Prediction of COVID-19 outbreak based on time-lagged relationship between outbound traffic from Wuhan and COVID-19 incidence in China: A time-series analysis (Preprint)

BACKGROUND Mass migration during the holiday season may have accelerated the spread of the 2019 coronavirus disease (COVID-19) outbreak across China. But the complex role of migration has not been fully accounted for in most epidemic models of COVID-19. OBJECTIVE To develop a simple and practical prediction model of COVID-19 based on the geo-temporal relationship between daily outbound traffic from Wuhan and the incidence of COVID-19 in 31 Chinese provinces during January-February 2020. METHODS We collected incidence of confirmed COVID-19 cases from National and Provincial Health Committee reports from January 10 to February 29, 2020. Volume of outbound traffic from Wuhan to other provinces was measured by Baidu Migration Index, a widely used metric that tracks migration based on cellphone location. We used cross-correlation function and autoregressive integrated moving average (ARIMA) model to examine time-lagged association between traffic volume and COVID-19 incidence by province. Contributors to the provincial variation in the temporal associations were investigated. Additionally, we estimated the reduction in cumulative incidence of COVID-19 cases following the travel ban for Wuhan. RESULTS Cross-correlation function analyses suggested that the volume of outbound traffic from Wuhan was positively associated with COVID-19 incidence in all provinces, with correlation coefficients between 0.22-0.78 (all P<0.05). Approximately 42% of provinces showed <1 week of lag between traffic volume and COVID-19 incidence, 39% with 1 week, and 19% with 2-3 weeks. Migration had more prolonged impacts in provinces closer to Wuhan and with more passenger influx from Wuhan, but affected economically advantaged provinces to a lesser extent. We further estimated that the travel ban may have prevented approximately 19,768 COVID-19 cases (95% CI: 13,589, 25,946) outside of Wuhan by February 29, 2020. CONCLUSIONS Outflowing migration from Wuhan facilitated the COVID-19 transmission to other parts of China with varying time-lagged effects dependent on provincial characteristics. The travel ban led to a significant reduction in COVID-19 outside of Wuhan.