Real-time forecasting of COVID-19-related hospital strain in France using a non-Markovian mechanistic model (preprint)

Background The COVID-19 pandemic emphasised the importance of access to reliable real-time forecasts for key epidemiological indicators. Given the strong heterogeneity between regions, providing forecasts at the local level is essential for health professionals. Methods We developed a SARS-CoV-2 transmission model in France, COVIDici, that performs parameter estimation using up-to-date vaccination coverage and hospital data to provide forecasts up to a four-week horizon based on the current epidemic trend. We present the model, its associated online tool and perform a retrospective evaluation of the forecasts provided from January to December 2021 by comparing to three standard statistical forecasting methods (auto-regression, exponential smoothing, and ARIMA) at the national and regional levels. Results COVIDici allowed simultaneous real-time visualisation of several indicators of the COVID-19 epidemic at the sub-national level. For anticipating risk of critical care unit overload, it performed worse compared to the baseline methods for forecasts under the three-week horizon, but had better point forecasts at the longest horizons (e.g. four weeks) for 8 of the 13 regions considered depending on the metric. Conclusions Effective communication between modelers and clinicians is essential for utilising forecasts for health care planning. Online visualisation tools and consideration of how metrics can be affected by distortion from non-pharmaceutical government interventions facilitate this dialogue.

Modelling long-term COVID-19 hospital admission dynamics using immune protection waning data (preprint)

Immune waning is key to the timely anticipation of COVID-19 long-term dynamics. We assess the impact of periodic vaccination campaigns using a compartmental epidemiological model with multiple age structures and parameterised using empiric time-dependent vaccine protection data. Despite the inherent uncertainty, we show that vaccination on its own, especially if restricted to individuals over 60 years old, seems insufficient to prevent a large number of hospital admissions.

Estimating time-dependent infectious contact: a multi-strain epidemiological model of SARS-CoV-2 on the island of Ireland (preprint)

Mathematical modelling plays a key role in understanding and predicting the epidemiological dynamics of infectious diseases. We construct a flexible discrete-time model that incorporates multiple viral strains with different transmissibilities to estimate the changing infectious contact that generates new infections. Using a Bayesian approach, we fit the model to longitudinal data on hospitalisation with COVID-19 from the Republic of Ireland and Northern Ireland during the first year of the pandemic. We describe the estimated change in infectious contact in the context of governmentmandated non-pharmaceutical interventions in the two jurisdictions on the island of Ireland. We take advantage of the fitted model to conduct counterfactual analyses exploring the impact of lockdown timing and introducing a novel, more transmissible variant. We found substantial differences in infectious contact between the two jurisdictions during periods of varied restriction easing and December holidays. Our counterfactual analyses reveal that implementing lockdowns earlier would have decreased subsequent hospitalisation substantially in most, but not all cases, and that an introduction of a more transmissible variant - without necessarily being more severe - can cause a large impact on the health care burden.

The importance of the population age-structure: insights from Covid-19 dynamics model structured by age, time since infection and acquired immunity (preprint)

The Covid-19 pandemic outbreak was followed by an huge amount of modeling studies in order to rapidly gain insights to implement the best public health policies. However, most of those compartmental models used a classical ordinary differential equations (ODEs) system based formalism that came with the tacit assumption the time spent in each compartment does not depend of the time already spent in it. To overcome this "memoryless" issue, a widely used workaround is to artificially increase and chain the number of compartments of an unique reality (e.g. many compartments for infected individuals). It allows for a greater heterogeneity and thus be closer to the observed situation, at the cost of rendering the whole model more difficult to apprehend and parametrize. We propose here an alternative formalism based on a partial differential equations (PDEs) system instead of ordinary differential equations, which provides naturally a memory structure for each compartment, and thus allows to keep a restrained number of compartments. We use such a model applied to the French situation, accounting for vaccinal and natural immunity. The results seem to indicate that the vaccination rate is not enough to ensure the end of the epidemic, but, above all, highlight a huge uncertainty attributable to the age-structured contact matrix.

Spread of SARS-CoV-2 Delta variant infections bearing the S:E484Q and S:T95I mutations in July and August 2021 in France (preprint)

Analysing 92,598 variant screening tests performed on SARS-CoV-2 positive samples collected in France between 1 July and 31 August 2021 shows an increase of Kappa-like infections. Full genome sequencing reveals that these correspond to Delta variants bearing the S:E484Q mutation. Most of these sequences belong to a phylogenetic cluster and also bear the S:T95I mutation. Further monitoring is needed to determine if this trend is driven by undocumented superspreading events or an early signal of future viral evolutionary dynamics.

Anticipating the hospital burden of future COVID-19 epidemic waves (preprint)

Forecasting SARS-CoV-2 epidemic trends with confidence more than a few weeks ahead is almost impossible as these entirely depend on political decisions. We address this problem by investigating the consequences for the health system of an epidemic wave of a given size. This approach yields semi-quantitative results that depend on the proportion of the population already infected and vaccinated. We introduce the COVimpact software, which allows users to visualise estimated numbers of ICU admissions, deaths, and infections stratified by age class at the French departmental, regional, or national level caused by the wave. We illustrate the usefulness of our approach by showing that for France, even with a 95% vaccination coverage, the current vaccine efficiency against the delta variant would make a large epidemic wave infecting 25% of the population difficult to sustain for the current hospital bed occupancy capacity. Overall, using the final epidemic wave size and ignoring detailed epidemiological dynamics yields valuable and practical insights to optimise public health response to epidemics.

The SARS-CoV-2 B.1.351/V2 variant is outgrowing the B.1.1.7/V1 variant in French regions in April 2021 (preprint)

SARS-CoV-2 variants threaten our ability to control COVID-19 epidemics. We analyzed 36,590 variant-specific RT-PCR tests performed on samples collected between April 12 and May 7, 2021 in France to compare variant spread. Contrarily to January to March 2021, we found that the V2 variant had a significant transmission advantage over V1 in some regions (15.1 to 16.1% in Ile-de-France and 16.1 to 18.8% in Hauts-de-France). This shift in transmission advantage is consistent with the immune evasion abilities of V2 and the high levels of immunization in these regions.

SARS-CoV-2 variants of concern are associated with lower RT-PCR amplification cycles between January and March 2021 in France (preprint)

SARS-CoV-2 variants raise concern regarding the mortality caused by COVID-19 epidemics. We analyse 88,375 cycle amplification (Ct) values from variant-specific RT-PCR tests performed between January 26 and March 13, 2021. We estimate that on March 12, nearly 85% of the infections were caused by the V1 variant and that its transmission advantage over wild type strains was between 38 and 44%. We also find that tests positive for V1 and V2/V3 variants exhibit significantly lower cycle threshold (Ct) values.

Epidemiological and clinical insights from SARS-CoV-2 RT-PCR cycle amplification values (preprint)

The SARS-CoV-2 pandemic has led to an unprecedented daily use of molecular RT-PCR tests. These tests are interpreted qualitatively for diagnosis, and the relevance of the test result intensity, i.e. the number of amplification cycles (Ct), is debated because of strong potential biases. We analyze a national database of tests performed on more than 2 million individuals between January and November 2020. Although we find Ct values to vary depending on the testing laboratory or the assay used, we detect strong significant trends with patient age, number of days after symptoms onset, or the state of the epidemic (the temporal reproduction number) at the time of the test. These results suggest that Ct values can be used to improve short-term predictions for epidemic surveillance.

Rapid SARS-CoV-2 variants spread detected in France using specific RT-PCR testing (preprint)

SARS-CoV-2 variants raise major concerns regarding the control of COVID-19 epidemics. We analyse 40,000 specific RT-PCR tests performed on SARS-CoV-2-positive samples collected between Jan 26 and Feb 16, 2021. We find a high transmission advantage of variants and show that their spread in the country is more advanced than anticipated.

Rt2: computing and visualising COVID-19 epidemics temporal reproduction number (preprint)

Analysing the spread of COVID-19 epidemics in a timely manner is essential for public health authorities. However, raw numbers may be misleading because of spatial and temporal variations. We introduce Rt2, an R-program with a shiny interface, which uses incidence data, i.e. number of new cases per day, to compute variations in the temporal reproduction number ([R]t), which corresponds to the average number of secondary infections caused by an infected person. This number is computed with the R0 package, which better captures past variations, and the EpiEstim package, which provides a more accurate estimate of current values. [R]t can be computed in different countries using either the daily number of new cases or of deaths. For France, these numbers can also be computed at the regional and departmental level using also daily numbers of hospital and ICU admissions. Finally, in addition to [R]t, we represent the incidence using a one-week sliding window to buffer daily variations. Overall, Rt2 provides an accurate and timely overview of the state and speed of spread of COVID-19 epidemics at different scales, using different metrics.

Wearing masks and establishing COVID-19 areas reduces secondary attack risk in nursing homes (preprint)

Background: COVID-19 is spreading rapidly in nursing homes (NHs). It is urgent to evaluate the effect of infection prevention and control (IPC) measures to reduce COVID spreading. Methods: We analysed COVID-19 outbreaks in 12 NH using rRT-PCR for SARS-CoV2. We estimated secondary attack risks (SARs) and identified cofactors associated with the proportion of infected residents. Results: The SAR was below 5%, suggesting a high efficiency of IPC measures. Mask-wearing or establishment of COVID-19 zones for infected residents were associated with lower SAR. Conclusions: Wearing masks and isolating potentially infected residents appear to limit SARS-CoV-2 spread in nursing homes.

Age-structured non-pharmaceutical interventions for optimal control of COVID-19 epidemic (preprint)

In an epidemic, individuals can widely differ in the way they spread the infection, for instance depending on their age or on the number of days they have been infected for. The latter allows to take into account the variation of infectiousness as a function of time since infection. In the absence of pharmaceutical interventions such as a vaccine or treatment, non-pharmaceutical interventions (e.g. social distancing) are of great importance to mitigate the pandemic. We propose a model with a double continuous structure by host age and time since infection. By applying optimal control theory to our age-structured model, we identify a solution minimizing deaths and costs associated with the implementation of the control strategy itself. This strategy depends on the age heterogeneity between individuals and consists in a relatively high isolation intensity over the older populations during a hundred days, followed by a steady decrease in a way that depends on the cost associated to a such control. The isolation of the younger population is weaker and occurs only if the cost associated with the control is relatively low. We show that the optimal control strategy strongly outperforms other strategies such as uniform constant control over the whole populations or over its younger fraction. These results bring new facts the debate about age-based control interventions and open promising avenues of research, for instance of age-based contact tracing.

Early phylodynamics analysis of the COVID-19 epidemics in France (preprint)

France was one of the first countries to be reached by the COVID-19 pandemic. Here, we analyse 196 SARS-Cov-2 genomes collected between Jan 24 and Mar 24 2020, and perform a phylodynamics analysis. In particular, we analyse the doubling time, reproduction number (Rt) and infection duration associated with the epidemic wave that was detected in incidence data starting from Feb 27. Different models suggest a slowing down of the epidemic in Mar, which would be consistent with the implementation of the national lock-down on Mar 17. The inferred distributions for the effective infection duration and Rt are in line with those estimated from contact tracing data. Finally, based on the available sequence data, we estimate that the French epidemic wave originated between mid-Jan and early Feb. Overall, this analysis shows the potential to use sequence genomic data to inform public health decisions in an epidemic crisis context and calls for further analyses with denser sampling.

Epidemiological monitoring and control perspectives: application of a parsimonious modelling framework to the COVID-19 dynamics in France (preprint)

SARS-Cov-2 virus has spread over the world creating one of the fastest pandemics ever. The absence of immunity, asymptomatic transmission, and the relatively high level of virulence of the COVID-19 infection it causes led to a massive flow of patients in intensive care units (ICU). This unprecedented situation calls for rapid and accurate mathematical models to best inform public health policies. We develop an original parsimonious model that accounts for the effect of the age of infection on the natural history of the disease. Analysing the ongoing COVID-19 in France, we estimate the value of the key epidemiological parameters, such as the basic reproduction number (R0), and the efficiency of the national control strategy. We then use our deterministic model to explore several scenarios posterior to lock-down lifting and compare the efficiency of non pharmaceutical interventions (NPI) described in the literature.

Optimal COVID-19 epidemic control until vaccine deployment (preprint)

Since Dec 2019, the COVID-19 epidemic has spread over the globe creating one of the greatest pandemics ever witnessed. This epidemic wave will only begin to roll back once a critical proportion of the population is immunised, either by mounting natural immunity following infection, or by vaccination. The latter option can minimise the cost in terms of human lives but it requires to wait until a safe and efficient vaccine is developed, a period estimated to last at least 18 months. In this work, we use optimal control theory to explore the best strategy to implement while waiting for the vaccine. We seek a solution minimizing deaths and costs due to the implementation of the control strategy itself. We find that such a solution leads to an increasing level of control with a maximum reached near the 16th month of the epidemics and a steady decrease until vaccine deployment. The average containment level is approximately 50\% during the 25-months period for vaccine deployment. This strategy strongly outperforms others with constant or cycling allocations of the same amount of resources to control the outbreak. This work opens new perspectives to mitigate the effects of the ongoing COVID-19 pandemics, and be used as a proof-of-concept in using mathematical modelling techniques to enlighten decision making and public health management in the early times of an outbreak.