Contrastive analysis of diagnostic tests evaluation without gold stand-ard: review article

Considering the advancement of medical sciences, diagnostic tests have been developed to distinguish patients from healthy population. Therefore, Determining and evaluation of the diagnostic accuracy tests is of great importance. The accuracy of a test under evaluation is determined through the amount of agreement between its results with the results of the gold standard, and this test accuracy can be defined based on sensitivity, specificity, positive predictive value, negative predictive value and the area under the receiver operative characteristic curve (AUC). Gold standard is an accurate and error- free method to determine the presence or absence of disease of interest and classify patients, which is not available in some diseases and situations as this method is costly or invasive. In these cases, reference standard is a best available replacement method to be used by physicians to diagnostic disease. However, in some situation, the acceptable reference standard is invasive or costly and does not exist or unreliable. It can be imperfect and results of the reference standard method are not necessarily error- free and cannot be applied to everyone in the study; all these cases point to the conditions in which the gold standard is not available. The use of reference standard including error causes to incorrect separation of patients from healthy population and thus, it cannot be a comparing measure for other diagnostic tests and its results are inaccurate. Therefore, other alternatives methods are needed for evaluation and determine the diagnostic accuracy tests when the gold standard does not exist. Imputation method, correct imperfect reference standard method, the construct reference standard method, latent class models, differential verification, composite reference standard and discrepant analysis are of these alternative methods. Each of these methods, considering its features, advantages, and limitations can be used to evaluate the accuracy of diagnostic test in the absence of gold standard. The present study gave an overview of methods to evaluation of diagnostic accuracy tests when there is no gold standard and the focus of this study was on explain the concept of these solutions, review and compare them and their strengths and weaknesses.