Trial registration and results reporting (ClinicalTrials.gov, EU CTR)

Registration means entering a trial's protocol summary in a public database before the first participant is enrolled: the question, the design, the arms, the eligibility criteria, the primary and secondary endpoints, the planned size and the sponsor. The record gets an identifier (an NCT number on ClinicalTrials.gov, an EU CT number in the European system, an ISRCTN, a CTRI number in India, an ACTRN in Australia and New Zealand) that appears in every paper about the trial and on this site's trial pages. The World Health Organization's registry platform pulls the national registries together. Since 2005 the major medical journals have refused to publish unregistered trials, United States law has required registration and, since 2017, the posting of summary results within a year of completion for most drug and device trials, and the European Union's Clinical Trials Regulation requires the same through its Clinical Trials Information System.

Registration exists because of two well-documented failures. Publication bias: trials with positive results were published and negative ones were not, so the literature overstated what treatments did. Outcome switching: a trial that missed its primary endpoint would report a secondary or post-hoc one as if it had been the plan. A registry entry with a date stamp makes both visible. It also lets patients and doctors find open trials, which is how the trial finder on this site works, and lets reviewers count the trials that finished years ago and never reported, a group that remains stubbornly large in audits of registries.

The corpus reflects the history. Phase 3 trials in the registration era carry an NCT or equivalent identifier on their pages; older trials such as CALGB 9343, START-B, CHHiP, PRIME II and IMPORT LOW predate the requirement or were registered only with ISRCTN, and the corpus rules that check for identifiers list them as exceptions for that reason. DYNAMIC is registered with the Australian registry, and CIRCULATE-Japan with Japan's. When reading any trial, the registry record is the place to check whether the endpoint reported was the endpoint planned, when the trial was registered relative to its start, and whether results have been posted where the paper has not appeared.