Large-scale evaluation of bacteriological-based method and qPCR performance for Brucellosis diagnosis in livestock using Bayesian latent class analysis

Abstract

The performance of direct tests, such as bacteriological culture and qPCR, for the diagnosis of brucellosis has been evaluated in a limited number of studies, often based on small sample sizes. Moreover, the absence of a gold standard makes this assessment even more challenging. A potential alternative for evaluating the performance of direct tests is Bayesian latent class analysis (BLCA), which does not require prior knowledge of disease status or a gold standard. This study aimed to estimate the sensitivity (Se) and specificity (Sp) of bacteriological culture for brucellosis diagnosis. In a brucellosis-endemic area, a large number of seronegative and seropositive buffaloes and cattle were tested using bacteriological culture and qPCR. BLCA was applied to estimate the performance of both tests. The median Se of bacteriological culture was estimated at 61.3%, compared to 70.9% of qPCR. The median Sp was 99.6% for bacteriological culture and 89.3% for qPCR. Bacteriological culture demonstrated a higher Positive Predictive Value (PPV) than qPCR in both buffaloes and cattle, whereas the Negative Predictive Value (NPV) of the two methods did not differ significantly. These results suggest that, in settings of low brucellosis prevalence, a positive bacteriological culture has a greater predictive value than qPCR .

Keywords: Bayesian modelling; Brucellosis; bacteriological culture; livestock; sensitivity; specificity.