ABSTRACT
The assessment of the age-specific seroprevalence profile from serological data is often performed using current infection status data. This data is obtained by dichotomizing the optical density measurements on the basis of a fixed cut-off associated with the employed diagnostic assay. This means that the information embedded in the individual antibody optical density is lost, as it is replaced by a binary variable indicating only whether the subject is still susceptible or immune to the infection. Moreover, due to the nature of the assay’s cut-off, this categorization is prone to misclassification and to the removal of those cases labeled as inconclusive. To overcome this issue, mixture models have been introduced for estimating the age-specific population seroprevalence, modeing directly the optical density measurements. In this chapter, we propose hierarchical Bayesian mixture models for the estimation of the age-specific seroprevalence for both pre- and post-vaccination epochs, using Markov chain Monte Carlo (MCMC) methods. For the former scenario, we discuss methods to estimate age-specific force on infection from the estimated seroprevalence. Data from varicella-zoster virus, parvovirus B19, and measles serological surveys in three European countries are employed to illustrate the proposed methodology.
