Why an automated tracker finds poor sharing of clinical trial results for an academic sponsor: a bibliometric analysis

• Published in: Scientometrics 2021-02, Vol.126 (2), p.1239-1248
• Main Authors: Decullier, E., Tang, P. V., Huot, L., Maisonneuve, H.
• Format: Article
• Language: English
• Subjects: Bibliometrics; Clinical trials; Computer Science; Corporate sponsorship; Documents; Hospices; Information Storage and Retrieval; Library Science; Visibility; Websites
• ISSN: 0138-9130; 1588-2861
• DOI: 10.1007/s11192-020-03775-0

Objective: a Researchers have a duty to make the results of their research available publicly. According to the FDA, clinical trial data must be made public less than 12 months after the end of the trial. The trialstracker website ( https://trialstracker.ebmdatalab.net/#/ ) ranks sponsors according to the proportion of unpublished clinical trials by extracting data from a prospective trial registry: clinicaltrials.gov. A low proportion of published trials could be explained by a failure to share results or the sponsor incorrectly filling out the clinicaltrials.gov database. The objective of this study was to assess for which of these reasons one academic sponsor was shown to have a high proportion of unpublished results. Methods: Bibliographic searches were performed and followed up with an email contact for the 104 HCL (Hospices Civils de Lyon, France) trials used by trialstracker to assess publication status. Results: Trialstracker considered that only 25 out of the 104 HCL trials had been published. By searching PubMed between February and April 2019, we rapidly identified publications for 27 further trials. A more advanced search and contact with the investigators allowed us to identify 24 more published trials. Overall, the proportion of trials published was 72.1% ( n  = 75) i.e. 3 times higher than the proportion provided by trialstracker. Even when restricted to the results found via a simple search, the proportion of publications was still higher with a two-fold increase. Conclusion: We found that trialstracker greatly underestimated the number of publications. All actors should therefore contribute to improving the visibility of clinical trial results by providing NCT numbers for all publications (investigator), and by updating clinicaltrials.gov (sponsor and investigator).