Quarantine and the risk of COVID-19 importation

Abstract

Using a stochastic model, we assess the risk of importation-induced local transmission chains in locations seeing few or no local transmissions and evaluate the role of quarantine in the mitigation of this risk. We find that the rate of importations plays a critical role in determining the risk that case importations lead to local transmission chains, more so than local transmission characteristics, i.e. strength of social distancing measures (NPI). The latter influences the severity of the outbreaks when they do take place. Quarantine after arrival in a location is an efficacious way to reduce the rate of importations. Locations that see no or low-level local transmission should ensure that the rate of importations remains low. A high level of compliance with post-arrival quarantine followed by testing achieves this objective with less of an impact than travel restrictions or bans.

Keywords: COVID-19; importation risk; mathematical modelling; quarantine.

Fig. 1.

Percentage of provinces, territories and health regions in Canada having declared new confirmed cases of COVID-19 in 2020 during a three-week period prior to dates shown. Time runs from three weeks after the first case in Canada on 25 January 2020 to 30 November 2020.

Fig. 2.

Spatialisation process: transport of cases from other locations; importation into the location of interest; local amplification; exportation to other locations. Red indicates control methods. See ES.1.1 for details.

Fig. 3.

Input−output setting: imports arrive through transport, are potentially funnelled through quarantine, then enter one of the infected compartments. The three types of outcomes considered are shown on the right. In the epidemic model, compartment S represents susceptible individuals, L₁ and L₂ are latent individuals, I₁ and I₂ are detected infectious individuals, A₁ and A₂ are undetected infectious individuals, D are deaths from detected infections or posthumously tied to the infection; finally R_I and R_A are recovered from detected and undetected infections, respectively.

Fig. 4.

Proportions of critical (dotted red) and non-critical successful (solid blue) importations as a function of the initial number of imported L₁ cases. Circles:, squares: 1.2 and triangles: 2.5.

Fig. 5.

Proportion of successful importations (SI, grey scale) and attack rates (blue scale) in a three-month period for different values of the average number of days 1/λ between two importations and , an indicator of NPI efficacy.

Fig. 6.

Evolution of the status of import cases during a (left) one- or (right) two-week quarantine period imposed upon arrival. Dark grey flows are individuals who are still a risk to the jurisdiction at the end of the quarantine period. Here, simulations were run for 2500 individuals of each type of unobservable cases L₁, L₂, A₁ and A₂ entering quarantine.

Fig. 7.

Probability c that the quarantine is successful as a function of its duration t_q. From dark to light grey: fraction π of undetected cases varying from 0.1 to 0.9 by steps of 0.1. Vertical bars show the most commonly used quarantine durations.