초록
Serological surveillance programs in animal populations are becoming increasingly important to estimate prevalence of a specific disease and subsequently to document disease-free status in a region or a country. For these purposes, the programs need to be based on both theoretical and economical aspects from the designing phase. From Aujeszky's disease (AD)-eradication program point of view, group of animals (aggregates, herds) not individual animal is the more important sampling unit of concern. In this study the authors therefore attempted to compute an appropriate sample size tailored to a current surveillance program against AD, assuming that the goal of this program is either herd-level prevalence estimation or documentation of AD-freedom. For prevalence estimation, assuming a finite population with imperfect sensitivity (Se) and specificity (Sp) of ELISA kit for AD diagnosis, the number of herds present, expected herd prevalence, and desired accuracy for a certain level of confidence, sample size was estimated at herd-level in the first stage and individual animal-level in the second stage. A two-stage sampling design was used to calculate a sample size to indicate AD-freedom. In this instance, the computation was based on the possible detection of a predetermined prevalence at a certain herd-level Se and Sp. This study indicated that the sample size varied with predetermined confidence, tolerance, Se and Sp at herd- and animal-level, and within- and among-herd prevalence. In general, smaller sample size was required to estimate AD prevalence than to document of AD-freedom. Compared to individual-based samples, two-stage sampling strategy requires a larger sample size to show disease-freedom. Statistical considerations including herd-level test characteristics when designing surveillance program also are further discussed.