Statistical aspects of quarantine interventions and incubation periods in epidemics

Quarantine of contacts of known SARS patients proved effective at controlling further transmissions during the 2003 outbreak. However, little work has been done on determining the optimal quarantine length. Our initial work analyzed the impact of using the maximum reported incubation period from an outbreak to guide quarantine policy. To do this, we quantified the probability of an infectious case transmitting disease after such quarantine. Methods were developed using order statistics theory with validation via a simulation study. Later we derived incubation-period-based methods to calculate the desired quarantine length, using the maximum likelihood estimation method built on truncated density function. The effectiveness of these latter methods was evaluated using simulation studies and applied to an actual in-flight SARS outbreak data. Because the incubation period is an important quantity in developing effective epidemic control programs, we proposed an E-M algorithm based method to estimate incubation period for epidemics where confirmed cases could be infected by more than one infectious source; this method was also applied to the in-flight SARS outbreak data. When a chain of infection happens, another possible problem arises that we only have information about the times of clinical illness onset but not the times of initial infection. We developed statistical approaches for using epidemic chains of symptomatic cases to estimate the distribution of incubation periods. We used the geometric distribution to approximate the date of infection, and derived estimators of the distribution parameters. We also examined imputation methods. We performed simulation to compare the performance of the various estimators.