受试者工作特征曲线法分析粪便幽门螺杆菌特异性抗原试验的诊断效率

目的 探讨粪便幽门螺杆菌特异性抗原(HpSA)在诊断Hp感染中的价值.方法 因上消化道症状就诊的患者353例,采用酶联免疫吸附法(ELISA)和酶免疫法(EIA)检测HpSA,以~(13)C-尿素呼气试验(~(13)C-UBT)为诊断Hp感染的"金标准".应用不拘分布形式参数法(拟合双正态模型)和非参数法分别绘制光滑受试者工作特征(ROC)曲线和经验ROC曲线,以曲线下面积(Az)评价HpSA试验的诊断价值.结果 ELISA和EIA法检测Hp感染阳性数与~(13)C-UBT检测比较差异无统计学意义(169例和170例比177例,P值均>0.05).ELISA和EIA法检测HpSA符合率98.58%(Kappa值=0.97).ELISA法检测HpSA的不拘分布形式参数法和非参数法Az分别为0.96和0.95,EIA法检测HpSA的不拘分布形式参数法和非参数法Az分别为0.97和0.94,与完全无诊断价值的机会线下面积(0.50)相比差异均有统计学意义(P值均<0.001);两种方法Az相关系数为0.61,面积差异无统计学意义(P>0.05).结论 ROC曲线作为一种统计学工具可以更加客观、真实的评价诊断试验.ELISA和EIA法检测HpSA诊断Hp感染其准确性和可靠性均较高,值得临床进一步推广.

Evaluation of effectiveness of Helicobacter pylori stool antigen test for Helicobacter pylori infection based on the receiver operating characteristic analysis

Objective To assess the efficacy of Helicobacter pylori (Hp) stool antigen (HpSA) test for detection of Hp infection.Methods The stool specimens collected ftom 353 patients with upper gastrointestinal symptoms were detected for HpSA by enzyme linked immunosorbent assay (ELISA) and enzyme immunoassay (EIA).The diagnostic accuracy of HpSA ELISA and EIA were evaluated in comparion with the results of a ~(13)C-Urea breath test (~(13)C-UBT) which was defined as gold standard.The relative operating characteristic (ROC) curve was constructed by using parametric distribution-free (PDF) approach and non-parametric method.The area under the ROC curve was used to evaluate the value of HpSA test in diagnosis of Hp infection.Results The chi-square test showed that positive Hp infection was detected in 1 69 cases by ELISA,170 cases by EIA and 177 cases by ~(13)C-UBT with no significant difference (P>0.05).The coincidence of ELISA and EIA in detecting HpSA was 98.58%(Kappa=0.97).The area under the ROC curve for ELISA detection of HpSA was 0.96 by PDF and 0.95 by non-parametric method,whereas it was 0.97 and 0.94,respectively,for EIA detection of HpSA.The area under the ROC curve was significant higher for ELISA and EIA than no diagnostic value of 0.50(P<0.001)).The correlation of area under the ROC curve between ELISA and EIA was 0.61(P>0.05).Conclusions ROC curve,as a statistical tool,is more objective and realistic in evaluating diagnostic tests.It is an accurate and reliable test for HpSA detection by both EIISA and EIA.which is helpful in clinical practice.