Open Access Open Badges Research article

Assessment of the diagnostic sensitivity and specificity of an indirect ELISA kit for the diagnosis of Brucella ovis infection in rams

Anne Praud1, Jean-Luc Champion2, Yannick Corde3, Antoine Drapeau3, Laurence Meyer4 and Bruno Garin-Bastuji3*

Author affiliations

1 National Veterinary School of Alfort (ENVA) / French Agency for Food, Environmental and Occupational Health Safety (ANSES), USC Epidemiology of Animal Infectious Diseases Unit (Epi-MAI), 94700, Maisons-Alfort, France

2 Groupement de Défense Sanitaire des Alpes de Haute Provence, 04000, Digne les Bains, France

3 ANSES, Animal Health Laboratory, EU/OIE/FAO Brucellosis Reference Laboratory, 94706, Maisons-Alfort, France

4 INSERM, Centre for research in Epidemiology and Population Health (CESP), U1018, Faculté de Médecine Paris-Sud, Le Kremlin-Bicêtre; AP-HP, Hopital Bicêtre, Epidemiology and Public Health Service, Université Paris-Sud, 94276, Le Kremlin-Bicêtre, France

For all author emails, please log on.

Citation and License

BMC Veterinary Research 2012, 8:68  doi:10.1186/1746-6148-8-68

Published: 28 May 2012



Brucella ovis causes an infectious disease responsible for infertility and subsequent economic losses in sheep production. The standard serological test to detect B. ovis infection in rams is the complement fixation test (CFT), which has imperfect sensitivity and specificity in addition to technical drawbacks. Other available tests include the indirect enzyme-linked immunosorbent assays (I-ELISA) but no I-ELISA kit has been fully evaluated.

The study aimed to compare an I-ELISA kit and the standard CFT. Our study was carried out on serum samples from 4599 rams from the South of France where the disease is enzootic. A Bayesian approach was used to estimate tests characteristics (diagnostic sensitivity, Se and diagnostic specificity, Sp). The tests were then studied together in order to optimise testing strategies to detect B. ovis.


After optimising the cut-off values in order to avoid doubtful results without deteriorating the concordance between the results of the two tests, the I-ELISA appeared to be slightly more sensitive than CFT (Se I-ELISA = 0.917 [0.822; 0.992], 95% Credibility Interval (CrI) compared to Se CFT = 0.860 [0.740; 0.967], 95% CrI). However, CFT was slightly more specific than I-ELISA (Sp CFT = 0.988 [0.947; 1.0], 95% CrI) compared to Sp I-ELISA =0.952 [0.901; 1.0], 95% CrI).

The tests were then associated with two different interpretation schemes. The series association increased the specificity of screening and could be used for pre-movement testing in rams from uninfected flocks. The parallel association increased sequence sensitivity, thus appearing more suitable for eradicating the disease in infected flocks.


The high sensitivity and acceptable specificity of this I-ELISA kit support its potential interest to avoid the limitations of CFT. The two tests could also be used together or combined with other diagnostic methods such as semen culture to improve the testing strategy. The choice of test sequence and interpretation criteria depends on the epidemiological context, screening objectives and the financial and practical constraints.

Brucella ovis; Diagnostic tests; CFT; I-ELISA; Sensitivity; Specificity; Bayesian approach