Cost and sensitivity of on-farm versus slaughterhouse surveys for prevalence estimation and substantiating freedom from disease.

Within the framework of Swiss surveillance for epizootic diseases, dairy cattle are sampled using bulk tank milk while non-dairy cattle are sampled on the farm. The latter method is costly, time-demanding and dangerous for the personnel. However, slaughterhouses could be an alternative sampling point for this population. To assess the cost-effectiveness and sensitivity of such an approach, surveillance using slaughterhouse sampling was modelled with data from the 2012 Swiss animal movement database (AMD). We simulated a cross-sectional study for bluetongue (BT), and surveillance programmes to substantiate freedom from infectious bovine rhinotracheitis (IBR) and enzootic bovine leucosis (EBL) (combined) to compare the outcome of random on-farm sampling versus slaughterhouse sampling. We found that, under Swiss conditions, slaughterhouse sampling results in low herd-level sensitivities because animals are sent by owners to slaughter individually and not in large groups, restricting the number of samples per herd. This makes slaughterhouse sampling inappropriate for prevalence surveys at the herd-level. However, for prevalence surveys at the animal-level and for substantiation of freedom from disease, slaughterhouse surveillance is equally or more cost-efficient than on-farm sampling.

Analysis of national serological surveys for the documentation of freedom from porcine reproductive and respiratory syndrome in Switzerland.

Results of national serological surveys for porcine reproductive and respiratory syndrome (PRRS) conducted in Switzerland in 2001 and 2004 were analyzed. In 2001, 41,124 breeding sows from 2,540 herds out of 6,406 were sampled, and in 2004, 7,498 animals were sampled from 1,074 herds out of 5,320. All serum samples were tested for PRRS using an ELISA developed at the Institute of Virology and Immunoprophylaxis (IVI), Switzerland with a sensitivity (Se) and specificity (Sp) of 94 and 97%, respectively. Positive samples were re-tested with a commercial ELISA (IDEXX) with Se of 100% and Sp of 99%. Samples positive in the second test were confirmed with the fluorescent antibody test (FAT). A stochastic model using data from the main survey conducted in 2001 was done to verify whether the sampling scheme used could detect at least one infected herd with 99% confidence level if the herd designed prevalence was at 0.1 or 0.2%. Additionally, a Bayesian approach was conducted to calculate the post-survey probability of freedom from PRRS using data from the 2001 and 2004 surveys. A Monte Carlo simulation with 5000 iteration was run for each model. Eleven samples in 2001 and six in 2004, all from different farms, could not be conclusively confirmed as negative by the FAT. All other samples were negative. Truly infected animals and herds were not predicted by a stochastic model at the 99% confidence level and 0.1% herd prevalence using data from the 2001 survey. However, it was demonstrated that the prior probability of freedom from PRRS increased from 89.3 to 99.2% after the 2001 survey. Upon completion of the 2004 survey, the probability of freedom from PRRS reached a value of 99.7%. Based on our results, we could conclude that the pig industry in Switzerland is free of PRRS virus with this level of confidence. Restricted import activities over the last decades are a possible explanation for the continuing absence of PRRS-infection in the Swiss swine population. Import requirements defined by the pig industry minimize the risk of introduction of PRRS-infected animals in the future.