key: cord-0913910-acp19wvf authors: Xue, Ling; Jing, Shuanglin; Sun, Wei; Liu, Maoxing; Peng, Zhihang; Zhu, Huaiping title: Evaluating the impact of the travel ban within mainland China on the epidemic of the COVID-19 date: 2021-04-07 journal: Int J Infect Dis DOI: 10.1016/j.ijid.2021.03.088 sha: 77baddd5331461247a1de81dc81e8c810339a606 doc_id: 913910 cord_uid: acp19wvf OBJECTIVES: The ongoing COVID-19 pandemic expanded its geographic distribution through the movement of humans and caused subsequent local outbreaks. Hence, it is essential to investigate how human mobility and travel ban affect the transmission and spatial spread while minimizing the impact on social activities and national economics. METHODS: We developed a mobility network model for spatial epidemics, explicitly taking into account time-varying inter-province and inner-province population flows, spatial heterogeneity in terms of disease transmission, as well as the impact of media reports. The model is applied to study the epidemic of the dynamic network of [Formula: see text] provinces of mainland China. The model was calibrated using the publicly available incidence and movement data. RESULTS: We estimated that the second outbreak occurred approximately on February 24, 2020, and the cumulative number of cases as of March 15, 2020, increased by 290.1% (95% CI: (255.3%, 324.9%)) without a travel ban in mainland China (excluding Hubei and Tibet). We found that inter-province travel contributes more to the increase of cumulative number of cases than intra-province travel. CONCLUSION: Our quantitative and qualitative research results suggest that the strict travel ban has successfully prevented a severe secondary outbreak in mainland China, which provides solutions for many countries and regions experiencing secondary outbreaks of COVID-19. The global outbreak caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has been declared a pandemic by the World Health Organization (WHO). The number of imported SARS-CoV-2 cases in many countries has been rising due to large-scale geographic spread (Candido et al., 2020 . The ongoing epidemic of COVID-19 presents significant challenges to the development of the economy. As of November 15, 2020, the COVID-19 epidemic has caused 53.7 million confirmed cases and 1.3 million deaths worldwide (2020f). However, by the beginning of March 2020, the COVID-19 epidemic has already been under control in mainland China (Xu and Li, 2020) , and most provinces have gradually resumed work and production. The Spring Festival was coming when the COVID-19 outbreak became a public health concern in China. To prevent the further spatial spread of COVID-19 caused by the massive population movement during the most important traditional holiday period, Wuhan of Hubei province started lockdown on January 23, 2020 (Tian et al., 2020) . Soon afterwards, other provinces also implemented travel bans between provinces, suspended local public transportation such as buses and subway, closed schools and entertainment places, banned public gatherings, encouraged the public to wash their hands, wear masks, and suggested other personal preventive actions (Lai et al., 2020) . In addition to the travel ban implemented in each province, Chinese public health authorities have also reinforced screening, identification, diagnosis, timely reporting and contact tracing of suspected or confirmed cases, isolation of infected individuals, quarantine of exposed individuals to minimize the risk of SARS-CoV-2 infection . China has allocated medical resources among more than 30 provinces, municipalities, and districts to optimize limited healthcare facilities and achieved centralized treatment of severely affected areas through unified adjustment, deployment, and mobilization (2020e). Implementing these non-pharmaceutical interventions (NPIs) has led to a rapid decline in daily new cases across the country. Therefore, it is crucial to understand how China successfully curbed the spread of the COVID-19 epidemic, which will provide valuable experiences and guidance for other countries suffering from the epidemic and prevent the second wave of the COVID-19 outbreak in China. Many previous mathematical models have studied the impact of non-pharmaceutical interventions (NPIs) on the spread of COVID-19 (Candido et al., 2020 , Chinazzi et al., 2020 , Gatto et al., 2020 , Jia et al., 2020 , Lai et al., 2020 , Sun et al., 2020 , Tian et al., 2020 , Wells et al., 2020 . As this highly infectious disease rapidly spreads, media reports have been active in community education on COVID-19 by disseminating useful information on pathogenesis, prevention, and mitigation (Liu Qian et al., 2020) . To the best of our knowledge, very few works has simultaneously taken into account the delay and strength of media reports, variation of isolation rates, the infection caused by asymptomatic infection, as well as inter-province and inner-province travel bans. To evaluate the impact of media reports, variation of isolation rates, inter-province, and inner-province travel bans on the epidemics in mainland China, we develop a network model consisting of 30 provinces of the country. We consider the heterogeneity among provinces in terms of population size, the timing and strength of initiating NPIs, and the mobility rate of humans. We fit the reported COVID-19 data to the models and compute the effective reproduction number for every 30 provinces. We conducted the comparative analysis to quantify the impact of travel restriction, media reports, and detection and isolation of cases on the dynamics of the COVID-19 epidemic using the Baidu travel data and Baidu Index data to project the corresponding effective containment strategy for each province. We also assessed the risk of postponing the travel ban on the spread of COVID-19. and Health Commission (HC) in each province reported detailed information on the patients infected by COVID-19, including gender, age, the timing of symptom onset, and hospitalization. We focused on the number of daily new cases. Public health authorities are increasingly advocating the use of virtual social networks to spread information about infectious diseases. Although it is difficult to measure the effectiveness of such strategies, many researchers have found that disease information has dramatically changed the awareness and behavior of online users and public (Vance et al., 2009) . The awareness of the public and the prevention of diseases can be reflected by the active acquisition of relevant information. The Baidu Search Index is provided by Baidu, which represents the levels of Internet users' attention to keyword searches that vary with time (Liu Xiaofan et al., 2020) . To characterize the impact of media reports on the COVID-19 outbreak, we collected the Baidu Search Index (2020b) shown in Figure. Large-scale population movements may cause the further spread of the epidemic (Meloni et al., 2011) . We collected real-time travel data using the Baidu Migration server (2020a) from January 10, 2020, to March 15, 2020, which provides real-time travel patterns in China based on mobile-phone positioning services. These data include the daily migration ratio and Baidu Migration Index between provinces, as shown in Figure. A2 of Appendix A. The explicit relationship between the population size and the Baidu Migration Index was not explained explicitly. Sanche et al. (2020) estimated that one unit of the Baidu Migration Index corresponds to approximately 44520 migrants, and we estimated the travel flow between the provinces of mainland China using this approximation. To study the relationship between media reports and the COVID-19 epidemic, we used the cross-correlation analysis (Wei, 2006) to reveal the statistically significant correlation between the Baidu Search Index and the number of daily new cases from January 10, 2020, to March 15, 2020. Cross-correlation is a standard method of estimating the degree to which two series are correlated. We used the collected data to reveal the range of cross-correlation coefficients and determine the specific value of the time lag when the cross-correlation coefficient reaches its maximum. Search Index in each province, as shown in Figure. A3 of Appendix A. The results show a statistically significant correlation between the number of daily new cases and the Baidu Search Index in each province. Figure. A3 shows that the local maximum cross-correlation coefficient occurs at lag=1 in Tianjin, Liaoning, Hainan, and Ningxia. For Beijing, Inner Mongolia, Gansu, and Xinjiang, the local maximum cross-correlation coefficient occurs at lag=0. The local maximum cross-correlation coefficient occurs at lag<0 in other provinces. Population flow from Hubei can be hypothesized to export the virus to other provinces, where local outbreaks are produced thereafter (Jia et al., 2020) . We found that there is a highly positive correlation between the cumulative number of cases in each province (excluding Hubei) as of March 15, 2020, and the transition probability from Hubei province (average of daily migration rates from Hubei to each province from January 10 to January 23, 2020) as shown in Figure. 1(A). Meanwhile, we also found that there is a moderately negative correlation between the dates of first documented arrivals of infected persons in 26 provinces (excluding Hubei, Shaanxi, Jiangxi, and Xinjiang) and transition probability from Hubei province (average of daily migration rates from Hubei to each province from January 10 to January 23, as shown in Figure. 1(B) ). The total population of each province is divided into six classes, namely, ( ), ( ) and ( ) representing the number of susceptible, exposed, asymptomatic infected, symptomatic infected, hospitalized, and recovered individuals in the n-th province. The total population in the n-th province at time is denoted by ( ). The model describing the dynamics of the epidemic in the n-th province is as follows (1) The detailed description of Model (1) is in Appendix B. Using Model (1), we quantified the impact of non-pharmacological interventions on the dynamics of the COVID-19 epidemic using the Baidu travel data, Baidu Index data, and the isolation rate to project the corresponding effective containment strategy for each province, as well as the possibility of further spread after lifting the travel ban and the delayed travel ban. We simulate the epidemic for each province from January 10, 2020, taking into account the underreported cases at the beginning of the epidemic in Hubei. Since daily numbers of new cases were not reported in Hubei Province from December 1, 2019, to January 18, 2020, we assume that the initially daily number of newly infected individuals is 0 in simulations. Detailed parameter estimation and fitting results are in Appendix B. We quantified the spread of COVID-19 without any travel ban. Hubei), respectively (see Figure. 2). We quantified how the transmission of COVID-19 varied with the inter-province travel ban, as well as the possibility of further spread of COVID-19 in each province after the inter-province travel ban was lifted on January 23, 2020 (see Figure. C3 and Figure. C4 of Appendix C). We simulated the scenarios when the inter-province travel ban is not implemented as is shown in Figure. (excluding Hubei and Tibet). The impact of the inter-province travel ban on the peak size, peak time, and cumulative cases of each province is summarized in Table C1 of Appendix C. We The peak size of the first wave increased by 28.76% in Hubei. The trough appeared on March 9. As of March 15, the cumulative number of cases increased by 40.30% in Hubei (see Figure. C8 of Appendix C). The timing of the travel ban in the early stages of the COVID-19 outbreak was critical, we quantified the possibility of the spread of COVID-19 in each province for the period after a travel ban was implemented on January 27, 2020, and January 30, 2020, respectively. Figure. C9 of Appendix C and Table 2 showed the spreading of the epidemic when the travel ban was delayed. days, respectively, also by March 15, 2020, the cumulative number of cases increased by 23.12% and 24.18%, respectively. Previous work revealed that media coverage significantly delayed and reduced the peak and final size of the epidemics (Cui et al., 2008) . We used Model (1) to quantify the spread of COVID-19 with media reports. We further confirmed that the intensity of different media reports plays an important role in the spread of COVID-19 (see Figure. C10, and Figure. C11 of Appendix C, Table 2 ). Due to the relaxation of the awareness, the behavioral parameter is reduced by 40%, the peak occurs about one day later, the peak size increases by 32.59%, and the final epidemic size increases by 57.78% in mainland China (excluding Hubei and Tibet) as of March 15, 2020. Due to the improvement of self-protection awareness, the behavioral parameter is increasing by 40%, the peak occurs about one day earlier, the peak size is reduced by 14.43%, and the final epidemic size is reduced by 18.14% in mainland China (excluding Hubei and Tibet) as of March 15, 2020. Due to the relaxation of the awareness, the behavioral parameter is reduced by 40%, the peak occurs about half a day later, the peak size increases by around 37.86%, and the final epidemic size increases by 58.18% in Hubei as of March 15, 2020. Due to the improvement of self-protection awareness, the behavioral parameter increasing by 40%, the peak occurs about one day earlier, the peak size is reduced by around 20.86%, and the final epidemic size is reduced by 22.31% in Hubei as of March 15, 2020 (see Figure. C10 of Appendix C and Table 2 ). The initial isolation rate reflects how soon the NPIs are implemented after infected cases are detected. We quantified the impact of different initial isolation rates on the dynamics of COVID-19 transmission. When the initial isolation rate 0 is increased twice by expanding test capacity, the peak occurs about two days earlier, and the final epidemic size is reduced by around 8.96% in mainland China (excluding Hubei and Tibet) as of March 15, 2020. When the initial isolation rate 0 is increased by four times by expanding test capacity, the peak occurs about four days earlier, and the final epidemic size is reduced by around 19.71% in mainland China (excluding Hubei and Tibet) as of March 15, 2020. When the initial isolation rate 0 is increased twice by the strengthening of testing capability, the peak occurs about two days earlier, and the final epidemic size is reduced by around 8.17% in Hubei as of March 15, 2020. When the initial isolation rate 0 is increased four times by strengthening the testing capability, the peak occurs about four days earlier, and the final epidemic size is reduced by around 18.95% in Hubei as of March 15, 2020 (see Figure. C12 of Appendix C and Table 3 ). Highly virulent emergent pathogens are more likely to spread with the progressing global urbanization and growing connectivity among metropolitan centers (Brockmann et al., 2006 , Moore et al., 2003 , Vespignani, 2009 ). Application of Georeferenced epidemiological data and modeling tools supply insights that inform decision-making for emergency management (Gatto et al., 2020) . To understand how human mobility between and within provinces, the other factors include media reports, and the initial isolation rate quarantine influence the dynamics of the spread of SARS-CoV-2, we developed a meta-population disease transmission model, where the human mobility is characterized by a dynamic directed network. The model takes into account the heterogeneity between different provinces, such as the timing of implementing NPIs, the hospitalization rate due to the availability of health care resources, and population behavior J o u r n a l P r e -p r o o f changes. We only analyzed the epidemic in mainland China for the period from January 10 to March 15, 2020, since the epidemic was almost dying out throughout the whole country. Although scattered outbreaks occurred in Heilongjiang province due to imported cases, and outbreaks in Beijing City and Dalian City caused by seafood, the scales of these outbreaks were small and thus were soon contained by mass screening, contact tracing, and quarantine within the regions. Our model has incorporated major key factors responsible for the temporal and spatial spread of COVID-19, such as media reports publicity, travel ban between and within provinces, and the rate at which infected individuals are isolated at the beginning of an outbreak within a province. People was quarantined when travel between provinces, but the quarantine policy did not apply for people traveling within a province. This may be the reason that the intra-province travel ban is more crucial in containing an epidemic which is consistent with that in available literature (Lai et al., 2020) . We found that enhancing media reports and increasing isolation rates could significantly reduce the final size of the epidemics, although the impact varies for different locations. Our findings provide insights that help support the planning of emergency management and decision-making for public health authorities. However, our work still has limitations. The efficacy of disease control relies on the availability of data needed for modeling and quantifying human mobility (Tizzoni et al., 2014) . We did not consider the imported cases from overseas and the role of environmental factors and their variations in different provinces in the transmission of COVID-19, which we keep as future work when such datas become available. The authors declare that they have no conflicts of interest. The final size refers to the cumulative number of cases as of March 15, 2020. The final size refers to the cumulative number of cases as of March 15, 2020. Baidu Search Index Chinese Center for Disease Control and Prevention National Health Commission of the People's Republic of China National school of development World Health Organization; 2020f The scaling laws of human travel Routes for COVID-19 importation in Brazil Early containment strategies and core measures for prevention and control of novel coronavirus pneumonia in China The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak The impact of media on the control of infectious diseases Spread and dynamics of the COVID-19 epidemic in Italy: Effects of emergency containment measures Population flow drives spatio-temporal distribution of COVID-19 in China The effect of human mobility and control measures on the COVID-19 epidemic in China Effect of non-pharmaceutical interventions to contain COVID-19 in China Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV-2) Health communication through news media during the early stage of the COVID-19 outbreak in China: A digital topic modeling approach Effect of Media Intervention in Epidemics: An Analysis Based on Epidemiology Models and Case Studies in the COVID-19 Pandemic Modeling human mobility responses to the large-scale spreading of infectious diseases Global urbanization and impact on health High contagiousness and rapid severe acute respiratory syndrome coronavirus 2 Multi-stage and multi-scale patch model and the case study of novel coronavirus When to lift the lockdown in Hubei province during COVID-19 epidemic? An insight from a patch model and multiple source data An investigation of transmission control measures during the first 50 days of the COVID-19 epidemic in China On the use of human mobility proxies for modeling epidemics Social internet sites as a source of public health information Predicting the behavior of techno-social systems Time series analysis. The Oxford Handbook of When will be the resumption of work in Wuhan and its surrounding areas during COVID-19 epidemic? A data-driven network modeling analysis Beware of the second wave of COVID-19 21) 110668(96150,125185)