Dynamic Estimation of Epidemiological Parameters of COVID-19 Outbreak and Effects of Interventions on Its Spread

Hongzhe Zhang; Xiaohang Zhao; Kexin Yin; Yiren Yan; Wei Qian; Bintong Chen; Xiao Fang

Using our approach detailed in the Method section, we estimated the true cumulative number of infections in Wuhan by each day for the period between January 19, 2020 and March 5, 2020. The input to our method is the cumulative number of infections in Wuhan by January 18, 2020 estimated in Imai et al. 12 , whose baseline estimate is 4,000 with a 95% confidence interval [1, 800] . Fig. 1 plots the true cumulative number of infections estimated by our method in a dotted blue line, in comparison to its respective official number reported by the government (solid blue line). As shown, the gap between these two curves is significant, especially at the beginning of the observation period measured by percentage. Such marked difference is partly attributable to the lack of testing and treatment capacities, especially at the beginning of the outbreak. In particular, we estimated that the true cumulative numbers of infections in Wuhan by January 23, 2020 (date of Wuhan lockdown) and March 5, 2020 were 11,239 [95% CI 4, 372] and 124,506 [95% CI 69,526-265,113], respectively. In comparison, their respective official numbers were 495 and 49,797. We also provide our estimated true cumulative number of infections in Wuhan by each day in the observation period (Supplementary Table 1) . Fig. 1 also presents the estimated true number of actively infected and quarantined cases by each day in the observation period (dotted red line) and its respective official number (solid red line). The former is computed by our method, which estimates the actual number of actively infected cases who are quarantined effectively, whereas the latter typically counts those actively infected and currently quarantined at hospitals. By March 5, 2020, our estimated true number of actively infected and quarantined cases was 44,778 [95% CI 24,049-112,697] whereas its official counterpart was 20,049. The gap between these two curves represents the number of actively infected people who are effectively quarantined but fail to be included in the government statistics. Many of these infected people could not be tested or officially admitted to hospital, but nevertheless conducted effective self-quarantine at home or other isolated places.