利用McNemar方案及ROC曲线评价进口品牌甲功检测结果临床诊断价值

目的: 评估不同进口品牌罗氏Cobas E602、西门子Centaur XP及贝克曼DXI800间血清甲状腺功能（甲功）检测项目临床诊断价值及诊断效能是否一致。方法:选择初诊甲状腺功能亢进症（甲亢）患者63例、初诊甲状腺功能减退症（甲减）患者66例及健康体检者70名作为对照组。采集3组未经治疗期血清标本，检测甲功5项（TSH、FT4、FT3、T4、T3），采用多因素联合诊断受试者工作特征曲线（ROC曲线）及WS/T505-2017《定性测定性能评价指南》提供的McNemar方案评价诊断价值及效能。结果:多因素联合诊断ROC曲线分析，3个品牌5个项目联合检测，甲亢患者的曲线下面积（AUC）在0.98~0.99之间，甲减在0.81~0.84之间，3个品牌比较差异无统计学意义（均P>0.05）；根据McNemar方案，以各品牌既定临界点（参考范围上下限）作为判断点，甲亢患者3个品牌诊断效能基本一致，仅TSH项目西门子具备更好临床敏感性的同时降低了临床特异性，甲减患者3个品牌互有优势，但FT3及FT4无论是甲亢或甲减，3个品牌临床敏感性及临床特异性均较低（13.70%~50.00%）。结论:Cobas E602、Centaur XP及DXI800检测血清甲功项目时，对于甲亢的诊断价值均较高并基本一致，建议使用该3个品牌的检验机构可以开展甲亢检测结果的互信；对于甲减的诊断价值虽然中等并一致，但FT3及FT4 3个品牌临床敏感性均较低。

Evaluation of the clinical diagnostic value of imported brand in the results of the thyroid function test by McNemar scheme and ROC curve

Objective: To evaluate the consistency of the clinical diagnostic value and diagnostic efficacy of serum thyroid function test items among various imported pharmaceutical brands, including Roche Cobas E602, Siemens Centaur XP and Beckman DXI800. Methods: We selected 63 cases diagnosed hyperthyroidism, 66 cases diagnosed hypothyroidism, as well as 70 cases not diagnosed either of the aforementioned diseases as the control group. We collected untreated serum samples from the three groups and tested five thyroid functions(TSH, FT4, FT3, T4, T3). The diagnostic value and efficacy of the treatment was evaluated based on the Receiver Operating Characteristic curve (ROC curve) of multi-factor combined diagnosis and the McNemar scheme highlighted in WS/T505-2017 Guide of Qualitative Determination Performance Evaluation. Results: Based on the analysis of the ROC curve, five projects cooperative detection with three brands, the AUC of patients with hyperthyroidism was between 0.98 and 0.99, and the AUC of patients with hypothyroidism was between 0.81 and 0.84; there was no statistical difference among the three brands(all P>0.05). Based on McNemar scheme, considering the established critical point (upper and lower bound of the reference range) of each brand, the diagnostic efficacy of the three brands for the hyperthyroid patients was virtually the same, except TSH, Siemens demonstrated better clinical sensitivity and lower clinical specificity. The three brands showed different advantages when used to treat hypothyroidism patients. Furthermore, the clinical sensitivity and specificity shown on FT3 and FT4 were relatively low(13.70%-50.00%) for all three brands, regardless of whether they were used against hyperthyroidism or hypothyroidism. Conclusion: In terms of the testing of serum thyroid function, Cobas E602, Centaur XP and DXI800 all have high diagnostic value and are consistent for detecting hyperthyroidism. Hence, we would suggest clinical laboratory organizations that utilize these three brands for hyperthyroidism testing to consider mutual acceptance of testing results. In addition, although the three brands all have medium-level diagnostic value and are consistent for detecting hypothyroidism, since the clinical sensitivity on FT3 and FT4 is low for all three, we believe that diagnosis efficacy still needs to be improved.