A model of bovine tuberculosis in the badger Melesmeles: the inclusion of cattle and the use of a live test


G.C. Smith, Central Science Laboratory, Sand Hutton, York YO41 1LZ, UK (fax 01904 462111; e-mailg.smith@csl.gov.uk).


  • 1 An individual-based stochastic simulation model was used to investigate the control of bovine tuberculosis (TB) in the European badger Meles meles by using a live test to determine the presence of infection. The model was an extension of earlier models, and nearly all population and epidemiological parameters were derived from one study site.
  • 2 This is the first TB model to examine sex differences in disease epidemiology, and the transmission of TB from badgers to cattle. The latter is an essential step if reactive badger control strategies are to be modelled.
  • 3 Heterogeneity was introduced to the simulation model by the use of a carrying capacity, which defined the maximum number of breeding females per social group.
  • 4 The prevalence of TB, and the number of simulated cattle herd breakdowns, was reduced for all control strategies using a live test, namely localised culling, ring culling and proactive culling. However, only proactive culling resulted in a marked reduction in these values within a few years.
  • 5 If trapping efficacy was increased above its current value (80%), this did not improve the effectiveness of these culling strategies.
  • 6 If the number of individual badgers caught and tested per social group was doubled from two to four animals per group, then the overall level of effectiveness of these strategies could be doubled.
  • 7 The effectiveness could be improved if the sensitivity of the live test was increased, but did not continue to show an improvement above a sensitivity of about 70%.
  • 8 Given the constraints of the current live test sensitivity (41%) and a trapping efficacy of 80%, proactive culling, following the testing of four individuals per group, led to an average of three cattle herd breakdowns per year in the simulation, compared with an average of 31 per year when simulating the live test trial as used between 1994 and 1996.