International trials registry is missing important information, study findsBMJ 2009; 339 doi: https://doi.org/10.1136/bmj.b3627 (Published 07 September 2009) Cite this as: BMJ 2009;339:b3627
All rapid responses
I concur with Dr. Ross’ findings elucidating the inadequacy of
information for registered clinical trials on clinicaltrials.gov. In my
own micro investigation of select eleven completed clinical trials,
currently under review, I have found such inadequacies across public trial
registers including clinicaltrials.gov. Typically, the trial registration
numbers are not recorded in the trial publications – be it journal
articles, or at times even in case of published trial results. Lack of
registration information makes it extremely difficult to track the trial
across public registries, journals, and results databases. While
registries at times do record the secondary trial IDs that might be
employed in trial publications; lack of comprehensive glossary of
secondary ID information on trial registers makes such information
Although clinicaltrials.gov provides a space to upload trial results,
none of the trial results in my sample were recoded on clinicaltrials.gov.
Nine out of a total of eleven trial results were published; six of which
were published on respective company web-sites while the remaining three
were published exclusively on third-party websites. Only three out of a
total of eleven trials were published in journals. Since trial results are
expected to be published irrespective of the outcome, I concluded that two
trials in my sample were non-transparent.
Triangulating the published information for trials across sources,
one out of nine published trials was found to be discrepant due to
inconsistency in the total number of enrolled participants and study
locations. Triangulation thus lead to a useful taxonomy of trials labeled
as transparent, non-transparent, and discrepant.
Trial start date for all the trials in my small sample were recorded,
however I found an average time lag of over 13 months between trial start
date and the date of trial registration on the public registry. This huge
time lag apparently seemed to undermine, the very purpose of registering a
trial on the public registry. Moreover, most of the trials in the sample
ran for duration of 3-6 months. This fact further raised suspicions about
retrospective and perhaps even selective registration on the registry.
Chan et al 2004(1) have already reported on the selective reporting
of outcomes in case of randomized trials. Yet, a rigorous analysis needs
to be carried out for every indicator on the public trial registries.
Expounding on the reported and expected measures for every indicator on
public trial registries will hopefully address the inadequacies of trial
registries, and bring them up to speed with the prerequisites to employ
these registries for ongoing monitoring and evaluation of clinical
(1) Chan AW, Hrobjartsson A, Haahr MT, Gotzsche PC, Altman DG. Empirical
evidence for selective reporting of outcomes in randomized trials:
comparison of protocols to published articles. JAMA. 2004; 291: 2457-65.
Competing interests: No competing interests