DATA · Q&A with Deborah Zarin, head of ClinicalTrials.gov
By Mark D. Uehling
March 8, 2005 | Not since the early years of GenBank have so many in academia and industry been so interested in a mundane federal database. ClinicalTrials.gov, run by the NIH's National Library of Medicine, was born as a matchmaking Web site for patients seeking to participate in federally sponsored clinical trials. Now, in the wake of the Vioxx recall and controversy over belatedly or never-published pediatric data from psychiatric trials, ClinicalTrials.gov is being buffeted by new demands from two directions.
First, the editors of medical journals are clamoring for ClinicalTrials.gov to morph into a scientifically useful repository of all trial data — something that the U.S. Congress permitted, but never required, in the Food and Drug Administration Modernization Act (FDAMA) of 1997. Pharmaceutical companies, meanwhile, are pressuring Congress to make sure that ClinicalTrials.gov remains a voluntary, partial archive of clinical trial data.
Who's in the middle? Federal employees seeking to comply with the law and balance the interests of the public and the industry. After just a few weeks on the job, Deborah Zarin, the new director of ClinicalTrials.gov, spoke to Bio·IT World's Mark D. Uehling. Zarin, a physician with background in outcomes research, supervises a staff of 15 to 20 employees and contractors at the NIH and FDA.
Q: What's your highest priority at the moment?
Zarin: There is growing recognition that we have more than one user group. Initially, it was consumers and their healthcare providers. The research, academia, policy-maker world is another user group that is very important. We have to work to increase confidence in the registry, both in terms of which trials are in, and which data elements are completed. The data get into ClinicalTrials.gov through 'trusted intermediaries' and a little help from the FDA?
We do closely collaborate with the FDA. We're implementing the process, but they have a strong role to play in terms of being the data provider for all the investigational new drug (IND) trials. The FDA certifies the data providers, as some in the industry might put their own data in there. There is a protocol registration system on the Internet that is a separate site — it's not a public site. Then you can just register your trial online. Whether a drug works or not — the result of a trial — was that something that Congress required to be listed in ClinicalTrials.gov?
There are a number of different kinds of data elements, including the title of the study, the inclusion-exclusion criteria for the subjects, where the trial is being conducted, issues of trial design, details about where the active research sites are, and whether it's recruiting or not. FDAMA says you have to register certain types of trials, and it has to include these data fields. We can't make more than those fields mandatory. Once a trial is put in your system, a sponsor is supposed to update the information to indicate if it is still actively recruiting patients. How would you characterize compliance with that aspect of the law?
There has been uneven compliance, both with initial registration and with keeping things up to date. FDAMA specified 'severe and life-threatening illnesses.' But it never fully defined 'severe and life-threatening.' Clearly, there is agreement that it includes cancer and AIDS. But what else it includes has never been clearly defined. It's hard to answer that. In the future, after additional legislation, how might a community physician or a member of the public know if all the trials for a particular drug or disease were listed at ClinicalTrials.gov?
That depends on what kind of enforcement mechanism, what kind of system, Congress would authorize. But it doesn't necessarily have to be that staff-intensive, as long as there is a clear system of how it's supposed to happen. It might be like filing taxes. In the future, how might the results from a clinical trial be structured and displayed by your search engine?
The most no-brainer way is linking to published articles. But that's a very complicated question. This would involve more detail than would be on a label. How do you do that? There are enormous volumes to be written on the results of a clinical trial. When you submit an article to a peer-reviewed journal, there is sometimes a lot of back and forth about what is the appropriate, dispassionate, fair way to report those data. Medical journal editors want each clinical trial registered prior to any publication of the trial results. Would that really help ordinary physicians?
Unless somebody mandates the inclusion of results for every clinical trial, there are going to be instances where sponsors choose not to publish — or maybe they try to publish and no one is accepting it. Who knows? But at a minimum, if things worked, you'd have the denominator. You could say, 'Gee, this drug company did 12 studies of this drug, and I can only find two published. Isn't that interesting?' Q: Critics of the industry are skeptical that the industry will comply with any new or existing law ...
Some of the issues that played out with antidepressants in kids — there were actually 20-odd studies that were done. Approximately eight were published. The FDA had all those studies. They weren't required, or even allowed, to publicize them because they are considered proprietary information. That's how the law is written. But imagine if, on a public Web site, all 20 studies were listed. Thoughtful physicians could say, 'Gee, where are the other 12 studies?' PHOTO BY RANDALL SCOTT