Article type
Year
Abstract
Background: To date, diagnostic studies form a gap in the Cochrane Database of Systematic Reviews, although initiatives in this direction are growing. The methodology and interpretation of diagnostic studies that could yield high-level evidence is more complicated and insufficiently standardized, which may frustrate systematic review and meta-analysis.
Objective: To present problems and possible solutions to include diagnostic studies in systematic reviews, as derived from an exemplary diagnostic RCT in vascular surgery.
Methods: In patients with severe leg ischemia, a (reference) diagnostic test is missing to decide which of these patients needs invasive (operative) vascular treatment and whether this therapy will ultimately relieve their ischemic pain. Hence, a diagnostic RCT was executed, in which the treatment policy was based either on the clinical eye of the surgeon or on the results of a new test combination[1]. Pain relief was the primary endpoint[2].
Results: A diagnostic RCT design was found to be useful, to overcome the lacking reference standard. Using this design minimized most causes of bias, allowed for a power-analysis, and included the patients final outcome. On the other hand, the effect of the diagnostic strategy was difficult to assess due to the influence of the treatment effect on the clinical outcome. Repeated testing was necessary to adjust treatment policy if the clinical condition changed. This precluded randomisation of only those patients with discordant test results[3]. Moreover, adherence to the treatment as prescribed by the test results was difficult and also influenced by the actual condition of the patient.
Conclusion: Any diagnostic study requires careful consideration of the study methodology and the endpoints chosen. Relevant guidelines as to the description of such trials have been proposed in the STARD initiative[4]. Despite the hurdles in the execution of a diagnostic RCT, its design appears preferable to enable meta-analysis of more diagnostic studies and presentation as a Cochrane systematic review, as long as these studies are homogeneous and their endpoints comparable.
References: 1. De Graaff JC, Ubbink DT, Tijssen JG, Legemate DA. Diagnostic randomized clinical trial is best solution for management issues in critical limb ischemia. J Clin Epidemiol. 2004;in press. 2. De Graaff JC, Ubbink DT, Legemate DA, Tijssen JG, Jacobs MJ.Evaluation of toe pressure and transcutaneous oxygen measurements in management of chronic critical leg ischemia: a diagnostic randomized clinical trial. J Vasc Surg. 2003;38 (3):528-34. 3. Bossuyt PM, Lijmer JG, Mol BW. Randomised comparisons of medical tests: sometimes valid, not always efficient. Lancet. 2000;356:1844-1847. 4. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: The STARD Initiative. Ann Intern Med. 2003;138(1):40-4.
Objective: To present problems and possible solutions to include diagnostic studies in systematic reviews, as derived from an exemplary diagnostic RCT in vascular surgery.
Methods: In patients with severe leg ischemia, a (reference) diagnostic test is missing to decide which of these patients needs invasive (operative) vascular treatment and whether this therapy will ultimately relieve their ischemic pain. Hence, a diagnostic RCT was executed, in which the treatment policy was based either on the clinical eye of the surgeon or on the results of a new test combination[1]. Pain relief was the primary endpoint[2].
Results: A diagnostic RCT design was found to be useful, to overcome the lacking reference standard. Using this design minimized most causes of bias, allowed for a power-analysis, and included the patients final outcome. On the other hand, the effect of the diagnostic strategy was difficult to assess due to the influence of the treatment effect on the clinical outcome. Repeated testing was necessary to adjust treatment policy if the clinical condition changed. This precluded randomisation of only those patients with discordant test results[3]. Moreover, adherence to the treatment as prescribed by the test results was difficult and also influenced by the actual condition of the patient.
Conclusion: Any diagnostic study requires careful consideration of the study methodology and the endpoints chosen. Relevant guidelines as to the description of such trials have been proposed in the STARD initiative[4]. Despite the hurdles in the execution of a diagnostic RCT, its design appears preferable to enable meta-analysis of more diagnostic studies and presentation as a Cochrane systematic review, as long as these studies are homogeneous and their endpoints comparable.
References: 1. De Graaff JC, Ubbink DT, Tijssen JG, Legemate DA. Diagnostic randomized clinical trial is best solution for management issues in critical limb ischemia. J Clin Epidemiol. 2004;in press. 2. De Graaff JC, Ubbink DT, Legemate DA, Tijssen JG, Jacobs MJ.Evaluation of toe pressure and transcutaneous oxygen measurements in management of chronic critical leg ischemia: a diagnostic randomized clinical trial. J Vasc Surg. 2003;38 (3):528-34. 3. Bossuyt PM, Lijmer JG, Mol BW. Randomised comparisons of medical tests: sometimes valid, not always efficient. Lancet. 2000;356:1844-1847. 4. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: The STARD Initiative. Ann Intern Med. 2003;138(1):40-4.