Elsevier

Veterinary Microbiology

Volume 125, Issues 1–2, 15 November 2007, Pages 187-192
Veterinary Microbiology

Short communication
Evaluation of three serological tests for brucellosis in naturally infected cattle using latent class analysis

https://doi.org/10.1016/j.vetmic.2007.05.012Get rights and content

Abstract

Serological methods are traditionally used in diagnosis of brucellosis. However, the comparative performance of these tests and their accuracy under the local environment in Zambia has not been assessed. Thus, the objective of our study was to evaluate the diagnostic performance of three serological tests for brucellosis; Rose Bengal Test (RBT), competitive ELISA (c-ELISA) and Fluorescence Polarisation Assay (FPA) in naturally infected cattle in Zambia without an appropriate reference test to classify animals into truly infected and non-infected.

Serological test results from a study to determine sero-prevalence were used to compare the performance of RBT, c-ELISA and FPA in diagnosing brucellosis in traditional cattle. Since none of the tests can be seen as a perfect reference test or gold standard, their performance in a population of naturally infected cattle was evaluated using latent class analysis which allows the sensitivity (Se) and specificity (Sp) to be estimated in the absence of a gold standard. The highest Se was achieved by the c-ELISA (97%; Credible Posterior Interval (CPI) = 93–100%) and the highest Sp by the FPA (93%; CPI = 85–99%), conversely these tests also had the lowest Sp and Se, respectively, with the RBT performing well in both the Se (93%; CPI = 84–98%) and Sp (81%; CPI = 61–97).

Introduction

Brucellosis is a disease of world-wide economic and public health concern, which is endemic in Sub-Saharan Africa, including Zambia (Acha and Szyfres, 2005, Bishop et al., 1994, McDermott and Arimi, 2002, Pappas et al., 2006). While isolation and identification of Brucella spp. is defined as the ‘gold standard’ for diagnosis of brucellosis (Alton et al., 1988, OIE, 2004), indirect testing of anti-Brucella spp. antibodies in serum, milk and other clinical specimen are routinely used in brucellosis control and eradication (Alton et al., 1988, Bishop et al., 1994, K.H. Nielsen et al., 1996, OIE, 2004). Previous studies of brucellosis in Zambia used the Rose Bengal Test (RBT) and Serum Agglutination Test (SAT) (Bell et al., 1977, Gallagher, 1973, Ghirotti et al., 1991, Orino et al., 1994, Suzuki et al., 1996). However, the performance the RBT or SAT as well as the more recent tests such as competitive ELISA (c-ELISA) and Fluorescence Polarisation Assay (FPA) has not been investigated in naturally infected cattle population reared under Zambian conditions. Thus, the objective of this study was to evaluate the diagnostic performance of three serological tests for brucellosis (RBT, FPA and c-ELISA) in naturally infected cattle. Since none of these tests could be considered as a perfect reference test, a latent class analysis allowing the estimation of test characteristics in the absence of a ‘gold standard’ was applied.

Section snippets

Sampling scheme

The sera used in this study were part of a survey conducted between August 2003 and November 2005 (Muma et al., 2006). Sera were randomly collected from traditionally reared cattle from three different areas of Zambia. Only animals aged ≥2 years and without history of vaccination against brucellosis (S19) were included. Sampling was done in two phases. The first phase comprised sera (n = 1245) collected between August 2003 and September 2004 while phase 2 comprised sera collected between

Results and discussion

Table 3 presents the tabulated results of the dichotomized tests for the two datasets DS1 and DS2. Based on these data, the Se and Sp of the RBT, FPA and c-ELISA as well as the prevalence in the subpopulations were estimated. In Table 4, the posterior means and 95% CPI of these estimates are given for DS1, DS2 and DS2 with the test result of FPA ignored.

Using latent class analysis the Se and Sp of three serological tests, RBT, FPA and ELISA, for brucellosis were estimated in a population of

Conclusion

The tests were applied on naturally infected cows presumably at all stages of the infection process (latent, incubation and chronic) and representing varying stages of age, pregnancy status, health status, etc. Therefore the test characteristics found in this study can be considered to be more indicative of the tests performance in the target population, than e.g., validation tests performed on sera from culture-negative and-positive animals, which tend to overestimate Se and Sp (Greiner and

Acknowledgements

This study was funded by the Norwegian Council for Higher Education's Programme for Development Research and Education (NUFU). The authors thank the Canadian Food Inspection Agency for providing the FPA reagents.

References (28)

  • L.M. Bell et al.

    Serological evidence of Brucella melitensis infection in goats and eland in Zambia

    Vet. Rec.

    (1977)
  • G.C. Bishop et al.

    Bovine brucellosis

  • A.J. Branscum et al.

    Estimation of diagnostic-test sensitivity and specificity through Bayesian modeling

    Prev. Vet. Med.

    (2005)
  • M.J. Corbel

    Characterization of antibodies active in Rose-Bengal plate test

    Vet. Rec.

    (1972)
  • Cited by (44)

    • A systematic review of current immunological tests for the diagnosis of cattle brucellosis

      2018, Preventive Veterinary Medicine
      Citation Excerpt :

      Thus, discrepancies can be attributed to a reduced DSp caused by an inappropriate cut-off, the different impact that the epidemiological situation has on different tests or both (Section 2). Indeed, this is a plausible explanation (Greiner and Gardner, 2000; Jacobson, 1996; Thrusfield, 2013) that has been discussed in some studies, although without reaching firm conclusions because of the lack of bases to propose other cut-offs (Bronsvoort et al., 2009; Matope et al., 2011; Muma et al., 2007). Likewise, the soundness of assumptions on prevalence is variable.

    • Evaluation of three serological tests for diagnosis of bovine brucellosis in smallholder farms in Pakistan by estimating sensitivity and specificity using Bayesian latent class analysis

      2018, Preventive Veterinary Medicine
      Citation Excerpt :

      Unknown disease status could be a potential bias in the current study, however the approach used to estimate the test parameters in smallholder farming structure is logistically sound. Also, LCA only assumes the status of the disease defined by the serological tests rather than truly infected animals that could lead to bias in estimation of test properties (Muma et al., 2007). Given that in the current study regions, farmers did not vaccinate their animals against brucellosis, most of the animals were exposed to the infection naturally in the field, and the serological responses appear to be stable after seroconversion.

    View all citing articles on Scopus
    View full text