Test Research Versus Diagnostic Research (Opinion‪)‬

The diagnostic workup starts with a patient presenting with symptoms or signs suggestive of a particular disease. The workup is commonly a consecutive process starting with medical history and physical examination and simple tests followed by more burdensome and costly diagnostic procedures. Generally, after each test all available results are converted (often implicitly) to a probability of disease, which in turn directs decisions for additional testing or initiation of appropriate treatment. Setting a diagnosis is a multitest or multivariable process of estimating and updating the diagnostic probability of disease presence given combinations of test results. Each test may be more or less burdensome to the patient, time-consuming, and/or costly. Different tests often provide to various degrees the same information because they are all associated with the same underlying disorder. Relevant for physicians is to know which tests are redundant and which have true, independent predictive value for the presence or absence of the target disease. Accordingly, studies of diagnostic accuracy should demonstrate which (subsequent) test results truly increase or decrease the probability of disease presence as estimated from the previous results, and to what extent. Various reviews have demonstrated that the majority of published studies of diagnostic accuracy still have methodologic flaws in design or analysis or provide results with limited practical applicability (1-3). This has been attributed to the absence of a proper methodologic framework for diagnostic test evaluations as, for example, exists for studies of therapies and etiologic factors and has motivated various researchers to establish frameworks for studies of diagnostic accuracy, such as the recent STARD Initiative (4-12). In our view, an issue that has received too little attention in most of these methodologic essays is the difference between test research and diagnostic research.