Prediction of Epidemic Spread of the 2019 Novel Coronavirus Driven by Spring Festival Transportation in China: A Population-Based Study

Changyu Fan, Linping Liu, Wei Guo, Anuo Yang, Chenchen Ye, Maitixirepu Jilili, Meina Ren, Peng Xu, Hexing Long, Yufan WangInternational Journal of Environmental Research and Public Health 2020 · DOI

TL;DR

This study uses migration data to predict the spread of the 2019-nCoV epidemic in China, showing a strong correlation between the number of Wuhan's migrant population returning home and confirmed cases. The findings highlight that Hubei Province and its surrounding areas are at high risk, suggesting potential underestimation of cases in some regions and informing targeted epidemic prevention efforts.

What Problem Does This Paper Address?

The rapid global spread of the 2019 novel coronavirus (2019-nCoV) presented a significant challenge for public health officials. A key factor in this spread was the massive movement of people during the Spring Festival holiday, particularly the large floating population originating from Wuhan, the initial epicenter. Understanding where these individuals returned to was crucial for predicting and controlling the epidemic's trajectory. However, at the time of the study, data on the precise distribution and characteristics of Wuhan's migrant population was limited, hindering effective real-time resource allocation and intervention strategies. This paper aims to fill that gap by estimating the scale and origins of Wuhan's migrant population and analyzing their potential role in epidemic spread.

Methods

The study employed a multi-pronged approach:

  • Data Source: Primarily utilized the 2013-2018 China Migrants Dynamic Survey (CMDS) and data from the Wuhan municipal government on its floating population.
  • Descriptive Statistical Analysis: Analyzed the distribution and social characteristics (age, education, family structure, household registration, reasons for migration) of Wuhan's floating population at provincial and prefectural levels.
  • Quasi-Experimental Analysis: Investigated the correlation between the size of Wuhan's floating population returning to their hometowns and the daily number of confirmed 2019-nCoV cases. This framework aimed to understand the impact of this mobile population on epidemic spread.
  • Epidemic Prediction Modeling: Developed models to predict epidemic trends at prefecture and province levels based on the proportion of Wuhan's floating population originating from these areas.

Datasets

  • 2013-2018 China Migrants Dynamic Survey (CMDS): A large-scale survey providing data on migrant populations across China.
  • Wuhan Municipal Government Data: Information on Wuhan's floating population.
  • "Dingxiangyuan" National Real-time Epidemic Website: Data on confirmed 2019-nCoV cases from January 25-31, 2020.

Key Insights and Ideas

  • Migration as a Key Driver: The study strongly emphasizes that the movement of Wuhan's large floating population during the Spring Festival was a primary driver of the epidemic's spread beyond the initial epicenter.
  • Hubei Province as a Hotspot: The concentration of Wuhan's migrant population within Hubei Province (75%) and its surrounding prefectural-level cities (66%) clearly identified these areas as having the highest risk of rapid epidemic spread.
  • Underestimation of Cases: The correlation analysis suggests that confirmed case numbers in some regions, particularly in Hubei Province, might have been underestimated, highlighting the need for enhanced surveillance.
  • Vulnerable Demographics: The study identified a significant proportion of middle-aged and older individuals within Wuhan's migrant population, who are at higher risk for severe illness, further increasing the epidemic's potential impact.
  • Rural Vulnerability: The high percentage of migrants with rural household registrations and lower educational levels points to potential challenges in health awareness and access to medical resources in their hometowns, exacerbating transmission risks.

Key Results

  • Geographic Distribution: Approximately 75% of Wuhan's floating population originated from Hubei Province, with a further 66% coming from prefectural-level cities surrounding Wuhan. The distribution showed a clear spatial pattern of "circle layers and echelons."
  • Social Characteristics: 84% migrated as nuclear families, 71% with 3-4 family members, 85% had rural household registration, and 84% migrated for work or business.
  • Correlation with Cases: A high correlation was found between the size of Wuhan's migrant population returning home and the daily number of confirmed cases.
  • Underestimation Evidence: The study provided evidence suggesting that confirmed cases in some Hubei provinces and cities might be underestimated, with rapid increases observed in certain regions.

Limitations and Future Work

  • Exclusion of Certain Populations: The study acknowledges limitations in not including other large-scale populations such as college students, short-term business travelers, transit passengers, and tourists.
  • Geographic Scope: The analysis excluded Hong Kong, Macao, and international migrants, limiting the estimation of population size and epidemic spread in these regions.
  • Modeling Limitations: The authors noted that their model did not include infectious disease analysis models like SIR, which could provide a more in-depth analysis of potential spread.
  • Temporal Scope: The findings are primarily applicable to the period around the end of the Spring Festival holiday, and the epidemic's spread may become more complex as populations return to work and study.
  • Future Work: The authors suggest using big data, such as transportation and mobile internet location information, to include short-term floating populations in future studies to better estimate population outflow and scale. They also emphasize the need for rigorous studies on population outflow from epidemic epicenters.

Summary updated

Abstract

After the 2019 novel coronavirus (2019-nCoV) outbreak, we estimated the distribution and scale of more than 5 million migrants residing in Wuhan after they returned to their hometown communities in Hubei Province or other provinces at the end of 2019 by using the data from the 2013-2018 China Migrants Dynamic Survey (CMDS). We found that the distribution of Wuhan's migrants is centred in Hubei Province (approximately 75%) at a provincial level, gradually decreasing in the surrounding provinces in layers, with obvious spatial characteristics of circle layers and echelons. The scale of Wuhan's migrants, whose origins in Hubei Province give rise to a gradient reduction from east to west within the province, and account for 66% of Wuhan's total migrants, are from the surrounding prefectural-level cities of Wuhan. The distribution comprises 94 districts and counties in Hubei Province, and the cumulative percentage of the top 30 districts and counties exceeds 80%. Wuhan's migrants have a large proportion of middle-aged and high-risk individuals. Their social characteristics include nuclear family migration (84%), migration with families of 3-4 members (71%), a rural household registration (85%), and working or doing business (84%) as the main reason for migration. Using a quasi-experimental analysis framework, we found that the size of Wuhan's migrants was highly correlated with the daily number of confirmed cases. Furthermore, we compared the epidemic situation in different regions and found that the number of confirmed cases in some provinces and cities in Hubei Province may be underestimated, while the epidemic situation in some regions has increased rapidly. The results are conducive to monitoring the epidemic prevention and control in various regions.
BibTeX
@article{fan2020prediction,
  author = {Changyu Fan and Linping Liu and Wei Guo and Anuo Yang and Chenchen Ye and Maitixirepu Jilili and Meina Ren and Peng Xu and Hexing Long and Yufan Wang},
  title = {{Prediction of Epidemic Spread of the 2019 Novel Coronavirus Driven by Spring Festival Transportation in China: A Population-Based Study}},
  journal = {International Journal of Environmental Research and Public Health},
  year = {2020},
  doi = {10.3390/ijerph17051679},
}
Figure