Bayesian latent class estimation of sensitivity and specificity parameters of diagnostic tests for bovine tuberculosis in chronically infected herds in Northern Ireland

• Published in: The veterinary journal (1997) Vol. 238; pp. 15 - 21
• Main Authors: Lahuerta-Marin, A., Milne, M.G., McNair, J., Skuce, R.A., McBride, S.H., Menzies, F.D., McDowell, S.J.W., Byrne, A.W., Handel, I.G., de C. Bronsvoort, B.M.
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01-08-2018
• Subjects: Bayesian latent class analysis; Bovine tuberculosis; Interferon-γ test; Sensitivity; Single intradermal cervical comparative tuberculin test; Specificity; Interferon-γ test; Sensitivity; Specificity; Single intradermal cervical comparative tuberculin test; Bayesian latent class analysis; Bovine tuberculosis
• ISSN: 1090-0233; 1532-2971
• DOI: 10.1016/j.tvjl.2018.04.019

•Performance characteristics of bovine tuberculosis (bTB) tests were determined in chronically infected herds.•Bayesian latent class analysis was used to estimate sensitivities and specificities.•The single intradermal cervical comparative tuberculin (SICCT) test may have a lower sensitivity than reported previously.•Combinations of SICCT and interferon-γ tests, and meat inspection data, affected estimates of sensitivity and specificity.•The ability of SICCT and interferon-γ tests to detect infected animals was lower in dairy herds than beef herds.
In the European Union, the recommended ante-mortem diagnostic methods for bovine tuberculosis (bTB) include the single intradermal cervical comparative tuberculin (SICCT) test and the interferon-gamma (IFN-γ) test as an ancillary test. The SICCT test has a moderate sensitivity (Se) and high specificity (Sp), while the IFN-γ test has good Se, but a lower Sp than the SICCT test. A retrospective Bayesian latent class analysis was conducted on 71,185 cattle from 806 herds chronically infected with bTB distributed across Northern Ireland (NI) to estimate the Se and Sp of the common ante-mortem tests and meat inspection. Analyses were also performed on data stratified by farming type and herd location to explore possible differences in test performance given the heterogeneity in the population. The mean estimates in chronically infected herds were: (1) ‘standard’ SICCT: Se 40.5–57.7%, Sp 96.3–99.7%; (2) ‘severe’ SICCT: Se 49.0%–60.6%, Sp 94.4–99.4%; (3) IFN-γ(bovine–avian) using a NI optical density (OD) cut-off difference of 0.05: IFN-γ(B–A)NI: Se 85.8–93.0%, Sp 75.6–96.2%; (4) IFN-γ(bovine–avian) using a standard ‘commercial’ OD cut-off difference of 0.1: IFN-γ(B–A)0.1: Se 83.1–92.1%, Sp 83.1–97.3%; and (5) meat inspection: Se 49.0–57.1% Se, Sp 99.1–100%. Se estimates were lower in cattle from dairy farms than from beef farms. There were no notable differences in estimates by location of herds. Certain population characteristics, such as production type, might influence the ability of bTB tests to disclose truly infected cases.