확률모형을 이용한 오제스키병 혈청학적 모니터링 프로그램 평가

The objective of this study was to analyze data from the planned national serological monitoring program for Aujeszky's disease (AD) using a simulation model to evaluate probable outcomes expected in the sample derived from the simulated herds at predefined within-herd prevalence and herd prevalence. Additionally, prevalence at animal- and herd-level estimated by the stochastic simulation model based on the distributions of the proportion of infected herds and test-positive animals was compared with those of data from a national serological survey in 2006, in which 106,762 fattening pigs from 5,325 herds were tested for AD using a commercial ELISA kit. A fixed value of 95% was used for test sensitivity, and the specificity was modeled with a minimum, most likely and maximum of 95, 97 and 99%, respectively. The within-herd prevalence and herd prevalence was modeled using Pert and Triang distributions, respectively with a minimum, most likely and maximum point values. In all calculations, population size of 1,000 was used due to lack of representative information. The mean number of infected herds and true test-positives was estimated to be 27 herds (median = 25; 95% percentile 44) and 214 pigs (median = 196; 95% percentile 423), respectively. When testing 20 pigs (mean of 2006 survey) in each herd, there was a 3.3% probability that the potential for false-positive reactions due to less than 100% specificity of the ELISA test would be detected. It was found that the model showed prevalence of 0.21% (99% percentile 0.50%) and 0.5% (99% percentile 0.99%) at animal- and herd-level, respectively. These rates were much similar to data from the 2006 survey (0.62% versus 0.83%). The overall mean herd-level sensitivity of the 2006 survey for fattening pigs was 99.9%, with only a 0.2% probability of failing to detect at least one infected herd.