Assessing heterogeneity of disease transmission from data on households' outbreaks
by
Ning Li
La Trobe University
Coauthors: Guoqi Qian, Ricahrd Huggins

Discrete epidemic models are appropriate for describing transmission of an infectious disease that has a virtually constant latent period. When observations from a set of similar households are available, it is possible to estimate the rate at which a susceptible in a household becomes infected. We consider the situation when the infection rate differs across households, possibly due to different levels of crowding or differences in the interaction of household members. There are usually two types of data upon which inferences may be based: the actual chains of infection, and the total number of cases. The first type of data enables a full likelihood be determined and estimated. The MLE, however, is not directly available for the second type of data, which is less detail but easier to collect. We demonstrate that Markov Chain Monte Carlo methods provide a solution. A hierarchical model is set up in a Bayesian framework. Posterior distributions of interest are computed using the Metropolis-Hastings within Gibbs sampler. The procedure is illustrated on a measles dataset.

Date received: March 27, 2002