Accurate estimation of the probability of observing a reactor to the skin test in officially tuberculosis free herds in a high prevalence region

Skin testing is a crucial tool for eradication of bovine tuberculosis (bTB) and maintenance of the officially tuberculosis-free (OTF) status at the herd and regional level. According to previous estimates, the specificity of skin tests under field conditions can be very high level at the animal level, close to 100%, though lower values have been also reported. Given that the cumulative effect of imperfect specificity can substantially increase the herd-level probability of reactors under OTF conditions, here we aimed to accurately estimate the performance of the single intradermal tuberculin test (SIT) in OTF herds in a high prevalence area using a Bayesian hierarchical model. The analysis of data from 3,039 cattle in 42 OTF beef herds revealed a median probability of disclosing a reactor using a severe interpretation of 1.24% (95% PPI = 0.083-8.3%), with an increasing risk in older animals. Still, simulation of the outcomes of the test at the herd-level revealed a median probability of observing ≥1 reactors per herd of 70.8%, for a herd-level specificity of 29.2% in our population (in contrast with field data indicating that >98% of Spanish herds test negative annually). Infection could not be confirmed post-mortem in any of the reactors. Our findings indicate that, while SIT remains a reliable tool for animal-level testing in OTF herds, interpretation of results at the herd-level can be complicated, particularly when potential exposure to bTB cannot be completely ruled out. Risk-based surveillance strategies could help to minimise the impact of precautionary OTF status suspensions.