Review Article

Accuracy of Tests for Diagnosis of Animal Tuberculosis: Moving Away from the Golden Calf (and towards Bayesian Models)

Table 4

Average posterior median (APM) and 95% average posterior probability intervals (APPI) of the Se and Sp of postmortem diagnostic tests for tuberculosis in the articles reviewed.

TestNSe estimatesAPM and 95% APPI Se estimates (%)§Sp estimatesAPM and 95% APPI Sp estimates (%)§Species

Culture91779.2 (64.8–88.7)1799.8 (98.0–100)
 Tissue81579.9 (65.0–94.0)1599.9 (99.3–100)Cattle, wild boar, swine, meerkat, badger
 Milk and blood1215.0 (6.6–30.0)297.5 (88.0–100)Cattle
Direct PCR102180.6 (54.4–92.3)2197.6 (95.0–99.7)
 Tissue91982.9 (69.1–92.3)1997.6 (95.4–99.7)Cattle, wild boar, swine,
 Milk and blood1248.5 (27.0–77.5)293.0 (69.0–99.5)Cattle
Meat inspection5953.7 (49.9–54.8)999.1 (98.4–99.7)Cattle
Histopathology21077.8 (55.2–94.6)1099.0 (96.8–100)Cattle
Culture+meat inspection1374.6 (62.0–87.1)398.5 (97.7–99.2)Cattle

Number of articles that evaluate each test. Number of trials from which sensitivity and specificity estimates were extracted. §Sensitivity and specificity average posterior median and 95% average posterior probability interval of each test. Nature of the test is indicated in bold.