Agreement between the caudal fold test and serological tests for the detection of Mycobacterium bovis infection in bison

Abstract

The objective of this study was to estimate agreement between the caudal fold test (CFT) and different serological tests for the detection of Mycobacterium bovis infection in bison by using prevalence-adjusted bias-adjusted kappa (PABAK). A total of 212 of wild wood bison from Wood Buffalo National Park were tested with the CFT as well as several serological tests: fluorescent polarization assay (FPA), multiantigen print immunoassay (MAPIA), rapid lateral-flow test (RT) and dual path platform test (DPP). For RT, 3 variations were conducted using 30μl of serum (RT 30), 20μl of serum (RT 20) and 20μl of serum considering only a strong reaction as positive (RT 20 ST). The McNemar's χ2 test was conducted to assess whether the proportion of positive test results to 2 different tests differed. Two measures of agreement between pair of tests were estimated: the Cohen's kappa statistic and PABAK. The apparent prevalence of tuberculosis in the sampled animals varied depending on the diagnostic test from 6.1% (FPA and DPP) to 47.2% (CFT). The prevalence estimated by CFT differed from the prevalence estimated by the other tests, whereas the prevalence estimated by FPA, MAPIA, RT 20 ST and DPP were not significantly different. The kappa and PABAK estimates calculated between CFT and the rest of the tests suggested poor to slight agreement between tests (k and PABAK<0.25 in all cases). The PABAK estimates for the pairwise combinations among serological tests were numerically greater than the kappa estimates (and significantly greater when FPA was compared to the rest of serological tests), and suggested substantial to almost perfect agreement (PABAK>0.75 in all cases). The disagreement between the skin and serological tests for the detection of M. bovis infection could be partly because the tests measure different immunological responses (cell-mediated vs. humoral) that are predominant at different stages of the infection, and partly due to inaccuracy of the tests. Further research is needed to evaluate the accuracy of diagnostic tests in order to establish a reliable case definition, combining different tests, to be used in the surveillance and control of tuberculosis in free-ranging bison populations.