Small sample estimation of relative accuracy for binary screening tests

摘要: Comparing the accuracy of two screening tests is ideally achieved by administering both as well a gold standard test to all study subjects. In practice, more ethical screen positive design often used, one that requires determination only for subjects on either under investigation. Although it not possible quantify absolute each with such design,the relative can be estimated. Since estimation has poor small sample properties, adjusted estimators based adding constants observed data have been proposed. The advantage they yield point and variance estimates in settings. However, we show through theory numerical examples beneficial or detrimental performance. Furthermore, performance estimator depends magnitude constant but also parameters cannot estimated from study, making selection an optimal difficult practice. We examine using comparing cervical cancer.