To determine the publication rate of motor-rehabilitation trials poststroke and the consistency between registry records and their corresponding main publications in trial design, primary objectives and outcomes, eligibility criteria, and sample size.
We searched 18 clinical trial registries to identify randomized controlled trials of motor-based stroke rehabilitation registered after July 2005 and completed before April 2017. Eligible trials included adults with stroke, with at least one outcome measure related to motor function. Information in the registry records was compared with that of their main publications, if any.
Three hundred twenty-three trials met our eligibility criteria; we were unable to find a peer-reviewed publication reporting the main findings for 46% (150/323) of these. Of the 169 trials with peer-reviewed articles published in English, 141 (83%) were consistent with the registry record in trial design, 100 (59%) were consistent in primary objectives, 71 (42%) were consistent in primary outcomes, 28 (17%) were consistent in eligibility criteria, and 74 (44%) were consistent in sample size.
Approximately half of motor-based stroke rehabilitation trials were not published, even more than 3 years after the end of the trial. When main publications were found, they substantially deviated from information in the registry record. These findings highlight the importance of trial registries for identifying unpublished stroke rehabilitation trials and of searching trial registries when conducting systematic reviews and meta-analysis to help ensure that reviews are unbiased.