Imagine making a decision about whether to buy a new home based on seeing only half the property. The real estate agent invites you in to look around — the bedrooms, the kitchen. But you’re not allowed to visit the dining area, and the basement is off-limits.
Do you buy the house? The choice may seem more like an uninformed guess. And that’s essentially how doctors are made to operate today when it comes to decisions about which medical interventions (drugs, tests, procedures) to use on us.
In medicine, clinical trials inform the choices health professionals and regulators make about the safety and effectiveness of treatments and medical devices. An estimated half of these studies are never published, and those with negative or unpromising results are more likely to remain hidden from view, a phenomenon called "publication bias."
Even legal and ethical mandates that have been enacted in recent years haven't been able to close the gap.
ClinicalTrials.gov — the world's largest database of clinical trials, run by the National Institutes of Health in the US — is designed to be a repository of all the world's clinical trials, past and present. A recent New England Journal of Medicine study found that just a small fraction of trial sponsors comply with a 2007 law that requires them to register their work in the database and report a summary of their results.
Among more than 13,000 clinical trials the researchers analyzed, only 13 percent of sponsors running trials had reported their findings within the required one-year window.
All of that unpublished data may contain information on patient harms or side effects, clues to failed studies or unpromising drugs that shouldn’t be retested, and information that independent researchers could use to vet the quality of research findings. It may also contain information about potentially helpful clinical trials people might want to enroll in.
Without this data, doctors are left to act like realtors selling houses they themselves have hardly seen — in decisions that are, quite literally, life or death.
The tipping point
The New England Journal of Medicine researchers did find, however, that compliance with the mandate improved over the five-year period. At five years, 41 percent of industry-funded trials, 28 percent of academic/non-government-funded trials, and 39 percent of government-funded trials had reported results.
These improvements are the result of a movement to fix the broken system: to make sure the results of studies don’t go missing and that all clinical trials are registered in public databases (like ClinicalTrials.gov) for the world to see.
"We were ready for a tipping-point event," said Deborah Zarin, director of ClinicalTrials.gov. "This is a statement that the expectations have changed: if you're a researcher or a research sponsor, there's going to be an expectation that you're going to make data available at the end of your study."
New recommendations about "patient-level" data
A recent report out of the Institute of Medicine — "Sharing Clinical Trial Data: Maximizing Benefits, Minimizing Risks" — basically says the status quo on clinical trails transparency is unacceptable. In the report, a group of stakeholders including doctors, journal editors, and patient groups wrote that data sharing needs to be the "expected norm" in medical research, and outlined what clinical trials information should be made transparent and when.
The IOM said that instead of just designing and planning a study, scientists need to plan and document how they're going to share the data from that study so it's usable to others who may want to re-analyze it.
One of the most provocative recommendations included a push for sharing "participant-level" data. Right now, only the summaries of results from studies are expected to be communicated and registered to clinical trials databases.
But as one New England Journal of Medicine editorial notes, a lot of information can get lost when researchers apply statistical methods to bring together and analyze all the quirks and individual findings from patients. Different methods for summarizing data can also lead to very different conclusions.
For example, an MRI scan on a patient needs to be read and interpreted by a doctor; notes from the emergency department inform whether a trial participant actually had a heart attack. When independent researchers look at these scans and notes, they may come to very different conclusions.
Right now, researchers usually lock away this kind of data at the end of a study. But the IOM said this needs to change. It recommended that six months after the publication of trial results (or 18 months after trial completion), researchers share the raw data about study participants through open databases.
The IOM also recommended that:
- Data sharing becomes an essential step in running a clinical trial: As soon as researchers register a trial and before they enroll the first patient, the people running the trial have a plan in place for which data will be shared and when.
- When a study is completed, within a year scientists should share a summary of their results, including any adverse events.
- Within 18 months of a trial's end, scientists should share the full data sets from the clinical trial (i.e., the participant-level data).
- Data-sharing initiatives should be transparent and overseen by independent panels of experts and the public.
The slow movement toward clinical trials data sharing
The IOM's suggestions are in line with other initiatives to push the clinical trials transparency agenda further.
The Department of Health and Human Services released a new plan late last year to make ClinicalTrials.gov even more comprehensive.
The AllTrials campaign, led by a group of crusading scientists and doctors in Europe, has been pushing for every clinical trial (past, present, and future) around the world to be registered, with their full methods and results reported.
More recently, drug companies including GlaxoSmithKline, Novartis, and Sanofi created a web portal where independent researchers can request anonymized patient-level data from trials. In this case, the drug companies act as the gatekeepers of the data.
Yale University has been collaborating with industry partners, including Medtronic and Johnson & Johnson, to open up raw data from clinical studies for independent researchers to see.
Because the transparency landscape in medicine is advancing, some think the IOM group could have been even more provocative.
"I would have liked the report to go further," said Harlan Krumholz, a professor of medicine and advocate for data transparency at Yale University, "with respect to shortening the time to sharing, being more inclusive of the types of studies in which data should be shared.
"That being said, several years ago it would have been hard to imagine an IOM report that makes clear the issues we face in the current culture and the need to change."