514 chapter 13 Were we overtaken by time, or should we have jumped the TIRADS train sooner? To investigate the diagnostic accuracy of [18F]FDG-PET/CT following preselection using TIRADS, we are currently planning a retrospective study of the EfFECTS trial cohort, in which – against our beliefs – we will attempt a blinded reassessment of all baseline ultrasound captures using TIRADS. Challenges in diagnostic randomized controlled trials Randomized controlled trials (RCTs) are considered the gold standard not only in therapeutic but also in diagnostic research, because they best compare clinically relevant health outcomes (i.e., clinical utility, patient outcome efficacy and societal efficacy levels according to the Fryback-Thornbury hierarchy) between groups of patients that undergo different diagnostic strategies [30-32, 34, 759]. Still, diagnostic studies mostly focus on diagnostic accuracy (i.e., test sensitivity and specificity) instead of on clinical utility [33, 759, 760]. Diagnostic RCTs, also called test-treatment trials, are rare. In 2012, it was estimated that only 37 of approximately 22,000 RCTs were diagnostic as compared to therapeutic [761]. To the best of our knowledge, newer estimations are not available, although numbers are hopefully increasing. Scarcity of diagnostic RCTs likely remains because the prevailing cohort study design is much easier to perform (i.e., two different diagnostics can be performed in a blinded fashion without one influencing the other or the outcome) and well accepted for diagnostic studies, and because patient outcomes are generally more difficult to measure and report than diagnostic accuracy [35, 760, 762]. In contrast to therapeutic studies, RCTs are not required prior to implementation of a new diagnostic [33]. Clinical practice guidelines do not seem to consider or demand evidence regarding clinical utility of tests, either, settling for diagnostic accuracy parameters as sufficient evidence instead [763]. In continuing absence of higher-level evidence from RCTs, this may be a vicious circle. Opponents have previously advocated that RCTs are not suitable for imaging studies, as they are too difficult, expensive, and time consuming, which I disagree with [764]. Despite major challenges in methodology, ethics, and outcome reporting, many of which I have experienced myself during the course of the EfFECTS trial, I wholeheartedly agree to the general call for more diagnostic RCTs [35]. Especially in these times where shared-decision making, patient outcomes, and cost-effectiveness of health care strategies are so relevant, merely gathering diagnostic accuracy data should not simply be considered sufficient for a new diagnostic to be implemented in clinical practice. Nonetheless, the methodology of diagnostic RCTs is highly complex and poses different ethical challenges. Some essential RCT elements, including study power, randomization, blinding, and correct analysis of predefined outcome measures can be more challenging to uphold in diagnostic RCTs than in RCTs of treatment interventions [34, 35, 765]. Therefore, when planning a diagnostic study, one should consider whether the randomized design is indeed the best design to measure
RkJQdWJsaXNoZXIy MTk4NDMw