Registering Clinical Trial Results

Abstract

IN 2000, THE NATIONAL INSTITUTES OF HEALTH (NIH) established ClinicalTrials.gov in response to congressional action 2 years earlier. Sponsors were required to register only those clinical trials assessing drugs for treatment of serious or life-threatening diseases. Compliance with even this limited scope of required registration was poor until 2 related events occurred in 2004: first, in June of that year, the New York State attorney general sued GlaxoSmithKline for failing to publish the negative results of a trial of paroxetine in pediatric patients. Second, in September, the International Committee of Medical Journal Editors announced that, effective July 2005, reports of clinical trials would be accepted for publication only if the trials had been properly registered. The number of trials registered on ClinicalTrials.gov increased substantially after these events. Before the US Food and Drug Administration (FDA) Amendments Act of 2007 (FDAAA), trial registration under US law did not require registering the results of clinical trials. Instead, registration was limited to disclosing the purpose of the trial, the eligibility criteria for patients in the trial, and the location of trial sites. In the FDAAA, Congress for the first time directed the NIH to require sponsors of clinical trials of drugs to post at least some results from those trials on ClinicalTrials.gov and also directed the NIH to adopt new regulations by September 2010 to require sponsors to post more complete results on ClinicalTrials.gov. The NIH is currently developing this regulation, which will be the important next step in the registration of clinical trials and an important opportunity for the NIH to improve the quality and reliability of clinical research of drugs. In addition, Congress directed the NIH to require sponsors to post the basic results data from clinical trials (other than phase 1 trials) of drugs that have been approved by the FDA. The basic results include the demographic and baseline characteristics of the patients, the number of patients lost to follow-up, the number excluded from the analysis, and the primary and secondary outcome measures. For the primary and secondary outcomes, results data must include a table of values along with the appropriate tests of the statistical significance of these values. A narrative summary of the clinical data, therefore, fails to satisfy this registration requirement. However, Congress permitted sponsors to delay posting even basic results until after the FDA approved the drug evaluated in the trial. If the FDA had previously approved the drug but the trial studied a new use of the drug “not included in the labeling of the approved drug,” the sponsor is permitted to delay posting the results from the trial until after the FDA approves (or rejects) the new use. For trials investigating an unapproved use of an approved drug, posting the results may be delayed up to 3 years after the trial is completed. This means that while a clinical trial may have been properly registered and the trial may have been long since completed and the data analyzed and reported to the FDA, the sponsor is not obligated to post even the basic results on ClinicalTrials.gov until the FDA acts on the drug or the new use under investigation in the trial. In the FDAAA, however, Congress directed the NIH, in adopting the new regulation expanding ClinicalTrials.gov, to consider whether to require sponsors to post results from trials of drugs not approved by the FDA. If the NIH decides that the new regulation should require posting results from trials of unapproved drugs, many of the delays that are now permissible would be eliminated. The NIH should decide in favor of this requirement. The other 2 most important issues facing the NIH in developing this regulation are, first, whether to require sponsors to include on ClinicalTrials.gov the full protocol for each trial, and, second, whether to require sponsors to do the same for all patient-level data for the primary and secondary outcomes. Even though sponsors are required to submit trial protocols to the FDA as part of the Investigational New Drug application process, the FDA traditionally has treated such applications, including protocols, as confidential proprietary information and does not make them public. Similarly, the FDA generally does not make public the patientlevel data submitted by sponsors. Instead, the FDA makes public, after approval, summaries of the data as part of its