Evaluation of three bulk tank milk enzyme-linked immunosorbent assays to estimate within-herd prevalence of bovine leukosis virus in dairy farms.

The aim of this study was to estimate the bovine leukosis virus (BLV) herd-level and within-herd prevalence (WHP) in dairy herds and to assess the ability of three bulk tank milk (BTM) ELISA-Ab tests (SVANOVIR, Bovichek, and IDEXX) to predict the BLV WHP. A cross-sectional study was performed on a convenience sample of 93 dairy herds from Québec, Canada, where individual milk samples from all lactating cows and a BTM sample were collected. Individual milk ELISA-Ab results (n = 7612 cows) from a previously validated kit were incorporated into a two-stage hierarchical Bayesian latent class model to estimate BLV herd-level prevalence and WHP. To mitigate potential test saturation, we evaluated the three BTM ELISA-Ab tests with and without sample dilutions. Adjusted BTM ELISA-Ab optical density values were linked to the WHP estimates using a zero-inflated beta regression, and dilution-specific predictive models were compared. The herd-level BLV prevalence was estimated at 87% (95% Bayesian credible interval [BCI]: 78, 93), and BLV WHP ranged from 0% to 90%. The best predictive model-dilution combinations were SVANOVIR 1/10 (nonmonotonic curvilinear), Bovichek 1/5 (linear), and IDEXX 1/50 (nonmonotonic curvilinear). Across tests, optical density values increased with WHP, consistent with higher antibody concentrations in bulk milk. BLV remains highly prevalent at the herd level, with substantial variability in WHP among positive herds. These results indicate that BLV WHP can be reliably estimated from a single BTM sample, reducing the need for individual cow testing.