Assessment of frequency and reporting of design changes among clinical drug trials published in influential medical journals

Background Modifications of primary outcome measures after trial initiation can undermine their scientific validity. We aimed to quantify these changes and to characterize potential predictors in clinical drug trials published in high impact factor general medicine journals. Methods The study cohort included all prospective, adult drug clinical trials published in the New England Journal of Medicine, JAMA and Lancet between June 2017 and May 2018. The matching ClinicalTrials.gov entries effective at trial initiation were compared with those effective on July 2018, thereby identifying amendments to primary outcome definitions and assessment timeframes and the planned sample size (defined as >10% change). The primary publications were reviewed for reporting of these amendments. Associations between identified changes and trial characteristics were explored using logistic regression. Results Of the 147 included trials, modifications to primary outcome measures were identified in 80 (54%). Primary outcome definitions, outcome assessment timeframes and the planned sample size were modified in 28 (19%), 12 (8%) and 65 (45%) of registry entries, respectively; of which 21 (75%), 11 (92%), and 33 (51%), respectively, were not reported in the related publications. There were no significant associations between modifications in registry entries and study characteristics. Conclusions Approximately half of trials published in influential medical journals present changes to the design while patient accrual is ongoing, and two thirds of them are unreported. Justification for such modifications should be reported more clearly. Reviewers, editors and readers should consult ClinicalTrials.gov for a more comprehensive report of the evolution of key study methods.