Statistical Analysis
We calculated empirical cumulative density functions of the incubation period and serial intervals by computing the cumulative fraction of all observations that fell below each observed value in the respective data sets.We estimated
the incubation period by identifying the earliest and latest time of possible exposure and the time of symptom onset for each case.Treating these times as interval-censored estimates of the incubation period for each person,we fit a log-normal distribution to these data using maximum-likelihood techniques.We then examined the robustness of our estimates with multiple definitions of onset and with the exclusion of particular cases.We estimated the serial interval by identifying the times of symptom onset in the patient and in the person who transmitted the infection(infected–infector pairs) and then fitting a log-normal distribution to these interval-censored data.21 We estimated the medians and 5th and 95th percentiles of the incubation period and the serial interval using the quantiles of the log-normal distribution fit to each data set(R statistical package, version 2.15.1, and coarse Data Tools library).