Objective: To evaluate discrepancies between trial registry entries and final reports of randomized controlled trials (RCTs) published in major general surgical journals.
Background: Health care decisions are based on published results in peer-reviewed journals. Mandatory trial registration was introduced to increase transparency and reduce publication and outcome reporting bias.
Methods: The discrepancy rate between trial registry entries and final reports of all RCTs published during 2010 in the Annals of Surgery, Archives of Surgery, and British Journal of Surgery was evaluated.
Results: Of 596 identified studies, 545 were excluded because they were not RCTs or interim reports/secondary analysis of RCTs or because of missing trial registry information.
In the remaining 51 RCTs, prospective registration was found in 9.8% (n = 5), registration during trial conduct in 33.3% (n = 17), and retrospective registration in 56.9% (n = 29), respectively.
For the primary and secondary outcomes, there was no discrepancy in 54.9% and 33.3%, complete omission in 7.8% and 31.3%, new introduction in 7.8% and 39.2%, a change in definition in 9.8% and 5.8%, downgrading from primary to secondary in 21.6%, and upgrading from secondary to primary in 13.7%. There were few discrepancies in randomization, blinding, and intervention and some in targeted sample size and inclusion/exclusion criteria.
Conclusions: When interpreting the results of surgical RCTs, the possibility of selective reporting, and thus outcome reporting bias, has to be kept in mind. For future trials, prospective registration should be strictly respected with the ultimate goal to increase transparency and contribute to high-level evidence reports for optimal patient care in surgery.
This study evaluated discrepancy rates between trial registry entries and final reports of randomized controlled trials published during 2010 in 3 major general surgery journals. Considerable discrepancies were found concerning outcomes, few regarding randomization, blinding, and interventions, and some concerning targeted sample size and inclusion/exclusion criteria. Potential reporting bias needs to be considered when interpreting individual study results.
*Department of Surgery, Basel University Hospital, Basel, Switzerland
†Department of Biostatistics, The University of Liverpool, Liverpool, United Kingdom.
Reprints: Rachel Rosenthal, MD, MSc, Department of Surgery, Basel University Hospital, Spitalstrasse 26, CH-4031 Basel, Switzerland. E-mail: firstname.lastname@example.org.
There is no funding associated with this research. Part of the work was undertaken as dissertation of R.R. to obtain a Master of Science in Clinical Research degree at the University of Liverpool.
Disclosure: The authors declare no conflicts of interest.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.annalsofsurgery.com).