用贝叶斯方法对日本血吸虫感染两种检测方法进行评价  

Evaluation of Two Tests for Detecting Schistosoma Japonicum Infection Using a Bayesian Approach

作　　者：王显红[1] 周晓农[1] 李远林 吕山[1] 李兰花[1] 贾铁武[1] 陈绍荣 杨忠 方文 陈凤 

机构地区：[1]中国疾病预防控制中心寄生虫病预防控制所,200025 [2]云南省大理州血吸虫病防治研究所,671000

出　　处：《中国卫生统计》2007年第4期361-363,共3页Chinese Journal of Health Statistics

基　　金：国家自然科学基金重大项目(编号30590373);联合国儿童基金会/联合国开发署/世界银行/世界卫生组织热带病研究与培训特别规划署资助项目(TDRA30298)

摘　　要：目的探讨估计日本血吸虫感染两种检测方法(ELISA和Kato-Katz)在现场应用时的灵敏度和特异度的新方法。方法采用贝叶斯(Bayes)统计方法,利用2005年云南省洱源县现场调查资料,对两种方法在流行区和非流行区应用时的检测效果分别进行了评价。结果在流行区,ELISA灵敏度较高(0.913,中位数,下同)而特异度较低(0.584),Kato-Katz法灵敏度低(0.148)而特异度很高(0.994);在非流行区,ELISA灵敏度为0.766,特异度为0.908,Kato-Katz法灵敏度和特异度与其在流行区的相似。结论在无金标准的情况下,Bayes方法能够对检测方法做出科学的评价。Objective To explore a novel approach to estimate the sensitivities and specificities of two tests, i. e., ELISA and Kato - Katz, for detecting schistosoma japonicum infection when used in the field. Methods A Bayesian approach was employed to evaluate the characteristics of the two tests when used in endemic and non - endemic areas, respectively, based on the data from the field investigation in Eryuan county of Yunnan province, 2005. Results In endemic areas, a high sensitivity with the median of 0. 913 and a low specificity with the median of 0. 584 preserited in ELISA, and a low sensitivity with the median of 0. 148 along with a very high specificity（0. 994） appeared in Kato - Katz. In non - endemic areas, a moderate sensitivity of 0. 766 with a high specmcity（0. 908） was showed in ELISA, and the sensitivity and specificity of Kato - Katz was similar to that in endemic areas, respectively. Conclusion The Bayesian approach can evaluate the test pmpenies appropriately when there is no gold standard.

关 键 词：贝叶斯统计 ELISA 改良加藤法 灵敏度 特异度 日本血吸虫 

分 类 号：R181.3[医药卫生—流行病学]