The use of a Bayesian latent class model to estimate the test characteristics of three liver fluke diagnostic tests under New Zealand field conditions

Abstract

The liver fluke Fasciola hepatica is a trematode parasite of farmed livestock with worldwide distribution, causing chronic production losses and possible death from hepatobiliary damage. The effective management of liver fluke infection requires diagnostic tests which can accurately identify infected animals at both the individual and herd level. However, the accuracy of liver fluke diagnostic tests performed on individual New Zealand cattle is currently unknown. The aim of this study was to use a Bayesian latent class model (LCM) to estimate the test characteristics of three liver fluke diagnostic tests, the coproantigen ELISA, the IDEXX antibody ELISA and the faecal egg count. One hundred and twenty dairy cows each from two dairy farms were blood and faecal sampled in April 2021. The samples were transported to Massey University, Palmerston North, and the three diagnostic tests completed following the respective manufacturer instructions. A Bayesian LCM model, adapted from the original Hui and Walter 2 tests 2 populations model, was built to estimate the test characteristics of the three diagnostic tests in the two dairy herds. The model was implemented in JAGS using Markov chain Monte Carlo sampling. The first 30,000 iterations were discarded as burn-in, and the next 200,000 iterations were used to construct the posterior distributions. Uninformed priors, beta (1,1), were used as the prior distributions for the prevalence estimation and informed beta priors, based on published results, were used as the prior distributions for estimating the sensitivity and specificity of each diagnostic test. Model convergence was confirmed by inspection of trace plots and examination of the results of the Gelman and Rubin test. The results found that the coproantigen ELISA test was the most accurate for diagnosing liver fluke infection in individual animals with a sensitivity = 0.98 (95 % CI 0.95–1.00) and specificity = 0.95 (95 % CI 0.81–1.00) compared to the IDEXX antibody ELISA test, sensitivity = 0.39 (95 % CI 0.32–0.47) and specificity = 0.86 (95 % CI 0.75–0.96) or the FEC, sensitivity = 0.23 (95 % CI 0.17–0.30) and specificity = 0.92 (95 % CI 0.86–0.97). Based on these results clinicians should be encouraged to use the coproantigen ELISA test to diagnose liver fluke infection in individual cattle.