, Volume 74, Issue 1, pp 44-55

Comparing two diagnostic tests when two tests are applied to same patients and test scores are given in categories

Purchase on Springer.com

$39.95 / €34.95 / £29.95*

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access

Abstract

Suppose that a new diagnostic test is developed for making sharp distinction between disease A and B. To show its superiority to the standard test, the study design is introduced that the new and standard test are applied blindly to each patient in two groups, one group is definitely known being contracted with disease A and the other definitely known being contracted with disease B, where test scores are given in categories by multiple readers (raters). The design inevitably limits the number of patients used for the comparison and dependency could be introduced between the results of two tests. Application of existing statistical methods is not easy to be justified since they are based on asymptotic distributions of test statistics. We develop in this paper a method based on conditional logistic regressions that is applicable to small size of data and is useful to show the superiority of the new test to standard test by adjusting for the effect of readers in the study design. The method is applied to the data for comparing a new test and standard test for differentiating between epidermal cyst and ganglion.