Table 3.
Results from the Bayesian latent class model estimating the sensitivity and specificity of three assays (conventional PCR, Biomeme’s commercially available C. burnetii Go-Strip Assay, and a C. burnetii PCR-HRM technique) and the estimated proportion of ticks carrying C. burnetii.
| Parameter | Posterior median estimate (%)* | 95% credibility interval |
|---|---|---|
| Sensitivity | ||
| Conventional PCR | 24% | 10–47% |
| Biomeme | 57% | 34–90% |
| PCRHRM | 86% | 56–99% |
| Specificity | ||
| Conventional PCR | 98% | 94–100% |
| Biomeme | 96% | 91–100% |
| PCRHRM | 94% | 86–100% |
| Proportion with C. burnetti | 16% | 0.08–28% |
*The model assumes independence across assays. Model estimates were generated from 169 tick samples collected from wildlife and cattle in Kenya between May 2011 and 2019. The Deviance Information Criterion (DIC) of the final model was 27.91 (95% credibility interval: 23.41–37.26%).