Age-adjusted ELISA cutoffs for Johne's disease in young adult dairy cattle using Bayesian nonparametric models

Johne's disease (JD), caused by Mycobacterium avium subsp. paratuberculosis, presents significant economic and health challenges for dairy farms globally. Enzyme-linked immunosorbent assay (ELISA) is commonly utilized for herd-level JD surveillance because of its cost-effectiveness and rapid execution, although its low sensitivity in young cattle complicates control efforts. This retrospective open-cohort study evaluated whether age-adjusted interpretation of longitudinal ELISA sample-to-positive (S/P) ratios during young adult age (24 to 48 months of age) could support risk stratification for subsequent JD seroconversion in a commercial dairy herd. A total of 500 dairy cows in a single herd with repeated serum ELISA testing between July 2018 and February 2026 were included. Age-adjusted cutoff patterns were estimated using linear, quadratic, and flexible spline Bayesian nonparametric models based on test-negative observations. Cows were classified according to longitudinal S/P ratio patterns during young adult age as below cutoff, non-consecutive exceedance, or consecutive exceedance. A landmark survival analysis from 48 months was performed to evaluate seroconversion risk. Overall, 43 of 500 cows (8.6%) became JD-seropositive. The performances across three models were comparable, with a high negative predictive value (96.5%) but low positive predictive value (about 11%). Using the linear model, cows with consecutive exceedances showed a higher hazard of seroconversion (hazard ratio 4.61, 95% CI: 1.22-17.40, p = 0.024) compared to the below cutoff group. These findings suggest that age-adjusted longitudinal S/P ratio patterns may be practical for subsequent JD seroconversion risk. However, the results are exploratory and require validation in diverse herds and against confirmed infection outcomes.