The personal sacrifices made by tens of thousands of patients, along with billions of dollars invested in drug development, are effectively being wasted because the results of failed experiments are so rarely made available to other scientists.
The finding, from a McGill University study, points to a vast domain of data on so-called stalled drugs that could inform related scientific research and help physicians make better choices on behalf of their patients.
Instead, "it mostly sits inside filing cabinets where it's inaccessible to people who might want to use it," said McGill bioethicist Jonathan Kimmelman, who led the study published this week in the BMJ (British Medical Journal). "We're putting in a lot of effort and exposing a lot of people to risk, and yet only a fraction of the information that we're collecting is getting out there."
To understand the scale of the problem better, Prof. Kimmelman and his colleagues looked at 177 registered clinical trials involving a range of drugs that reached the level of a phase III trial – the final stage of testing before a drug can be approved by the U.S. Food and Drug Administration (FDA). Typically, FDA approval determines whether a drug will reach consumers in much of the world.
The McGill team found that in cases where a phase III trial was conducted but the drug was not approved, the results were published only 37 per cent of the time.
"We went in thinking that we would see less publication of data on unlicensed drugs, but we never imagined we would see this low a level," Prof. Kimmelman said.
A total of 20,135 human subjects participated in the 63 per cent of the trials that were never published.
The picture changed somewhat for drugs that won approval. The team found that the trial results of those drugs were published 75 per cent of the time. Yet this is also a concern, Prof. Kimmelman said, because it implies that complete data on one-quarter of the drugs coming to market are not publicly available. This problem has come to light in controversies over the risks and efficacy of a number of high-profile medications, including Vioxx, Paxil and Tamiflu.
But the McGill investigation throws a spotlight on a related issue: Far less is known about drugs that are not approved for a particular use, even though that information could be scientifically important.
"It's really the first look at this submerged part of the iceberg of drug development," said Michelle Mello, a Stanford University law professor who specializes in health regulations and research ethics. "These are drugs that have not reached market yet but are nonetheless generating a lot of useful information that can tell us something about the risks of similar products, or products that might be approved in other jurisdictions."
There are several reasons stalled drug trial results may go unreported. When the work is conducted on behalf of a drug company, there could be incentive to suppress failed results because the details may aid competitors. And negative results can be harder to publish because editors of scientific journals find them less interesting and newsworthy.
For busy researchers aiming to maximize the impact of their science, the most likely reason not to publish may simply be a lack of time and resources. That was the conclusion of a systematic review of authors who presented abstracts of findings at scientific conferences but never followed up by publishing full-length articles on the work. The review, which appeared last month in the Journal of Clinical Epidemiology, recommended that scientists be required to have a dissemination plan for their data before a clinical trial begins, and that all data ultimately be posted in an open-access format.
The FDA currently requires that the results of clinical trials for approved drugs be deposited in a public registry, although this rule is not strictly enforced. New regulations under consideration by the U.S. government could extend that requirement to unapproved drugs.
Gordon Guyatt, a clinical epidemiologist at McMaster University, who was not involved in the McGill study, noted that research ethics boards, which are required for oversight of all human trials, should insist that data be made public regardless of the outcome.
"I believe it's a violation of ethics to enroll patients, put them through everything you put them through and then not report the results," he said.
Prof. Kimmelmen added that when research is supported by public dollars, funding agencies could play a stronger role in improving publication rates of clinical trials.
"It's said that we live in the age of information," he said. "But the fact is that much of the information we're producing is not in the public domain."