Economic decision analysis model of a paratuberculosis test and cull program
Collins, M.T.; Morgan, I.R.
Journal of the American Veterinary Medical Association 199(12): 1724-1729
1991
ISSN/ISBN: 0003-1488 PMID: 1813465 Document Number: 372914
A spreadsheet computer program was written to perform decision tree analysis for control of paratuberculosis, when testing all adults in a herd and culling all animals with positive test results. The program incorporated diagnostic test sensitivity, specificity, and test cost with the cost or value of each of the 4 possible outcomes; true-positive, true-negative, false-positive, and false-negative test results. The program was designed to repeat the analysis for the independent variable pretest paratuberculosis prevalence (0 to 100%). Model output was represented graphically as profit or loss in dollars against pretest prevalence. The threshold was defined as the pretest prevalence at which benefit cost equaled zero. Reed-Frost disease modelling techniques were used to predict the number of Mycobacterium paratuberculosis-infected replacement heifers resulting from infected cows during a control programme. Sensitivity analysis was performed on variables of the decision tree model: test sensitivity, specificity, test cost, and factors affecting the cost of paratuberculosis to a commercial dairy. A test and cull programme was profitable when paratuberculosis caused >= 6% decrease in milk production if the pretest prevalence was >6% test sensitivity was 50%, test specificity was 98%, and the testing cost was $4/cow. Test specificities >98% did not greatly affect the threshold for tests with a 50% sensitivity and costing $4/cow. Test sensitivity had minimal effect of the threshold. Using a diagnostic test with a 50% sensitivity and 98% specificity as an example, test cost was shown to affect the threshold prevalence at which the test and cull programme became profitable. Any test costing >= $8/cow was roughly equivalent in profit threshold. Given the characteristics of most diagnostic tests for paratuberculosis in use today, and assuming that paratuberculosis causes at least a 6% decrease in milk production of cows, the decision analysis model suggested a test and cull programme should be profitable when pre-test paratuberculosis herd prevalence is >5%.