Bayesian estimation of seroprevalence of small ruminant lentiviruses in sheep from Poland

In Poland, no systematic survey of ruminant lentiviruses (SRLVs) infection was performed, neither at the national nor at the regional level and only limited knowledge exists on the prevalence of SRLVs among sheep. The aim of the present study was to establish the true prevalence of SRLVs infection in sheep from Poland at the animal and herd-levels. The blood samples used for this study were the fraction of samples collected by Veterinary Inspection during an official sampling for the national monitoring program for brucellosis. Under this program the animals and herds were randomly selected using the data available from ARMA (Agency for Restructuring and Modernisation of Agriculture). The sampling unit was the herd and the target population included at least 5% of sheep, over 6 months old, from each of 16 voievodships (provinces) of Poland. Two-stage cluster sampling design was performed in this study offering the possibility to determine the prevalence of SRLVs infection, when only a fraction of herds and a fraction of animals in the herds are tested. In total, 8233 sheep serum samples coming from 832 herds were tested by indirect ELISA. 1474 (17.9%) samples were positive and 261 (31.4%) herds with at least one seropositive animals were identified. The overall true prevalence estimated by the Bayesian framework was 9.3% (95% CI 6.8, 11.3) and 33.3% (95% CI 26.5, 38.2) on the animal and herd level, respectively. Large variation in the animal and herd prevalence between the voivodships was observed. True prevalence on the herd level varied from 0.0% (95% CI 0.0, 0.0) to 71.6% (95% CI 67.6, 75.9) whereas true prevalence on the animal level ranged from 0.0% (95% CI 0.0, 0.0) to 55.3% (95% CI 50.0, 61.2). The true prevalence of SRLVs infection at animal and herd level increased according to herd size as was proved by posterior probabilities (POPR).