ABSTRACT
Background: The current outbreak of coronavirus disease 2019 (COVID-19) has quickly spread across countries and become a global crisis. However, one of the most important clinical characteristics in epidemiology, the distribution of the incubation period, remains unclear. Different estimates of the incubation period of COVID-19 were reported in recent published studies, but all have their own limitations. In this study, we propose a novel low-cost and accurate method to estimate the incubation distribution. Methods: We have conducted a cross-sectional and forward follow-up study by identifying those asymptomatic individuals at their time of departure from Wuhan and then following them until their symptoms developed. The renewal process is hence adopted by considering the incubation period as a renewal and the duration between departure and symptom onset as a forward recurrence time. Under mild assumptions, the observations of selected forward times can be used to consistently estimate the parameters in the distribution of the incubation period. Such a method enhances the accuracy of estimation by reducing recall bias and utilizing the abundant and readily available forward time data. Findings: The estimated distribution of forward time fits the observations in the collected data well. The estimated median of incubation period is 8.13 days (95% confidence interval [CI]: 7.37-8.91), the mean is 8.62 days (95% CI: 8.02-9.28), the 90th percentile is 14.65 days (95% CI: 14.00-15.26), and the 99th percentile is 20.59 days (95% CI: 19.47, 21.62). Compared with results in other studies, the incubation period estimated in this study is longer. Interpretation: Based on the estimated incubation distribution in this study, about 10% of patients with COVID-19 would not develop symptoms until 14 days after infection. Further study of the incubation distribution is warranted to directly estimate the proportion with long incubation periods.