REFERENCE TEST ERRORS BIAS THE EVALUATION OF DIAGNOSTIC-TESTS FOR ISCHEMIC HEART-DISEASE
REFERENCE TEST ERRORS BIAS THE EVALUATION OF DIAGNOSTIC-TESTS FOR ISCHEMIC HEART-DISEASE
复制标题
DOI：
10.1007/bf02595925
复制
发表时间：
1988-09-01
影响因子：
5.7
通讯作者：
BARON, AE
中科院分区：
文献类型：
--
作者：
BOYKO, EJ;ALDERMAN, BW;BARON, AE
通常认为灵敏度和特异性是固有的诊断测试性质，其不随疾病流行而改变，然而，L's Gart和Buck已经表明，当诊断测试与灵敏度和/或特异性小于100%的参考测试（“金标准”）比较时，这些诊断测试性质随疾病流行而改变。[4] Hui和Waiters，Vacek，e Thibodeau，7和DeNeei的研究证实并扩展了这一发现。虽然这一结果在20多年前就被首次证实，但它在很大程度上没有被注意到，因为关于诊断测试评估的主要著作没有提到它。9- 1 z许多证据表明，冠状动脉造影术，经常被用作缺血性心脏病的参考测试，的敏感性和特异性低于100%。研究人员发现，冠状动脉造影术与常规冠状动脉造影术相比，反应性充血反应，直接测量人体心肌缺血的新方法。与尸检结果相比，在一项研究中，42名受试者中有17名在该试验后57天内死亡，14名受试者的冠状动脉造影低估了显著狭窄，而在另一项研究中，9%的受试者发生了血管造影高估狭窄。受过训练的观察者通常对冠状动脉造影片的解释不一致，即，17在实践中，这种不一致导致疾病状态分类的可变性和不完善的参考测试性能。我们描述了参考测试误差对诊断测试灵敏度和特异性测量的预期影响。然后，我们重新审视了几项研究的令人困惑的发现，
IT IS COMMONLY TAUGHT that sensitivity and specificity are inherent diagnostic test properties that do not change with disease prevalence, l's Gart and Buck have shown, however, that these diagnostic test properties change with disease prevalence when the diagnostic test is compared with a reference test (" gold standard") whose sensitivity and/or specificity is less than 100%. 4 Studies by Hui and Waiters, s Vacek, e Thibodeau, 7 and DeNeei s have affirmed and extended this finding. A/though this result was first demonstrated more than 20 years ago, it has gone largely unnoticed, as major writings on the evaluat/on of diagnostic tests make no mention of it. 9-1zMuch evidence suggests that coronary angiography, which is frequently used as the reference test for ischemic heart disease, has sensitivity and specificity less than 100 per cent. Researchers have shown that coronary angiography frequently misclassified both presence and absence of significant stenosis when compared with the reactive hyperemic response, a new method for directly measuring myocardial ischemia in humans. Is Compared with autopsy results, in one study coronary angiography underestimated significant stenosis in 17 of 42 subjects who died within 57 days after this test, 14 while in another study angiographic overestimation of stenosis occurred among 9% of subjects. iS Trained observers often have disagreed on the interpretation of coronary angiograms, le, 17 In practice this disagreement leads to variability in classification of disease status, and imperfect reference test performance. We describe the expected effects of reference test errors on the measurement of diagnostic test sensitivity and specificity. We then reexamine the puzzling findings of several studies comparing non-