This presentation is part of: Oral Concurrent Session B - Methods in Judgment and Decision Making
RANDOMISATION TURNS DIAGNOSTIC RESEARCH INTO INTERVENTION RESEARCH
K.G.M. Moons, PhD, Cornee J. Biesheuvel, MSc, and Diederick E. Grobbee, Prof. dr. University Medical Center Utrecht, Julius Center for Health Sciences and Primary Care, Utrecht,, Netherlands
In almost every system to grade epidemiological studies according to their level of evidence, randomized studies or meta-analyses of randomized studies receive the highest classification. Although the use of such hierarchies may help to separate the wheat from the chaff, it has also led to misconception and abuse. The popular belief that only randomized studies produce results applicable to clinical practice does a disservice to patient care, clinical investigation and the education of health care professionals. Indeed the ultimate goal of medical care, including diagnostic testing, is to improve patient outcome. Hence, it has widely been advocated that after establishing a test’s diagnostic accuracy, the impact of the test on patient outcome must also be quantified. As studies on patient outcome commonly require a randomized approach to prevent confounding, a randomized design for diagnostic test evaluations is increasingly advocated. However, to demonstrate the beneficial effect of a diagnostic procedure or strategy on patient outcome, we believe that randomisation is by no means a prerequisite. Using randomized studies in diagnostic research certainly changes an essential characteristic of this type of clinical research. It turns diagnostic accuracy (or classification) research into intervention or etiologic research. The nature of the diagnostic question and the object of research determine the appropriate study design. In our view, a test’s effect on patient outcome can be inferred and indeed considered as quantified 1) if the test is meant to include or exclude a disease for which an established reference is available, 2) if a cross-sectional accuracy study has shown the test’s ability to adequately detect the presence or absence of that disease based on the reference, and finally 3) if other (randomised) therapeutic studies have provided evidence on efficacy of the optimal management of this disease. In such instances diagnostic research does not require an additional randomized comparison between two or more test-treatment strategies (one with and one without the test under study) to establish the test’s effect on patient outcome. We will additionally discuss when a randomized design is needed to properly infer on a test’s value to change patient outcome. We conclude that in many instances, a cross-sectional accuracy study will be sufficient to quantify the clinical value of diagnostic tests on patient outcome.
