Abstract
The number of diagnostic test accuracy (DTA) studies concerning biomarkers have gradually increased during the past years. However, study designs remain imperfect, and the statistical methods used are not meaningful in some published studies. Here, we introduce recommendations for designing DTA studies, including consecutive enrollment of participants with uniform inclusion and exclusion criteria, blinded testing and interpretation, prespecified thresholds, and the use of one reference standard for all subjects. In addition, we also describe more relevant statistical methods in DTA studies, including decision curve analysis (DCA), nomograms, diagnostic model and scale, net reclassification index (NRI), and the integrated discriminatory index (IDI). This review may help clinicians to better design DTA studies that investigating biomarkers.