Article type
Year
Abstract
Background: Publication and selective outcome reporting bias are recognized concerns in the medical literature. Prospective registration of trials may provide transparency in clinical trial conduct and completeness of reporting of the studied outcomes.
Objectives: To determine the completeness of outcomes reported in published reports of randomized controlled trials (RCT) in biomedical journals indexed in MEDLINE; whether this is associated with statistical significance.
Methods: Cohort of reports of RCTs which were prospectively registered in the Netherlands Trial Register (NTR) as planned or ongoing with a latest expected end date of 31 August, 2008. We compared registry entries with the subsequent publications of the trial results identified in MEDLINE. Primary and secondary outcomes specified in the registry entries were compared with those reported in journal publications. Statistical significance of the published outcomes was recorded from the publications.
Results: 599 NTR trials entries with 279 publications were identified. Primary outcomes differed between registry entries and publications for 33% of the trials; secondary outcomes differed for 53%. Statistically significant primary outcomes did not have a higher change than nonsignificant outcomes, but statistically significant effects on secondary outcomes did (complete results will be available at the symposium). On average, only 54% of the reported outcomes were statistically significant.
Conclusions: Incomplete reporting of outcomes within published articles of randomised trials indexed in MEDLINE is still common, but is not always associated with statistical non-significance. Submitted manuscripts could be compared with trial registry entries as a component of peer review.
Objectives: To determine the completeness of outcomes reported in published reports of randomized controlled trials (RCT) in biomedical journals indexed in MEDLINE; whether this is associated with statistical significance.
Methods: Cohort of reports of RCTs which were prospectively registered in the Netherlands Trial Register (NTR) as planned or ongoing with a latest expected end date of 31 August, 2008. We compared registry entries with the subsequent publications of the trial results identified in MEDLINE. Primary and secondary outcomes specified in the registry entries were compared with those reported in journal publications. Statistical significance of the published outcomes was recorded from the publications.
Results: 599 NTR trials entries with 279 publications were identified. Primary outcomes differed between registry entries and publications for 33% of the trials; secondary outcomes differed for 53%. Statistically significant primary outcomes did not have a higher change than nonsignificant outcomes, but statistically significant effects on secondary outcomes did (complete results will be available at the symposium). On average, only 54% of the reported outcomes were statistically significant.
Conclusions: Incomplete reporting of outcomes within published articles of randomised trials indexed in MEDLINE is still common, but is not always associated with statistical non-significance. Submitted manuscripts could be compared with trial registry entries as a component of peer review.