Strongly heterogeneous transmission of COVID-19 in mainland China: local and regional variation

Yuke Wang; Peter F.M. Teunis

For 112 confirmed cases in Tianjin between 21 January 2020 and 12 February 2020, the transmission network could be estimated based on the dates of symptom onset, augmented by contact information between cases. Table 1 shows some summary statistics of demographic and contact information for those cases in Tianjin. As contacts between many of the cases could be identified, so that many elements of the transmission probability matrix were known, joint estimation of the remaining unknown contact probabilities and the serial interval distribution was feasible. With a prior for the serial interval distribution set as Gamma(4,2), 1,000 updates were performed, alternating between probability transmission matrix and serial interval distribution parameters, with 100 iterations for each update and the best fit parameters were chosen as those with the highest posterior probability. Figure 2 shows an estimated (posterior mode) transmission network. Using the available contact information, there appeared to be considerable variation in the sizes of clusters of cases. There were two major clusters: one among crew members of Tianjin railway and another one among sales representatives and customers in Baodi department store. Figure 1b shows the posterior mode serial interval distribution as a gamma distribution (shape parameter 3·16, scale parameter 1·52), with a mean of 4·8 days. This best fit serial interval distribution has been used in all following analyses. Figure 3 shows estimated reproduction numbers by day for the initial 425 confirmed cases in Wuhan. "The market" node as an infectious source was linked to as many as 13·6 cases, on average. For those 425 initial cases, the mean reproduction number was 2·5 (until 31 December 2019). The estimated numbers of imported cases from Wuhan varied among provinces in mainland China. Provinces bordering Hubei province include Henan, Hunan, Anhui, and Jiangxi (213, 175, 163, 146), and provinces with close economic ties with Hubei province include Zhejiang and Guangdong (271 and 228, respectively). These provinces all had high estimated numbers of cases imported from Wuhan by 3 February 2020 (Figure 4b ).