Prospective registration was associated with a reduced risk of bias for randomized controlled trials: a meta-research study

Abstract

ObjectivesTo investigate the association between trial registration and the risk of bias (RoB). Study Design and SettingsRandomized controlled trials (RCTs) from systematic reviews of medication-related harm published between January 1, 2015, and January 1, 2020, were used, assessing first post and start dates on open registries along with RoB. Multivariable logistic regression analyses for both individual safeguards and overall RoB by registration status (ie, nonregistered, retrospectively registered, and prospectively registered) were conducted. ResultsA total of 2030 RCTs were identified from 151 systematic reviews; 65.46% (851/1300) were registered prospectively and 34.54% (449/1300) retrospectively among 64.04% (1300/2030) registered RCTs. Regression analysis indicated that when compared to nonregistered trials, prospective registration was associated with safeguards against bias using randomization sequence generation (odds ratio [OR] = 1.51, 95% CI: 1.09, 2.07), allocation concealment (OR = 1.69, 95% CI: 1.22, 2.36), blinding of outcome assessors (OR = 1.65, 95% CI: 1.14, 2.38), as well as lower overall RoB (OR = 2.04, 95% CI: 1.19, 3.50). When comparing prospectively and retrospectively registered trials, prospective registration was more likely to have lower overall and individual RoB, but this was not statistically significant. Prospective registration was associated with blinding of participants (OR = 1.70, 95% CI: 1.26, 2.30) and health-care providers (OR = 1.68, 95% CI: 1.25, 2.28), but not lower overall RoB as compared to retrospective registration. ConclusionProspectively registered trials were more likely than nonregistered trials to implement adequate procedures against bias. Prospectively registered trials may also serve as an indicator of lower overall RoB in RCTs.