Predicting the Spread of the COVID-19 Across Cities in China with Population Migration and Policy Intervention
27 Pages Posted: 1 Apr 2020More...
Background: The coronavirus disease 2019 (COVID-19) that broke out in Wuhan has been rapidly spreading to cities in China and the other countries globally. Policy makers in infected countries are facing unprecedented challenges in making tough decisions on when and what degrees of measures should be implemented. In this study, we developed a dynamic infectious disease model to estimate the spread and numbers of infected cases across the cities in China using a real-time human mobility dataset. The impact on the scales of the spreading by different measures and time associated with such measures were explored.
Methods: An infectious disease model, defined by five health states: susceptible, unconfirmed infectious, confirmed, recovered, and dead (SICRD), was constructed to simulate the spread of the epidemics across 371 prefectural cities in China over 200 days from January 1, 2020 to July 18, 2020. Inter-city human movement was captured by a large-scale human mobility dataset which was extracted from the mobile phone network from January 1, 2020 to February 7, 2020. Basic reproduction number and transition probabilities between states were estimated from reported cases for the same period. The median and 95% credible interval (CrI) of these parameters and simulation results were obtained from 360 repeated experiments. The base case analysis assumed no intervention / doing nothing (hypothetical scenario). The actual (effective) intervention scenario represented the real measures implemented in Wuhan (city-wide quarantine, strictest travel ban, and school closure etc.) on January 23, 2020. An earlier intervention (ten days earlier) , a lesser effective intervention, an earlier but lesser effective intervention, and time duration associated with such measures on the impacts of the spreading of the virus were explored.
Findings: The basic reproductive number was estimated as 2·22 (95% CrI: 2·15-2·33). In our base case analyses, the estimated cumulative numbers of infected cases were 11·2 (95% CrI: 11·0-11·4), 6·18 (5·87-6·40), 27·1 (26·4-27·8), and 21·1 (20·7-21·6) (in millions) in Wuhan, Huanggang, Chongqing, and Shanghai, respectively, and 1·18 (95% CrI: 1·16-1·22) billion in the whole of mainland China. When a tough effective intervention was implemented on the January 23 and lasted eight weeks (until March 22, 2020) the model estimated the corresponding infected cases in these four cities were reduced to 89,600 (95% CrI: 44,200-289,800), 19,300 (12,500-31,300), 2,390 (1,970-3,250), and 2,080 (1,710-2,830), respectively, represented 0·8%, 0·3%, 0·01%, and 0·01% of respective cases in base cases. Nationally the estimated cumulative infected population reduced to 0·28 million (95% CrI: 0·18-0·34), 0·02% of 1·18 billion of the base case. If the time of the actual intervention in Wuhan had been brought forward by ten days (January 13, 2020 instead of January 23, 2020), the estimated number of infected cases nationally for the same period would be 65,200 (95% CrI: 42,000-77,500), which was a quarter of the estimated cases in the actual scenario. In the lesser effective scenario, the estimated infected cases was 2·74 million (95% CrI: 1·68-6·51) which was ten times compared with the effective intervention in the actual scenario. The deaths was 0·063 million(95% CrI: 0·039-0·150). If the lesser effective intervention was implemented ten days earlier than the actual scenario (January 13, 2020 instead of January 23, 2020), the number of infected cases would be 0·652 million (95% CrI: 0·403-1·490), which was nearly a quarter of without early implementation. We estimated the optimal duration of the current quarantine measure in China should be maintained for about two months (from January 23, 2020 to March 22, 2020), otherwise, the epidemic might be resurgent.
Interpretation: Human mobility places a crucial role in the city-wide spreading of COVID-19. Without intervention, a substantial proportion of the population will be infected. An early effective intervention can significantly reduce the number of infected cases, but it needs to be sustained for a certain period in order to fully control the epidemic. There are important implications for countries outside China.
Funding Statement: J.Z. was supported by the National Natural Science Foundation of China (NSFC) (no. 61673070), L.D. was supported by the NSFC (no. 41801299), and Z.H was supported by NSFC (no. 61374165). G.Y. was supported by the Newton Fund through a UK-China AMR Partnership Hub award (Grant reference no: MR/S013717/1), and the National Institute for Health Research (NIHR) Health Technology Assessment Programme (Reference Number 13/34/64).
Declaration of Interests: The authors declare no competing interests.
Ethics Approval Statement: Data was publicly available.
Keywords: COVID-19, Epidemics, Intervention
Suggested Citation: Suggested Citation