By Melinda Young, Author

When IRB members and research ethicists consider the debate over sharing clinical trial data with the public and other researchers, there is one example they might wish to consider: Oseltamivir, known as Tamiflu.

Tamiflu, which was designed to treat influenza symptoms, was widely stockpiled by governments around the world more than a decade ago. Out of concern of a worldwide pandemic flu, they wanted something in their arsenals to help slow the spread of the disease and help alleviate discomfort in those infected.

Published studies suggested the benefits of the drug outweighed the expense and patient side effects. Scientific skeptics challenged those conclusions and spent several years pushing Tamiflu’s manufacturer, Roche, to release Tamiflu trials data.

The Cochrane Collaboration published results of a review of the unpublished trial data, showing that although the drug shortened flu-like illness by less than a day, it had no effect on hospitalizations or secondary infections, such as pneumonia.1

“There seems to be marginal evidence of its effectiveness,” says Mark Schreiner, MD, executive vice-chair of the committee for protection of human subjects (CPHS) at the Children’s Hospital of Philadelphia (CHOP).

In Europe, governments spent billions of dollars to stockpile a drug of questionable value, Schreiner says.

Greater clinical trial data transparency could prevent costly mistakes, and it’s something that already is provided for pediatric research through provisions in the Best Pharmaceuticals for Children Act, he adds.

“The pediatrics act requires study reports,” he says. “You can find information online about lots of studies that failed because they were mandated to be there since the early 2000s, and that’s the level of transparency that is missing in most drugs developed first for adults.”

The International Committee of Medical Journal Editors (ICMJE) published an editorial in June 2017, saying there is an ethical obligation to share interventional clinical trial data.2

Beginning July 1, 2018, manuscripts with clinical trial results that are submitted to ICMJE journals must contain a data-sharing statement.2

Data-sharing statements must include the following2:

  • Will individual de-identified participant data be shared?
  • What data will be shared?
  • Are related documents available, such as study protocol and statistical analysis plan?
  • When, and for how long, will data become available?
  • What access criteria data will be shared, and by whom?2

“What ICMJE is about is not just sharing data so it can be used, but it’s about increasing the transparency in the process so that everybody can learn from what’s been done,” Schreiner says.

It’s also about maximizing the utility of the data from clinical trials.

“To maximize societal benefits from the sacrifice or willingness of volunteers to take part in clinical trials, we should make data available at some point in time so other investigators can explore or use this for other purposes,” Schreiner explains.

This cooperative approach might be good for society, but some people in the research community fear it could allow someone else to swoop in and benefit from their hard work.

For instance, suppose a drug trial failed for its intended indication. But a data analysis reveals that some participants have a genetic variant that led to a positive response, says Barbara Engel, MD, PhD, CPHS chair at CHOP.

“So you do another trial, focusing on people with those variants,” Engel says. “With some justification, researchers might worry that someone else will get hold of this data and do this trial before they get a chance to.”

The ICMJE’s perspective is that if society is going to fund research and volunteers are willing to take risks to benefit society, then potential gains from the research should be maximized, Schreiner notes.

“In an ideal world, this is a cooperative venture between people who understand data and people who ask the question, and I’d hope this would be a partnership rather than marauders swooping in,” he says.

IRBs and those who work in human research protection need to make sure risks and benefits are appropriate when reviewing studies, but they do not have a defined role for follow-up studies that use de-identified data.

“It’s our expectation researchers will use this data, and it may inform future research, but how it’s specifically defined is not always known going into the initial review,” says Amy Schwarzhoff, MBA, CIP, director of human subjects research at CHOP.

However, IRBs can ensure that study sponsors intend to publish results.

“We ask people to discuss their plans for publication,” Schreiner says. “We have, in fact, deferred studies from pharmaceutical companies that say they own data and no one will be able to publish it.”

Sponsors sometimes do this because they don’t want to give away their secrets until there’s a confirmatory study, he adds. “People have novel ideas, and they believe if details come out, someone could scoop them.”

But what can happen is that one company runs a clinical trial in a new class of drugs. The study fails and is not published. Then another company has a similar agent in the same new class, runs the study, and it also fails and is not published. By the time a fifth clinical trial is begun in the same class of failed agents, hundreds of people have volunteered their time and risked their health. And the latest people to enroll have no way of knowing that every other trial of that class of drugs has failed, Schreiner explains.

“We always joked there should be a journal of failed experiments,” Engel says. “If failed clinical trials were published, we might learn from these.”


  1. Jefferson T, Jones MA, Doshi P, et al. Neuraminidase inhibitors for preventing and treating influenza in adults and children. Cochrane Database of Systematic Reviews. 2014;4.
  2. Taichman DB, Sahni P, Pinborg A, et al. Annals Int Med. June 6, 2017, editorial:1-3.