Combining models to generate consensus medium-term projections of hospital admissions, occupancy and deaths relating to COVID-19 in England (preprint)

Mathematical modelling has played an important role in offering informed advice during the COVID-19 pandemic. In England, a cross government and academia collaboration generated Medium-Term Projections (MTPs) of possible epidemic trajectories over the future 4-6 weeks from a collection of epidemiological models.In this paper we outline this collaborative modelling approach and evaluate the accuracy of the combined and individual model projections against the data over the period November 2021-December 2022 when various Omicron subvariants were spreading across England. Using a number of statistical methods, we quantify the predictive performance of the model projections for both the combined and individual MTPs, by evaluating the point and probabilistic accuracy. Our results illustrate that the combined MTPs, produced from an ensemble of heterogeneous epidemiological models, were a closer fit to the data than the individual models during the periods of epidemic growth or decline, with the 90% confidence intervals widest around the epidemic peaks. We also show that the combined MTPs increase the robustness and reduce the biases associated with a single model projection. Learning from our experience of ensemble modelling during the COVID-19 epidemic, our findings highlight the importance of developing cross-institutional multi-model infectious disease hubs for future outbreak control.

SARS-CoV-2, influenza A/B and respiratory syncytial virus positivity and association with influenza-like illness and self-reported symptoms, over the 2022/23 winter season in the UK: a longitudinal surveillance cohort (preprint)

Background Syndromic surveillance often relies on patients presenting to healthcare. Community cohorts, although more challenging to recruit, could provide additional population-wide insights, particularly with SARS-CoV-2 co-circulating with other respiratory viruses. Methods We estimated positivity and incidence of SARS-CoV-2, influenza A/B, and RSV, and trends in self-reported symptoms including influenza-like illness (ILI), over the 2022/23 winter season in a broadly representative UK community cohort (COVID-19 Infection Survey), using negative-binomial generalised additive models. We estimated associations between test positivity and each of symptoms and influenza vaccination, using adjusted logistic and multinomial models. Findings Swabs taken at 32,937/1,352,979 (2.4%) assessments tested positive for SARS-CoV-2, 181/14,939 (1.2%) for RSV and 130/14,939 (0.9%) for influenza A/B, varying by age over time. Positivity and incidence peaks were earliest for RSV, then influenza A/B, then SARS-CoV-2, and were highest for RSV in the youngest and for SARS-CoV-2 in the oldest age-groups. Many test-positives did not report key symptoms: middle-aged participants were generally more symptomatic than older or younger participants, but still only ~25% reported ILI-WHO and ~60% ILI-ECDC. Most symptomatic participants did not test positive for any of the three viruses. Influenza A/B-positivity was lower in participants reporting influenza vaccination in the current and previous seasons (odds ratio=0.55 (95% CI 0.32,0.95)) versus neither season. Interpretation Symptom profiles varied little by aetiology, making distinguishing SARS-CoV-2, influenza and RSV using symptoms challenging. Most symptoms were not explained by these viruses, indicating the importance of other pathogens in syndromic surveillance. Influenza vaccination was associated with lower rates of community influenza test positivity. Funding UK Health Security Agency, Department of Health and Social Care, National Institute for Health Research.

An approximate diffusion process for environmental stochasticity in infectious disease transmission modelling (preprint)

Modelling the transmission dynamics of an infectious disease is a complex task. Not only it is difficult to accurately model the inherent non-stationarity and heterogeneity of transmission, but it is nearly impossible to describe, mechanistically, changes in extrinsic environmental factors including public behaviour and seasonal fluctuations. An elegant approach to capturing environmental stochasticity is to model the force of infection as a stochastic process. However, inference in this context requires solving a computationally expensive ``missing data" problem, using data-augmentation techniques. We propose to model the time-varying transmission-potential as an approximate diffusion process using a path-wise series expansion of Brownian motion. This approximation replaces the ``missing data" imputation step with the inference of the expansion coefficients: a simpler and computationally cheaper task. We illustrate the merit of this approach through two examples: modelling influenza using a canonical SIR model, and the modelling of COVID-19 pandemic using a multi-type SEIR model.

A comparison of two frameworks for multi-state modelling, applied to outcomes after hospital admissions with COVID-19 (preprint)

We compare two multi-state modelling frameworks that can be used to represent dates of events following hospital admission for people infected during an epidemic. The methods are applied to data from people admitted to hospital with COVID-19, to estimate the probability of admission to ICU, the probability of death in hospital for patients before and after ICU admission, the lengths of stay in hospital, and how all these vary with age and gender. One modelling framework is based on defining transition-specific hazard functions for competing risks. A less commonly used framework defines partially-latent subpopulations who will experience each subsequent event, and uses a mixture model to estimate the probability that an individual will experience each event, and the distribution of the time to the event given that it occurs. We compare the advantages and disadvantages of these two frameworks, in the context of the COVID-19 example. The issues include the interpretation of the model parameters, the computational efficiency of estimating the quantities of interest, implementation in software and assessing goodness of fit. In the example, we find that some groups appear to be at very low risk of some events, in particular ICU admission, and these are best represented by using "cure-rate" models to define transition-specific hazards. We provide general-purpose software to implement all the models we describe in the "flexsurv" R package, which allows arbitrarily-flexible distributions to be used to represent the cause-specific hazards or times to events.

Trends in COVID-19 hospital outcomes in England before and after vaccine introduction, a cohort study (preprint)

Widespread vaccination campaigns have changed the landscape for COVID-19, vastly altering symptoms and reducing morbidity and mortality. We estimate trends in mortality by month of admission and vaccination status among those hospitalised with COVID-19 in England between March 2020 to September 2021, controlling for demographic factors and hospital load. Among 259,727 hospitalised COVID-19 cases, 51,948 (20.0%) experienced mortality in hospital. Hospitalised fatality risk ranged from 40.3% (95% confidence interval 39.4-41.3%) in March 2020 to 8.1% (7.2-9.0%) in June 2021. Older individuals and those with multiple co-morbidities were more likely to die or else experienced longer stays prior to discharge. Compared to unvaccinated people, the hazard of hospitalised mortality was 0.71 (0.67-0.77) with a first vaccine dose, and 0.56 (0.52-0.61) with a second vaccine dose. Compared to hospital load at 0-20% of the busiest week, the hazard of hospitalised mortality during periods of peak load (90-100%), was 1.23 (1.12-1.34). The prognosis for people hospitalised with COVID-19 in England has varied substantially throughout the pandemic and according to case-mix, vaccination, and hospital load. Our estimates provide an indication for demands on hospital resources, and the relationship between hospital burden and outcomes.

Evaluating the impact of local tracing partnerships on the performance of contact tracing for COVID-19 in England (preprint)

Assessing the impact of an intervention using time-series observational data on multiple units and outcomes is a frequent problem in many fields of scientific research. In this paper, we present a novel method to estimate intervention effects in such a setting by generalising existing approaches based on the factor analysis model and developing a Bayesian algorithm for inference. Our method is one of the few that can simultaneously: deal with outcomes of mixed type (continuous, binomial, count); increase efficiency in the estimates of the causal effects by jointly modelling multiple outcomes affected by the intervention; easily provide uncertainty quantification for all causal estimands of interest. We use the proposed approach to evaluate the impact that local tracing partnerships (LTP) had on the effectiveness of England's Test and Trace (TT) programme for COVID-19. Our analyses suggest that, overall, LTPs had a small positive impact on TT. However, there is considerable heterogeneity in the estimates of the causal effects over units and time.

Genomic epidemiology of SARS-CoV-2 in a UK university identifies dynamics of transmission (preprint)

Understanding the drivers for spread of SARS-CoV-2 in higher education settings is important to limit transmission between students, and onward spread into at-risk populations. In this study, we prospectively sequenced 482 SARS-CoV-2 isolates derived from asymptomatic student screening and symptomatic testing of students and staff at the University of Cambridge from 5 October to 6 December 2020. We performed a detailed phylogenetic comparison with 972 isolates from the surrounding community, complemented with epidemiological and contact tracing data, to determine transmission dynamics. After a limited number of viral introductions into the university, the majority of student cases were linked to a single genetic cluster, likely dispersed across the university following social gatherings at a venue outside the university. We identified considerable onward transmission associated with student accommodation and courses; this was effectively contained using local infection control measures and dramatically reduced following a national lockdown. We observed that transmission clusters were largely segregated within the university or within the community. This study highlights key determinants of SARS-CoV-2 transmission and effective interventions in a higher education setting that will inform public health policy during pandemics.

Hospitalisation risk for COVID-19 patients infected with SARS-CoV-2 variant B.1.1.7: cohort analysis (preprint)

Objective: To evaluate the relationship between coronavirus disease 2019 (COVID-19) diagnosis with SARS-CoV-2 variant B.1.1.7 (also known as Variant of Concern 202012/01) and the risk of hospitalisation compared to diagnosis with wildtype SARS-CoV-2 variants. Design: Retrospective cohort, analysed using stratified Cox regression. Setting: Community-based SARS-CoV-2 testing in England, individually linked with hospitalisation data. Participants: 839,278 laboratory-confirmed COVID-19 patients, of whom 36,233 had been hospitalised within 14 days, tested between 23rd November 2020 and 31st January 2021 and analysed at a laboratory with an available TaqPath assay that enables assessment of S-gene target failure (SGTF). SGTF is a proxy test for the B.1.1.7 variant. Patient data were stratified by age, sex, ethnicity, deprivation, region of residence, and date of positive test. Main outcome measures: Hospitalisation between 1 and 14 days after the first positive SARS-CoV-2 test. Results: 27,710 of 592,409 SGTF patients (4.7%) and 8,523 of 246,869 non-SGTF patients (3.5%) had been hospitalised within 1-14 days. The stratum-adjusted hazard ratio (HR) of hospitalisation was 1.52 (95% confidence interval [CI] 1.47 to 1.57) for COVID-19 patients infected with SGTF variants, compared to those infected with non-SGTF variants. The effect was modified by age (P<0.001), with HRs of 0.93-1.21 for SGTF compared to non-SGTF patients below age 20 years, 1.29 in those aged 20-29, and 1.45-1.65 in age groups 30 years or older. Conclusions: The results suggest that the risk of hospitalisation is higher for individuals infected with the B.1.1.7 variant compared to wildtype SARS-CoV-2, likely reflecting a more severe disease. The higher severity may be specific to adults above the age of 30.

Decreasing hospital burden of COVID-19 during the first wave in Regione Lombardia: an emergency measures context (preprint)

Background: The aim of this study is to quantify the hospital burden of COVID-19 during the first wave and how it changed over calendar time; to interpret the results in light of the emergency measures introduced to manage the strain on secondary healthcare. Methods: : This is a cohort study of hospitalised confirmed cases of COVID-19 admitted from February-June 2020 and followed up till 17th July 2020, analysed using a mixture multi-state model. All hospital patients with confirmed COVID-19 disease in Regione Lombardia were involved, admitted from February-June 2020, with non-missing hospital of admission and non-missing admission date. Results: : The cohort consists of 40,550 patients hospitalised during the first wave. These patients had a median age of 69 (interquartile range 56-80) and were more likely to be men (60%) than women (40%). The hospital-fatality risk, averaged over all pathways through hospital, was 27.5% (95% CI 27.1-28.0%); and steadily decreased from 34.6% (32.5-36.6%) in February to 7.6% (6.3-10.6%) in June. Among surviving patients, median length of stay in hospital was 11.8 (11.6-12.3) days, compared to 8.1 (7.8-8.5) days in non-survivors. Averaged over final outcomes, median length of stay in hospital decreased from 21.4 (20.5-22.8) days in February to 5.2 (4.7-5.8) days in June. Conclusions: : The hospital burden, in terms of both risks of poor outcomes and lengths of stay in hospital, has been demonstrated to have decreased over the months of the first wave, perhaps reflecting improved treatment and management of COVID-19 cases, as well as reduced burden as the first wave waned. The quantified burden allows for planning of hospital beds needed for current and future waves of SARS-CoV-2.

Quantifying efficiency gains of innovative designs of two-arm vaccine trials for COVID-19 using an epidemic simulation model (preprint)

Clinical trials of a vaccine during an epidemic face particular challenges, such as the pressure to identify an effective vaccine quickly to control the epidemic, and the effect that time-space-varying infection incidence has on the power of a trial. We illustrate how the operating characteristics of different trial design elements may be evaluated using a network epidemic and trial simulation model, based on COVID-19 and individually randomised two-arm trials with a binary outcome. We show that "ring" recruitment strategies, prioritising participants at high risk of infection, can result in substantial improvement in terms of power, if sufficiently many contacts of observed cases are at high risk. In addition, we introduce a novel method to make more efficient use of the data from the earliest cases of infection observed in the trial, whose infection may have been too early to be vaccine-preventable. Finally, we compare several methods of response-adaptive randomisation, discussing their advantages and disadvantages in this two-arm context and identifying particular adaptation strategies that preserve power and estimation properties, while slightly reducing the number of infections, given an effective vaccine.

Trends in risks of severe events and lengths of stay for COVID-19 hospitalisations in England over the pre-vaccination era: results from the Public Health England SARI-Watch surveillance scheme (preprint)

Background: Trends in hospitalised case-fatality risk (HFR), risk of intensive care unit (ICU) admission and lengths of stay for patients hospitalised for COVID-19 in England over the pre-vaccination era are unknown. Methods: Data on hospital and ICU admissions with COVID-19 at 31 NHS trusts in England were collected by Public Health England's Severe Acute Respiratory Infections surveillance system and linked to death information. We applied parametric multi-state mixture models, accounting for censored outcomes and regressing risks and times between events on month of admission, geography, and baseline characteristics. Findings: 20,785 adults were admitted with COVID-19 in 2020. Between March and June/July/August estimated HFR reduced from 31.9% (95% confidence interval 30.3-33.5%) to 10.9% (9.4-12.7%), then rose steadily from 21.6% (18.4-25.5%) in September to 25.7% (23.0-29.2%) in December, with steeper increases among older patients, those with multi-morbidity and outside London/South of England. ICU admission risk reduced from 13.9% (12.8-15.2%) in March to 6.2% (5.3-7.1%) in May, rising to a high of 14.2% (11.1-17.2%) in September. Median length of stay in non-critical care increased during 2020, from 6.6 to 12.3 days for those dying, and from 6.1 to 9.3 days for those discharged. Interpretation: Initial improvements in patient outcomes, corresponding to developments in clinical practice, were not sustained throughout 2020, with HFR in December approaching the levels seen at the start of the pandemic, whilst median hospital stays have lengthened. The role of increased transmission, new variants, case-mix and hospital pressures in increasing COVID-19 severity requires urgent further investigation.

Changes in UK hospital mortality in the first wave of COVID-19: the ISARIC WHO Clinical Characterisation Protocol prospective multicentre observational cohort study (preprint)

Background Mortality rates of UK patients hospitalised with COVID-19 appeared to fall during the first wave. We quantify potential drivers of this change and identify groups of patients who remain at high risk of dying in hospital. Methods The International Severe Acute Respiratory and emerging Infection Consortium (ISARIC) WHO Clinical Characterisation Protocol UK recruited a prospective cohort admitted to 247 acute UK hospitals with COVID-19 in the first wave (March to August 2020). Outcome was hospital mortality within 28 days of admission. We performed a three-way decomposition mediation analysis using natural effects models to explore associations between week of admission and hospital mortality adjusting for confounders (demographics, comorbidity, illness severity) and quantifying potential mediators (respiratory support and steroids). Findings Unadjusted hospital mortality fell from 32.3% (95%CI 31.8, 32.7) in March/April to 16.4% (95%CI 15.0, 17.8) in June/July 2020. Reductions were seen in all ages, ethnicities, both sexes, and in comorbid and non-comorbid patients. After adjustment, there was a 19% reduction in the odds of mortality per 4 week period (OR 0.81, 95%CI 0.79, 0.83). 15.2% of this reduction was explained by greater disease severity and comorbidity earlier in the epidemic. The use of respiratory support changed with greater use of non-invasive ventilation (NIV). 22.2% (OR 0.94, 95%CI 0.94, 0.96) of the reduction in mortality was mediated by changes in respiratory support. Interpretation The fall in hospital mortality in COVID-19 patients during the first wave in the UK was partly accounted for by changes in case mix and illness severity. A significant reduction was associated with differences in respiratory support and critical care use, which may partly reflect improved clinical decision making. The remaining improvement in mortality is not explained by these factors, and may relate to community behaviour on inoculum dose and hospital capacity strain. Funding NIHR & MRC

Short-term forecasts to inform the response to the COVID-19 epidemic in the UK (preprint)

Background: Short-term forecasts of infectious disease can create situational awareness and inform planning for outbreak response. Here, we report on multi-model forecasts of Covid-19 in the UK that were generated at regular intervals starting at the end of March 2020, in order to monitor expected healthcare utilisation and population impacts in real time. Methods: We evaluated the performance of individual model forecasts generated between 24 March and 14 July 2020, using a variety of metrics including the weighted interval score as well as metrics that assess the calibration, sharpness, bias and absolute error of forecasts separately. We further combined the predictions from individual models to ensemble forecasts using a simple mean as well as a quantile regression average that aimed to maximise performance. We further compared model performance to a null model of no change. Results: In most cases, individual models performed better than the null model, and ensembles models were well calibrated and performed comparatively to the best individual models. The quantile regression average did not noticeably outperform the mean ensemble. Conclusions: Ensembles of multi-model forecasts can inform the policy response to the Covid-19 pandemic by assessing future resource needs and expected population impact of morbidity and mortality.

A prospective study of risk factors associated with seroprevalence of SARS-CoV-2 antibodies in healthcare workers at a large UK teaching hospital (preprint)

Background The COVID-19 pandemic continues to grow at an unprecedented rate. Healthcare workers (HCWs) are at higher risk of SARS-CoV-2 infection than the general population but risk factors for HCW infection are not well described. Methods We conducted a prospective sero-epidemiological study of HCWs at a UK teaching hospital using a SARS-CoV-2 immunoassay. Risk factors for seropositivity were analysed using multivariate logistic regression. Findings 410/5,698 (7.2%) staff tested positive for SARS-CoV-2 antibodies. Seroprevalence was higher in those working in designated COVID-19 areas compared with other areas (9.47% versus 6.16%) Healthcare assistants (aOR 2.06 [95%CI 1.14-3.71]; p=0.016) and domestic and portering staff (aOR 3.45 [95% CI 1.07-11.42]; p=0.039) had significantly higher seroprevalence than other staff groups after adjusting for age, sex, ethnicity and COVID-19 working location. Staff working in acute medicine and medical sub-specialities were also at higher risk (aOR 2.07 [95% CI 1.31-3.25]; p=0.002). Staff from Black, Asian and minority ethnic (BAME) backgrounds had an aOR of 1.65 (95% CI 1.32-2.07; p<0.0001) compared to white staff; this increased risk was independent of COVID-19 area working. The only symptoms significantly associated with seropositivity in a multivariable model were loss of sense of taste or smell, fever and myalgia; 31% of staff testing positive reported no prior symptoms. Interpretation Risk of SARS-CoV-2 infection amongst HCWs is heterogeneous and influenced by COVID-19 working location, role, age and ethnicity. Increased risk amongst BAME staff cannot be accounted for solely by occupational factors. Funding Wellcome Trust, Addenbrookes Charitable Trust, National Institute for Health Research, Academy of Medical Sciences, the Health Foundation and the NIHR Cambridge Biomedical Research Centre.

A2B-COVID: A method for evaluating potential SARS-CoV-2 transmission events (preprint)

Identifying linked cases of infection is a key part of the public health response to viral infectious disease. Viral genome sequence data is of great value in this task, but requires careful analysis, and may need to be complemented by additional types of data. The Covid-19 pandemic has highlighted the urgent need for analytical methods which bring together sources of data to inform epidemiological investigations. We here describe A2B-COVID, an approach for the rapid identification of linked cases of coronavirus infection. Our method combines knowledge about infection dynamics, data describing the movements of individuals, and novel approaches to genome sequence data to assess whether or not cases of infection are consistent or inconsistent with linkage via transmission. We apply our method to analyse and compare data collected from two wards at Cambridge University Hospitals, showing qualitatively different patterns of linkage between cases on designated Covid-19 and non-Covid-19 wards. Our method is suitable for the rapid analysis of data from clinical or other potential outbreak settings.

Nowcasting CoVID-19 Deaths in England by Age and Region (preprint)

Understanding the trajectory of the daily numbers of deaths in people with CoVID-19 is essential to decisions on the response to the CoVID-19 pandemic. Estimating this trajectory from data on numbers of deaths is complicated by the delay between deaths occurring and their being reported to the authorities. In England, Public Health England receives death reports from a number of sources and the reporting delay is typically several days, but can be several weeks. Delayed reporting results in considerable uncertainty about the number of deaths that occurred on the most recent days. In this article, we estimate the number of deaths per day in each of five age strata within seven English regions. We use a Bayesian hierarchical model that involves a submodel for the number of deaths per day and a submodel for the reporting delay distribution. This model accounts for reporting-day effects and longer-term changes over time in the delay distribution. We show how the model can be fitted in a computationally efficient way when the delay distribution is same in multiple strata, e.g. over a wide range of ages.

Real-time Nowcasting and Forecasting of COVID-19 Dynamics in England: the first wave? (preprint)

England has been heavily affected by the SARS-CoV-2 pandemic, with severe 'lock-down' mitigation measures now gradually being lifted. The real-time pandemic monitoring presented here has contributed to the evidence informing this pandemic management. Estimates on the 10th May showed lock-down had reduced transmission by 75%, the reproduction number falling from 2.6 to 0.61. This regionally-varying impact was largest in London of 81% (95% CrI: 77%-84%). Reproduction numbers have since slowly increased, and on 19th June the probability that the epidemic is growing was greater than 50% in two regions, South West and London. An estimated 8% of the population had been infected, with a higher proportion in London (17%). The infection-to-fatality ratio is 1.1% (0.9%-1.4%) overall but 17% (14%-22%) among the over-75s. This ongoing work will be key to quantifying any widespread resurgence should accrued immunity and effective contact tracing be insufficient to preclude a second wave.

Transmission dynamics and control measures of COVID-19 outbreak in China: a modelling study (preprint)

COVID-19 is reported to have been effectively brought under control in China at its initial start place. To understand the COVID-19 outbreak in China and provide potential lessons for other parts of the world, in this study we combine a mathematical modelling with multiple datasets to estimate its transmissibility and severity and how it was affected by the unprecedented control measures. Our analyses show that before 29th January 2020, the ascertainment rate is 6.9%(95%CI: 3.5 - 14.6%); then it increased to 41.5%(95%CI: 30.6 - 65.1%). The basic reproduction number (R0) was 2.23(95%CI: 1.86 - 3.22) before 8th February 2020; then it dropped to 0.04(95%CI: 0.01 - 0.10). This estimation also indicates that the effect on transmissibility of control measures taken since 23rd January 2020 emerged about two weeks late. The confirmed case fatality rate is estimated at 4.41%(95%CI: 3.65 - 5.30%). This shows that SARS-CoV-2 virus is highly transmissible but less severe than SARS-CoV-1 and MERS-CoV. We found that at the early stage, the majority of R0 comes from the undetected infected people. This implies that the successful control in China was achieved through decreasing the contact rates among people in general populations and increasing the rate of detection and quarantine of the infected cases.