两冠心病诊断试剂盒双正态ROC曲线下面积及其等效性或非劣效性检验

Application of Equivalence or Non-Inferiority Test for Diagnostic Accuracy under the Bi-normal Distribution Based on the Areas of ROC in Two OxLDL Kits for Coronary Heart Disease Diagnostic

目的探讨在双正态假定下,应用标准化差法进行定量资料ROC曲线下面积的估计及其等效性检验或非劣效性检验,比较两氧化低密度脂蛋白试剂盒在诊断冠心病中的价值。方法从ROC曲线的定义出发,根据模型中参数的统计学意义,完成ROC曲线的构建、曲线下面积的估计,并利用标准化差结合等效性检验、非劣效性检验原理,进行参数检验,或在Bootstrap基础上利用可信区间法得到结论。结果两试剂盒均显示氧化低密度脂蛋白在冠心病诊断中具有较高的准确性。从非劣效性检验的结果可以看出,CHN试剂盒在冠心病诊断上非劣于已经投入临床使用的SWZ试剂盒。结论两试剂盒具有较高的临床推广价值,且具有较高性价比的CHN试剂盒在国内临床市场有较好的前景。同时为类似问题的解决提供了方法学参考。

Objective To assess equivalence or non-inferiority in accuracy between two OxLDL kits for coronary heart disease diagnostic,plot ROC curves and compare the areas using equivalence or non-inferiority test.Methods According to the theory of ROC and the statistical meaning of the parameters,the bi-normal model ROC curves were plotted,the areas under the curves were estimated and compared using the standardized difference approach and the standardized difference confidence interval based on bootstrap for assessing equivalence or non-inferiority in diagnostic accuracy.Results The different kits show conformably that the OxLDL is suitable for Coronary Heart Disease diagnostic,the kit of CHN is non-inferior to the SWZ.Conclusion Results from the trial have shown that the accuracy of OxLDL for coronary heart disease diagnostic.At the same time,explore a right way to put forward equivalence or non-inferiority test using the standardized difference approach under the bi-normal distribution and confidence interval approach based on bootstrap,provide a theoretical reference for similar problems.