The COVID-19 outbreak has been a serious public health threat worldwide. Individually documented case descriptions of COVID-19 are published by Chinese provinces (excluding Hubei). We use these descriptions to study the transmission characteristics in China, and how they are influenced by public awareness and control measures.
Dates for infection, symptom onset, quarantine, hospitalization and diagnosis, and sources of infection are tabulated from published cases descriptions. We use MCMC to estimate the exponential growth rate of cases infected in Hubei, and parametrize the distributions for the incubation period, and periods from symptom onset to hospitalization and diagnosis. Graph of infection is constructed and used to estimate the reproduction number in each generation.
The median incubation period is 6.2 days overall, and 7.3% patients have an incubation period longer than the recommended 14-day quarantine period. The median period from symptom onset to hospitalization is 4.4 days before the lockdown of Wuhan city, and 2.1 days after the lockdown. The median period from symptom onset to diagnosis is 8.8 days before the lockdown, and 4.8 days after the lockdown. The number of cases in Hubei doubled every 3 days before the lockdown. In other provinces, reproduction number decreased from 1.51 in the first-generation patients to less than 0.21 in later generation patients.
We parametrized incubation period, and the periods from symptom onset to hospitalization and diagnosis. Some (possibly a majority of) patients were infected before the symptom onset of their sources of infection, causing a low success rate of official (not self) quarantines. Social distancing and self-quarantine greatly reduced the reproduction number, and shorted the periods from symptom onset to hospitalization and diagnosis.