Summary
1. EXECUTIVE SUMMARY
The pathogenesis, epidemiology and options for control of Johne's disease in sheep were reviewed and mathematical models developed to simulate the spread of Johne’s disease within infected flocks, and between flocks on a regional basis. The models also allow the evaluation and comparison of various control options at both flock and regional levels.
1.1 OJD Flock Model
This model simulates spread of OJD within an infected flock. Within the simulated flock, sheep progress between states of susceptible (SUS), Immune (IM), Latent (LT), Light shedding (LS), Heavy shedding (HS) and Clinical cases (CC) and age groups according to defined rules and the input values specified for the simulation. Animals may die or be culled according to the rules of the model.
The user can specify the numbers of sheep in each age-group and state at the beginning of the simulation, as well as values for the various parameters affecting transmission and progression of infection. The model simulates spread within the flock for a specified number of years. Each simulation can be run for a number of iterations, with Monte Carlo simulation used to provide random variation between iterations.
The effects of OJD on flock productivity can be simulated as effects on lambing percentage and on wool production and quality, in addition to the losses due to mortality/premature culling of affected sheep. Although these effects have still not been quantified, the model allows examination of ‘what-if’ scenarios to estimate the cost of OJD in infected flocks.
The model also allows simulation of various control strategies, including management control, vaccination and test & cull options. Simulations with and without the selected control options are run simultaneously and are compared within the model.
Results for each simulation are summarised as the mean, standard deviation and percentiles of the multiple iterations for each output variable of interest. Outputs from the model include the percentage of sheep in each state/age group at the end of each simulated year, the number and percentage of sheep dying from OJD each year, numbers of sheep tested or vaccinated each year and the amount of wool produced and return from wool sales each year.
1.2 OJD Regional Model
This model simulates spread of OJD between properties within a region. Properties are simulated on a grid of farms, and spread may occur by local spread between adjoining farms, or through the movement of replacement sheep between farms. Spread is simulated for a specified number of years, and the simulation is repeated for a number of iterations to provide a distribution of output values. Each iteration will produce different output values because of the use of probability distributions for some input values, and Monte Carlo simulation to randomise some model processes.
The model simulates spread within each infected flock once it becomes infected, as well as simulating the effect of vaccination on prevalence in the flock. Therefore, the longer a flock has been infected the higher the prevalence and the greater the risk of spread. Conversely, once a flock starts vaccinating, prevalence declines progressively, with a corresponding decline in the risk of spread to other flocks.
Control options that can be used include vaccination, surveillance and quarantine and movement controls on sale/purchase of sheep. The effect of various control strategies can be evaluated by comparing simulations with and without the strategy in place, or with variations of several proposed strategies.
Results for each simulation are summarised as the mean, standard deviation and percentiles of the multiple iterations for each output variable of interest. Outputs from the model include the percentage of flocks infected at the end of each simulated year, and the percentage of flocks tested, quarantined or vaccinated each year.
1.3 Validation and sensitivity analysis of models
Both models were validated by comparison of model outputs with existing data, and appear to provide realistic estimates of the spread of infection both within and between flocks, depending on the input values used. Because of the use of Monte Carlo methods, results varied considerably between iterations with the same input values, due to random chance. This was particularly apparent with the regional spread model, where the prevalence of infected flocks was closely linked to the simulated number of infected studs.
Model output also varied substantially between simulations depending on the input values used. For both models, the most important variables contributing to this variation where age and breed susceptibility, contact rate between susceptible sheep and potentially infected faeces and the probability of an infected sheep progressing from the latent (LT) state to become a light shedder (LS).
1.4 Conclusion
There is still inadequate data available to accurately estimate the true values for many of the parameters involved in spread and progression of OJD infection at both within-flock and regional levels. However, these models provide an opportunity to investigate the effects of assumed realistic values on the rate of spread of infection. In addition, the models allow estimation of the likely costs of disease, and the effectiveness and cost-benefit of proposed control strategies, particularly at the farm level.
As more precise estimates of the values of key parameters become available, the models will allow a rapid assessment of the likely impact of these values on our understanding of the disease.