4.1. COVID-19 Transmission in China is Highly Correlated with Factors that Return Migrants to Their Birthplace
Most of the differential spatial transmission of COVID-19 in China is due to migrants returning from affected regions. Long-distance transmission of COVID-19 at this time, from cities such as Harbin and Wenzhou, is due to the large numbers of migrants from Wuhan in these cities and the probability that the migrants will engage in long-distance travel. A large number of studies used the number of transport networks to examine the possibility of transmission outbreaks. However, this method is inaccurate in predicting migrants’ homecoming in China. The transportation used for returning home in China in the spring may consist of multiple forms, including self-driving, which may be different from other forms of migration. Therefore, it is not feasible to use only a single transportation method to make predictions.
China is a country undergoing rapid urbanization, and during this process, there is a large migrant population. In this study, we take advantage of the important characteristic of migration in the Spring Festival travel rush with a clear direction, that is, returning to the birthplace, and propose the idea of using the information of residence place and birthplace in the migrant population questionnaire to build a Residence place–Birthplace probabilities network. The bivariate correlation analysis and regression results in
Table 1 and
Table 2 show that this method can satisfactorily explain the spatial disease transmission mode, and why the disease does not occur in cities with high population and density in China. It can also simulate the risk and the number of confirmed diagnoses in other prefecture-level cities after the transmission of COVID-19.
4.2. Different Emergency Management and Control Plans for Different Disease Outbreak Scenarios in Immigrant Cities
Because of the characteristics of human networking, the transmission of disease is spatially heterogeneous. Spatially localized mass treatment is a crude approach compared to detailed contact tracing [
31,
32,
33], but it may be implemented more quickly in practice. However, its broad nature poses problems: the number of individuals and regions affected by the intervention may be larger, with corresponding burdens that must be accommodated; if the intervention is harmful at the individual level (such as isolation), individuals and economic development will suffer unnecessary losses. It should be considered how to better integrate all focused interventions. Although a large number of recent reports have predicted how many people would contract pneumonia [
4,
23,
24,
25], there are few studies on the different threat levels of similar emerging epidemic for cities across the country, and the extent of this threat has not been quantified. In terms of COVID-19 (
Figure 1), it is not appropriate for China to carry out nationwide intervention by stopping work and implementing school closures.
For cities not in the disease region, when there is a disease situation in other cities, we need to consider the origin of immigrants from different places by
, and the probability of migrants in this city going to different cities by
through the R-B matrix. The use of the two sets of migration probabilities is helpful to formulate the post-disease spatially differential control policy. According to our forecast, as shown in
Figure 2, in Kunming, Hefei, and Changsha cities, most of the migration originates from the province or nearby cities, which has little impact on other cities in China and is relatively simple to control. However, in Wenzhou, Guangzhou [
34], Dongguan, Shenzhen, or other cities, most of the migrant population originates from different provinces that are distributed all over China. Therefore, there is a wide range of diffusion, and it has a large impact on other cities and is relatively difficult to control. In short, we believe that a minimum amount of risk management and control should be implemented for large-scale intervention, and we must implement different levels of emergency plans according to the risk of disease transmission.
4.3. Policy Intervention and Disease Prevention in the Important Migrant Birthplace Regions
Accurate geographic migration networks enable early and efficient planning of interventions to make the best use of scarce resources or manpower. We identified these high-risk node cities in
Figure 3. On the one hand, most migrant birthplace regions tend to be economically and socially underdeveloped, with large birth populations, and are important nodes in the network of emerging epidemic [
16,
33]. By constructing a matrix, we identified these important migrant birthplaces by
, and predicted the overall risk of transmission to other cities under the same circumstances if similar emerging epidemic scenarios occur in other large immigrant cities. The results in
Figure 3 show that the regions with the highest probability are in the center of China, and these cities are located in Henan and Anhui provinces, which are economically underdeveloped but have large populations. On the other hand, some important migrant concentration node cities calculated through the index
are also worth noting. These cities are often world-class global cities with more developed economies and very close links with other immigrant cities in China.
Figure 3 shows that these cities with high risk include Beijing, Shanghai, Guangzhou, Shenzhen and Chongqing. Therefore, in general, it is most effective to make emergency plans for these high-risk node regions under disease spread situations.
The following emergency plans need to be prioritized in these important node regions: (1) Plans for the deployment of vaccines, the provision of detection reagents, the provision of protective equipment (masks, disinfectants), and the provision of disease prevention personnel [
17,
33] will interfere with disease transmission through biosecurity measures. (2) Carrying out the construction of emergency isolation facilities (such as temporary infectious disease hospitals or reconstruction plans), plans for the control and isolation of populations returning to their hometowns, plans for suspension of school and work [
1], and strict adherence to universal precautions in health care settings are critical in controlling the transmission of disease through isolation measures in these node cities.5. Conclusions
The outbreak of COVID-19 in China has attracted wide attention from all over the world. The impact of COVID-19 has been significant, raising concerns regarding public health risks in China and worldwide. A large migrant population left Wuhan and returned to their hometowns for the Spring Festival, which resulted in the transport of a large amount of virus to other regions.
In this study, we constructed a matrix of the residence–birthplace (R-B matrix) of migrant populations by using the data from the China migrant population survey in 2017. This matrix was used to analyze the confirmed cases of COVID-19 from February 1–15, 2020 after the outbreak in Wuhan. Then, we used this R-B matrix to simulate an outbreak scenario in 22 immigrant cities in China, and proposed risk prevention measures after the outbreak. It was found that similar emerging epidemic scenarios would occur in the cities of Wenzhou, Guangzhou, Dongguan, or Shenzhen, with a wider disease transmission. We also used the matrix to determine that cities in Henan province, Anhui province, and Municipalities (such as Beijing, Shanghai, Guangzhou, Shenzhen, Chongqing) have a high risk level of disease carriers after the outbreak due to a high influx or outflow of people. Thus, we should give priority to formulating policies that will ensure that an emergency response plan is launched first after the occurrence of a similar emerging epidemic outbreak to check, control, and isolate people who leave or return after the outbreak.
Our method has several limitations. First, the method in this study was only obtained from the perspective of the migrant population, but a small proportion of the registered population also returned to their birthplace during the Spring Festival after the outbreak of the disease. Second, COVID-19 was transmitted by cross-provincial, cross-city, and trans-national travel. In addition, inadequate criteria for early outbreaks may have contributed to some of the errors in the number of people diagnosed, and therefore, further research is needed. However, the migrant population contributed greatly to disease spread, and we believe that limiting migration and isolating people is one of the most effective ways to control an emerging epidemic.
Our suggestions are very helpful and relevant, and would assist the Chinese government in its future management of disease. Priority should be given to research on the migration places of migrant populations in immigrant cities so that targeted policies can be formulated, especially population control during the Spring Festival. Simulation of the transmission of secondary infections in migrant birthplaces can also contribute to long-term infection research. In addition, it is necessary to model the number of people infected at different times.
In summary, this study provides a new perspective for the examination of epidemiological and behavioral data by proposing migrant networks. We have shown that we did not analyze traffic data (e.g., number of flights or train frequencies), but rather, we analyzed the social network causes of traffic migration. This is more predictive, and even in complex situations, simple tools can be designed to provide a method for predicting the transmission of disease and measuring risk. It is hoped that by providing simple and accurate approximations, a constantly evolving understanding of the complexity of the real world will be gained, and the method can continue to assist in understanding and controlling diseases.