| Sumario: | This article is the fifth in a series in which I describe several of the more common statistical errors in the biomedical literature. The first article focused on 10 errors in descriptive statistics and in interpreting probability, or P values (AMWA J. 2003;18: 67-71); the second article described 9 errors in interpreting differences between groups (AMWA J. 2003;18: 103-106); the third article addressed 5 errors in presenting statistical information in figures and tables (AMWA J. 2004;19:9-11); and the fourth article focused on 3 errors in correlation and regression (AMWA J. 2005;20:10-11). Here, I describe 4 errors in reporting the results of diagnostic tests.Diagnostic test characteristics--sensitivity, specificity, predictive values, likelihood ratios, and so on--are often misunderstood and are therefore sometimes reported and interpreted incorrectly. In this article, I describe the most common characteristics and how to report them.
|
|---|