[目的]了解新疆乌苏市羊棘球蚴病真实流行率。[方法]首先用两种酶联免疫吸附试验试剂盒，对采自乌苏市屠宰场的172份羊血清同时进行抗体检测，汇总交叉结果，通过建立贝叶斯模型，确定先验分布，完成Gibbs抽样及相关参数计算，评价两种试剂盒（A和B）对羊棘球蚴病的诊断价值，选择诊断价值较高的试剂盒，对乌苏市羊棘球蚴病个体真实流行率进行测量。其次，采用横断面研究，采取两阶段抽样策略。第一阶段为乡镇抽样，采用随机抽样，预定流行率为85%，CI为90%，可接受误差为15%；第二阶段为个体抽样，采用按养殖大小成比例抽样（PPS），预期流行率为30%，CI为90%，可接受误差为5%。采集的1 787份血清，利用诊断价值取高的试剂盒进行检测，根据检测结果、敏感性和特异性，计算真实流行率。[结果] 通过贝叶斯模型分析，试剂盒A和B的敏感性分别为91.8%、76.6%，特异性分别为95.4%、80.7%，试剂盒A的诊断价值较高。乌苏市羊棘球蚴病个体表观流行率（AP）为48.6%，统计分析得出个体真实流行率（TP）为50.5%，95%置信区间（CI）为48.2%~52.8%。[结论] 贝叶斯估计是抽样信息对先验分布做出调整的结果，其更接近参数的真实情况。本研究结果既可以作为先验信息，也可以作为其他血清学检查方法的评价和现场应用提供参考信息。本研究首次通过两阶段抽样策略对羊棘球蚴病的个体流行率进行了测量，相对以往的监测方法更加科学，更接近研究时段内该地区的真实情况。在今后的羊棘球蚴病流行病学调查、监测和检测工作中可逐步推广ELISA诊断方法。
[Objective] In order to recognize the true prevalence of ovine Echinococcosis in Wusu city of Xinjiang. [Methods] Two kinds of ELISA kits（kit A and kit B）were used to detect antibodies existing in 172 collected sheep serum samples from slaughter houses. After a series of steps including summarizing of the cross-reaction results，confirming prior distribution by establishing Bayesian model and completing Gibbs sampling and related parameter calculations，the value of the two kinds of kits in diagnosis was evaluated to choose the one with higher value，so as to measure the true prevalence of individuals. Next，adopting ways of cross-sectional study and two-stage sampling strategy，1 787 serum samples were tested using kit A and the true prevalence rates were calculated on basis of the test results，sensitivity and specificity. As to the two-stage sampling，first was random sampling in the counties，with an estimated prevalence of 85%，90% CI，and the acceptable error of 15%. The second stage was individual sampling by PPS strategy（sampling in proportion to the production scale），with an estimated prevalence of 30%，90% CI，and the acceptable error of 5%. [Results] Based on the results of Bayesian model analysis，the sensitivity of the kit A and B were 91.8% and 76.6%，respectively，their specificity were 95.4% and 80.7%，respectively. Hence kit A showed higher diagnostic value. The apparent prevalence（AP）of individuals was 48.6%. Through statistical analysis，the true prevalence（TP）of 50.5% was obtained（95% CI：48.2%~52.8%）. [Conclusion] Because Bayesian estimation is the result of adjusting sampling information in view of prior distribution，it is more close to the truth of parameters. The analysis results could be considered as both prior information and an assessment for other serological detection methods，as well as providing information for field application. In this research，two-stage sampling strategy was used for the first time to measure the individual prevalence rate of ovine echinococcosis. Compared with the past methods，it was more scientific and much closer to the real situation of the region during the research phase. As a conclusion，the ELISA method could be gradually applied in fields of epidemiological survey，surveillance and detection of ovine echinococcosis.