Estimating the probability of freedom of classical swine fever virus of the East-Belgium wild boar population.

Abstract A report of the Scientific Committee on Animal Health and Animal Welfare of the European Commission (CEC, 1999.) includes recommendations for setting up monitoring programmes for classical swine fever (CSF) infection in a wild-boar population, based on the assumption that one would detect at least 5% prevalence in a CSF-infected wild-boar population. This assumption, however, is not science based. We propose an alternative method to provide evidence for a wild-boar population being free of CSF and evaluate the efficiency of a surveillance programme that was implemented in Belgium in 1998. In our study, the probability of freedom of CSF-virus was estimated based on 789 samples; these were collected from wild-boars within the surveillance programme (within the three provinces which include 95% of the Belgian wild-boar population) and examined by three diagnostics methods (antibody detection, virus detection and virus RNA detection). A Bayesian framework was used for the estimation, accounting for the diagnostic test characteristics without the assumption of the presence of a gold standard. The median probability of freedom of CSF-virus was estimated at 0.970, with a 95% credibility interval of 0.149–1.000. Independent on the choice of the prior information, the posterior distributions for the probability of freedom of CSF-virus were always skewed close to the upper boundary of 1. This represents a big gain of knowledge since we did not use any prior information for the probability of freedom of CSF-virus and took the uncertainty about the accuracy of the diagnostic methods into account.