Mathematical Modelling of COVID-19 Transmission and Mitigation Strategies in the Population of Ontario, Canada
Total Page:16
File Type:pdf, Size:1020Kb
Early release, published at www.cmaj.ca on April 8, 2020. Subject to revision. RESEARCH HEALTH SERVICES Mathematical modelling of COVID-19 transmission and mitigation strategies in the population of Ontario, Canada Ashleigh R. Tuite PhD MPH, David N. Fisman MD MPH, Amy L. Greer PhD MSc n Cite as: CMAJ 2020. doi: 10.1503/cmaj.200476; early-released April 8, 2020 See related article at www.cmaj.ca/lookup/doi/10.1503/cmaj.200606 ABSTRACT BACKGROUND: Physical-distancing ing and less restrictive physical dis- height of the epidemic peak relative to interventions are being used in Canada tancing. Interventions were either the base case, with restrictive physical to slow the spread of severe acute implemented for fixed durations or distancing estimated to have the great- respiratory syndrome coronavirus 2, dynamically cycled on and off, based est effect. Longer duration interventions but it is not clear how effective they on projected occupancy of intensive were more effective. Dynamic interven- will be. We evaluated how different care unit (ICU) beds. We present medi- tions were projected to reduce the pro- nonpharmaceutical interventions could ans and credible intervals from 100 portion of the population infected at the be used to control the coronavirus dis- replicates per scenario using a 2-year end of the 2-year period and could ease 2019 (COVID-19) pandemic and time horizon. reduce the median number of cases in reduce the burden on the health care ICU below current estimates of Ontario’s system. RESULTS: We estimated that 56% (95% ICU capacity. credible interval 42%–63%) of the METHODS: We used an age-structured Ontario population would be infected INTERPRETATION: Without substantial compartmental model of COVID-19 over the course of the epidemic in the physical distancing or a combination of transmission in the population of base case. At the epidemic peak, we moderate physical distancing with Ontario, Canada. We compared a base projected 107 000 (95% credible interval enhanced case finding, we project that case with limited testing, isolation and 60 760–149 000) cases in hospital (non- ICU resources would be overwhelmed. quarantine to scenarios with the fol- ICU) and 55 500 (95% credible interval Dynamic physical distancing could lowing: enhanced case finding, restric- 32 700–75 200) cases in ICU. For fixed- maintain health-system capacity and tive physical-distancing measures, or duration scenarios, all interventions also allow periodic psychological and a combination of enhanced case find- were projected to delay and reduce the economic respite for populations. he coronavirus disease 2019 (COVID-19) pandemic repre- acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is highly sents a global public health emergency unparalleled in transmissible.4–7 It causes moderate to severe clinical outcomes recent time. In the 2 months since the initial World in about 20% of all recognized infected individuals.5,8,9 In the THealth Organization report describing the COVID-19 outbreak absence of a vaccine, public health responses have focused on concentrated in Wuhan, China,1 the number of confirmed cases the use of nonpharmaceutical interventions.10 These nonphar- has risen sharply from 282 to more than 330 000, with 14 510 maceutical interventions include “case-based” measures such as reported deaths across all regions of the globe.2 The first testing, contact tracing, isolation (of infected cases) and quaran- imported case of COVID-19 in Ontario, Canada, was reported on tine (of exposed cases); and “non-case-based” measures such as Jan. 25, 2020, and community transmission was first docu- reducing the probability of transmission given an effective con- mented on Mar. 1, 2020, in British Columbia, Canada.3 tact (e.g., hand hygiene and cough etiquette) and physical- This pathogen represents a substantial challenge for public distancing measures to reduce the contact rate in the popula- health, pandemic planning and health care systems. Severe tion. Physical distancing minimizes opportunities for © 2020 Joule Inc. or its licensors CMAJ 1 person-to-person transmission of the virus to occur. These sion within health care settings. For simplicity, we assumed that physical-distancing measures include some combination of all deaths occurred in cases requiring intensive care. We included school closure, teleworking, cancellation of group activities and cases in hospital (non-ICU) and requiring intensive care to esti- events, and a general overall reduction in community contacts. mate health care requirements over the course of the epidemic. Although these measures are expected to be effective in reducing The model was constructed in R.13 transmission of SARS-CoV-2, they are also associated with sub- stantial economic costs and societal disruption. Model parameters RESEARCH Epidemiologic models can contribute important insight for The model was stratified by 5-year age groups using 2019 popu- public health decision-makers by allowing for the examination lation estimates.14 Contacts within and between age groups were of a variety of “what-if” scenarios. The Canadian Pandemic based on the POLYMOD study,15 using contact data specific for Influenza Plan for the Health Sector (the backbone of which the United Kingdom. The model was further stratified by health informs COVID-19 pandemic preparedness and response) iden- status to account for differential vulnerability to severe infection tifies 2 main objectives for responding to a pandemic: to min- among those with underlying health conditions. We obtained imize serious morbidity and mortality, and to minimize societal comorbidity estimates by age from the Canadian Community disruption.11 The overarching goal of pandemic response is to Health Survey (CCHS)16 for Ontario and included the following find a combination of nonpharmaceutical interventions that conditions: hypertension, heart disease, asthma, stroke, diabe- would minimize the number of cases requiring in-patient med- tes and cancer. For younger age groups (< 12 yr), we used esti- ical care (e.g., hospital and intensive care unit [ICU] admis- mates from Moran and colleagues.17 A limitation of the CCHS is sions) and deaths, while also minimizing the level of societal that it may undersample individuals from socioeconomically dis- disruption. Societal disruption could be reduced by limiting the advantaged populations. overall duration that the intervention needs to be in force to Parameters describing the natural history and clinical course achieve the associated reductions in morbidity and mortality. A of infection were derived from published studies (Table 1, full challenge for pandemic response is that, in a fully susceptible details in Appendix 1). The rate of growth of epidemics is gov- population, although nonpharmaceutical interventions may erned by reproduction numbers, or the number of secondary slow disease transmission while they are in place, once the infections caused by a primary infectious case. For a pandemic intervention is lifted (or compliance with the intervention disease, in which prior immunity is absent, the operative repro- becomes low), the transmission of the pathogen rebounds duction number is referred to as the basic reproduction number 10,12 23 rapid ly. In the case of COVID-19, it may not be possible to (R0). To capture variability in transmission, specifically the minimize morbidity and mortality, and societal and economic observation that the basic reproduction number for COVID-19 is disruption at the same time. overdispersed, with some cases transmitting to many others Given these considerations, we used a transmission dynamic (superspreader events), while many other cases transmit much model of COVID-19 to explore the potential impact of case-based less, we have added volatility to the transmission term.24–26 This and non-case-based nonpharmaceutical interventions in the causes each model run to have a different outcome owing to sto- population of Ontario, Canada. Our analysis focuses on identify- chasticity (i.e., random variation between model runs). The ing strategies that keep the number of projected severe cases model was initiated with 750 prevalent cases (based on 150 (hospital and ICU admissions) within a range that would not reported cases in Ontario on Mar. 19, 2020, and an assumed overwhelm the Ontario health care system, while also consider- reporting rate of 20%), that were randomly distributed across the ing the amount of time these interventions would be in place. infectious compartments. Methods Interventions Testing was assumed to move individuals with nonsevere symp- Model overview toms from the infectious to isolated compartments. Isolated We developed an age-structured compartmental model that cases were assumed to have reduced transmission compared describes COVID-19 transmission in the province of Ontario, Can- with nonisolated cases. Physical-distancing measures were ada. We used a modified “susceptible-exposed-infectious- assumed to reduce the number of contacts per day across the recovered” framework that incorporated additional compart- entire population. Details of parameters that were varied under ments to account for public health interventions, different different interventions are included in Table 2. For the base case, severities of clinical symptoms and risk of hospital admission. An we assumed that there was a degree of testing and isolation overview of the model compartments and movements between occurring and that a proportion of exposed cases were quaran- them is