THE PERFORMANCES OF TWO DIAGNOSTICS TESTS: MCNEMAR AND NEWCOMBE GRAPHICAL APPROACH | [Pakistan Journal of Statistics • 2017]

Author(s):

1. Mustafa Agah Tekindal: (PhD) (Biostatistics), Izmir, Turkey

2. Can Ateş: Faculty of Medicine Department of Biostatistics, Van Yüzüncü Yıl University Van, Turkey.

3. Özlem Güllü Kaymaz: Faculty of Sciences, Department of Statistics, Ankara University Ankara, Turkey

4. Yasemin Yavuz: Faculty of Medicine, Department of Biostatistics, Ankara University Ankara, Turkey.

Abstract:

This study aims to address a simultaneous comparison of sensitivity and specificity of two different tests by taking into consideration of prevalence and clinical significance level of the tests under the condition that the result of the gold standard test is known. The proposed method involves the point and interval estimation of the weighted mean (for f) concerning the specificity and sensitivity differences of two tests. Getting a value between 0 and 1, the parameter l represents the clinical cost of false positive and false negative and the occasion posing serious problems for the prevalence value belonging to the population to which tests are administered. Data is acquired from a study carried out at the Department of Cytopathology of Ankara University Medical School. In that study, the impact of the use of various antibodies in determining thyroid nodules as benign or malign via Preoperative Fine-Needle Aspiration Biopsy in order to achieve diagnostic accuracy was investigated. Considering the prevalence of Newcombe graphic approach that compares both the sensitivity and specificity, it can be concluded that Newcombe graphical approach gives better results than the classical McNemar test for the same sample group used in the study Simultaneously comparing the sensitivity and specificity of tests by taking into consideration the clinical costs of false positive/false negative values and prevalence, the Newcombe graphic approach method is also quite advantageous in terms of clinical interpretation.

Page(s): 301-314

DOI: DOI not available

Published: Journal: Pakistan Journal of Statistics, Volume: 33, Issue: 4, Year: 2017

Keywords:

comparison of paired ratios , simultaneous designs , Sensitivity and specificity , Newcombe graphical approach

References:

References are not available for this document.

Citations

Citations are not available for this document.

Citations

Downloads

Views