Early dynamics of transmission and control of COVID-19: a mathematical modelling study

Adam J Kucharski; Timothy W Russell; Charlie Diamond; Yang Liu; John Edmunds; Sebastian Funk; Rosalind M Eggo

In the model, individuals were divided into four infection classes ( Figure 1 ): susceptible, exposed (but not yet infectious), infectious, and removed (i.e. isolated, recovered or otherwise no longer infectious). The model accounted for delays in symptom onset and reporting, as well as uncertainty in case observation (see Appendix for full model details). The incubation period was assumed to be Erlang distributed with mean 5.2 days (16) and delay from onset-to-isolation Erlang distributed with mean 2.9 days (2,15). The delay from onset-to-reporting was assumed to be exponentially distributed with mean 6.1 days (2). Once exposed to infection, a proportion of individuals travelled internationally and we assumed that the probability of cases being exported from Wuhan to a specific other country depended on the number of cases in Wuhan, the number of outbound travellers (assumed to be 3300 per day before travel restrictions were introduced on 23rd January 2020 and zero after), the relative connectivity of different countries (18) , and the relative probability of reporting a case outside Wuhan, to account for differences in clinical case definition, detection and reporting within Wuhan and internationally. We considered the 20 countries outside China most at risk of exported cases in the analysis.