Quality of Randomized Controlled Trials Reporting in the Primary Treatment of Brain Tumors

Authors:
Publisher website
Google Scholar
Abstract
Purpose: To assess the reporting quality of randomized controlled trials (RCTs) in the primary treatment of brain tumors and to identify significant predictors of quality. Patients and Methods: Two investigators searched MEDLINE, EMBASE, and bibliographies of retrieved articles for RCTs in the primary treatment of brain tumors published between January 1990 and December 2004. We assessed the quality of overall reporting and key methodologic factors reporting (allocation concealment, blinding, and intention to treat [ITT]). Two investigators also rated articles independently using items from the revised Consolidated Standards of Reporting Trials statement. A generalized estimated equation was used to generate regression models that identified significant factors associated with quality of reporting. Results: We retrieved 74 relevant RCTs that randomly assigned 14,498 brain tumor patients. The quality of overall reporting has improved during the last 15 years, but eight of the 15 methodologic items were reported in less than 50% of trials. In the appraisal of the reporting quality of key methodologies, allocation concealment, blinding, and adherence to the ITT principle were reported in less than 30% of articles. Multivariable regression models revealed that an impact factor more than 1.66, publication after 1995, and sample size more than 280 were significant factors associated with better overall reporting, whereas complete industrial funding, impact factors more than 2.64, and positive primary outcomes were predictors of higher ratings of the three most important methodologic qualities. Conclusion: Despite improvement in general reporting quality, key methodologies that safeguard against biases may still benefit from better description. Significant factors associated with better reporting may act as surrogates for other characteristics.

This publication has 22 references indexed in Scilit: