Assessing blinding in trials of psychiatric disorders: a meta-analysis based on blinding index.

The assessment of blinding in RCTs is rarely performed. Currently most studies that do report data on evaluation of blinding merely report percentages of correct guessing, not taking into account correct guessing by chance. Blinding assessment using the blinding index (BI) has never been performed in a systematic review on studies of major psychiatric disorders. This study is a systematic review of psychiatric randomized control trials using the BI as a chance-corrected measurement of blinding, a tool to analyze and understand the patterns of blinding across studies of major psychiatric disorders with available data. Of 2467 psychiatric RCTs from 2000 to 2010, 66 reported on blinding and 40 studies were found to have enough information on evaluation of blinding to be analyzed using the BI. The experimental treatment groups had an average BI value of 0.14 and the control groups had an average BI value of 0.00. The most common BI scenario was random-random, indicating ideal blinding. A positive correlation between effect size and more correct guesses was also found. Overall, based on BI values and the most common blinding scenario, the published articles on major psychiatric disorders from 2000 to 2010, which reported on blinding assessment for patients, were effectively blinded.