Agent-based model for Johne's disease dynamics in a dairy herd

Abstract

Johne's disease is an infectious gastrointestinal disease in ruminants caused by Mycobacterium avium subsp. paratuberculosis that causes diarrhea, emaciation, decreased milk production and eventually death. The disease is transmitted in utero and via milk and colostrums to calves, and fecal-orally to all age classes. Financial losses due to the disease are estimated to be over $200 million in the US dairy industry. The goal of this study was to evaluate the cost effectiveness of control measures based on diagnosis with a sensitive ELISA, EVELISA. An agent-based, discrete time model was developed to simulate Johne's disease dynamics in a US dairy herd. Spatial aspects of disease transmission were taken into account by using six spatial compartments. The effects on disease prevalence were studied with and without transmission routes included in the model. Further, using the model, cost effectiveness of ELISA-based Johne's disease control was evaluated. Using the parameters we collected and assumed, our model showed the initial prevalence of Johne's disease (33.1 ± 0.2%) in the farm increased to 87.7 ± 1.7% in a 10 year-simulation. When ELISA-based control measures were included in the simulation, the increase in prevalence was significantly slowed down, especially when EVELISA was used. However, the level of the prevalence was still higher than the initial level after 10 year simulation even with the ELISA-based diagnostic intervention. The prevalence was further reduced when quarterly ELISA testing was included. The cost analysis showed that the quarterly ELISA and EVELISA testing could bring $44.8 and $51.5/animal/year more revenues, respectively, to a dairy farm.

Figure 1

The scheme by which individuals move through different spatial compartments. Modes of disease transmission that occur in each compartment are indicated. Green squares indicate location of animals.

Figure 2

Population dynamics of JD in a dairy farm simulated by the JD agent-based model. A: All transmission; B: no fecal-oral; C: no milk transmission; D: no colostrums transmission; E: no vertical transmission. Red: Total infected animals; Blue: Exposed animals; Green: Low shedding animals; Purple: high shedding animals.

Figure 3

Influence of assumed parameters on transmission of MAP. In this study, assumed values were used for infection rates for fecal-oral (F), milk (M), C (colostrums) and V (vertical) transmissions. Simulations were run with halved or doubled each infection rate. Each bar represents days that required for the total prevalence (exposed + low shedding + high shedding animals) to reach 50%. The error bars indicate standard deviation of data obtained by 10 simulations. Statistical significance among the group was detected by ANOVA test. Asterisks indicate that a statistical significant between the data and the original data (N) was detected by pair-wise t- test with Bonferroni and Holm adjustments.

Figure 4

Population dynamics of JD in a dairy farm simulated by the JD agent-based model. A: No testing; B: ELISA (once/year); C: EVELISA (once/year); D: ELISA (4 times/year); E: EVELISA (4 times/year). Red: Total infected animals; Blue: Exposed animals; Green: Low shedding animals; Purple: high shedding animals.