BIRTH-DEFECTS REGISTERED BY DOUBLE SAMPLING - A BAYESIAN-APPROACH INCORPORATING COVARIATES AND MODEL UNCERTAINTY

In double-sampling schemes, a large sample is classified by using one method, and a subsample is also classified with a supplementary method . In the application discussed here, we are attempting to identify inf ants in Norway born with Down's syndrome by using both a national birt h registry and a regional registry. Usual methods for analysing such d ata assume that one classification method is perfect, which is not the case here. We develop a Bayesian approach that allows for error in bo th registries, includes covariates (here, the age of the mother) and e xplicitly accounts for our lack of knowledge about the complexity of t he relationships between the variables considered. Markov chain Monte Carlo methods are used to approximate the posterior. In the data consi dered here, the error rates of the two registries appear to be substan tial. Despite a strong relationship between maternal age and risk of D own's syndrome, the inclusion of the maternal age covariate does not s ubstantially change the overall estimates.