Estimating Neospora caninum prevalence in wildlife populations using Bayesian inference

Area of Science:

• Wildlife epidemiology
• Veterinary parasitology
• Statistical modeling in ecology

Background:

• Disease prevalence studies in wildlife are crucial for disease modeling and epidemiology but are often understudied.
• Standard laboratory tests for disease screening in wildlife are frequently validated only for domestic animals, potentially leading to inaccurate prevalence estimates.
• Accurate diagnostic test validation is essential for reliable wildlife disease surveillance.

Purpose of the Study:

• To determine the sensitivity and specificity of a competitive enzyme-linked immunosorbent assay (cELISA) for Neospora caninum exposure in American bison, Père David's deer, and white-tailed deer.
• To estimate the true prevalence of Neospora caninum exposure in these three wildlife populations using Bayesian latent class analysis (LCA).
• To assess the impact of test accuracy on apparent versus true disease prevalence in wildlife.

Main Methods:

• Bayesian latent class analysis (LCA) was employed to estimate the diagnostic accuracy (sensitivity and specificity) of the cELISA test.
• The study evaluated the cELISA test in three distinct wildlife populations in southeastern Ohio: American bison, Père David's deer, and white-tailed deer.
• True prevalence of Neospora caninum exposure was calculated for each species using the estimated test characteristics derived from LCA.

Main Results:

• True Neospora caninum prevalence varied significantly across species: 0.1%–3.1% in American bison, 51.0%–53.8% in Père David's deer, and 40.0%–45.9% in white-tailed deer.
• The cELISA test demonstrated high accuracy in American bison and Père David's deer, closely matching manufacturer-reported values (96% sensitivity, 99% specificity).
• Sensitivity of the cELISA in white-tailed deer was highly variable, ranging from 78.9% to 99.9%, indicating potential over- or underestimation of prevalence without LCA.

Conclusions:

• Apparent prevalence derived from cELISA results did not accurately reflect true Neospora caninum prevalence in white-tailed deer and Père David's deer.
• Significant differences in true prevalence among sympatric species highlight varied Neospora caninum epidemiology, even in shared environments.
• Bayesian LCA provides a robust statistical framework to overcome limitations in validating diagnostic tests for wildlife, enhancing the accuracy of disease surveillance and epidemiological studies.