introduction of a new treatment and predictors of survival that are rare in the population. Next, the updating strategies were applied to patient data from the QResearch database, an electronic health records database from general practices in the UK, to study the updating of a model for predicting 70-day covid-19 related mortality. We found that a dynamic updating process outperformed one-time discrete updating in the simulations. Bayesian dynamic updating has the advantages of making use of knowledge from previous updates and requiring less data compared to refitting.

Analysis of uptake, effectiveness and safety of COVID-19 vaccinations in pregnancy using the QResearch database: research protocol and statistical analysis plan (preprint)

Background The COVID-19 pandemic has affected millions of people globally with major health, social and economic consequences, prompting development of vaccines for use in the general population. However, vaccination uptake is lower in some groups, including in pregnant women, because of concerns regarding vaccine safety. There is evidence of increased risk of adverse pregnancy and neonatal outcomes associated with SARS-CoV-2 infection, but fear of vaccine-associated adverse events on the baby both in short and longer term is one of the main drivers of low uptake for this group. Other vaccines commonly used in pregnancy include influenza and pertussis. These both have reportedly higher uptake compared with COVID-19 vaccination, which may be because they are perceived to be safer. In this study, we will undertake an independent evaluation of the uptake, effectiveness and safety of COVID-19 vaccinations in pregnant women using the QResearch primary care database in England. Objectives A. To determine COVID-19 vaccine uptake in pregnant women compared to uptake of influenza and pertussis vaccinations. B. To estimate COVID-19 vaccine effectiveness in pregnant women by evaluating the risk of severe COVID-19 outcomes following vaccination. C. To assess the safety of COVID-19 vaccination in pregnancy by evaluating the risks of adverse pregnancy and perinatal outcomes and adverse events of special interest for vaccine safety after COVID-19 vaccination compared with influenza and pertussis vaccinations. Methods This population-based study uses the QResearch database of primary health care records, linked to individual-level data on hospital admissions, mortality, COVID-19 vaccination, SARS-CoV-2 testing data and congenital anomalies. We will include women aged 16 to 49 years with at least one pregnancy during the study period of 30th December 2020 to the latest date available. Babies born during the study period will be identified and linked to the mothers record, where possible. We will describe vaccine uptake in pregnant women by trimester and population subgroups defined by demographics and other characteristics. Cox proportional hazards multivariable regression will be used to identify factors associated with vaccine uptake. The effectiveness of COVID-19 vaccines in pregnant women will be assessed using time varying Royston-Palmar regression analyses to determine unadjusted and adjusted hazard ratios for the occurrence of severe COVID-19 outcomes after each vaccine dose compared with unvaccinated individuals. For the safety analysis, we will we use logistic regression analyses to determine unadjusted and adjusted odds ratios for the occurrence of maternal (e.g. miscarriage, ectopic pregnancy and gestational diabetes) and perinatal outcomes (e.g. stillbirth, small for gestational age and congenital anomalies) by vaccination status compared to unvaccinated individuals. For the adverse events of special interest for vaccine safety (e.g. venous thromboembolism, myocarditis and Guillain Barre syndrome), we will use time varying Royston-Palmar regression analyses to determine unadjusted and adjusted hazard ratios for the occurrence of each outcome by vaccination status to unvaccinated individuals. Ethics and dissemination QResearch is a Research Ethics Approved Research Database with ongoing approval from the East Midlands Multi-Centre Research Ethics Committee (Ref: 18/EM/0400). This study was approved by the QResearch Scientific Committee on 9th June 2022. This research protocol has been developed with support from a patient and public involvement panel, who will continue to provide input throughout the duration of the study. Research findings will be submitted to pre-print servers such as MedRxIv, academic publication and disseminated more broadly through media releases and community groups and conference presentations.

Impact of vaccination on COVID-19-associated admissions to critical care in England: a population cohort study of linked data (preprint)

Introduction This study aims to explore the impact of COVID-19 vaccination on critical care by examining associations between vaccination and admission to critical care with COVID-19 during England's Delta wave, by age group, dose, and over time. Methods We used linked routinely-collected data to conduct a population cohort study of patients admitted to adult critical care in England for management of COVID-19 between 1 May and 15 December 2021. Included participants were the whole population of England aged 18 years or over (44.7 million), including 10,141 patients admitted to critical care with COVID-19. The intervention was vaccination with one, two, or a booster/three doses of any COVID-19 vaccine. Results Compared with unvaccinated patients, vaccinated patients were older (median 64 years for patients receiving two or more doses versus 50 years for unvaccinated), with higher levels of severe comorbidity (20.3% versus 3.9%) and immunocompromise (15.0% versus 2.3%). Compared with patients who were unvaccinated, those vaccinated with two doses had a relative risk reduction (RRR) of between 90.1% (patients aged 18-29, 95% CI, 86.8% to 92.7%) and 95.9% (patients aged 60-69, 95% CI, 95.5% to 96.2%). Waning was only observed for those aged 70+, for whom the RRR reduced from 97.3% (91.0% to 99.2%) to 86.7% (85.3% to 90.1%) between May and December but increased again to 98.3% (97.6% to 98.8%) with a booster/third dose. Conclusion Important demographic and clinical differences exist between vaccinated and unvaccinated patients admitted to critical care with COVID-19. While not a causal analysis, our findings are consistent with a substantial and sustained impact of vaccination on reducing admissions to critical care during England's Delta wave, with evidence of waning predominantly restricted to those aged 70+.

Uptake of Sotrovimab for prevention of severe COVID-19 and its safety in the community in England (preprint)

Sotrovimab is a neutralising monoclonal antibody (nMAB), currently administrated in England to treat extremely clinically vulnerable COVID-19 patients. Trials have shown it to have mild or moderate side effects, however safety in real-world settings has not been yet evaluated. We used national databases to investigate its uptake and safety in community patients across England. We used a cohort study to describe uptake and a self-controlled case series design to evaluate the risks of 49 pre-specified suspected adverse events in the 2-28 days post-treatment. Between December 11, 2021 and May 24, 2022, there were 172,860 COVID-19 patients eligible for treatment. Of the 22,815 people who received Sotrovimab, 21,487 (94.2%) had a positive SARS-CoV-2 test and 5,999 (26.3%) were not on the eligible list. Between treated and untreated eligible individuals, the mean age (54.6, SD: 16.1 vs 54.1, SD: 18.3) and sex distribution (women: 60.9% vs 58.1%; men: 38.9% vs 41.1%) were similar. There were marked variations in uptake between ethnic groups, which was higher amongst Indian (15.0%; 95%CI 13.8, 16.3), Other Asian (13.7%; 95%CI 11.9, 15.8), White (13.4%; 95%CI 13.3, 13.6), and Bangladeshi (11.4%; 95%CI 8.8, 14.6); and lower amongst Black Caribbean individuals (6.4%; 95%CI 5.4, 7.5) and Black Africans (4.7%; 95%CI 4.1, 5.4). We found no increased risk of any of the suspected adverse events in the overall period of 2-28 days post-treatment, but an increased risk of rheumatoid arthritis (IRR 3.08, 95% CI 1.44, 6.58) and of systematic lupus erythematosus (IRR 5.15, 95% CI 1.60, 16.60) in the 2-3 days post-treatment, when we narrowed the risk period. FundingNational Institute of Health Research (Grant reference 135561)

QCovid 4 - Predicting risk of death or hospitalisation from COVID-19 in adults testing positive for SARS-CoV-2 infection during the Omicron wave in England (preprint)

To (a) derive and validate risk prediction algorithms (QCovid4) to estimate risk of COVID-19 mortality and hospitalisation in UK adults with a SARS-CoV-2 positive test during the Omicron pandemic wave and (b) evaluate performance with earlier versions of algorithms developed in previous pandemic waves and the high-risk cohort identified by NHS Digital in England. Design Population-based cohort study using the QResearch database linked to national data on COVID-19 vaccination, high risk patients prioritised for COVID-19 therapeutics, SARS-CoV-2 results, hospitalisation, cancer registry, systemic anticancer treatment, radiotherapy and the national death registry. Settings and study period 1.3 million adults in the derivation cohort and 0.15 million adults in the validation cohort aged 18-100 years with a SARS-CoV-2 positive test between 11th December 2021 and 31st March 2022 with follow up to 30th June 2022. Main outcome measures Our primary outcome was COVID-19 death. The secondary outcome of interest was COVID-19 hospital admission. Models fitted in the derivation cohort to derive risk equations using a range of predictor variables. Performance evaluated in a separate validation cohort. Results Of 1,297,984 people with a SARS-CoV-2 positive test in the derivation cohort, 18,756 (1.45%) had a COVID-19 related hospital admission and 3,878 (0.3%) had a COVID-19 death during follow-up. Of the 145,404 people in the validation cohort, there were 2,124 (1.46%) COVID-19 admissions and 461 (0.3%) COVID-19 deaths. The COVID-19 mortality rate in men increased with age and deprivation. In the QCovid4 model in men hazard ratios were highest for those with the following conditions- kidney transplant (6.1-fold increase), Downs syndrome (4.9-fold); radiotherapy (3.1-fold); type 1 diabetes (3.4-fold), chemotherapy grade A (3.8-fold), grade B (5.8-fold), grade C (10.9-fold), solid organ transplant ever (2.4-fold), dementia (1.62-fold), Parkinsons disease (2.2-fold), liver cirrhosis (2.5-fold). Other conditions associated with increased COVID-19 mortality included learning disability, chronic kidney disease (stages 4 and 5), blood cancer, respiratory cancer, immunosuppressants, oral steroids, COPD, coronary heart disease, stroke, atrial fibrillation, heart failure, thromboembolism, rheumatoid or SLE, schizophrenia or bipolar disease sickle cell or HIV or SCID, type 2 diabetes. Results were similar in the model in women. COVID-19 mortality risk was lower among those who had received COVID-19 vaccination compared with unvaccinated individuals with evidence of a dose response relationship. The reduced mortality rates associated with prior SARS-CoV-2 infection were similar in men (adjusted hazard ratio (HR) 0.51 (95% CI 0.40, 0.64)) and women (adjusted HR 0.55 (95%CI 0.45, 0.67)). The QCOVID4 algorithm explained 76.6% (95%CI 74.4 to 78.8) of the variation in time to COVID-19 death (R2) in women. The D statistic was 3.70 (95%CI 3.48 to 3.93) and the Harrells C statistic was 0.965 (95%CI 0.951 to 0.978). The corresponding results for COVID-19 death in men were similar with R2 76.0% (95% 73.9 to 78.2); D statistic 3.65 (95%CI 3.43 to 3.86) and C statistic of 0.970 (95%CI 0.962 to 0.979). QCOVID4 discrimination for mortality was slightly higher than that for QCOVID1 and QCOVID2, but calibration was much improved. Conclusion The QCovid4 risk algorithm modelled from data during the UK Omicron wave now includes vaccination dose and prior SARS-CoV-2 infection and predicts COVID-19 mortality among people with a positive test. It has excellent performance and could be used for targeting COVID-19 vaccination and therapeutics. Although large disparities in risks of severe COVID-19 outcomes among ethnic minority groups were observed during the early waves of the pandemic, these are much reduced now with no increased risk of mortality by ethnic group.

Risk of COVID-19 related deaths for SARS-CoV-2 Omicron (B.1.1.529) compared with Delta (B.1.617.2) (preprint)

Objective To assess the risk of death involving COVID-19 following infection from Omicron (B.1.1.539/BA.1) relative to Delta (B.1.617.2). Design Retrospective cohort study. Setting England, UK, 1 December 2021 to 25 January 2022. Participants 1,035,163 people aged 18-100 years who tested positive for SARS-CoV-2 in the national surveillance programme, and had an infection identified as either Omicron- or Delta compatible. Main outcome measures Death involving COVID-19 as identified from death certification records. The exposure of interest was the SARS-CoV-2 variant identified from NHS Test and Trace PCR positive tests taken in the community (pillar 2) and analysed by Lighthouse laboratories. Cause-specific Cox proportional hazard regression models were adjusted for sex, age, vaccination status, previous infection, calendar time, ethnicity, Index of Multiple Deprivation rank, household deprivation, university degree, keyworker status, country of birth, main language, region, disability, and comorbidities. Additionally, we tested for interactions between variant and sex, age, vaccination status and comorbidities. Results The risk of death involving COVID-19 was 67% lower for Omicron compared to Delta and the reduction in the risk of death involving COVID-19 for Omicron compared to Delta was more pronounced in males than in females and in people under 70 years old than in people aged 70 years or over. Regardless of age, reduction of the risk of death from Omicron relative to Delta more was more pronounced in people who had received a booster than in those having received only two doses. Conclusions Our results support early work showing the relative reduction in severity of Omicron compared to Delta in terms of hospitalisation and extends this research to assess COVID-19 mortality. Our work also highlights the importance of the vaccination booster campaign, where the reduction in risk of death involving COVID-19 is most pronounced in individuals who had received a booster. What is already known on this topic The Omicron variant, which refers to the whole lineage (BA.1, BA.2, BA.3) had already been shown to be more transmissible than the Delta variant, but there is emerging evidence suggests that the risk of hospitalisation and risk of death within 28 days after a SARS-COV-2 test is lower. However, with a highly transmissible infection and high levels of population testing, definition of death within 28 days is more likely to be susceptible to misclassification bias due to asymptomatic or co-incidental infection. There is no study so far comparing the risk of COVID-19 death as identified from death certification records, with the cause of death assessed by the physician who attended the patient in the last illness. What this study adds Using data from a large cohort of COVID-19 infections that occurred in December 2021, we examined the difference in the risk COVID-19 death, as identified from death certification records, between the Delta and Omicron BA.1 variant. Our study shows that risk of death involving COVID-19 was reduced by 67% following infection with the Omicron BA.1 variant relative to the Delta variant after adjusting for a wide range of potential confounders, including vaccination status and comorbidities. Importantly, we found that the relative risk of COVID-19 mortality following Omicron versus Delta infection varied by age and sex, with lower relative risk in younger individuals and for males than females. The reduction in risk of death involving COVID-19 was also most pronounced in individuals who had received a booster.

Risk of myocarditis following sequential COVID-19 vaccinations by age and sex (preprint)

ABSTRACT In an updated self-controlled case series analysis of 42,200,614 people aged 13 years or more, we evaluate the association between COVID-19 vaccination and myocarditis, stratified by age and sex, including 10,978,507 people receiving a third vaccine dose. Myocarditis risk was increased during 1-28 days following a third dose of BNT162b2 (IRR 2.02, 95%CI 1.40, 2.91). Associations were strongest in males younger than 40 years for all vaccine types with an additional 3 (95%CI 1, 5) and 12 (95% CI 1,17) events per million estimated in the 1-28 days following a first dose of BNT162b2 and mRNA-1273, respectively; 14 (95%CI 8, 17), 12 (95%CI 1, 7) and 101 (95%CI 95, 104) additional events following a second dose of ChAdOx1, BNT162b2 and mRNA-1273, respectively; and 13 (95%CI 7, 15) additional events following a third dose of BNT162b2, compared with 7 (95%CI 2, 11) additional events following COVID-19 infection. An association between COVID-19 infection and myocarditis was observed in all ages for both sexes but was substantially higher in those older than 40 years. These findings have important implications for public health and vaccination policy. Funding Health Data Research UK.

An External Validation of the QCovid Risk Prediction Algorithm for Risk of Mortality from COVID-19 in Adults: National Validation Cohort Study in Scotland (preprint)

Background: The QCovid algorithm is a risk prediction tool for COVID-19 hospitalisation and mortality that can be used to stratify patients by risk into vulnerability groups . We carried out an external validation of the QCovid algorithm in Scotland.Methods: We established a national COVID-19 data platform using individual level data for the population of Scotland (5.4 million residents). Primary care data were linked to reverse-transcription polymerase chain reaction (RT-PCR) virology testing, hospitalisation and mortality data. We assessed the performance of the QCovid algorithm in predicting COVID-19 hospitalisation and deaths in our dataset for two time periods: 1 March, 2020 to 30 April, 2020, and 1 May, 2020 to 30 June, 2020.Findings: Our dataset comprised 5,384,819 individuals, representing 99% of the estimated population (5,463,300) resident in Scotland in 2020. The algorithm showed excellent calibration in both time periods with close correspondence between observed and predicted risks. Harrell ’s C for deaths in males and females in the first period was 0.946 (95% CI: 0.941 - 0.951) and 0.925 (95% CI: 0.919 - 0.931) respectively. Harrell’s C for hospitalisations in males and females in the first period was 0.809 (95% CI: 0.801 - 0.817) and 0.816 (95% CI: 0.808 - 0.823) respectively.Interpretation: The QCovid algorithm shows high levels of external validity in predicting the risk of COVID- 19 hospitalisation and death in the population of Scotland.Funding: Medical Research Council, National Institute for Health Research Health Technology Assessment Programme, funded through the UK Research and Innovation Industrial Strategy Challenge Fund Health Data Research UK.Declaration of Interests: Dr. Hippisley-Cox reports grants from MRC, grants from Wellcome Trrust, grants from NIHR, during the conduct of the study; other from ClinRisk Ltd, outside the submitted work. Dr. Sheikh reports grants from NIHR, grants from MRC, grants from HRR UK, during the conduct of the study. All other authors report no conflict of interest.Ethics Approval Statement: Ethical permission for this study was granted from South East Scotland Research Ethics Committee 02 [12/SS/0201]. The Public Benefit and Privacy Panel Committee of Public Health Scotland, approved the linkage and analysis of the de-identified datasets for this project [1920-0279].

Analysis of severe outcomes associated with the SARS-CoV-2 Variant of Concern 202012/01 in England using ICNARC Case Mix Programme and QResearch databases (preprint)

ABSTRACT Background A new, more transmissible variant of SARS-CoV-2, variant of concern (VOC) 202012/01 or lineage B.1.1.7, has emerged in the UK. We estimate the risk of critical care admission, mortality in critical ill patients, and overall mortality associated with VOC B.1.1.7 compared with the original variant. We also compare clinical outcomes between these variants ‘ groups. Methods We linked a large primary care (QResearch), the national critical care (ICNARC CMP) and the COVID-19 testing (PHE) database and extracted two cohorts. The first was used to explore the association between VOC B.1.1.7 and critical care admission and 28-day mortality. The second to determine the risk of mortality in critically ill patients with VOC B.1.1.7 compared to those without. We used Royston-Parmar models adjusted for age, sex, region, other socio-demographics and comorbidities (asthma, COPD, type I and II, hypertension). We reported information on types and duration of organ supports for the two variants ‘ groups. Findings The first cohort included 198,420 patients. Of these, 80,494 had VOC B.1.1.7, 712 were critically ill and 630 died by 28 days. The second cohort included 3432 critically ill patients. Of these, 2019 had VOC B.1.1.7 and 822 died at the end of critical care. Using the first cohort, we estimated adjusted hazard ratios for critical care admission and mortality to be 1.99 (95% CI: 1.59, 2.49) and 1.59 (95% CI: 1.25-2.03) for VOC B.1.1.7 compared with the original variant group, respectively. The adjusted hazard ratio for mortality in critical care, estimated using the second cohort, was 0.93 (95% CI 0.76-1.15) for patients with VOC B.1.1.7, compared to those without. Interpretation VOC B.1.1.7 appears to be more severe. Patients with VOC B.1.1.7 are at increased risk of critical care admission and mortality compared with patients without. For patients receiving critical care, mortality appears independent of virus strain. RESEARCH IN CONTEXT Evidence before this study A new variant of the SARS-CoV-2 virus, variant of concern (VOC) 202012/01, or lineage B.1.1.7, was detected in England in September 2020. The characteristics and outcomes of patients infected with VOC B.1.1.7 are not yet known. VOC B.1.1.7 has been associated with increased transmissibility. Early analyses have suggested infection with VOC B.1.1.7 may be associated with a higher risk of mortality compared with infection with other virus variants, but these analyses had either limited ability to adjust for key confounding variables or did not consider critical care admission. The effects of VOC B.1.1.7 on severe COVID-19 outcomes remain unclear. Added value of this study This study found a 60% higher risk of 28-day mortality associated with infection with VOC B.1.1.7 in patients tested in the community in comparison with the original variant, when adjusted for key confounding variables. The risk of critical care admission for those with VOC B.1.1.7 is double the risk associated with the original variant. For patients receiving critical care, the infecting variant is not associated with the risk of mortality at the end of critical care. Implications of all the available evidence The higher mortality and rate of critical care admission associated with VOC B.1.1.7, combined with its known increased transmissibility, are likely to put health care systems under further stress. These effects may be mitigated by the ongoing vaccination programme.

Influenza, Shingles and Pneumococcal Vaccine Uptake, Offer and Refusal in Adult Populations at High-Risk for COVID-19: A UK Population-Based Cohort Study (preprint)

Background: As COVID-19 vaccination programs are being rolled out globally, we studied the ethnic, deprivation, household size and comorbidity ‘patterning’ of existing vaccination programs in populations at high-risk for COVID-19, to inform risk-stratified vaccination strategies and mitigate health inequalities. Methods: A population-level cohort study of UK adults aged 65 years or older, using a large primary care database. We used multivariable logistic regression to assess uptake of influenza, pneumococcal and shingles vaccination across ethnic groups, deprivation quintile, household size, and comorbidities, computing odds ratios (OR) adjusted for age, sex, demographics, body mass index and smoking. Offers and refusals of each vaccination type were analysed in those not receiving them. Findings: The cohort comprised 2,054,463 patients from 1,318 general practices. 1,452,014 (70.7%) patients received influenza vaccine, 1,391,228 (67.7%) received pneumococcal vaccine, and 690,783 (53.4%) received shingles vaccine. Compared to Whites, influenza vaccination uptake was lower in Pakistani (adjusted odds ratio (OR) 0.82; 99% confidence interval: 0.74-0.90), Black Caribbean (OR 0.46; 0.43-0.48), Black African (OR 0.63; 0.58-0.68), Chinese (OR 0.70; 0.64-0.76) and ‘Other ethnic group’ (OR 0.65; 0.63-0.69). The Black Caribbean group had higher vaccination refusal than the White group for influenza vaccination (OR 1.17; 1.05-1.30). Vaccination uptake was lower among the more deprived and those living in household sizes above 3 or more persons, with some significant interactions between ethnicity and comorbidities. Uptake of all three vaccines was higher in those with asthma, COPD, type 2 diabetes, hypertension and learning disability, whilst lower in those with dementia. Interpretation: Whilst uptake and refusal of influenza, shingles and pneumococcal vaccination are patterned by ethnicity, deprivation, household size and comorbidities, vaccination offer is mostly patterned by comorbidities. This information can inform national policies to ensure equitable implementation of COVID-19 vaccination programs to avoid exacerbating health inequalities.Funding Statement: This project was funded by the Medical Research Council (Grant Ref: MR/V027778/1).Declaration of Interests: PST reports previous consultation with AstraZeneca and Duke-NUS outside the submitted work. KK is a Member of the Scientific Advisory Group for Emergencies (SAGE), Member of Independent SAGE, Director of the University of Leicester Centre for Black Minority Health and Trustee of the south Asian Health Foundation. JHC is a member of several SAGE committees and chair of the risk stratification subgroup of the NERVTAG. She is unpaid director of QResearch and founder and former medical director of ClinRisk Ltd (outside the submitted work). MP, AKC, HDM, DS, TAR, FZ, BRS, SJG, CC, CG have no interests to declare.

An external validation of the QCovid risk prediction algorithm for risk of mortality from COVID-19 in adults: national validation cohort study in England (preprint)

BackgroundTo externally validate a risk prediction algorithm (QCovid) to estimate mortality outcomes from COVID-19 in adults in England. MethodsPopulation-based cohort study using the ONS Public Health Linked Data Asset, a cohort based on the 2011 Census linked to Hospital Episode Statistics, the General Practice Extraction Service Data for pandemic planning and research, radiotherapy and systemic chemotherapy records. The primary outcome was time to COVID-19 death, defined as confirmed or suspected COVID-19 death as per death certification. Two time periods were used: (a) 24th January to 30th April 2020; and (b) 1st May to 28th July 2020. We evaluated the performance of the QCovid algorithms using measures of discrimination and calibration for each validation time period. FindingsThe study comprises 34,897,648 adults aged 19-100 years resident in England. There were 26,985 COVID-19 deaths during the first time-period and 13,177 during the second. The algorithms had good calibration in the validation cohort in both time periods with close correspondence of observed and predicted risks. They explained 77.1% (95% CI: 76.9% to 77.4%) of the variation in time to death in men in the first time-period (R2); the D statistic was 3.76 (95% CI: 3.73 to 3.79); Harrells C was 0.935 (0.933 to 0.937). Similar results were obtained for women, and in the second time-period. In the top 5% of patients with the highest predicted risks of death, the sensitivity for identifying deaths in the first time period was 65.9% for men and 71.7% for women. People in the top 20% of predicted risks of death accounted for 90.8% of all COVID-19 deaths for men and 93.0% for women. InterpretationThe QCovid population-based risk algorithm performed well, showing very high levels of discrimination for COVID-19 deaths in men and women for both time periods. It has the potential to be dynamically updated as the pandemic evolves and therefore, has potential use in guiding national policy. FundingNational Institute of Health Research RESEARCH IN CONTEXTO_ST_ABSEvidence before this studyC_ST_ABSPublic policy measures and clinical risk assessment relevant to COVID-19 need to be aided by rigorously developed and validated risk prediction models. A recent living systematic review of published risk prediction models for COVID-19 found most models are subject to a high risk of bias with optimistic reported performance, raising concern that these models may be unreliable when applied in practice. A population-based risk prediction model, QCovid risk prediction algorithm, has recently been developed to identify adults at high risk of serious COVID-19 outcomes, which overcome many of the limitations of previous tools. Added value of this studyCommissioned by the Chief Medical Officer for England, we validated the novel clinical risk prediction model (QCovid) to identify risks of short-term severe outcomes due to COVID-19. We used national linked datasets from general practice, death registry and hospital episode data for a population-representative sample of over 34 million adults. The risk models have excellent discrimination in men and women (Harrells C statistic>0.9) and are well calibrated. QCovid represents a new, evidence-based opportunity for population risk-stratification. Implications of all the available evidenceQCovid has the potential to support public health policy, from enabling shared decision making between clinicians and patients in relation to health and work risks, to targeted recruitment for clinical trials, and prioritisation of vaccination, for example.

Protocol for the development and evaluation of a tool for predicting risk of short-term adverse outcomes due to COVID-19 in the general UK population (preprint)

Introduction: Novel coronavirus 2019 (COVID-19) has propagated a global pandemic with significant health, economic and social costs. Emerging emergence has suggested that several factors may be associated with increased risk from severe outcomes or death from COVID-19. Clinical risk prediction tools have significant potential to generate individualised assessment of risk and may be useful for population stratification and other use cases. Methods and analysis: We will use a prospective open cohort study of routinely collected data from 1205 general practices in England in the QResearch database. The primary outcome is COVID-19 mortality (in or out-of-hospital) defined as confirmed or suspected COVID-19 mentioned on the death certificate, or death occurring in a person with SARS-CoV-2 infection between 24th January and 30th April 2020. Our primary outcome in adults is COVID-19 mortality (including out of hospital and in hospital deaths). We will also examine COVID-19 hospitalisation in children. Time-to-event models will be developed in the training data to derive separate risk equations in adults (19-100 years) for males and females for evaluation of risk of each outcome within the 3-month follow-up period (24th January to 30th April 2020), accounting for competing risks. Predictors considered will include age, sex, ethnicity, deprivation, smoking status, alcohol intake, body mass index, pre-existing medical co-morbidities, and concurrent medication. Measures of performance (prediction errors, calibration and discrimination) will be determined in the test data for men and women separately and by ten-year age group. For children, descriptive statistics will be undertaken if there are currently too few serious events to allow development of a risk model. The final model will be externally evaluated in (a) geographically separate practices and (b) other relevant datasets as they become available. Ethics and dissemination: The project has ethical approval and the results will be submitted for publication in a peer-reviewed journal.

Associations between COVID-19 infection, tobacco smoking and nicotine use, common respiratory conditions and inhaled corticosteroids: a prospective QResearch-Case Mix Programme data linkage study January-May 2020 (preprint)

Introduction Epidemiological and laboratory research seems to suggest that smoking and perhaps nicotine alone could reduce the severity of COVID-19. Likewise, there is some evidence that inhaled corticosteroids could also reduce its severity, opening the possibility that nicotine and inhaled steroids could be used as treatments. Methods In this prospective cohort study, we will link English general practice records from the QResearch database to Public Health England's database of SARS-CoV-2 positive tests, Hospital Episode Statistics, admission to intensive care units, and death from COVID-19 to identify our outcomes: hospitalisation, ICU admission, and death due to COVID. Using Cox regression, we will perform sequential adjustment for potential confounders identified by separate directed acyclic graphs to: 1. Assess the association between smoking and COVID-19 disease severity, and how that changes on adjustment for smoking-related comorbidity. 2. More closely characterise the association between smoking and severe COVID-19 disease by assessing whether the association is modified by age (as a proxy of length of smoking), gender, ethnic group, and whether people have asthma or COPD. 3. Assess for evidence of a dose-response relation between smoking intensity and disease severity, which would help create a case for causality. 4. Examine the association between former smokers who are using NRT or are vaping and disease severity. 5. Examine whether pre-existing respiratory disease is associated with severe COVID-19 infection. 6. Assess whether the association between chronic obstructive pulmonary disease (COPD) and asthma and COVID-19 disease severity is modified by age, gender, ethnicity, and smoking status. 7. Assess whether the use of inhaled corticosteroids is associated with severity of COVID-19 disease. 8. To assess whether the association between use of inhaled corticosteroids and severity of COVID-19 disease is modified by the number of other airways medications used (as a proxy for severity of condition) and whether people have asthma or COPD. Conclusions This representative population sample will, to our knowledge, present the first comprehensive examination of the association between smoking, nicotine use without smoking, respiratory disease, and severity of COVID-19. We will undertake several sensitivity analyses to examine the potential for bias in these associations.