A framework for multicriteria decision-aid analyses in animal health surveillance applied to periodic screening for French bovine tuberculosis.

In animal health surveillance, decision-makers must allocate limited financial and human resources, choosing appropriate protocols that consider technical, economic and human aspects (i.e. test sensitivity, cost benefits and policy acceptability respectively). Choosing in an objective manner and considering all these criteria can be challenging, especially where criteria have a tendency to be at odds with one another. In France, there are several mandatory protocols implemented to screen for bovine tuberculosis (TB), each with advantages and drawbacks concerning effectiveness, cost and acceptability. Previous studies have developed scenario tree models in order to evaluate mandatory periodic screening protocols and alternative protocols. Using these previously developed models, we estimated protocol sensitivity, costs at the level of State and farmer, probability of false suspects, and probability of culling an uninfected animal, which influence stakeholders' acceptability. We then assessed the level of difficulty for protocol implementation for veterinarians, farmers and State through the use of surveys. Using these criteria, we rank the protocols with the PROMETHEE method, a multicriteria decision-aid method, by considering the relative importance of each criterion from the decision-maker point of view in four administrative areas with contrasting epidemiological context. This method can be considered a tool to aid decision-makers in choosing the appropriate protocol to apply to a heard while considering the technical and socio-economic facets of the problem. Additionally, by adapting the criteria to specific issues with regards to decision-making, there is potential for applying the PROMETHEE method to other animal health surveillance problems.

Relative Contribution of Each Component of the French Ante-Mortem Surveillance System for Bovine Tuberculosis in Its Overall Sensitivity.

The aim of this study was to assess the contribution to the sensitivity of the French ante-mortem surveillance system for bovine tuberculosis in cattle of each of the system's components (periodic screening, epidemiological investigations, and screening exchanged animals), on a local scale defined by administrative areas. These components were individually assessed in previous studies by scenario tree modeling. We used scenario tree modeling at the herd level and combined the results to evaluate the overall sensitivity of the ante-mortem surveillance system. The probability to detect at least one infected herd was consistent with the location of the outbreaks detected in 2016. In areas with a high apparent incidence, the probability of an infected herd to be detected was satisfactory (for an infected herd there was a 100% probability to be detected over a two-year period). Periodic screening was the most important component for the overall sensitivity in infected areas. In other areas, where periodic screening had stopped, tracing-on epidemiological investigation was the most sensitive component of the system. Screening exchanged animals had a negligible part in the overall sensitivity of the surveillance system.