Quality of reporting of randomization methodology in nephrology trials

Abstract

Steven Fishbane1, Azzour D. Hazzan1, Shayan Shirazian2, Ezra Israel1 and Giovanni F. Strippoli3–6 1Hofstra North Shore-LIJ School of Medicine, Great Neck, New York, USA; 2Winthrop-University Hospital, Mineola, New York, USA; 3Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy; 4University of Sydney School of Public Health, Sydney, Australia; 5Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy; and 6Diaverum Medical Scientific Office, Lund, Sweden. Correspondence: Steven Fishbane, Hofstra North ShoreLIJ School of Medicine, 100 Community Drive, 2nd Floor, Great Neck, New York 11021, USA. E-mail: sfishbane@nshs.edu T he randomized controlled trial (RCT) is the ideal study design to address intervention questions.1 In an RCT, patients are allocated to an intervention or control group on the basis of chance. The strengths of this methodology include that it minimizes confounding by producing groups that are comparable in terms of baseline characteristics. In theory, this provides comparable prognoses of the groups prior to intervention. The reliability of an individual RCT, however, is affected by the methodological aspects of randomization procedures, which are reflected by the completeness and quality of reporting. When reporting is inadequate or superficial, confidence in the credibility of study procedures and results is diminished. There are certain attributes that define the quality of the procedures and reporting of RCTs. The Consolidated Standards of Reporting Trials (CONSORT) Statement, a widely accepted set of standards for reporting of clinical trials, lists four minimum requirements for reporting of randomization methodology.2 The first is a clear explanation of the method by which the random sequence is generated. Second is an explanation of the type of randomization as simple ran domization, permuted block (to avoid imbalances in allocation), stratification (to balance the distribution of certain baseline risk factors), or a combination of these techniques. Third is allocation concealment, the method of preventing study personnel from having awareness of treatment assignment before enrolling patients. Finally, there needs to be full reporting on the methods of implementation of randomization procedures.2 It is well worth noting that the CONSORT Statement goes well beyond reporting on randomization, and other aspects are highly valuable.2 Clinicians, guideline groups, and policy makers rely on the results of RCTs. It is clear, therefore, that the quality of design, conduct, and reporting of RCTs is a fundamental step in determining valid and applicable results. Studies with suboptimal quality may be misleading as to the true effects of interventions. The ability to advance clinical care may hence be hindered, while both resources and the volunteerism of study subjects may be squandered. High-quality reporting of methodology increases the credibility, applicability, and generalizability of methods and results. It should be noted that nephrology publishes the fewest RCTs of any medical specialty.3 When the number of RCTs in a field is less, then the contribution of each individual study to the accumulated literature and knowledge base is proportionately greater. As a result there is an even heightened burden of responsibility for high-quality reporting. The purpose of this Editorial is to discuss quality of randomization reporting in clinical trials in nephrology. We reviewed nephrology RCTs published in 13 nephrology and four general medical journals between 1 July 2010 and 30 June 2011 (see Supplementary Methods and Supplementary Figure S1 online) to explore these domains. A total of 74 articles met the inclusion criteria, and their characteristics are reported in Table 1. Sixty of 74 (82%) were published in nephrology journals, and 52 of 74 (70%) had sample sizes of fewer than 200 patients. The four CONSORT indicators of the quality of randomization could be ascertained from all articles. The method of sequence generation was not reported in 44 of 74 studies (59.5%) (Table 2). Randomization type was not reported in 29 of 74 publications (39.2%). When the type was reported, some form of stratification, permuted block, or a combination was used in 48.6% of studies (permuted block alone in 12.2%, stratification alone in 20.3%). Most studies that did not report randomization type used simple randomization, on the basis of responses provided by the authors of 16 studies. The method of allocation concealment was not reported in 43 of 74 studies (58.1%); when it was reported, central allocation was the most common procedure. Finally, in the vast majority of cases, information was not provided on how and by whom randomization was implemented. Quality of reporting of randomization methodology in nephrology trials