Sensitivity analysis was used to test the two assumptions: (i) that the existence of a latent period (,7-21 days) can be ignored; (ii) that the specific choice of a 30-day censoring interval for true infection times was appropriate given the length of the asymptomatic period (,100 days).