Simulated economic effects of improving the sensitivity of a diagnostic test in paratuberculosis control.

Low sensitivity (Se) of diagnostic tools is often mentioned as a major problem in the control of paratuberculosis (PTB) and much effort is put into the improvement of these tests. The hypothetical perspectives of improving the Se of a milk-antibody ELISA (hereafter: milk-ELISA) used in test-&-cull strategies against PTB in dairy cattle were investigated by simulations. The current Se varies between 10 and 80%, increasing with increasing lactation stage, parity and infection stage. We simulated the effects on a dairy herd's production of improving this Se to 80% (independent of these factors) and assumed no concomitant decrease in specificity. By using a PTB model called PTB-Simherd, 12 scenarios were simulated to study three test-&-cull strategies in each of four herds with 200 dairy cows. To show the maximal effect of using test-&-cull with such an improved test we simulated three strategies: (1) no testing, (2) testing with milk-ELISA test with the current Se and culling of positive cows immediately and (3) testing with milk-ELISA test with a Se improved to 80% and culling positive cows immediately. The four herds were defined by a moderate (25%) or high (80%) initial true within-herd prevalence (including young stock), and a poor or good heat-detection success of 40 or 60%. We assumed that these factors influenced the effects of improving the Se of the milk-ELISA. Management both concerning calf management and in general was specified to represent a typical Danish herd. Using an improved milk-ELISA was predicted to reduce the prevalence of PTB more effectively than the current ELISA, and over 10 years bring the production of a herd with moderate initial prevalence up to a production level comparable to a non-infected herd (unlike if the current ELISA had been used). In a herd with high initial prevalence (80%) milk production was increased more by using the improved milk-ELISA, but after 10 years the replacement rate was still very high causing problems with having enough recruitment animals-especially in high-prevalence herds with poor reproductive performance. Economically important measurements in all four herds benefited from the improvement of the test over a 10-year period. However, in the first 3-5 years the improved test would be more expensive to use than the current test, due to increased replacement (reduced net annual revenue per cow euro15 on average) but after that, net annual revenue increased continuously; after 10 years it was euro70-90 higher, than if the current milk-ELISA was used. Also, the milk-ELISA test with its current Se seemed to be profitable already after 2 years in high-prevalence herds using a test-&-cull strategy based on the milk-ELISA alone.