Methods for Evaluating Medical Tests and Biomarkers

Joris de Groot, Christiana Naaktgeboren, Hans Reitsma, Carl Moons Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, 3584 CG Utrecht, Netherlands Correspondence: Joris de Groot (j.degroot-17@umcutrecht.nl) A major contributor to the rising problem of overdiagnosis, with the subsequent risk of overtreatment, is the development of highly sensitive diagnostic technologies that challenge and sometimes expand prevailing disease definitions. Although the value of such new technology might be that it identifies new, milder, earlier or even other abnormalities, it is uncertain whether these “abnormalities” provide the same diagnostic and prognostic information, or require the same treatment as the original targeted disease. It is often unclear which of the newly detected abnormalities are benign and how many people might be diagnosed and treated unnecessarily as a result of widespread introduction of the new test. Failure to investigate the clinical relevance of broadening disease definitions which include these newly detected abnormalities may therefore lead to overdiagnosis and overtreatment. Spiral CT used in diagnosing pulmonary embolism, detecting small subsegmental embolisms, has been mentioned as an example of such situation. On-going technological advancements in medicine will only further increase the development of new diagnostic technologies that challenge existing disease definitions. We show why traditional cross-sectional diagnostic accuracy studies are insufficient to evaluate such new tests and how methodology for assessing their performance should catch up and keep pace with present-day technological developments. It is crucial to improve data analysis and presentation of current diagnostic studies, to make better use of existing data, or ultimately perform test-treatment trials to answer the question whether introduction of a new high sensitive test will in fact improve patient relevant outcomes, or rather induce overdiagnosis and overtreatment.