摘要
目的利用金标准为等级变量时诊断试验的评价方法 ,评价氧化低密度脂蛋白ELISA检测试剂盒在冠心病诊断中的诊断价值。方法共入选1190例观察对象,根据金标准检测结果分为3个不同状态,从ROC曲线下面积的定义出发,利用R软件,获得金标准为等级变量时氧化低密度脂蛋白ROC曲线下面积的非参数估计值。结果无论国产或瑞士产氧化低密度脂蛋白试剂盒,均说明氧化低密度脂蛋白对冠心病的不同疾病或健康状态具有区分能力,与AUC=0.5比较均具有统计学意义(P<0.001),且随着状态间的差距越大,其区分能力增强。两试剂盒与金标准相比的准确性分别为0.8797和0.8883,且具有统计学意义(P<0.001)。结论本研究为类似的研究提供了方法学参考。
Objective The study is to introduce the method for assessing the accuracies of diagnostic tests when the gold standard is ordinal scale,and evaluate oxidized low density lipoprotein( OxLDL) for coronary heart disease diagnosing. Methods 1190 objects were split into 3 different states,suspicious,confirmed patients and normal,by the gold standard,and nonparametric method of estimating and comparing the area under receiver operating characteristic( ROC ) curves was conducted to the situation in which the gold standard for coronary heart disease is ordinal scale. Results The results show ed that both the Chinese and Swiss Ox- LDL kits,the accuracies is high enough to differ 3 different states and statistically significant compared with 0. 5 ( P 〈 0. 001 ) ,the summary measures of diagnostic test accuracy interpreted to the area under the ROC curve are 0. 8797 and 0. 8883,and tests for comparing two diagnostic tests’accura- cies in the paired designs are presented. Conclusion The OxLDL is suit- able for coronary heart disease diagnosing,in another word,the OxLDL hasthe ability to classify different states,and the method described here is useful for assessing and comparing summary test accuracy when the gold standard is not binary scale.
出处
《中国卫生统计》
CSCD
北大核心
2012年第2期172-174,共3页
Chinese Journal of Health Statistics