Table 6.

Sensitivity and specificity estimates for MAP diagnosis in 913 beef cows in the absence of a gold standard using a three-test, two population Bayesian latent class model with non-informative priors and using a subset of the data including only cows for which all three diagnostic tests were completed.

	Median	95% CrI	Proportion posterior distribution overlap a	Mean difference in posterior estimates b
Sensitivity
Pooled PCR	0.54	0.36, 0.72
Individual PCR	0.96	0.80, 1.00
ELISA	0.36	0.22, 0.52
Pooled PCR vs. Individual PCR			0.02	−0.40
Pooled PCR vs. ELISA			0.19	0.17
Individual PCR vs. ELISA			0.004	0.57
Specificity
Pooled PCR	0.999	0.998, 1.00
Individual PCR	0.98	0.96, 1.00
ELISA	0.98	0.96, 0.99
Pooled PCR vs. Individual PCR			0.03	0.02
Pooled PCR vs. ELISA			0.002	0.02
Individual PCR vs. ELISA			0.59	0.003
Prevalence*
West	0.04	0.02, 0.06
East	0.13	0.08, 0.17
Covariance
Se (individual/pooled)	0.00	−0.02, 0.04
Sp (individual/pooled)	0.00	0.00, 0.00

^aProportion of overlap between the posterior chains for estimates of parameters of interest from the BLCM where 0 reflects completely separate and 1 reflects completely overlapping distributions (49).

^bMean of difference between pairs of posterior samples from the BLCMs (first parameter listed-second parameter).

^*The prevalence reported here reflects a non-representative subsample of the overall data (913 of 3,171 total samples selected).