Evaluation of test characteristics of 2 ELISA tests applied to bulk tank milk and claw-trimming records for herd-level diagnosis of bovine digital dermatitis using latent class analysis

Abstract

Bovine digital dermatitis (DD) is an infectious claw disease with a negative effect on animal welfare and production. Treponema spp. is the main causative agent, and infected animals produce specific antibodies. Our aim was to estimate sensitivity (Se) and specificity (Sp) of 2 ELISA research tests, Medicago's ELISA test and GD Animal Health's in-house ELISA test, for detection of DD-associated Treponema antibodies in bulk tank milk. We used bulk tank milk samples from 154 Norwegian dairy cattle herds, 96 from an expected high-prevalence region and 58 from a low-prevalence region. Both tests were evaluated separately against herd-level (aggregated) claw-trimming records extracted from the Norwegian Dairy Herd Recording System using latent class models in a Bayesian analysis. Cutoff values were selected using an explorative approach, and both noninformative priors for all parameters and informative β priors for distribution of Se and Sp of claw trimming were explored. The estimated (median) true herd-level prevalence of digital dermatitis varied between 24 and 30% in the high-prevalence region and between 3 and 6% in the low-prevalence region. For Medicago's ELISA test, an Se (95% posterior credible interval) of 0.57 (0.32; 0.94) could be achieved without compromising Sp, and for GD Animal Health's in-house ELISA test, an Se of 0.60 (0.37; 0.92) was achieved. Our study showed that both ELISA tests can detect antibodies against DD-associated Treponema spp. in bulk tank milk. However, neither of the 2 ELISA tests produced satisfactory sensitivity without compromising specificity. Based on these results, inspection at claw trimming in a chute is necessary for surveillance and control of DD at the herd level in Norway, although these ELISA tests of bulk tank milk might be a useful supplement.