- Defining Clinical Trial Transparency
- Clinical Trial Spin and Suppression
- Benefits and Costs of Transparency
- Legislative Strategies to Encourage Transparency
- Bottom-Up Strategies for Clinical Trial Transparency
- Conclusions: Towards Greater Transparency
- Clinical Trial Transparency References
Is current legislation sufficient to ensure the clinical trials of high integrity? I will review the US legislation that governs clinical trial reporting, starting with HOPE act in the 1980s and leading to the the FDA Amendments Act 2007 (FDAAA).
The HOPE Act of 1988 was the first major law intended to increase the transparency of clinical trials. Section 2317 of the HOPE Act directed the Department of Health and Human Services to “collect and disseminate information about HIV research, treatment, and prevention”  The HOPE Act led to the establishment of the first public, computerized databases for clinical trials, ACTIS, which recorded clinical trials for HIV/AIDS, and was intended to help both patients become aware of clinical trials of new HIV drugs and help physicians and researchers to more easily recruit subjects.
A decade passed before the next major legislative initiative to improve clinical trial transparency. The Food and Drug Administration Modernization Act of 1997, known as FDAMA, contained important new directives to expand clinical trial registration. Section 113 of FDAMA established clinicaltrials.gov and required registration of all clinical studies for “serious or life-threatening conditions,” which meant that not all clinical trials were necessarily submitted. FDAMA did not insist on the publication of trial results — FDAMA allowed for results to be posted, but these results were required to have already been published in a peer reviewed journal. The data that FDAMA required to be submitted to ClinicalTrials.gov includes: “a description of the purpose of each experimental drug; patient eligibility criteria; a description of the location of clinical trial sites; and a point of contact for patients wanting to enroll in the clinical trial.” Clinical Trials.gov was originally intended primarily as a clearinghouse to allow patients better access to clinical trials, rather than explicitly aiming to encourage dissemination of trial results.
FDAMA provided a valuable first step, but by 2005 it was apparent that not all parties were in full compliance with FDAMA. An FDA investigation compared the trials submitted to ClinicalTrials.gov to the primary registrations of trial protocols at CDER. Only 35% (239/688) of the protocols that should have been submitted to ClinicalTrials.gov were actually in the database. Companies are also able to submit voluntary listings for trials that are not strictly required by FDAMA. Only 35 voluntary listings had been contributed to the database, representing only a miniscule fraction of all Phase I, II, and III trials that had been performed between 1998 and 2005. NIH, which also performs many clinical trials, had a much higher rate of registration with ClinicalTrials.gov (90%) vs. industry (30%). This indicates that industrial sponsors of trials are still somewhat reluctant to make even basic information about trials easily accessible. It may be necessary to exact stricter penalties for non-compliance with the FDAMA requirements, or to encourage participation positively by offering benefits such as accelerated review to companies that file these mandatory submissions properly.
The most recent legislation affecting reporting of clinical trials is the FDA Amendment Act of 2007, or FDAAA. Section VIII of FDAAA mandates registration of all trials other than phase I trials for drugs, biologics, and devices, going far beyond the “drugs for serious or life-threatening conditions” that were the main target of FDAMA. The requirements of FDAAA are narrower than recommended by policy documents from the ICMJE and the World Health Organization ,
In addition to requiring clinical trials to be registered, FDAAA requires that clinical trial registrations on clinicaltrials.gov be linked to any results already available on the FDA website, to make it easier for researchers to access results that are already available. Going beyond mere clinical trial registration, FDAAA required that the “Basic Results Database” be set up by the National Institutes of Health by September 2008. This database includes information about a) participant demographics and baseline characteristics; b) primary and secondary outcomes and statistical analyses; c) disclosure of agreements between sponsors and nonemployees that restrict researchers from publishing results (although such agreements are not prohibited). FDAA also sets time limits to ensure that trial data is reported promptly; data must be made available in the Basic Results Database within 12 months of trial completion or 30 days of FDA approval. Data on adverse events must be collected and reported within 2 years. Importantly, FDAAA specified a three year process to evaluate its performance and assess the need for additional regulations to further increase transparency. The FDAAA doesn’t attempt to solve every possible issue – rules governing informed consent, research ethics and conduct are not covered, as they are assumed to be regulated by existing laws such as CFR 45 Part 46 and the Belmont Report.
FDAAA only works prospectively, for trials started after the bill became law. The vast majority of drug trial data remains locked inside the FDA and pharmaceutical companies. This represents a rich dataset that could be useful for evaluating the safety of current drugs (almost all of which were approved based on studies done before FDAAA) and predicating the safety and efficacy of new ones. Although it may not be practical or cost-effective to mandate the wholesale publishing of all of this data, it would be beneficial to have some provision (other than the slow and relatively ineffective Freedom of Information Act requests) for researchers to access this important data .
The FDAAA also does not require the submission of Phase I data. This is a major limitation, because Phase I studies constitute the largest group of investigations and explicitly assess safety, which is of primary concern to the FDA. The results of Phase I studies are also closely watched by investors, and the failure to publicly disclose trial protocols and results can allow companies to sweep failures under the rug while trumpeting successes, and could give investors a misleading impression of a company’s success rate and prospects.
FDAAA does not cover observational studies, such as post-approval safety monitoring. As more such studies are performed and more frequently used to guide clinical decisions, it will be important that the results of observational studies also be recorded in an accessible database like ClinicalTrials.gov.
Another area that the FDAAA does not address is provision of greater visibility into regulatory decision-making at the FDA. Under FDAAA, the responsibility is on the trial sponsor to submit data to ClinicalTrials.gov. Many of the problems with drug safety in recent years, however, have involved data submitted to the FDA.
One of the open questions surrounding open results databases is the extent of narrative descriptions of results that should be allowed. There is considerable danger that a trial submitter would attempt to describe their results in as favorable a light as possible if given the opportunity to explicate their results in written narrative. Such writing would likely be the first thing to be read by a user of the ClinicalTrials.gov database, and would have undue influence because of the difficulty of understanding and interpreting raw data. On the other hand, if narrative summaries are not allowed, the database becomes harder to use and understand. One possible solution to this dilemma is to permit multiple narrative descriptions of results, layered on top of the core tabular database. If the trial sponsor wishes to provide a narrative description, they may – however such a description could be commented on or challenged by other (validated and qualified) users of the database. If others had alternative interpretations of a dataset, they would be able to post and comment on those interpretations as well. Encouraging both availability of the data and dialogue should help to mitigate the risk that narrative summaries could be used to mislead users.
Drug companies have responded to demands for greater transparency in public meetings with FDA and NIH , and have suggested that no additional data submission requirements should be imposed until trials have been conducted to prove value of result disclosure. Of course, until results are disclosed systematically, it is virtually impossible to conduct such a trial. The other major concern raised is that result submission to public databases might constitute prior publication, and prevent results from being submitted to academic journals. However, journals have already agreed that database submissions to ClinicalTrials.gov do not constitute publications, and will not affect publication decisions. The industry has also created its own database, ClinicalStudyResults.org, although it is far from clear whether these efforts are sincere, or an attempt to create an illusion of voluntary transparency to avoid additional regulation.