Two female scientists in a NIH lab discussing research

NIH wants to expand the sharing of summary data from clinical trials, such as test results being reviewed here at NIH’s clinical center.

Clinical Center, National Institutes of Health

NIH aims to beef up clinical trial design as part of new data sharing rules

Drug companies and academic researchers will have to step up their public reporting of clinical trial results under new federal policies released today. The National Institutes of Health (NIH) in Bethesda, Maryland, also laid out a new plan for submitting clinical trial proposals that aims to beef up the rigor of the studies.

Researchers can no longer submit an unsolicited idea, but must respond to a request for applications that will include specific design requirements. The goal is to cut down on the number of “small crappy studies,” that don’t include sufficient numbers of patients or veer off from the original study plan, NIH staffers say. The agency wants to “reengineer the process by which clinical investigators develop ideas for new trials,” NIH officials explain in a commentary today in The Journal of the American Medical Association (JAMA).

NIH is timing these changes with the release today of a final U.S. Department of Health and Human Services (HHS) regulation that expands requirements that sponsors of trials regulated by the U.S. Food and Drug Administration (FDA) submit summary results no more than 1 year after a trial ends to ClinicalTrials.gov, the NIH-run public database. Companies will now have to report results not only for approved products, but also for mid- and late-stage (phase II and III) trials of FDA-regulated drugs and devices that haven’t yet been approved and may never reach the market. That will help increase the efficiency of research by letting others know about trials that failed, officials say.

A new NIH policy will also require submission of summary results to ClinicalTrials.gov for all clinical research supported by the agency, including early phase I safety trials and behavioral research. At the moment, even many major academic medical centers aren’t publishing trial results of NIH-funded trials within the required time frame.

The data reporting policies largely mirror proposals released for comment in 2014. One noteworthy addition, however, is that the final HHS rule requires that sponsors submit their protocol and original statistical analysis plan along with the summary results. That will dissuade researchers from trying new ways of analyzing their data to get a more interesting result, or “P-value hacking, where people sort of shop around for a statistical test to give them the P value that they love,” said NIH Director Francis Collins in a call with reporters.

HHS declined, however, to add narrative summaries to the bare-bones results tables now in ClinicalTrials.gov because it would have been difficult to ensure that the summaries were not biased to suggest that a treatment worked. Instead, officials will continue to encourage patient advocates and other groups to build on the data summaries.

The agencies also didn’t feel ready to require that trial sponsors share data for individual patients because experts are still working out how to do that without compromising privacy. University of California, San Francisco, medical informatics researcher Ida Sim, who is co-leading one such project, agrees with that decision. “I think this is an appropriate scope at this moment,” says Sim, who served on a 2015 Institute of Medicine panel urging broader sharing of clinical trial results.

The new rules go into effect 18 January 2017 and trial sponsors will have 90 days to begin complying; companies can request an exemption for up to 2 years if FDA hasn’t yet approved sale of the drug. The NIH policy covers studies funded after 18 January 2017. Trial sponsors who fail to comply could face FDA fines and suspension of NIH funding for clinical research.

NIH today also announced new guidance for clinical trial proposals, such as a suggested protocol template and training requirements. In addition, instead of sending their trial proposals to NIH as investigator-initiated grant applications, investigators will need to respond to a so-called funding opportunity announcement (FOA). That means NIH will be able to include “review criteria that focus on the rationale, design, and operational and analysis plans,” such as whether sample sizes are adequate, the JAMA article explains. The change will also ensure that proposed trials are routed to peer review panels with the expertise to evaluate those technical details, NIH officials say.

In the past, NIH says, a scientifically interesting proposal could sail through peer review and receive funding, even though the design was weak. The changes follow on other new NIH review standards to improve the rigor and reproducibility of NIH-funded preclinical research.

“We want to be sure that we are receiving and having the chance to review the most effective applications for clinical trials. It doesn’t work very well for them to sort of slide in randomly through various doorways without having some standardization of exactly what information is going to be included and also a clear path for its review by experts,” Collins explained.

The changes may trigger some pushback. But Collins says that researchers shouldn’t worry that the FOAs will be used to limit the scope of the clinical trials. Researchers will still be able to study “a broad range of important medical problems,” he says.

*Correction, 16 September, 1:31 p.m.: The article has been clarified to indicate that the final rule requiring that results from certain U.S. Food and Drug Administration-regulated clinical trials be posted in ClinicalTrials.gov was issued by HHS, not FDA.