key: cord-1054993-utlponna authors: Zhang, Yuzhen; Jiang, Bin; Yuan, Jiamin; Tao, Yanyun title: The impact of social distancing and epicenter lockdown on the COVID-19 epidemic in mainland China: A data-driven SEIQR model study date: 2020-03-06 journal: nan DOI: 10.1101/2020.03.04.20031187 sha: f40eaec892778166e4f8f8eeb576893e82a591be doc_id: 1054993 cord_uid: utlponna The outbreak of coronavirus disease 2019 (COVID-19) which originated in Wuhan, China, constitutes a public health emergency of international concern with a very high risk of spread and impact at the global level. We developed data-driven susceptible-exposed-infectious-quarantine-recovered (SEIQR) models to simulate the epidemic with the interventions of social distancing and epicenter lockdown. Population migration data combined with officially reported data were used to estimate model parameters, and then calculated the daily exported infected individuals by estimating the daily infected ratio and daily susceptible population size. As of Jan 01, 2020, the estimated initial number of latently infected individuals was 380.1 (95%-CI: 379.8~381.0). With 30 days of substantial social distancing, the reproductive number in Wuhan and Hubei was reduced from 2.2 (95%-CI: 1.4~3.9) to 1.58 (95%-CI: 1.34~2.07), and in other provinces from 2.56 (95%-CI: 2.43~2.63) to 1.65 (95%-CI: 1.56~1.76). We found that earlier intervention of social distancing could significantly limit the epidemic in mainland China. The number of infections could be reduced up to 98.9%, and the number of deaths could be reduced by up to 99.3% as of Feb 23, 2020. However, earlier epicenter lockdown would partially neutralize this favorable effect. Because it would cause in situ deteriorating, which overwhelms the improvement out of the epicenter. To minimize the epidemic size and death, stepwise implementation of social distancing in the epicenter city first, then in the province, and later the whole nation without the epicenter lockdown would be practical and cost-effective. The coronavirus disease 2019 (COVID-19), initially taken as "pneumonia of unknown etiology", emerged in December 2019, Wuhan, Hubei Province, China. The causative pathogen was announced by the Chinese Center for Disease Control and Prevention (China CDC) on Jan 08, 2020, to be a novel coronavirus [1] , lately named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [2] . COVID-19 broke out in Wuhan in January 2020, and spread to the whole Hubei Province, the rest of China and abroad with astonishing speed. On Jan. 31, 2020, the World Health Organization (WHO) announced that COVID-19 constitutes a "public health emergency of international concern". As of Feb 28, there were 7,8961 cases confirmed in China, and 4,691 cases confirmed in 51 other countries. On that day, WHO increased the assessment of the risk of spread and risk of impact of COVID-19 to very high at the global level [3] . Apart from the intrinsic infectivity of the virus, population mobility and epidemic prevention and control measures could affect the prevalence scale. Unfortunately, the prevalence of COVID-19 encountered the Spring Festival Migration of China, the world's largest annual human migration as hundreds of millions of people rush home for family reunions. In addition, the epicenter Wuhan is the capital of Hubei Province of China. It has a population of more than 15 million, including resident and floating population, and it situates at the transportation hub in the central China area. As of Jan 23, 2020, more than 5 million people were migrating out of Wuhan according to official accounts [3] . Emergency monitoring and close contact management in Wuhan was carried out since Jan 03, 2020; China CDC Level 2 emergency response activated on Jan 06, and Level 1 emergency response activated on Jan 15 [1] . On Jan 20, 2020, COVID-19 was included in the statutory report of Class B infectious diseases, managed as Class A infectious diseases by the National Health Commission of China. The Chinese government locked down the Wuhan city on Jan 23, and then locked down other cities of Hubei Province immediately after. By Jan 25, 30 provincial governments in China activated first-level public health emergency response. Hence, In addition to strict quarantine management, substantial social distancing measures to limit population mobility and to reduce within-population contact rates were executed almost in the whole country. For example, public activities were canceled; communities adopted enclosed management; the national holiday of Spring Festival and the winter vacation were extended so that work resumption and school re-opening could be extensively postponed. In addition, people were required to wear facemasks in public. However, with all those efforts, the prevalence of COVID-19 was escalating. Epidemic prevention and control strategies need to be re-examined. Vaccine and antiviral drug development is the ultimate way to defeat a virus, but it is time-consuming. Non-pharmaceutical interventions to interrupt transmission could be implemented immediately, gaining time for pharmaceutical development. Briefly, there were three steps of non-pharmaceutical interventions for reducing contact rates between susceptible individuals and infected individuals. First, quarantine management, i.e. quarantining the infected, the suspicious and their close contacts; second, social distancing to confine within-population contact; third, locking down the epicenter to prevent further exportation of infected and latently infected individuals to other regions. Quarantine management is a fundamental measure ought to be taken once the human-human transmission is confirmed. Theoretically, if substantial social distancing and/or epicenter lockdown were implemented early enough, there would be no prevalence or no spreading. But realistically, it takes time for preliminary investigation. Besides, rigorous measures would bring about deep social influences and economic consequences. So, it is challenging to choose the right response at the right scale in the right area at the right time [4] , especially when the transmission pattern and clinical characteristics were not fully understood. The importance of non-pharmaceutical control measures requires further research to quantify their impact [3] . Mathematical models are useful to evaluate the possible effects on epidemic dynamics of preventive measures, and to improve decision-making in global health [5, 6] . In this study, we developed data-driven susceptible-exposed-infectious-quarantine-recovered (SEIQR) models to simulate the epidemic under the interventions of social distancing and epicenter lockdown, and to evaluate the impact of earlier interventions on the epidemic size and death number totally and respectively in Wuhan, Hubei Province, and 12 other provinces and municipalities. Population migration data combined with officially reported data were used to estimate model parameters, the number of the latently infected individuals, and daily exportation of the infected and latently infected from Wuhan. Our work provided evidence for the decision-making concerning prevention and control of the ongoing COVID-19 epidemic in other countries and future infectious disease epidemic. In this study, SEIQR models were developed to simulate the COVID-19 epidemic in China. We built three SEIQR models to simulate the epidemic in Wuhan, the rest of Hubei Province, and other provinces and municipalities in China, respectively. Other provinces and municipalities studied here include nine provinces (i.e. Zhejiang, Guangdong, Henan, Hunan, Anhui, Jiangxi, Jiangsu, Shangdong and Sichuan) and three municipalities (i.e. Chongqing, Beijing and Shanghai) in China, which covered 90% infections in mainland China. For convenience, "Hubei" represents the rest of Hubei Province excluding Wuhan, and "other provinces" represents the above-mentioned provinces and municipalities in the text. daily latent infected ratio the daily exposed individuals exported from Wuhan to Hubei the daily exposed individuals exported from Wuhan to other provinces lockdown social distancing estimated latent infected individuals Figure 1 . The framework of building models for Wuhan, Hubei and other provinces. We added the intervention measures, i.e. social distancing and epicenter lockdown, to models. These models were able to simulate the COVID-19 epidemic with interventions activated on different dates. Figure 1 describes the framework of building estimation models. The reported confirmed, recovered and death cases were used to build initial models for Wuhan, Hubei and other provinces. Since there could be considerable under-reporting bias in the epicenter for lots of reasons. We assumed that the cases reported by other provinces were more accurate. Therefore, the model built on the cases reported by other provinces could estimate more accurately the number of infections. The reproductive number, infectious period and mean latent of models built for Wuhan and Hubei should be revised based on the latently infected individuals. The number of latently infected individuals in Wuhan before Jan 23 was estimated by the latently infected ratio and the daily population migration data from Wuhan to the rest of China. In the revised models of Wuhan and Hubei, parameters involving social distancing measures such as the removed rate, the decline factor of the reproductive number and the isolating ratio of susceptible individuals inherited from the initial models. The daily latent infected ratio is calculated based on the daily number of exposed individuals in Wuhan and the daily susceptible population size. Through the daily latent infected ratio and the daily number of individuals leaving Wuhan, the daily number of exposed individuals exported to Hubei and other provinces can be estimated. Wuhan has a population of around 15.39 million, including a resident population of 9 million, and nearly 6.39 million floating population exported before Jan 23 [7] . The population exported from Wuhan is the main source of imported cases in the rest of China. The travelers flowed from the rest of Hubei to other provinces, between two of the other provinces, and heading to Wuhan were not considered in this study. We assumed that the population exported from Wuhan was susceptible and exposed population. During the period of time from Jan 01 to Jan 22, 2020, the number of exported population from Wuhan was derived from two sources, a) domestic passenger number by air, train and road provided by Cao, Z., et al [7] , and b) international outbound passenger number by air departed from Tianhe Airport (Wuhan, China) provided by a commercial APP named "Flight Steward". Officially reported case data by the National Health Commission were used to estimate the parameters of SEIQR models. As of Dec 31, 2019, there were 47 confirmed COVID-19 cases in Wuhan [1] . We took that 47 cases as the initial number of infected individuals in the model. From Jan 1 to 23, 2020, the estimated latent infected ratio in Wuhan is 0.12% [7] . As of Jan 23, the accumulated number of latently infected individuals in Wuhan and exported infected individuals estimated by the latently infected ratio were 10800 and 7677.6, respectively. The estimated number of latently infected individuals by Jan 01, 2020 in Wuhan was 380.9 (95%-CI: 379.8~381.0). The estimated base population size of daily susceptible individuals was 2*106. The base reproductive number r0 of Wuhan and Hubei was set to 2.2 (95%-CI: 1.4~3.9) referring to the study of Liu, Q., et al [1] . When the interventions continue for 30 days, r0 was reduced to 1.58 (95%-CI: 1.34~2.07). In other provinces, the estimated average base reproductive number r0 was 2.52 (95%-CI: 2.43~2.63), and with 30-day interventions, r0 was reduced to 1.65 (95%-CI: 1.63~1.69). The estimated infectious period TI was 2.26 days (95%-CI: 2.14~2.39) in Wuhan and Hubei, and was 3.75 days (95%-CI: 3.43~4.13) in other provinces. The estimated mean latent was 10.1 days (95%-CI: 8.82~11.78) in Wuhan and Hubei, and was 11.01 days (95%-CI: 9.51~13.06) in other provinces. The estimated removed rate was 0.0125 (95%-CI: 0.008~0.017) in Wuhan, 0.0175 (95%-CI: 0.0126~0.0224) in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Reasonably, earlier activating substantial social distancing at high-level (1*λ0), which is the actual social distancing strength implemented in China, could lead to the best results. showed the estimated epidemic controlled with the current strategy (simultaneously activated nationwide social distancing and epicenter lockdown on Jan 23) for comparison. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint . https://doi.org/10.1101/2020.03.04.20031187 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The results of virtual experiments vary. Some research suggested social distancing is effective, under the condition of early activation and long-lasting implement of combined measures [8] , strict implement [6, 9] , and spontaneously adopted [10] ; some concluded the effectiveness was mild [11] ; some argued that moderate social distancing can worsen the disease outcome, [6, 9] But real-world studies are rare. Here we quantified the impact of social distancing on the epidemic size in the real world, providing evidence that social distancing is effective to interrupt the transmission of the respiratory pathogen, especially with early implementation and to a substantial extent. Activating social distancing at the possible earliest time in the epicenter could gain time for the preparation of regions out of epicenter, and reduce wide range socioeconomic impact. The third question is the necessity and the timing of the epicenter lockdown. Unexpectedly, our results showed that the significant decrease of the epidemic size by earlier social distancing would be partially neutralized by earlier epicenter lockdown. Further investigation suggested that the influences of epicenter lockdown on epidemic size and death differed between Wuhan and non-Wuhan regions. Earlier lockdown of Wuhan would deteriorate the situation in situ, but would largely lower the infection number in the rest of Hubei province and other provinces. The reasons may be as follows. After lockdown, a portion of the infected and the latently infected (the exposed in the model) would accumulate in Wuhan instead of exporting outside. Theoretically, it would not change the total infection or death number, but we found the conversion rate from the close contacts to the exposed is much higher in Wuhan than in other provinces, as listed in table S1 of Supplement Material, so is the death rate. Quickly surging infected patients in Wuhan exerted extraordinary pressure on local healthcare systems, resulting in an acute shortage of medical resources and healthcare workers. Then the fast-growing infection and resource shortage fueled each other in a vicious cycle. Numerous close contacts could not be quarantined for medical observation. Plentiful infected patients could not be hospitalized or quarantined for treatment. Then they would contact more susceptible persons and infect more people. Besides, unhospitalised and under-treated patients suffered a higher death rate. Taken together, the accumulation of Several issues to be noted when interpreting our work. One is about the comprehensiveness of the interventions. In addition to the mainly discussed strategies of social distancing and epidemic lockdown, measures implemented in China also included forceful medical supports to Wuhan, reinforced quarantine management, and public health measures, e.g. requiring people to wear masks in public, raising public awareness for hand hygiene. Although these measures were not specifically discussed, their effectiveness has been illustrated by the adjustment of daily removed rate, and the decline of the base productive number r0 in our model. Another is about the time range of the interventions evaluated in this paper. We choose Jan 3, 2020 as the earliest available time point to implement interventions of social distancing and/or epicenter lockdown for two reasons. Practically, it is the date Wuhan started close contact management [1] , a kind of preliminary quarantine management. Taking the move before this date could be better, but scarcely possible, so beyond the scope of our discussion. Technically, the available population migration data [7] covers the period from Jan 01 to Jan 22, so the timing of Wuhan lockdown investigated in this study is confined within this period. The effect of abolishing the epicenter lockdown was estimated under the assumption that the scale of the population migrated out of Wuhan would keep stable. Normally, the great migration departing big cities like Wuhan would pause around Jan 25, the day of Spring Festival, then the migration would continue with a reverse direction back to cities. However, under the circumstances of the epidemic prevailing, if the lockdown was abolished, individuals trying to get away from Wuhan might increase, leading to more imported cases in other regions than expected. In addition, the application of our results should be tailored in a different situation. The non-Hubei provinces we studied were the ten provinces with most infections besides Hubei, plus the two municipalities of most importance in mainland China. These provincial regions had a large inflow of people from Wuhan, and had a high local population density. So there were more imported cases, and easy to trigger widespread local transmission. For those regions with little people flow from the epicenter and with a population of low density, it could be possible that reinforced quarantine management without stringent social distancing could be effective enough to interrupt local transmission. An extreme example is that there was only one COVID-19 case in Tibet Autonomous Region of China, and the one was an imported case from Wuhan. He was timely quarantined for treatment and discharged after recovery. 23 close contacts were quickly traced and quarantined for medical observation, and then ruled out of infection. For other regions in and out of China, the specific measures may be adjusted to the different social-political-economic environment. Based on the epidemic of COVID-19 in mainland China, we developed data-driven SEIQR models to investigate the impact of social distancing and epicenter lockdown on the epidemic dynamics. Activating social distancing at the possible earliest time with comprehensive and rigorous measures could significantly reduce the epidemic size of COVID-19. A stepwise implementation of social distancing in the epicenter city first, then in the province, and later the whole nation could be more practical and cost-effective. The . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.04.20031187 doi: medRxiv preprint decision of epicenter lockdown depends on the primary goal of epidemic control. To minimize the epidemic size and death, the epicenter city should not be locked down. To better confine the epidemic to the origin region, and mitigate the impact on nationwide socio-economy, the epicenter lockdown should be taken earlier. Gratefully, the epidemic in mainland China has been declining steadily, especially in non-Wuhan regions. However, the epidemic out of China may just begin. Several countries, i.e. Korea, Italy, Iran, and Japan, etc. are confronted with an ongoing epidemic outbreak. With caution, the COVID-19 epidemic may even evolve into a global pandemic. Besides, other virulent infectious diseases may attack humans again in the future. We sincerely hope that our work could help the decision-making of epidemic prevention and control strategy for other countries in this COVID-19 epidemic and for future infectious disease epidemics. Tailored and sustainable approaches should be adopted in a different situation, striking a balance among the control of infection and death number, confining epidemic regions, and maintaining socioeconomic vitality. Early transmission dynamics in wuhan, china, of novel coronavirus-infected pneumonia Severe acute respiratory syndrome-related coronavirus: The species and its viruses -a statement of the coronavirus study group Report of the who-china joint mission on coronavirus disease 2019 (covid-19) A new twenty-first century science for effective epidemic response Modeling infectious disease dynamics in the complex landscape of global health Individual preventive social distancing during an epidemic may have negative population-level outcomes Incorporating human movement data to improve epidemiological estimates for 2019-ncov Simulation suggests that rapid activation of social distancing can arrest epidemic development due to a novel strain of influenza Controlling epidemic spread by social distancing: Do it well or not at all. BMC Public Health Spontaneous social distancing in response to a simulated epidemic: A virtual experiment Game theory of social distancing in response to an epidemic The authors declare no conflict of interest.