Does lock-down of Wuhan effectively restrict early geographic spread of novel coronavirus epidemic during chunyun in China? A spatial model study

Since early December 2019, an increasing number of atypical pneumonia have been reported in Wuhan [1], a city with a population of 11 million in the central part of China. In response to this outbreak, the Chinese Center for Disease Control and Prevention (China CDC) conducted an epidemiologic and etiologic investigation on December 31, 2019 [2], and found that human-to-human transmission occurred since the middle of December 2019 [3], and isolated and confirmed a novel strain of coronavirus on 7 January 2020 [4]. Along with the increasing number of COVID-19 cases in China, the geographic spread at a meta-population level (e.g. between cities in China, Asia-pacific regions, or northern hemisphere countries) has been reported [5].

When human-to-human transmission demonstrated a population health threat, use of restrictive measures such as isolation of cases and quarantine of contacts became one of the apparent emergency responses to the COVID-19 outbreak [6]. However, a substantial challenge has emerged when chunyun, the largest human migration on the planet, beginning on Jan 10, 2020 with billions of trips made for family reunions to celebrate the Spring Festival during the national holidays from January 24 to 31 [7]. Geographic spread of COVID-19 would have potentially be accelerated by chunyun under this circumstance, and therefore prevention and control of the COVID-19 infiltration into local communities across the nation would require immediate actions to restrict human movements. On January 23 and 24, 2020, the Chinese central government implemented the metropolis-wide lock-down of Wuhan and its surrounding satellite cities [3]. In addition to the Wuhan lock-down, the central government announced the extension of the national holidays, the Spring Festival, and set the back to work date as of February 10, 2020 (except Wuhan) [8].

In the face of this unprecedented threat and lack of effective countermeasures, the authorities invoked the lock-down of Wuhan as a means to interrupt the geographic spread of COVID-19 and expected to achieve the disease control goals [9]. Although several studies investigated the impact of city lockdown on the disease spread [10‐12], population based evidence is inadequate regarding their individual roles of city lockdown and chunyun as well as their aggregated role on the spread of COVID-19 in real world settings. This study aimed to evaluate the effectiveness of Wuhan lock-down for preventing the spread of COVID-19 at an early stage, and examine whether the effectiveness would vary according to the presence or absence of chunyun and lock-down.

During the period of January 1 to February 9, 2020, a total of 40,278 confirmed COVID-19 cases from 319 municipalities in mainland China were reported (Fig. 1). Across China, the population mobility had been increasing since the start of chunyun and reached the greatest on January 21 and then decreased afterward (Fig. 2). While the patterns of population inflow and outflow in China were similar, these patterns for population inflow to and outflow from Wuhan presented in different ways. The population outflow from Wuhan showed a generally increasing trend up to January 22 whereas the population inflow to Wuhan remained almost constant. It is noteworthy for Wuhan that the population outflow was always greater than the population inflow until January 26 (shortly after the announcement of lock-down). Change of population outflow from Wuhan was also noteworthy for the sharp increase after announcement of human-to-human transmission of COVID-19 on January 20.

Figures 3a and b illustrates a reasonably good model fit and the time-varying estimates of basic reproductive number (R0_t) of COVID-19 during the study period. Although we assumed R0_t varies over time, it stayed at the same level from January 1 to 25 and fell from 3.47 to 3.24 from January 26 onwards. Note that it slightly increased to 3.27 on February 9. The modelled latent and infectious time of COVID-19 was 6.11 days (95%CI: 3.13, 10.63) and 3.26 days (95%CI: 1.06,5.16), respectively.

Table 1 displays cumulative number of COVID-19 cases in China during the study period under different modelling scenarios. Under Scenario 1, the COVID-19 epidemic would have resulted in 3.84% less cases than the baseline by February 9, indicating that chunyun facilitated the spread of this infectious disease. Compared with the baseline scenario, scenario 2 would have produced 32.46% more COVID-19 cases, demonstrating the protective effectiveness of Wuhan lock-down. Under Scenario 3, the COVID-19 epidemic would have resulted in 20.22% more cases than the baseline in the absence of chunyun and lock-down.

Figure 4 demonstrates the geographic distribution of change in cumulative COVID-19 cases comparing different scenarios with the baseline scenario (the presence of both chunyun and lock-down) by February 9, 2020. Under scenario 1 (Fig. 4a), the majority of cities showed a relatively sharp change in case reduction despite the nuance expression in a few populous cities, indicating that chunyun was not a common stimulus for the COVID-19 spread across the nation. Under scenario 2 (Fig. 4b), all the cities would have had greater number of cases in the absence of lock-down, in particular, those in northeast, south and west China would have an increase over 100%, indicating the protective effect of Wuhan lock-down on preventing additional disease penetration towards all the other cities in China. Under scenario 3 (Fig. 4c), the protective effect of Wuhan lock-down varied in space and was offset by the presence of chunyun, especially for those corridor cities near Wuhan. Note that areas with over 100% increase of cases (in dark red color) under this scenario were mainly located within the five city groups shown in Fig. 4d.

In this retrospective analysis of 40,278 confirmed COVID-19 cases in China, we modelled 3 exposure scenarios using publically available data reported on a daily basis by local public health authorities. These scenarios differed in the exposure to chunyun, the largest population mobility on the earth, and Wuhan lock-down, the unprecedented control of 11 million people’s movement in response to the rapid spread of COVID-19 from the city. Of the simulations of three exposure scenarios, the lock-down of Wuhan remarkably demonstrated the protective effects by preventing 32.46% COVID-19 incidents by February 9, 2020, whereas chunyun contributed towards the observed geographic spread and would have produced 3.84% more cases by the same period. Although the impact of the presence of both chunyun and lock-down of Wuhan on the COVID-19 spread was heterogeneous in space, the majority of cities had been protected with risk of a subsequent outbreak mitigated by the lock-down in spite of the chunyun offset. In addition to previous findings allowing for the effects of fleeing population [11], flight restriction [12], and limited medical resources [10], we further demonstrated the individual role for chunyun and its combined role with the lock-down of Wuhan in comparison with the absence of both events. These findings complemented the growing evidence of impact of population mobility on the geographic spread of COVID-19, justified the invocation of Wuhan lock-down instead of simply claiming the precautionary principle, and provided additional vigilance against inter-city human movement at the critical early phase for next-stage informed decision making with respect to COVID-19 prevention and control in the communities.

In this study, we also investigated the epidemiological characteristics of COVID-19 outbreak in China. We first estimated its instantaneous R0_t, and found a decrease from 3.47 to 3.24 on January 26, which was probably due to the observed decrease in population mobility (Fig. 2). In addition, our estimate of latent and infectious periods were 6.11 and 3.26 days, respectively, with the former being comparable with the existing clinical studies [17, 18] and the latter, being rarely reported. Compared with the previously reported basic reproductive number (R0) which were fixed and ranged from 2.24 to 3.80 [3, 18‐20], our estimate was derived from a time varying model, which was somewhat more reliable as we have taken into account a larger sample size as well as population mobility between cities that would potentially affect both the contact rates between individuals and subsequently the estimation of COVID-19 spread. Notwithstanding the decrease of R0_t, there was an increasing trend on February 9 which corresponded to the time when migrants returned to study or work. Considering the contribution of chunyun towards geographic spread of COVID-19, this rebound trend of R0_t implied that the epidemic would perhaps spread more rapidly and move into the next stage of becoming endemic. Therefore, continuing efforts with rigorous prevention and control measures are required and expected to be strictly implemented in the communities including social distancing and home quarantine.

We estimated the COVID-19 occurrence under three scenarios and their respective changes in relation to the baseline incidents for each city. We found that there was evidently spatial heterogeneity in effects of chunyun and/or Wuhan lock-down. In principle, chunyun contributed towards the COVID-19 spread in China (Table 1) but not specifically for each locality (Fig. 4a), while the Wuhan lock-down restricted the subsequent spread towards the other cities across the nation (Table 1, Fig. 4b). In the absence of both chunyun and lock-down (Fig. 4c), more evident reduction in COVID-19 incidents would have occurred for five major urban agglomerations. These urban agglomerations (Fig. 4d) consist of two kinds of city clusters, i.e., well-developed megacities, and the others nested in the undeveloped regions of China. There is an obvious speculation that these clusters would experience similar outgoing and ingoing travel demands all year round, and therefore the effects of chunyun would be nuance. It is noteworthy that corridor cities near Wuhan (Fig. 4c) did not benefit much from the Wuhan lock-down as its protective effect was offset by chunyun. These corridor cities would perhaps become a case reservoir prior to the lock-down of Wuhan. Therefore, these corridor cities should be given a priority for continuing efforts in allocation of healthcare resources. Additional strategies should be developed based on their area-level characteristics.

Evidence of comparative effectiveness of large-scale lock-down on 11-million populations is rare because the outbreak of contagious disease of this kind is highly unusual and so is to have reliable national data. Using a larger sample size than previous studies and clear evidence of exposure, our findings are somewhat robust. However, there are some limitations. First, the use of deterministic SEIR model would be acceptable given the successful control of COVID-19 in China [21]. Yet, our analysis did not consider the under-reporting of cases during the study period, due in part to the pre-symptomatic transmission and asymptomatic individuals are neglected [22], which may result in an underestimation bias of our results. Iterated filtering algorithm such as Maximum Likelihood Estimation or Bayesian approach could be another option [23], especially in other settings with much more uncertainties than that of China. Second, the mathematical model is a deterministic model which cannot capture stochastic effects during the spread of COVID-19 under the influence of multiple socio-environmental factors. This may bias the number of incident cases of a modelled scenario. Third, we considered municipal cities as the spatial components, however, these publically available data were sparse at the city level for modelling a city-specific reproduction number, and therefore we assumed it the same across all cities. Although that the reproduction number for the entire country we calculated in this study is not substantially different from the others using a general SEIR model, it is possible that the transmissibility of COVID-19 might change from place to place. The basic reproduction number in the current setting could be underestimated, especially when the restricted travels of all SEIR individuals would underestimate the number of active cases in each region [24]. Considering that fewer people travel making each region behaving as epidemiologically isolated region in the presence of lockdown, future studies would focus on the identification of hot-spot cities at risks of becoming endemic allowing for geographical variation in reproduction number estimation. Fourth, the three simulated scenarios were based on different restrictive measures of population mobility, lacking control of other confounding factors, and therefore results should be interpreted with caution.