Is the dissemination bias still a problem?




Poster session 2 Thursday: Evidence synthesis - methods / improving conduct and reporting


Thursday 14 September 2017 - 12:30 to 14:00


All authors in correct order:

Ballesteros M1, Urrutia G1, Bonfill X1, Djulbegovic B2, Gich I1, Roque M1
1 Iberoamerican Cochrane Centre, Spain
2 Moffitt Cancer Center, United States
Presenting author and contact person

Presenting author:

Xavier Bonfill Cosp

Contact person:

Abstract text
Background: A randomised-controlled trial (RCT) should only be considered completed once it is published.
Objectives: To establish the publication rate of RCTs and to analyse their determinants. We also estimate the location and time-lag bias and describe the bibliometric characteristics of the publications.
Methods:This retrospective cohort observational study detected any protocol and publication resulting from RCTs involving cancer-related drug products authorised by the Spanish Agency of Medicines and Medical Devices, between 1999 and 2003. Searches were conducted on MEDLINE, EMBASE, CENTRAL, and Google Scholar until 2015. We also consulted conference proceedings and clinical trials registries platforms. Whenever the database searches were unsuccessful we contacted the national coordinator in Spain, the study sponsor and the research ethics committee of the coordinating institution.
Results:We identified 168 publications of 303 RCTs. The publication rate was 55.4% after a mean follow-up period of 12 years. After including other forms of publication it increasing to 57.1%. The only factor associated to the likelihood of non-publication was the study setting favouring only-national RCTs (OR 2.7; 95% CI 1.5−4.8). We were able to determine the cause of nonpublication in 57 (42.2%), 38 (66.6%) of them having been closed prematurely for different reasons. The overall mean length of time since the authorisation to the publication in the journal was 6.5 years (ranging from 2 to 14 years). The time to publication was shorter for studies sponsored by the pharmaceutical industry, those with favourable results, and those involving less than 1000 patients. No differences were seen according to the study setting. All index publications were disseminated in international journals, 78.6% of them in specialised oncology journals. The mean impact factor of the index publications was 13.3 (ranging from 1.3 to 53.4).
Conclusions: About half of the RCTs on cancer during the target period have not been published. The national setting is a factor associated to non-publication whereas the direction of results determines its dissemination (impact factor and timely publication).