Skip to Main Content

STAT downloaded data on all registered trials on as of Sept. 11, 2017 — more than 250,000 studies. We filtered the data to remove any trials whose findings were not required by law to be reported. This included trials completed prior to 2008, “Phase 1” safety trials, those completed less than a year earlier, and trials exempted by the Food and Drug Administration to protect proprietary commercial information during the approval process for new treatments.

NIH provided STAT a list of these exempt trials — which is available only by request — but it later said the list might be incomplete because companies were not required to request a reporting delay to qualify for one. Some studies failed to disclose the location of trial sites; in those cases, we assumed that at least one site was in the United States, making the trial subject to the reporting law.

That left about 18,700 trials for which results were due. STAT examined whether these trials’ results were submitted within the legal deadline — one year after a trial’s completion or termination. Each organization’s performance was then compared with its performance as of Sept. 11, 2015.

The National Institutes of Health, which operates, requires that each study specify a “responsible party” to submit data. It can be the trial sponsor — such as a drug company or the NIH itself — or the principal investigator, who is typically a grant recipient, contract research scientist, or drug company employee.

In its 2015 investigation of reporting, STAT considered responsible parties and research collaborators equally responsible for providing results. This approach emphasized each researcher’s ethical obligation to ensure that results are made public. Recent changes in how NIH structures data made it impractical to compare 2015 to 2017 performance using this earlier approach. For this article, STAT followed the letter of the law, reviewing each research organization’s performance based only on trials for which it was designated the responsible party.

A more detailed description of our methodology can be found here.