Bayesian latent class evaluation of three tests for the screening of subclinical caprine mastitis in Bangladesh.

Abstract

Routine monitoring for subclinical infection is one of the key mastitis control approaches. However, the accuracy of the most commonly used screening tests has not yet been established. The aim of the present study was therefore to evaluate the accuracy of three screening tests, namely California mastitis test (CMT), white side test (WST), and surf field mastitis test (SFMT) for the screening of subclinical caprine mastitis. A cross-sectional study based on 484 randomly collected milk (242 goats) samples from three districts of Bangladesh was conducted for the screening of subclinical mastitis by the aforementioned tests. The Bayesian latent class model was implemented in WinBUGS to estimate the tests' characteristics and true prevalence of subclinical mastitis. The Bayesian posterior estimates of sensitivities with a 95% credible intervals (CrIs) were 98.60% (95.18-99.95%), 98.28% (94.56-99.92%), and 89.98% (83.39-95.03%), and specificities with 95% CrIs were 99.19% (98.11-99.96%), 99.27% (97.34-99.98%), and 99.28% (97.35-99.98%), respectively for CMT, WST, and SFMT. The true prevalence of subclinical caprine mastitis was estimated to be 43.49% (95% CrI 37.46-48.98%). The positive predictive values (PPV) of the three tests were similar. The serial and parallel interpretation of any test pairs increased the PPV and negative predictive value respectively close to 100%. Based on the simplicity, cost and performance as well WST and SFMT simultaneously could be recommended for the screening of caprine subclinical mastitis in Bangladesh.