Publication and Registration Bias in Randomized Clinical Trials

The research literature has identified the impact of publication bias in randomized clinical trials (RCTs). The prospective registration of RCTs provides a means to mitigate publication bias by allowing researchers to explore a source of trial information, that may or may not be published. This thesis offers a critical appraisal of the compliance in prospective registration and results publication of RCTs. This thesis includes three studies. The first study evaluated the publication bias and prospective registration in digital health RCTs in ClinicalTrials.gov. To expand the scope of this study beyond a single trial registry, the second study examined the global prospective registration, the reporting of the trial registration number (TRN), and the publication rates of digital health RCTs registered in any of the World Health Organization (WHO) registries. To improve the external validity beyond the realm of digital health RCTs, the third study provided an empirical analysis of the prospective registration, the inclusion of TRN, and the selective registration bias in all RCTs registered in any of the WHO registries, and published in any PubMed-indexed journal in 2018, ten years after the adoption of the 7th revision of the Declaration of Helsinki (DoH). The overall results showed that (1) nearly one third of all digital health RCTs remain unpublished (27% and 34.5% in the first and second study respectively), (2) the compliance in prospective registration of RCTs remains low (29.3%, 38% and 41.7% in the first, second and third study respectively), and (3) the compliance in reporting the TRN in published RCTs was low (49%, 52% and 71.2% in the first, second and third study respectively), and (4) a new form of bias was detected “Selective Registration Bias”, with an increased (biased) tendency of the investigators to register a trial only when intending to publish the trial. The majority of RCTs (95.7% and 85.1% in the second and third study respectively), that were registered retrospectively, were registered within one year before submitting the RCT manuscript for journal publication. Addressed to editors, investigators and ethics committees, this thesis proposed several recommendations to inform the future development of best practices to mitigate the publication and registration bias in evidence-based medical research. This thesis proposed several recommendations to inform the future development of best practices to mitigate the publication and registration bias in evidence-based medical research.

Department dc:contributor.department

Dalla Lana School of Public Health

Year dc:date.issued

2019

Author dc:creator

• Al-Durra, Mustafa

Advisor dc:contributor.advisor

• Cafazzo, Joseph A

• Clinical Trials
• Digital Health
• Health Informatics
• Publication Bias
• Randomzied Controlled Trials
• Trial Registry and Registration

Handle dc:identifier.uri

http://hdl.handle.net/1807/102706

OAI identifier oai:identifier

oai:utoronto.scholaris.ca:1807/102706