During evaluation of the integrity of 172 clinical trial publications, we identified 138 unique trial registration documents linked to 157 publications. Eighty-eight (64%) registrations were retrospective. Discrepancies in key trial characteristics - ethics oversight, study timeline, study location, participant number and sample size - between 68 registration documents and their linked publications were reported to journals, publishers and a national regulatory body. Subsequently, revisions to 119/138 registration documents were lodged at the registry website, a median (IQR) of 44 (32-56) months after initial registration. Revisions were made to 56 of the 68 registration documents included in the report: there was a median of 8 (IQR 6-9) changes per document. 79-93% of revised documents contained ≥1 alteration to the primary outcome(s), the secondary outcome(s), and the participant inclusion/exclusion criteria. Changes in each of study location, study timeline, participant age, sample size, and randomization method were made in ≥33% of revised documents. Eight months after journals, publishers and the regulatory body were apprised of the revisions, none of the affected publications has been corrected with an , expression of concern, or retraction. These results call into question whether regulators and publishers regard trial registration documents as helpful in ensuring publication integrity.
Keywords: Research integrity; clinical trials; publishing ethics; trial registration.