Combining livestock and wildlife surveillance data to assess the likelihood of freedom from bovine tuberculosis in New Zealand

Bovine tuberculosis (Tb) has historically occurred in livestock throughout most of New Zealand, and also in wildlife over about 40% of the country. Using a test-and-cull strategy to eliminate Tb from livestock, intensive control of the major wildlife host (brushtail possums), and surveillance of key spillover hosts (wild ferrets, pigs and deer), the Animal Health Board has reduced the number of infected cattle herds by over 90% since 1994. Many formerly infected areas appear to now be free of the disease, so a key question is whether the disease control programme can be stopped in those places. We have developed a Bayesian framework for combining spatially explicit data derived from livestock testing and surveys of Tb prevalence in key wildlife hosts, and using those data to updating or refine an estimate of the probability of Tb freedom from wildlife provided by a priori prediction derived from a model of how Tb levels in possum are likely to have been affected by the duration and intensity of control. This paper outlines the conceptual framework, and some preliminary “case-study” steps toward implementation.