Endemic means change as SARS-CoV-2 evolves (preprint)

COVID-19 has become endemic, with dynamics that reflect the waning of immunity and re-exposure, by contrast to the epidemic phase driven by exposure in immunologically naive populations. Endemic does not, however, mean constant. Further evolution of SARS-CoV-2, as well as changes in behaviour and public health policy, continue to play a major role in the endemic load of disease and mortality. In this paper, we analyse evolutionary models to explore the impact that newly arising variants can have on the short-term and longer-term endemic load, characterizing how these impacts depend on the transmission and immunological properties of variants. We describe how evolutionary changes in the virus will increase the endemic load most for persistently immune-escape variants, by an intermediate amount for more transmissible variants, and least for transiently immune-escape variants. Balancing the tendency for evolution to favour variants that increase the endemic load, we explore the impact of vaccination strategies and non-pharmaceutical interventions (NPIs) that can counter these increases in the impact of disease. We end with some open questions about the future of COVID-19 as an endemic disease.

The Role of Vaccine Status Homophily in the COVID-19 Pandemic: A Cross-Sectional Survey with Modeling (preprint)

Background: Vaccine homophily describes non-heterogeneous vaccine uptake within contact networks. This study was performed to determine observable patterns of vaccine homophily, associations between vaccine homophily, self-reported vaccination, COVID-19 prevention behaviours, contact network size, and self-reported COVID-19, as well as the impact of vaccine homophily on disease transmission within and between vaccination groups under conditions of high and low vaccine efficacy. Methods: Residents of British Columbia, Canada, aged [≥]16 years, were recruited via online advertisements between February and March 2022, and provided information about vaccination status, perceived vaccination status of household and non-household contacts, compliance with COVID-19 prevention guidelines, and history of COVID-19. A deterministic mathematical model was used to assess transmission dynamics between vaccine status groups under conditions of high and low vaccine efficacy. Results: Vaccine homophily was observed among the 1304 respondents, but was lower among those with fewer doses (p<0.0001). Unvaccinated individuals had larger contact networks (p<0.0001), were more likely to report prior COVID-19 (p<0.0001), and reported lower compliance with COVID-19 prevention guidelines (p<0.0001). Mathematical modelling showed that vaccine homophily plays a considerable role in epidemic growth under conditions of high and low vaccine efficacy. Further, vaccine homophily contributes to a high force of infection among unvaccinated individuals under conditions of high vaccine efficacy, as well as elevated force of infection from unvaccinated to vaccinated individuals under conditions of low vaccine efficacy. Interpretation: The uneven uptake of COVID-19 vaccines and the nature of the contact network in the population play important roles in shaping COVID-19 transmission dynamics.

Cov2clusters: genomic clustering of SARS-CoV-2 sequences (preprint)

Background: The COVID-19 pandemic remains a global public health concern. Advances in rapid sequencing has led to an unprecedented level of SARS-CoV-2 whole genome sequence (WGS) data and sharing of sequences through global repositories that has enabled almost real-time genomic analysis of the pathogen. WGS data has been used previously to group genetically similar viral pathogens to reveal evidence of transmission, including methods that identify distinct clusters on a phylogenetic tree. Identifying clusters of linked cases can aid in the regional surveillance and management of the disease. In this study, we present a novel method for producing stable genomic clusters of SARS-CoV-2 cases, cov2clusters, and compare the sensitivity and stability of our approach to previous methods used for phylogenetic clustering using real-world SARS-CoV-2 sequence data obtained from British Columbia, Canada, Results: We found that cov2clusters produced more stable clusters than previously used phylogenetic clustering methods when adding sequence data through time, mimicking an increase in sequence data through the pandemic. Our method also showed high sensitivity when compared to epidemiologically informed clusters. These clusters often contained a high number of cases that were identical or near identical genetically. Conclusions: This new approach presented here allows for the identification of stable clusters of SARS-CoV-2 from WGS data. Producing high-resolution SARS-CoV-2 clusters from sequence data alone can a challenge and, where possible, both genomic and epidemiological data should be used in combination.

Genomic epidemiology offers high resolution estimates of serial intervals for COVID-19 (preprint)

Estimating key aspects of transmission is crucial in infectious disease control. Serial intervals - the time between symptom onset in an infector and infectee - are fundamental, and help to define rates of transmission, estimates of reproductive numbers, and vaccination levels needed to prevent transmission. However, estimating the serial interval requires knowledge of individuals' contacts and exposures (who infected whom), which is typically obtained through resource-intensive contact tracing efforts. We develop an alternate framework that uses virus sequences to inform who infected whom and thereby estimate serial intervals. The advantages are many-fold: virus sequences are often routinely collected to support epidemiological investigations and to monitor viral evolution. The genomic approach offers high resolution and cluster-specific estimates of the serial interval that are comparable with those obtained from contact tracing data. Our approach does not require contact tracing data, and can be used in large populations and over a range of time periods. We apply our techniques to SARS-CoV-2 sequence data from the first two waves of COVID-19 in Victoria, Australia. We find that serial interval estimates vary between clusters, supporting the need to monitor this key parameter and use updated estimates in onward applications. Compared to an early published serial interval estimate, using cluster-specific serial intervals can cause estimates of the effective reproduction number Rt to vary by a factor of up to 2-3. We also find that serial intervals estimated in settings such as schools and meat processing/packing plants tend to be shorter than those estimated in healthcare facilities.

COVID-19 endgame: from pandemic to endemic? Vaccination, reopening and evolution in a well-vaccinated population (preprint)

COVID-19 remains a major public health concern, with large resurgences even where there has been widespread uptake of vaccines. Waning immunity and the emergence of new variants will shape the long-term burden and dynamics of COVID-19. We explore the transition to the endemic state, and the endemic incidence, using a combination of modelling approaches. We compare gradual and rapid reopening and reopening at different vaccination levels. We examine how the eventual endemic state depends on the duration of immunity, the rate of importations, the efficacy of vaccines and the transmissibility. These depend on the evolution of the virus, which continues to undergo selection. Slower reopening leads to a lower peak level of incidence and fewer overall infections: as much as a 60% lower peak and a 10% lower total in some illustrative simulations; under realistic parameters, reopening when 70% of the population is vaccinated leads to a large resurgence in cases. The long-term endemic behaviour may stabilize as late as January 2023, with further waves of high incidence occurring depending on the transmissibility of the prevalent variant, duration of immunity, and antigenic drift. We find that long term endemic levels are not necessarily lower than current pandemic levels: in a population of 100,000 with representative parameter settings (Reproduction number 5, 1-year duration of immunity, vaccine efficacy at 80% and importations at 3 cases per 100K per day) there are over 100 daily incident cases in the model. The consequent burden on health care systems depends on the severity of infection in immunized or previously infected individuals.

COVID-19 clusters in schools: frequency, size, and transmission rates from crowdsourced exposure reports (preprint)

The role of schools in the spread of the COVID-19 pandemic is controversial, with some claiming they are an important driver of the pandemic and others arguing that transmission in schools is negligible. School cluster reports that have been collected in various jurisdictions are a source of data about transmission in schools. These reports consist of the name of a school, a date, and the number of students known to be infected. We provide a simple model for the frequency and size of clusters in this data, based on random arrivals of index cases at schools who then infect their classmates with a highly variable rate, fitting the overdispersion evident in the data. We fit our model to reports for several jurisdictions in the US and Canada, providing estimates of mean and dispersion for cluster size, whilst factoring in imperfect ascertainment. Our parameter estimates are robust to variations in ascertainment fraction. We use these estimates in three ways: i) to explore how uneven the distribution of cases is among different clusters in different jurisdictions (that is, what fraction of cases are in the 20% largest clusters), ii) to estimate how long it will be until we see a cluster a given size in jurisdiction, and iii) to determine the distribution of instantaneous transmission rate {beta} among different index case. We show how these latter distribution can be used in simulations of school transmission where we explore the effect of different interventions, in the context of highly variable transmission rates.

Mathematical modeling of COVID-19 in British Columbia:an age-structured model with time-dependent contact rates (preprint)

Following the emergence of COVID-19 at the end of 2019, several mathematical models have been developed to study the transmission dynamics of this disease. Many of these models assume homogeneous mixing in the underlying population. However, contact rates and mixing patterns can vary dramatically among individuals depending on their age and activity level. Variation in contact rates among age groups and over time can significantly impact how well a model captures observed trends. To properly model the age-dependent dynamics of COVID-19 and understand the impacts of interventions, it is essential to consider heterogeneity arising from contact rates and mixing patterns. We developed an age-structured model that incorporates time-varying contact rates and population mixing computed from the ongoing BC Mix COVID-19 survey to study transmission dynamics of COVID-19 in British Columbia (BC), Canada. Using a Bayesian inference framework, we fit four versions of our model to weekly reported cases of COVID-19 in BC, with each version allowing different assumptions of contact rates. We show that in addition to incorporating age-specific contact rates and mixing patterns, time-dependent (weekly) contact rates are needed to adequately capture the observed transmission dynamics of COVID-19. Our approach provides a framework for explicitly including empirical contact rates in a transmission model, which removes the need to otherwise model the impact of many non-pharmaceutical interventions. Further, this approach allows projection of future cases based on clear assumptions of age-specific contact rates, as opposed to less tractable assumptions regarding transmission rates.

Estimation of SARS-CoV-2 antibody prevalence through integration of serology and incidence data (preprint)

Serology tests for SARS-CoV-2 provide a paradigm for estimating the number of individuals who have had infection in the past (including cases that are not detected by routine testing, which has varied over the course of the pandemic and between jurisdictions). Classical statistical approaches to such estimation do not incorporate case counts over time, and may be inaccurate due to uncertainty about the sensitivity and specificity of the serology test. In this work, we provide a joint Bayesian model for case counts and serological data, integrating uncertainty through priors on the sensitivity and specificity. We also model the Phases of the pandemic with exponential growth and decay. This model improves upon maximum likelihood estimates by conditioning on more data, and by taking into account the epidemiological trajectory. We apply our model to the greater Vancouver area, British Columbia, Canada with data acquired during Phase 1 of the pandemic.

Vaccine Rollout Strategies: The Case for Vaccinating Essential Workers Early (preprint)

In planning for upcoming mass vaccinations against COVID-19, many jurisdictions have proposed using primarily age-based rollout strategies, where the oldest are vaccinated first and the youngest last. In the wake of growing evidence that approved vaccines are effective at preventing not only adverse outcomes, but also infection (and hence transmission of SARS-CoV-2), we propose that such age-based rollouts are both less equitable and less effective than strategies that prioritize essential workers. We demonstrate using modelling that strategies that target essential workers earlier consistently outperform those that do not, and that prioritizing essential workers provides a significant level of indirect protection for older adults. This conclusion holds across numerous outcomes, including cases, hospitalizations, deaths, prevalence of Long COVID, chronic impacts of COVID, quality adjusted life years lost and net monetary benefit lost. It also holds over a range of possible values for the efficacy of vaccination against infection. Our analysis focuses on regimes where the pandemic continues to be controlled with distancing and other measures as vaccination proceeds, and where the vaccination strategy is expected to last for over the coming 6-8 months - for example British Columbia, Canada. In such a setting with a total population of 5M, vaccinating essential workers sooner is expected to prevent over 200,000 infections, over 600 deaths, and to produce a net monetary benefit of over $500M. 20-25% of the quality adjusted life years lost, and 28-34% of the net monetary benefit lost, are due to chronic impacts of COVID-19.

Quantifying transmissibility of COVID-19 and impact of intervention within long-term health care facilities (preprint)

Estimates of the basic reproduction number (R0) for Coronavirus disease 2019 (COVID-19) are particularly variable in the context of transmission within locations such as long-term health care (LTHC) facilities. We sought to characterise the heterogeneity of R0 across known outbreaks within these facilities. We used a unique comprehensive dataset of all outbreaks that have occurred within LTHC facilities in British Columbia, Canada. We estimated R0 with a Bayesian hierarchical dynamic model of susceptible, exposed, infected, and recovered individuals, that incorporates heterogeneity of R0 between facilities. We further compared these estimates to those obtained with standard methods that utilize the exponential growth rate and maximum likelihood. The total size of an outbreak varied dramatically, with a range of attack rates of 2%-86%. The Bayesian analysis provides more constrained overall estimates of R0 = 2.83 (90% CrI [credible interval] 0.25-7.19) than standard methods, with a range within facilities of 0.66-10.06. We further estimated that intervention led to 67% (56%-73%) of all cases being averted within the LTHC facilities. Understanding the risks and impact of intervention are essential in planning during the ongoing global pandemic, particularly in high-risk environments such as LTHC facilities.

Modelling the impact of household size distribution on the transmission dynamics of COVID-19 (preprint)

Under the implementation of non-pharmaceutical interventions such as social distancing and lockdowns, household transmission has been shown to be significant for COVID-19, posing challenges for reducing incidence in settings where people are asked to self-isolate at home and to spend increasing amounts of time at home due to distancing measures. Accordingly, characteristics of households in a region have been shown to relate to transmission heterogeneity of the virus. We introduce a stochastic epidemiological model to examine the impact of the household size distribution in a region on the transmission dynamics. We choose parameters to reflect incidence in two health regions of the Greater Vancouver area in British Columbia and simulate the impact of distancing measures on transmission, with household size distribution the only different parameter between simulations for the two regions. Our result suggests that the dissimilarity in household size distribution alone can cause significant differences in incidence of the two regions, and the distributions drive distinct dynamics that match reported cases. Furthermore, our model suggests that offering individuals a place to isolate outside their household can speed the decline in cases, and does so more effectively where there are more larger households.

Fundamental Limitations of Contact Tracing for COVID-19 (preprint)

Contact tracing has played a central role in COVID-19 control in many jurisdictions and is often used in conjunction with other measures such as travel restrictions and social distancing mandates. Contact tracing is made ineffective, however, by delays in testing, calling, and isolating. Even if delays are minimized, contact tracing can only prevent a fraction of onward transmissions from contacts. Without other measures in place, contact tracing alone is insufficient to prevent exponential growth in the number of cases. Even when used effectively with other measures, occasional bursts in call loads can overwhelm contact tracing systems and lead to a loss of control. We propose embracing approaches to COVID-19 control that broadly test individuals without symptoms, in whatever way is economically feasible, either with fast cheap tests that can be deployed widely, with pooled testing, or with screening of judiciously chosen groups of high-risk individuals. Only by ramping up testing of asymptomatic individuals can we avoid the inherent delays that limit the efficacy of contact tracing.

Long-Term Persistence of Spike Antibody and Predictive Modeling of Antibody Dynamics Following Infection with SARS-CoV-2 (preprint)

Background: Antibodies to Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) have been shown to neutralize the virus in-vitro and prevent disease in animal challenge models upon re-exposure. However, current understanding of SARS-CoV-2 humoral dynamics and longevity is conflicting.Methods: The Co-Stars study prospectively enrolled 3679 healthcare workers to comprehensively characterize the kinetics of SARS-CoV-2 spike (S), receptor-binding-domain (RBD) and nucleoprotein (N) antibodies in parallel. Participants screening seropositive had serial monthly serological testing for maximum 7 months with the Mesoscale Discovery Assay. Survival analysis determined the proportion of sero-reversion while two hierarchical Gamma models predicted the upper- and lower-bounds of long-term antibody trajectory.Results: A total of 1163 monthly samples were provided from 349 seropositive participants. At 200 days post-symptoms, 99% of participants had detectable S-antibodies compared to 75% with detectable N-antibodies. S-antibody was predicted to remain detectable in 95% of participants until 465 days [95%CI 370-575] using a ‘continuous-decay’ model and indefinitely using a ‘decay-to-plateau’ model to account for antibody secretion by long-lived plasma cells. S-antibody titers correlated strongly with surrogate neutralization in-vitro (R2=0.72). N-antibodies, however, decayed rapidly with a half-life of 60 days [95%CI 52-68].Conclusions: The Co-STAR's study data presented here provides evidence for long-term persistence of neutralizing S-antibodies. This has important implications for the duration of functional immunity following SARS-CoV-2 infection. In contrast, the rapid decay of N-antibodies must be considered in future seroprevalence studies and public health decision-making. This is the first study to establish a mathematical framework capable of predicting long-term humoral dynamics following SARS-CoV-2 infection.Trial Registration: NCT04380896.Funding Statement: GOSH charity, Wellcome Trust (201470/Z/16/Z and 220565/Z/20/Z). GOSH NIHR Funded Biomedical Research Centre.Declaration of Interests: The authors have declared that no competing interests exist.Ethics Approval Statement: This study was approved by the UK Health Research Authority (www.hra.nhs.uk). Written informed consent was obtained from all participants before recruitment to the study.

Long-Term Persistence of Spike Antibody and Predictive Modeling of Antibody Dynamics Following Infection with SARS-CoV-2 (preprint)

Background: Antibodies to Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) have been shown to neutralize the virus in-vitro. Similarly, animal challenge models suggest that neutralizing antibodies isolated from SARS-CoV-2 infected individuals prevent against disease upon re-exposure to the virus. Understanding the nature and duration of the antibody response following SARS-CoV-2 infection is therefore critically important. Methods: Between April and October 2020 we undertook a prospective cohort study of 3555 healthcare workers in order to elucidate the duration and dynamics of antibody responses following infection with SARS-CoV-2. After a formal performance evaluation against 169 PCR confirmed cases and negative controls, the Meso-Scale Discovery assay was used to quantify in parallel, antibody titers to the SARS-CoV-2 nucleoprotein (N), spike (S) protein and the receptor-binding-domain (RBD) of the S-protein. All seropositive participants were followed up monthly for a maximum of 7 months; those participants that were symptomatic, with known dates of symptom-onset, seropositive by the MSD assay and who provided 2 or more monthly samples were included in the analysis. Survival analysis was used to determine the proportion of sero-reversion (switching from positive to negative) from the raw data. In order to predict long-term antibody dynamics, two hierarchical longitudinal Gamma models were implemented to provide predictions for the lower bound (continuous antibody decay to zero, 'Gamma-decay') and upper bound (decay-to-plateau due to long lived plasma cells, 'Gamma-plateau') long-term antibody titers. Results: A total of 1163 samples were provided from 349 of 3555 recruited participants who were symptomatic, seropositive by the MSD assay, and were followed up with 2 or more monthly samples. At 200 days post symptom onset, 99% of participants had detectable S-antibody whereas only 75% of participants had detectable N-antibody. Even under our most pessimistic assumption of persistent negative exponential decay, the S-antibody was predicted to remain detectable in 95% of participants until 465 days [95% CI 370-575] after symptom onset. Under the Gamma-plateau model, the entire posterior distribution of S-antibody titers at plateau remained above the threshold for detection indefinitely. Surrogate neutralization assays demonstrated a strong positive correlation between antibody titers to the S-protein and blocking of the ACE-2 receptor in-vitro [R2=0.72, p<0.001]. By contrast, the N-antibody waned rapidly with a half-life of 60 days [95% CI 52-68]. Discussion: This study has demonstrated persistence of the spike antibody in 99% of participants at 200 days following SARS-CoV-2 symptoms and rapid decay of the nucleoprotein antibody. Diagnostic tests or studies that rely on the N-antibody as a measure of seroprevalence must be interpreted with caution. Our lowest bound prediction for duration of the spike antibody was 465 days and our upper bound predicted spike antibody to remain indefinitely in line with the long-term seropositivity reported for SARS-CoV infection. The long-term persistence of the S-antibody, together with the strong positive correlation between the S-antibody and viral surrogate neutralization in-vitro, has important implications for the duration of functional immunity following SARS-CoV-2 infection.

COVID-19's unfortunate events in schools: mitigating classroom clusters in the context of variable transmission (preprint)

Widespread school closures occurred during the COVID-19 pandemic. Because closures are costly and damaging, many jurisdictions have since reopened schools with control measures in place. Early evidence indicated that schools were low risk and children were unlikely to be very infectious, but it is becoming clear that children and youth can acquire and transmit COVID-19 in school settings and that transmission clusters and outbreaks can be large. We describe the contrasting literature on school transmission, and argue that the apparent discrepancy can be reconciled by heterogeneity, or ``overdispersion'' in transmission, with many exposures yielding little to no risk of onward transmission, but some unfortunate exposures causing sizeable onward transmission. In addition, respiratory viral loads are as high in children and youth as in adults, pre- and asymptomatic transmission occur, and the possibility of aerosol transmission has been established. We use a stochastic individual-based model to find the implications of these combined observations for cluster sizes and control measures. We consider both individual and environment/activity contributions to the transmission rate, as both are known to contribute to variability in transmission. We find that even small heterogeneities in these contributions result in highly variable transmission cluster sizes in the classroom setting, with clusters ranging from 1 to 20 individuals in a class of 25. None of the mitigation protocols we modeled, initiated by a positive test in a symptomatic individual, are able to prevent large transmission clusters unless the transmission rate is low (in which case large clusters do not occur in any case). Among the measures we modeled, only rapid universal monitoring (for example by regular, onsite, pooled testing) accomplished this prevention. We suggest approaches and the rationale for mitigating these ``unfortunate events'', even if they are expected to be rare.

Humoral Response Dynamics Following Infection with SARS-CoV-2 (preprint)

Introduction: Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) specific antibodies have been shown to neutralize the virus in-vitro. Understanding antibody dynamics following SARS-CoV-2 infection is therefore crucial. Sensitive measurement of SARS-CoV-2 antibodies is also vital for large seroprevalence surveys which inform government policies and public health interventions. However, rapidly waning antibodies following SARS-CoV-2 infection could jeopardize the sensitivity of serological testing on which these surveys depend. Methods: This prospective cohort study of SARS-CoV-2 humoral dynamics in a central London hospital analyzed 137 serial samples collected from 67 participants seropositive to SARS-CoV-2 by the Meso-Scale Discovery assay. Antibody titers were quantified to the SARS-CoV-2 nucleoprotein (N), spike (S-)protein and the receptor-binding-domain (RBD) of the S-protein. Titers were log-transformed and a multivariate log-linear model with time-since-infection and clinical variables was fitted by Bayesian methods. Results: The mean estimated half-life of the N-antibody was 52 days (95% CI 42-65). The S- and RBD-antibody had significantly longer mean half-lives of 81 days (95% CI 61-111) and 83 days (95% CI 55-137) respectively. An ACE-2-receptor competition assay demonstrated significant correlation between the S and RBD-antibody titers and ACE2-receptor blocking in-vitro. The time-to-a-negative N-antibody test for 50% of the seropositive population was predicted to be 195 days (95% CI 163-236). Discussion: After SARS-CoV-2 infection, the predicted half-life of N-antibody was 52 days with 50% of seropositive participants becoming seronegative to this antibody at 195 days. Widely used serological tests that depend on the N-antibody will therefore significantly underestimate the prevalence of infection following the majority of infections.

Event-specific interventions to minimize COVID-19 transmission (preprint)

Coronavirus disease 2019 (COVID-19) is a global pandemic with over 11 million cases worldwide. Currently there is no treatment and no vaccine. Interventions such as hand washing, masks, social distancing, and "social bubbles" are used to limit community transmission, but it is challenging to choose the best interventions for a given activity. Here, we provide a quantitative framework to determine which interventions are likely to have the most impact in which settings. We introduce the concept of "event R", the expected number of new infections due to the presence of a single infected individual at an event. We obtain a fundamental relationship between event R and four parameters: transmission intensity, duration of exposure, the proximity of individuals, and the degree of mixing. We use reports of small outbreaks to establish event R and transmission intensity in a range of settings. We identify principles that guide whether physical distancing, masks and other barriers to transmission, or social bubbles will be most effective. We outline how this information can be obtained and used to re-open economies with principled measures to reduce COVID-19 transmission.

Long time frames to detect the impact of changing COVID-19 control measures (preprint)

Background: Many countries have implemented population-wide interventions such as physical distancing measures, in efforts to control COVID-19. The extent and success of such measures has varied. Many jurisdictions with declines in reported COVID-19 cases are moving to relax measures, while others are continuing to intensify efforts to reduce transmission. Aim: We aim to determine the time frame between a change in COVID-19 measures at the population level and the observable impact of such a change on cases. Methods: We examine how long it takes for there to be a substantial difference between the cases that occur following a change in control measures and those that would have occurred at baseline. We then examine how long it takes to detect a difference, given delays and noise in reported cases. We use changes in population-level (e.g., distancing) control measures informed by data and estimates from British Columbia, Canada. Results: We find that the time frames are long: it takes three weeks or more before we might expect a substantial difference in cases given a change in population-level COVID-19 control, and it takes slightly longer to detect the impacts of the change. The time frames are shorter (11-15 days) for dramatic changes in control, and they are impacted by noise and delays in the testing and reporting process, with delays reaching up to 25-40 days. Conclusion: The time until a change in broad control measures has an observed impact is longer than is typically understood, and is longer than the mean incubation period (time between exposure than onset) and the often used 14 day time period. Policy makers and public health planners should consider this when assessing the impact of policy change, and efforts should be made to develop rapid, consistent real-time COVID-19 surveillance.

How much leeway is there to relax COVID-19 control measures? (preprint)

Following successful widespread non-pharmaceutical interventions aiming to control COVID-19, many jurisdictions are moving towards reopening economies and borders. Given that little immunity has developed in most populations, re-establishing higher contact rates within and between populations carries substantial risks. Using a Bayesian epidemiological model, we estimate the leeway to reopen in a range of national and regional jurisdictions that have experienced different COVID-19 epidemics. We estimate the risks associated with different levels of reopening and the likely burden of new cases due to introductions from other jurisdictions. We find widely varying leeway to reopen, high risks of exceeding past peak sizes, and high possible burdens per introduced case per week, up to hundreds in some jurisdictions. We recommend a cautious approach to reopening economies and borders, coupled with strong monitoring for changes in transmission.

Estimating the impact of COVID-19 control measures using a Bayesian model of physical distancing (preprint)

Extensive physical distancing measures are currently the primary intervention against coronavirus disease 2019 (COVID-19) worldwide. It is therefore urgent to estimate the impact such measures are having. We introduce a Bayesian epidemiological model in which a proportion of individuals are willing and able to participate in distancing measures, with the timing of these measures informed by survey data on attitudes to distancing and COVID-19.We fit our model to reported COVID-19 cases in British Columbia, Canada, using an observation model that accounts for both underestimation and the delay between symptom onset and reporting. We estimate the impact that physical distancing (also known as social distancing)has had on the contact rate and examine the projected impact of relaxing distancing measures. We find that distancing has had a strong impact, consistent with declines in reported cases and in hospitalization and intensive care unit numbers. We estimate that approximately 0.78 (0.66-0.89 90% CI) of contacts have been removed for individuals in British Columbia practising physical distancing and that this fraction is above the threshold of 0.45 at which prevalence is expected to grow. However, relaxing distancing measures beyond this threshold re-starts rapid exponential growth. Because the extent of underestimation is unknown, the data are consistent with a wide range in the prevalence of COVID-19 in the population; changes to testing criteria over time introduce additional uncertainty. Our projections indicate that intermittent distancing measures - if sufficiently strong and robustly followed - could control COVID-19 transmission, but that if distancing measures are relaxed too much, the epidemic curve would grow to high prevalence.