MP46-12 STATISTICAL METHODS IN RANDOMIZED CONTROLLED TRIALS IN THE UROLOGICAL LITERATURE: A SYSTEMATIC REVIEW

Abstract

You have accessJournal of UrologyGeneral & Epidemiological Trends & Socioeconomics: Value of Care: Cost and Outcomes Measures II1 Apr 2016MP46-12 STATISTICAL METHODS IN RANDOMIZED CONTROLLED TRIALS IN THE UROLOGICAL LITERATURE: A SYSTEMATIC REVIEW Eugene B Cone, Vikram Narayan, Daniel Smith, Philip Dahm, and Charles D Scales Eugene B ConeEugene B Cone More articles by this author , Vikram NarayanVikram Narayan More articles by this author , Daniel SmithDaniel Smith More articles by this author , Philip DahmPhilip Dahm More articles by this author , and Charles D ScalesCharles D Scales More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2016.02.310AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES Given that randomized controlled trials (RCTs) provide the highest level of clinical evidence it is critical that their statistical methodology be sound and clearly reported. We performed a systematic review of reported statistical methods of RCTs published in 2013 as compared to RCTs in 2004 and 1996 with the objective of quantifying any improvement in statistical reporting. METHODS All RCTs involving human subjects published in 4 leading urology journals in 2013 were identified for formal statistical review, and compared to a prior statistical analysis of studies from 1996 and 2004 with the same inclusion criteria. Two independent reviewers abstracted data regarding statistical tests used and specificity of methodology reported using a standardized evaluation form. We settled disagreements by consensus and third party referee. Chi-square, students T, and ANOVA were used to analyze the results. Our objective was to compare statistical methods for urologic clinical trials in 2013 as compared to 2004 and 1996. RESULTS A total of 82 RCTs published in 2013 met inclusion criteria and were compared to 65 and 87 RCTs from 1996 and 2004, respectively. Similar to earlier years, the majority of RCTs used two armed designs (84.1%) with the median sample size (IQR) per arm increasing from 32 (19, 94) in 1996 to 45.5 (25, 116) in 2004 and 61 (39, 131) in 2013 (p<0.001). Reporting of important statistical criteria has improved substantially, including sample size justification, which statistical tests were used, and effect size estimates for primary outcomes (see Table). Importantly, reporting of other key aspects of the statistical methodology such as multiplicity adjustment (5%), test sidedness (53%), and statistical software used (54%) continue to be low. The most common statistical tests used in 2013 continue to be chi square (42.7%) and T tests (45.1%), while regression analysis is becoming increasingly common. CONCLUSIONS The quality of statistical reporting in the urological literature has improved over the last 2 decades. However, considerable opportunities to improve statistical reporting remain and should be addressed by educational efforts in applied statistics for authors, reviewers and journal editors. © 2016FiguresReferencesRelatedDetails Volume 195Issue 4SApril 2016Page: e621 Advertisement Copyright & Permissions© 2016MetricsAuthor Information Eugene B Cone More articles by this author Vikram Narayan More articles by this author Daniel Smith More articles by this author Philip Dahm More articles by this author Charles D Scales More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...