The effect of travel restrictions on the geographical spread of COVID-19 between large cities in China: a modelling study

Abstract

Background: To contain the spread of COVID-19, a cordon sanitaire was put in place in Wuhan prior to the Lunar New Year, on 23 January 2020. We assess the efficacy of the cordon sanitaire to delay the introduction and onset of local transmission of COVID-19 in other major cities in mainland China.

Methods: We estimated the number of infected travellers from Wuhan to other major cities in mainland China from November 2019 to February 2020 using previously estimated COVID-19 prevalence in Wuhan and publicly available mobility data. We focused on Beijing, Chongqing, Hangzhou, and Shenzhen as four representative major cities to identify the potential independent contribution of the cordon sanitaire and holiday travel. To do this, we simulated outbreaks generated by infected arrivals in these destination cities using stochastic branching processes. We also modelled the effect of the cordon sanitaire in combination with reduced transmissibility scenarios to simulate the effect of local non-pharmaceutical interventions.

Results: We find that in the four cities, given the potentially high prevalence of COVID-19 in Wuhan between December 2019 and early January 2020, local transmission may have been seeded as early as 1-8 January 2020. By the time the cordon sanitaire was imposed, infections were likely in the thousands. The cordon sanitaire alone did not substantially affect the epidemic progression in these cities, although it may have had some effect in smaller cities. Reduced transmissibility resulted in a notable decrease in the incidence of infection in the four studied cities.

Conclusions: Our results indicate that sustained transmission was likely occurring several weeks prior to the implementation of the cordon sanitaire in four major cities of mainland China and that the observed decrease in incidence was likely attributable to other non-pharmaceutical, transmission-reducing interventions.

Keywords: COVID-19; China; Cordon sanitaire; Delay; Mobility; Modelling; Outbreaks; SARS-CoV-2; Travel restrictions; Wuhan.

Fig. 1

Total domestic travel outflow from Wuhan under 4 travel pattern scenarios

Fig. 2

a Estimated median number of daily infected arrivals and b estimated cumulative number of infected arrivals from Wuhan for the four chosen cities (Beijing, Chongqing, Hangzhou, and Shenzhen, left to right) for Chunyun vs. non-Chunyun and cordon sanitaire imposed vs. no cordon sanitaire. The shaded area indicates the 95% uncertainty interval. The vertical dashed line indicates the date the cordon sanitaire was imposed

Fig. 3

Median daily incidence of COVID-19 (shaded areas indicate 50% and 95% confidence intervals) in Beijing, for Chunyun vs. non-Chunyun and cordon sanitaire imposed (red, solid) vs. no cordon sanitaire (blue, dashed) and for varying values of the effective reproduction number R_e, where R_e = 2.2 (no change, unmitigated local outbreak), reduced from 2.2 by 50% to 1.1 (mitigation of outbreak, R_e > 1), and 75% to 0.55 (suppression of outbreak, R_e < 1)

Fig. 4

Estimated date on which the probability of an outbreak exceeds 95% in the 4 cities of interest, for Chunyun vs. non-Chunyun and cordon sanitaire imposed vs. no cordon sanitaire and for varying values of the overdispersion parameter k [15, 21, 22]. Median (and 95% CI) estimated cumulative number of infections on 1 March in the four cities of interest, Chunyun vs. non-Chunyun, cordon sanitaire imposed vs. no cordon sanitaire, and for varying values of R_e, where R = 2.2 (no change, unmitigated local outbreak), reduced from 2.2 by 50% to 1.1 (mitigation of outbreak, R > 1), and 75% to 0.55 (suppression of outbreak, R < 1)