Investigation of within- and between-herd variability of bovine leukaemia virus bulk tank milk antibody levels over different sampling intervals in the Canadian Maritimes.

Bulk tank milk (BTM) samples are used to determine the infection status and estimate dairy herd prevalence for bovine leukaemia virus (BLV) using an antibody ELISA assay. BLV ELISA variability between samples from the same herd or from different herds has not been investigated over long time periods. The main objective of this study was to determine the within-herd and between-herd variability of a BTM BLV ELISA assay over 1-month, 3-month, and 3-year sampling intervals. All of the Canadian Maritime region dairy herds (n = 523) that were active in 2013 and 2016 were included (83.9% and 86.9% of total herds in 2013 and 2016, respectively). BLV antibody levels were measured in three BTM samples collected at 1-month intervals in early 2013 as well as two BTM samples collected over a 3-month interval in early 2016. Random-effects models, with fixed effects for sample replicate and province and random effects for herd, were used to estimate the variability between BTM samples from the same herd and between herds for 1-month, 3-month, and 3-year sampling intervals. The majority of variability of BTM BLV ELISA results was seen between herds (1-month, 6.792 ± 0.533; 3-month, 7.806 ± 0.652; 3-year, 6.222 ± 0.528). Unexplained variance between samples from the same herd, on square-root scale, was greatest for the 3-year (0.976 ± 0.104), followed by the 1-month (0.611 ± 0.035) then the 3-month (0.557 ± 0.071) intervals. Variability of BTM antibody levels within the same herd was present but was much smaller than the variability between herds, and was greatest for the 3-year sampling interval. The 3-month sampling interval resulted in the least variability and is appropriate to use for estimating the baseline level of within-herd prevalence for BLV control programs. Knowledge of the baseline variability and within-herd prevalence can help to determine effectiveness of control programs when BTM sampling is repeated at longer intervals.