Original ArticleVarious randomized designs can be used to evaluate medical tests
Introduction
For scientific purposes, it is worth knowing whether or not a result from a medical test corresponds to the truth. Can the test result be trusted? Does a positive result truly reflect presence of the target disease? These are the first questions that come to the mind in the evaluation of medical tests. Yet, from a patient perspective, mere knowledge about the present true state of things is not enough. Patients with complaints want to get better and those without want to maintain their health.
There are multiple ways in which medical tests can affect patients' health. In addition to the direct effects from the diagnostic procedure, there is the information generated by the test. Providing information on the likely cause or other aspects of one's health problems can have both a positive or a negative effect, albeit usually a limited one [1]. Patients want to be informed about the origin of their complaints, even in the absence of a cure. Such information may enable them to find better ways of coping with the complaints, for example, by developing strategies to limit the disabling impact on their daily activities. Yet, the main effects of medical tests on patient outcome will be the result of subsequent clinical decisions guided by the results of the test. Test results can lead to additional testing, or to starting, modifying, or withholding therapeutic interventions.
In many cases then, it is not only just the present health state that is of interest but also the future course of disease. It then follows that the value of information from using medical tests lies not only in the past (where these complaints come from) or the present (do they correspond to disease), but also in the future. The relevance of diagnostic information is closely related to prognosis and depends on associations between test results and future health and between testing and the outcome of treatment.
In this article, we discuss different methods for evaluating the prognostic value of tests. We start from an evaluation of the prognostic accuracy of tests, and then move on to the ability of tests to predict the outcome of treatment. Section 2 with a presentation of randomized designs for evaluating test–treatment combinations and their effect on patient outcome. Section 3 contains an elaboration of these methods for comparing and evaluating multiple tests.
Section snippets
Evaluating a single test
Several studies have examined the need for duplex ultrasonography (US) in patients with a cervical bruit without further symptoms of cerebrovascular disease. To answer this question, an assessment has to be made of the value of duplex US as a test. Such an evaluation may want to look at the amount of agreement between the index test and the clinical reference standard, the best available method to identify the true condition of the carotid arteries. In this case, the reference test will most
Comparing tests and test strategies
In many clinical situations, there are multiple tests available to examine the presence of the target condition. When one wants to compare 2 competing tests, the designs introduced earlier for the evaluation of a single test have to be adapted slightly.
Conclusions
The examples of published randomized diagnostic trials in this article show that it is feasible to perform thorough evaluations of tests in medicine. Other examples of such trials addressed the evaluation of different diagnostic techniques for ventilation-assisted pneumonia, the comparison of multi-detector row CT with digital subtraction angiography, and the use of transcutaneous oxygen and toe pressure to detect critical limb ischemia [22], [23], [24]. Additional examples can be found in
References (32)
- et al.
Likelihood ratios with confidence: sample size estimation for diagnostic test studies
J Clin Epidemiol
(1991) - et al.
Randomised comparisons of medical tests: sometimes invalid, not always efficient
Lancet
(2000) - et al.
Helicobacter pylori test-and-eradicate versus prompt endoscopy for management of dyspeptic patients: a randomised trial
Lancet
(2000) - et al.
Magnetic-resonance pelvimetry in breech presentation
Lancet
(1998) - et al.
The diagnostic randomized clinical trial is the best solution for management issues in critical limb ischemia
J Clin Epidemiol
(2004 Nov) - et al.
Computed tomography or magnetic resonance imaging for axillary symptoms following treatment of breast carcinoma? A randomized trial
Clin Radiol
(1993) - et al.
Cardiotocography only versus cardiotocography plus PR-interval analysis in intrapartum surveillance: a randomised, multicentre trial. FECG Study Group
Lancet
(2000) - et al.
Randomised study of screening for colorectal cancer with faecal-occult-blood test
Lancet
(1996) - et al.
The evaluation of diagnostic tests: evidence on technical and diagnostic accuracy, impact on patient outcome and cost-effectiveness is needed
J Clin Epidemiol
(2007) - et al.
Adverse psychological events occurring in the first year after predictive testing for Huntington's disease. The Canadian Collaborative Study Predictive Testing
J Med Genet
(1996)
Predictive power of duplex ultrasonography in asymptomatic carotid disease
Ann Intern Med
Utility of routine exercise treadmill testing early after percutaneous coronary intervention
BMC Cardiovasc Disord
Clinical trial designs for predictive marker validation in cancer treatment trials
J Clin Oncol
Antithrombotic treatment of ischemic stroke among patients with occlusion or severe stenosis of the internal carotid artery: a report of the Trial of Org 10172 in Acute Stroke Treatment (TOAST)
Neurology
Low molecular weight heparinoid, ORG 10172 (danaparoid), and outcome after acute ischemic stroke: a randomized controlled trial. The Publications Committee for the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) Investigators
JAMA
Lack of effect of aspirin in asymptomatic patients with carotid bruits and substantial carotid narrowing. The Asymptomatic Cervical Bruit Study Group
Ann Intern Med
Cited by (64)
Benefit assessment of diagnostics – Dealing with incomplete evidence
2023, Zeitschrift fur Evidenz, Fortbildung und Qualitat im GesundheitswesenMetaepidemiological Inventory of Diagnostic Studies on Urodynamics
2020, European Urology FocusPressure ulcer/injury classification today: An international perspective
2020, Journal of Tissue ViabilityCitation Excerpt :From an evidence-based practice perspective accuracy and relative and absolute measurement errors are important characteristics of PU/PI classification, but they do not indicate clinical effectiveness [9,50,56]. Interventional research designs may be needed to evaluate the clinical value of a measurement procedure such as PU/PI classification [57,58]. Depending on the geographic region, health care setting and administrative requirements, different PU/PI classification systems are, and will continue to be used.
GRADE guidelines: 21 part 1. Study design, risk of bias, and indirectness in rating the certainty across a body of evidence for test accuracy
2020, Journal of Clinical EpidemiologyEffect of Xpert MTB/RIF on clinical outcomes in routine care settings: individual patient data meta-analysis
2019, The Lancet Global HealthCitation Excerpt :As in randomised controlled trials (RCTs) of tuberculosis therapeutics,10 surrogate markers for patient-important outcomes for diagnostic-test evaluation—such as test accuracy, turnaround time, time to diagnosis, and time to treatment—might not translate onward to clinical outcomes and could be subject to substantial heterogeneity.11 Although therapeutic outcomes are similarly of primary interest in diagnostic RCTs,12,13 they have received less methodological attention than tuberculosis therapeutic trials.14 Five diagnostic RCTs15–19 across 41 sites in South Africa, Brazil, Zimbabwe, Zambia, and Tanzania, assessing the effect of Xpert among people tested for tuberculosis in routine-care settings have been done.
Issues and Challenges in the Systematic Evaluation of Biomarker Tests
2018, Precision Medicine: Tools and Quantitative Approaches
Adapted from: Lijmer JG, Bossuyt PMM. Diagnostic Testing and Prognosis: The randomized controlled trial in test evaluation research. In: Knottnerus JA, (Ed) The Evidence Base of Clinical Diagnosis. London: BMJ Books, 2001.