Diagnostic assessment without cut-offs: Application of serology for the modelling of bovine digital dermatitis infection

摘要: Abstract Bovine digital dermatitis (BDD) is an epidermitis which a leading cause of infectious lameness. The only recognized diagnostic test foot inspection, labour-intensive procedure. There no universally recognized, standardized lesion scoring system. As small lesions are easily missed, inspection has limited sensitivity. Furthermore, interpretation subjective, and prone to observer bias. Serology more convenient carry out potentially sensitive indicator infection. By carrying 20 serological assays using lesion-associated Treponema spp. isolates, three serogroups were identified. reliability the tests was established by assessing level agreement concordance correlation coefficient. Subsequently, ELISA suitable for routine use developed. benchmark validation conventionally determination key parameters, sensitivity specificity. This requires imposition cut-off point. For with outcomes on continuous scale, degree result differs from this disregarded. Bayesian statistical methodology been developed enables assay also be interpreted thereby optimizing information inherent in test. Using cross-sectional study dataset carried 8 representative dairy farms UK, probability infection, P ( I ), each individual animal estimated absence ‘Gold Standard’ modelling as latent variable determined status, L well serology, S . Covariate data (foot hygiene score age) utilized estimate ) when performed. Informative prior distributions elicited where possible. model predictive inference, computing estimates independently data. A detailed informative analysis farm-level distribution infection could thus Also, biases associated subjective status minimized. Model outputs showed that young stock unlikely infected, whereas cows tended have high or low probabilities being infected. Estimates considerably higher animals than those without. Associations identified between both covariates cows, but not stock. Under condition assumptions valid larger population, results work can generalized inference.