medRxiv preprint doi: https://doi.org/10.1101/2020.12.10.20246827; this version posted December 11, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license . Projecting the impact of a two-dose COVID-19 vaccination campaign in Ontario, Canada Thomas N. Vilches,1 Kevin Zhang,2,† Robert Van Exan,3 Joanne M. Langley,4 Seyed M. Moghadas5 1 Institute of Mathematics, Statistics and Scientific Computing, University of Campinas, Campinas SP, Brazil 2 Faculty of Medicine, University of Toronto, Toronto, Ontario, M5S 1A8 Canada 3 Immunization Policy & Knowledge Translation, Trent Lakes, Ontario, K0M 1A0, Canada 4 Canadian Center for Vaccinology, Dalhousie University, IWK Health Centre and Nova Scotia Health Authority, Halifax, Nova Scotia, B3K 6R8 Canada 5 Agent-Based Modelling Laboratory, York University, Toronto, Ontario, M3J 1P3 Canada †Corresponding author: [email protected] Abstract Background: Results of phase III vaccine clinical trials against COVID-19, although encouraging and well above initial expectations, have only reported on efficacy against disease and its severity. We evaluated the impact of vaccination on COVID-19 outbreak and disease outcomes in Ontario, Canada. Methods: We used an agent-based transmission model and parameterized it with COVID-19 characteristics, demographics of Ontario, and age-specific clinical outcomes derived from outbreak data. We implemented a two-dose vaccination program, prioritizing healthcare workers and high-risk individuals, with 40% vaccine coverage and vaccine efficacy of 95% against disease. Vaccines were distributed at a rate of 30 doses per day per 10,000 population with a 6- day schedule per week. We projected the impact of vaccination on attack rates, hospitalizations, and deaths. For scenario analyses, we varied the vaccine efficacy against infection, under the assumption of 5% pre-existing population immunity. Results: With no protection against infection, a two-dose vaccination campaign with a time interval of 21 days between doses reduced attack rate, hospitalizations, and deaths by 44.6% (95% CrI: 34.5% - 54.3%), 63.4% (95% CrI: 56.1% - 69.9%), and 70.0% (95% CrI: 62.6% - 75.8%), respectively. These reductions were improved with increased vaccine efficacy against infection, with similar estimated ranges in the corresponding scenarios with a 28-day time interval between vaccine doses. Conclusions: Vaccination can substantially mitigate ongoing COVID-19 outbreaks, even when vaccines offer limited protection against infection. This impact is founded upon a relatively strong vaccine efficacy against disease and severe outcomes. Keywords: COVID-19; vaccination; outbreak simulation NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. medRxiv preprint doi: https://doi.org/10.1101/2020.12.10.20246827; this version posted December 11, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license . Introduction Despite unprecedented public health measures, such as stay-at-home orders, school closures, and physical distancing (1–4), the novel coronavirus disease 2019 (COVID-19) continues to have severe global health and economic consequences (5,6). In Canada, these measures have been comparatively effective in flattening initial outbreaks (7,8). However, the devastating outcomes of the second wave and the high level of susceptibility to COVID-19 (8) have underscored the need for a safe and effective vaccine to control ongoing outbreaks. The target product profile (TPP) by the World Health Organization (WHO) provided a roadmap for potential COVID-19 vaccine candidates (9). The TPP indicated a preference for candidates that demonstrate a population-based efficacy of at least 70%, and a point estimate of 50%, against transmission and/or severe disease outcomes. Results from phase III vaccine clinical trials have been encouraging thus far, with Pfizer-BioNTech and Moderna reporting an efficacy of over 90% against symptomatic disease (10,11), exceeding the TTP target range. As of December 9, 2020, Health Canada has authorized the Pfizer-BioNTech vaccine for distribution (12), with 3 other candidates under review (13). Therefore, there is an urgent need to understand the potential population-level impact of vaccination campaigns with vaccine prioritization (14). We sought to evaluate the impact of a COVID-19 vaccination campaign, based on a scenario with two doses distributed either 21 days or 28 days apart, on attack rate and adverse clinical outcomes in Ontario. We extended an agent-based model of disease transmission (15) to include vaccination with an age-specific uptake distribution similar to that of past seasonal influenza epidemics and the 2009 H1N1 pandemic (16,17). We evaluated a roll-out strategy that prioritizes higher risk adults (i.e., healthcare workers, elderly, and comorbid individuals), followed by the general population, to minimize transmission and severe outcomes (14). Our results indicate that vaccination, even with a vaccine that offers limited protection against infection, could have a large impact on reducing hospitalizations and deaths in Ontario. Methods Model structure We extended a previously established agent-based COVID-19 transmission model (15) and included vaccination to simulate outbreak scenarios. The natural history of COVID-19 was implemented in the model by considering individual status as susceptible; latently infected (not yet infectious); asymptomatic (infected and infectious but with no symptoms); pre-symptomatic (infected, infectious and in the stage before symptomatic illness); symptomatic with either mild or severe/critical illness; recovered (and not infectious); and dead (Appendix, Figure A1). We binned the model population into five age groups of 0-4, 5-19, 20-49, 50-64, and 65+ years old based on the demographics of Ontario, Canada (18), and parameterized the model with estimates of the proportion of the population with comorbidities associated with severe COVID- 19 (Appendix Table A1) (19,20). Interactions between individuals were informed using an empirically determined contact network (21). The daily number of contacts for each individual 1 medRxiv preprint doi: https://doi.org/10.1101/2020.12.10.20246827; this version posted December 11, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license . was sampled from a negative-binomial distribution with age-dependent mean and standard deviation (Appendix, Tables A2). Disease dynamics We implemented disease transmission in a probabilistic manner whereby susceptible individuals were exposed to infectious individuals (i.e., asymptomatic, pre-symptomatic, or symptomatic stages of the disease). Infected individuals entered a latent period as part of an average incubation period of 5.2 days (22). For those who went on to develop symptomatic disease, the incubation period included a pre-symptomatic stage prior to the onset of symptoms (23). The duration of the pre-symptomatic stage was sampled from a Gamma distribution with a mean of 2.3 days (23). The infectious period post-symptom onset was sampled from a Gamma distribution with an average of 3.2 days (24). We considered an age-dependent probability of developing mild, severe, or critical illness after symptom onset. Infected individuals who did not develop symptoms remained asymptomatic after the latent period until recovery. Asymptomatic individuals were infectious for an average of 5 days, which was sampled from a Gamma distribution (24,25). Based on the number of secondary cases generated during each stage of the disease (26), we parameterized the infectivity of asymptomatic, mild symptomatic, and severe symptomatic stages to be 11%, 44%, and 89% relative to the pre-symptomatic stage (27). We assumed that recovered individuals could not be re-infected during the same outbreak scenario. Infection outcomes In our model, mild symptomatic cases recovered without the need for hospitalization. Persons with severe illness or illness requiring critical care used hospital beds in this model. We parameterized the model for the use of intensive care unit (ICU) and non-ICU beds based on recent COVID-19 hospitalization data stratified by the presence of comorbidities in Ontario (20). We assumed that all symptomatic cases who were not hospitalized self-isolated during the symptomatic period. For self-isolated cases, daily contacts were sampled from an age- dependent contact matrix derived from a representative sample population during COVID-19 lockdown (28). The time from symptom onset to hospital admission was uniformly sampled in the range of 2 to 5 days (15,29). The lengths of non-ICU and ICU stays were sampled from Gamma distributions with means of 12.4 and 14.4 days, respectively (30,31). Vaccination We implemented a two-dose vaccination campaign, achieving 40% vaccine coverage of the population over the course of the campaign. Vaccine distribution mirrored that of the age- specific vaccine coverage estimates for seasonal and 2009 pandemic influenza (16). We reviewed vaccine program distribution rates for influenza to determine the roll-out capacity (17,32–34). With the expected shortage of vaccines in the initial roll-out, we assumed that 30 individuals per 10,000 population are vaccinated per day in Ontario. A 40% vaccine coverage was achieved within 40 weeks with a 6-day immunization schedule per week. Vaccination was sequential with prioritization of: (i) healthcare workers, individuals with comorbidities, and those aged 65 and older (i.e., protection cohort); followed by (ii) individuals 2 medRxiv preprint doi: https://doi.org/10.1101/2020.12.10.20246827; this version posted December 11, 2020.