Diagnostic test evaluation of four serological assays for Brucella antibody detection in Bactrian camels (Camelus bactrianus) in Inner Mongolia, China: A Bayesian latent class analysis.

Abstract

Brucellosis remains an important zoonosis, but evidence on the diagnostic accuracy of serological assays in Bactrian camels is limited. We conducted a cross-sectional study involving 1091 Bactrian camels from 24 herds in Alxa Right Banner, Inner Mongolia, China. Serum samples were tested using the Rose Bengal test (RBT), indirect ELISA (i-ELISA), fluorescence polarization assay (FPA), and complement fixation test (CFT). A Bayesian latent class model accounting for conditional dependence among the antibody-based assays was used to estimate individual-animal sensitivity (Se) and specificity (Sp), the true prevalence of antibodies to Brucella spp. within each herd, and individual-animal predictive performance under serial and parallel interpretation rules. FPA showed the highest sensitivity (median 0.946; 95% posterior credible interval [PCI]: 0.919-0.966), followed by CFT (0.857; 0.818-0.891), RBT (0.843; 0.803-0.878), and i-ELISA (0.781; 0.737-0.820). Specificity was highest for CFT (0.995; 0.987-0.999), followed by RBT (0.987; 0.976-0.995), i-ELISA (0.959; 0.942-0.973), and FPA (0.927; 0.903-0.949). The sample-size-weighted true prevalence of antibodies to Brucella spp. among the sampled camels was 36.4% (95% PCI: 33.9-38.8%). Parallel interpretation of RBT and FPA achieved a sensitivity of 0.989 (95% PCI: 0.981-0.994). The principal assay rankings were stable under alternative prior assumptions, single-herd omissions, and alternative conditional-dependence specifications. No single assay achieved both high sensitivity and high specificity. At the individual-animal level, parallel interpretation of RBT and FPA may be considered when minimizing false-negative classifications is the priority, whereas CFT may be preferred when higher specificity is required.