[Design and analysis of a study on reliability and validity of diagnostic tests in urological clinical research]

Diagnosis is a process characterized by uncertainty, which the researcher can approach through knowledge based on the theory of probability, and, as such, it is part of the general decision-making process. To use a test is a process that helps to confirm or refuse the initial probability of the patient to be sick. In this article, we analyze the essential characteristics of diagnostic tests and the principles of effective evaluation of the studies about these tests. We explain in detail what reliability of an instrument is, the types of reliability that there are, and how to analyze them. So that, we introduce concepts such as Pearson's correlation coefficient, intra-class correlation coefficient, and Cohen's Kappa coefficient. Another topic described in this article is measurement of validity, defining the types of validity to be considered on an instrument. Finally, we treat the general principles that rule the design and execution of a study for the evaluation of diagnostic tests, the structure followed by these studies, and mainly the analysis, intention of which is to evaluate to what extent that diagnostic test differentiates between subjects with and without the disease studied. We explore the concepts of sensitivity, specificity, predictive values, and probability ratios, and ROC curves are explained.