Test characteristics of the tuberculin skin test and post-mortem examination for bovine tuberculosis diagnosis in cattle in Northern Ireland estimated by Bayesian latent class analysis with adjustments for covariates

Abstract

The single intradermal comparative cervical tuberculin (SICCT) test and post-mortem examination are the main diagnostic tools for bovine tuberculosis (bTB) in cattle in the British Isles. Latent class modelling is often used to estimate the bTB test characteristics due to the absence of a gold standard. However, the reported sensitivity of especially the SICCT test has shown a lot of variation. We applied both the Hui–Walter latent class model under the Bayesian framework and the Bayesian model specified at the animal level, including various risk factors as predictors, to estimate the SICCT test and post-mortem test characteristics. Data were collected from all cattle slaughtered in abattoirs in Northern Ireland in 2015. Both models showed comparable posterior median estimation for the sensitivity of the SICCT test (88.61% and 90.56%, respectively) using standard interpretation and for post-mortem examination (53.65% and 53.79%, respectively). Both models showed almost identical posterior median estimates for the specificity (99.99% vs. 99.80% for SICCT test at standard interpretation and 99.66% vs. 99.86% for post-mortem examination). The animal-level model showed slightly narrower posterior 95% credible intervals. Notably, this study was carried out in slaughtered cattle which may not be representative for the general cattle population.

Bayesian Estimation of Diagnostic Accuracy of Three Diagnostic Tests for Bovine Tuberculosis in Egyptian Dairy Cattle Using Latent Class Models

“…In the current study, Se estimates of SCT were relatively similar to 95.2% Se reported in dairy cows in Thailand [39], and higher than the 88.6% [28] and 57.7% [27] Se reported in Northern Ireland, the 80.3% Se reported in some French departments [40], and the 60.8% Se reported in Irish dairy farms [41]. However, it should be noted that the high Se estimate of SCT reported here could be due to the features of the test population.…”

“…Since, to the best of our knowledge, no available literature exists on the diagnostic Se and Sp estimates for M. bovis infection from similar farm management conditions in developing countries, vague priors for Se [ Se 1 ∼ uni f orm(0.560, 0.933)] and Sp [ Sp 1 ∼ uni f orm(0.593, 1.000)] of SCT (test 1) were specified based on diagnostic information from developed settings [10,27,28].…”

Tuberculin PPD Potency Assays in Naturally Infected Tuberculous Cattle as a Quality Control Measure in the Irish Bovine Tuberculosis Eradication Programme

“…When the United Kingdom (UK) switched from using Weybridge Tuberculin PPD to Dutch Tuberculin PPD, it was found that the tuberculin manufacturing source influenced both the Se and Sp of the SICTT (63). Data from international studies indicate a sensitivity range, at individual animal level, of 68–96.8% and 96–98.8% specificity for the CFT (80–91% sensitivity and 75.5–96.8% specificity), for SIT and, for the SICTT (55.1–93.5% sensitivity and 88.8–100% specificity) (7, 60, 64). The caudal fold has been repeatedly determined to be the least sensitive site available for intradermal test and hence, the CFT requires higher potency tuberculin to achieve an acceptable Se for use in bTB control/eradication programmes; the mid third of the neck proved to be the most sensitive site for the intradermal test (4).…”

Bayesian Methods in Veterinary Clinical Epidemiology

“…In contrast to the Hui-Walter (HW) model, the LR latent class model allows inclusion of multilevel data. This approach has been applied under a Bayesian framework in different epidemiologic studies (e.g., McInturff et al, 2004;Koop et al, 2013;Hartnack et al, 2013;Paul et al, 2014;O'Hagan et al, 2019;Fernandes et al, 2019) and yields not only estimates for test characteristics but also estimates for the effect of the risk factors. Studies that used both HW and LR showed that LR models tended to provide more precise posterior estimates for test sensitivity and specificity (Koop et al, 2013;O'Hagan et al, 2019).…”

“…This approach has been applied under a Bayesian framework in different epidemiologic studies (e.g., McInturff et al, 2004;Koop et al, 2013;Hartnack et al, 2013;Paul et al, 2014;O'Hagan et al, 2019;Fernandes et al, 2019) and yields not only estimates for test characteristics but also estimates for the effect of the risk factors. Studies that used both HW and LR showed that LR models tended to provide more precise posterior estimates for test sensitivity and specificity (Koop et al, 2013;O'Hagan et al, 2019). However in empirical examples, evaluation of these two methods with various settings of population characteristics can be difficult concerning the amount of data collection.…”