目的通过ROC曲线分析本实验室两种乙型肝炎表面抗原(HBsAg)酶联免疫试验(ELISA)的准确性,确定试验的最佳临界值;通过分析单独或平行使用两种试验的阳性符合率、阴性符合率,评估不同检测方式的效能,为制定正确的HBsAg检测策略提供依据。方法对HBsAg中和试验阳性标本和ELISA双试剂及核酸检测(NAT)均阴性的标本共778份,用2种试验同时检测,以检测OD值建立ROC曲线。通过分析曲线下面积(AUC)以及确定最大Youden指数获得两种试验准确性信息和最佳临界值;在此基础上模拟2种试验单独使用或平行使用方式,计算其阳性检出率和阴性检出率并分析相互之间的差异。结果通过分析2种试验ROC曲线下面积(AUC)可知A试验的检测准确性明显优于B试验,差异具有统计学意义(P〈0.05);A试验和B试验最佳检测临界值OD值分别为0.063,0.035;在该最佳检测临界值下,2种试验阳性符合率为A 94.44%(459/486),B 89.3%(434/486),平行使用2种试验联合检测(A+B)的阳性符合率为94.86%(461/486),试验A与A+B的阳性符合率在统计学上没有明显差异(u=0.2917,u〈1.96,P〉0.05),而试验B检测与A+B相比,其阳性符合率之间存在统计学上的明显差异(u=3.2095,u〉1.96,P〈0.05)。3种检测方案的阴性符合率分别为,A:91.1%(266/292)B:93.15%(272/292)A+B:87.67%(256/292),试验A与B的阴性符合率在统计学上没有明显差异(u=0.919);但2种试验单独使用和联合使用的阴性符合率之间存在明显差异(u1=4.366,u2=2.249)。结论针对采用2种试验进行HBsAg检测的实验室,可利用ROC曲线确定每种试验的最佳临界值,评估每种试验的准确性。对于目前血站实验室采用的HBsAg ELISA试验,存在单一试验性能等同于多个试验组合检测性能的现象。实验室在拟采用单一试验进行HBsAg检测策略时,可以以此作为依据。
Objective To analyze the diagnostic accuracy and optimal cut-off points of two kinds of commercially available kits ( A and B) for HBsAg used in our laboratory using ROC curves, then to discuss detection sensitivity and specificity of them. Methods 486 cases positive specimens and 292 negative ones determined by neutralization confirmatory test were implemented EIA tests simultaneously with A and B reagents respectively and the OD value of each sample was used to draw the ROC curves with SPSS 15.0. Diagnostic accuracy ,cut-off point value of the two kinds of regent were then acquired on the basis of the curves. Afterwards, detection sensitivity and specificity of regent A and B was calculated respectively by analyzing their positive detection rate and negative diagnosis rate in different testing strategy. Results The area under the ROC curves (AUC) (95% CI) was 0. 981 (0. 969 -0. 989) for regent A and 0. 968(0. 953 -0. 979) for regent B. There was statistically significant differences (P 〈0.05) for diagnostic efficiency of the two kinds of regent;The optimal cut-off points (OD value) for diagnosis were 0. 063 ( sensitivity :94.44%, specificity :91.1% ) for regent A and 0. 035 ( sensitivity: 89. 3%, specificity :93.15% ) for regent B which were determined by the maximum Youden index. As for the positive detection rate, regent A was 94.44% (459/486) ,regent B was 89. 3% (434/486) and A + B was 94. 86% (461/486) in this condition,and the negative coordination rate was 91.1% (266/292) for A,93.15% (272/292) for B and 87.67% (256/292) for A + B. So A testing strategy should he selected to prevent misdiagnosis of regent B if using only one kind of regent in the process of HBsAg screening. Conclusion ROC can be used to determine the diagnosis accuracy and optimal cut-off points for A and B HBsAg EIA regents,which can help to optimize the performance of HBsAg screening method in use in blood screening lab. And it also can provide an auxiliary judgm