Evaluation of PFC and AGID as flockscreening tests for OJD National Ovine Johne's Disease Control and Evaluation Program

Summary

A Monte Carlo simulation model was developed to estimate the sensitivity of pooled faecal culture (PFC) and the agar-gel immuno-diffusion test (AGID) as flock-screening tests for ovine Johne’s disease under a range of scenarios. The flock-sensitivity of a test is the level of confidence of detecting a specified prevalence of infection. Outputs from the model are probability distributions for the flock-sensitivities of the two tests for a given scenario. The model allows direct comparison of the tests under a variety of conditions, and considers the effects of:

Variations in animal-level sensitivity of the tests;
Variations in flock size, sample size and prevalence of infection;
Variations in animal-level sensitivity with type of lesion (paucibacillary or multibacillary);
Variations in the proportions of animals with paucibacillary/multibacillary lesions; and
Reduction in animal-level sensitivity due to pooling effects in pooled faecal culture;
Uncertainty as to the true values of input parameters such as animal-level test sensitivity and proportion of sheep with different lesion types.
Comparison of model outputs with results from a field trial of pooled faecal culture, and with calculated estimates, confirmed that the model provides reasonable estimates of flock-sensitivity of the tests.

The model was used to estimate:

the flock-sensitivities of current testing strategies;
sample sizes required for pooled faecal culture and serology to provide desired levels of flock-sensitivity for surveillance and market-assurance testing; and
the sample size required for serology to provide equivalent flock-sensitivity to pooled faecal culture under a range of scen​arios.
The mean flock-sensitivities for a Check Test (sample size = 100) were 67% and 42% for PFC and AGID respectively, and for a Sample Test (sample size = 350 for PFC, 500 for AGID) were 98% and 93% respectively. When large flocks were sampled, sample sizes of 300, 350 and 450 provided a flock-sensitivity for PFC of about 95%, 98% and 99% respectively, to detect infection if present at a prevalence of 2% in the sampled population. Sample sizes for the AGID to provide equivalent sensitivity to PFC were generally 2 – 3 times the PFC sample size, depending on the assumed animal-level sensitivities of the tests.

When whole-flock testing was simulated, AGID flock-sensitivity was generally poor for smaller flock-sizes and low prevalence or animal-level sensitivities, whereas PFC flock-sensitivity remained high. The flock-sensitivity for PFC was ≥98% for all combinations of flock size, prevalence, percentage of paucibacillary lesions and animal-level sensitivities tested compared to ≥68% for the AGID.

Although the AGID appears to perform reasonably well in higher prevalence flocks, its flock-sensitivity in low prevalence or recently infected flocks is likely to be very low, unless sample sizes 2 – 3 times those used for PFC are used. This is particularly important in Australia at present, as the majority of flocks being investigated outside the Residual Zone (endemic area) are likely to be relatively recently infected and still have only a low prevalence of infection. In these circumstances, PFC should be the preferred test, and larger sample sizes or whole-flock testing should be considered to maximise flock-sensitivity.

In higher prevalence flocks, such as many of those in the Residual Zone, the AGID will provide a satisfactory flock-sensitivity and a more-rapid result than PFC, particularly if biased sampling is used to maximise animal-level sensitivity and provided sample sizes are adequate.