The study, a review of 323 articles published last year in leading medical journals, found that only 147 of the clinical trials — 45.5 percent — were properly registered before the end of the trial in a way that clearly stated the main outcomes being assessed. Even among the articles that were registered, almost a third had discrepancies between the outcomes described in the registry and the ones ultimately reported.
Of the trials that were not registered properly, 89 — more than half — were never entered in the National Institutes of Health’s clinical trials registry. The report appears in the Sept. 2 issue of The Journal of the American Medical Association.
“We’re asking people to tell readers honestly and transparently, ‘What did you do, and what did you find?’ ” said an author of the paper, David Moher, senior scientist at the Ottawa Hospital Research Institute in Canada.
Dr. Moher added that bias could easily be introduced when investigators changed the primary outcome they were measuring — moving the goalpost, so to speak. That might happen, for example, if a clinical trial is designed to determine whether a drug increases survival rates but investigators change the primary outcome to assessing if the drug improves quality of life.