The sensitivity and specificity of fecal and cecal culture for the detection of Campylobacter in Dutch broiler flocks quantified by Bayesian analysis

Dutch broiler flocks are routinely tested for the presence of thermotolerant Campylobacter spp. using a standard cultural procedure for fecal and cecal samples. The objective of this study was to estimate the sensitivity and specificity of fecal and cecal culture for detection of Campylobacter colonization in broiler flocks in absence of a gold standard. Data from 1600 flocks were used from two different populations, whereby only flocks with both fecal and cecal culture results were included in the analysis. Latent class analysis using Bayesian inference was applied to generate the test characteristics of fecal and cecal culture. Two statistical models assuming conditional dependence of both tests on Campylobacter status were used to compare the results. On flock level, the sensitivity of the fecal culture was found to be 21% (95% CI: 12, 31) and 23% (95% CI: 13, 60), and the specificity was 98% (95% CI: 94, 99) and 97% (95% CI: 92, 99) for the two models, respectively. The sensitivity of the cecal culture was 64% (95% CI: 37, 89) and 66% (95% CI: 39, 90), and the specificity was 98% (95 CI: 94, 99) and 95% (95% CI: 72, 99) in respective models. The implications of a low sensitivity as in the case of the fecal culture is important for the design and interpretation of monitoring programmes and may result in excessive false negative test results. Although cecal culture is the more sensitive test, substantial misclassification of infected flocks may still occur.