Yale Docs Propose Model to Aid True, Evidence-Based Medicine

October 13, 2011

October 13, 2011 (New Haven, Connecticut) — Two experts from Yale University are calling for a complete overhaul of the system by which clinical-trial data are disseminated, stressing that the totality of evidence with regard to benefits vs harms for any particular product is never made available to the public, thereby severely restricting the ability of doctors and patients to make decisions about treatment [1].

A new approach is needed "that serves the needs of society and respects the concerns of industry," say Drs Harlan M Krumholz and Joseph Ross (Yale University School of Medicine, New Haven, CT) in a commentary in the October 12, 2011 issue of the Journal of the American Medical Association. Their appeal is backed by the Cochrane Collaboration, which is also calling for all data from clinical trials to be made publicly available [2].

"Once a drug is being sold, physicians and their patients are often making decisions with only a fraction of the available evidence, so we are left with a great deal more uncertainty about what drugs do than people know," Krumholz, a cardiologist, told heartwire . "We have this problem to a great extent in cardiology," he noted, but the concept applies equally to all branches of medicine.

We called it evidence-based medicine, but we didn't have all the evidence.

He and Ross have proposed a model based on companies providing data through an intermediary to two independent groups, which at the end of their analyses will make all the data on a particular drug or device available in the public domain for others to access. And they have their first volunteer; they are working with Medtronic on its product to promote bone growth, rhBMP-2, which has been the subject of safety concerns.

"My hope is that people will look back at this era of medicine and say, 'Let me get this straight, you were practicing evidence-based medicine, but it was semi–evidence-based medicine?' And we will have to say, 'You're right. We called it evidence-based medicine but we didn't have all the evidence,' " Krumholz says.

"Trust Us, and Trust the FDA" Is No Longer Enough

Krumholz gave two recent examples in cardiology where having all the data in the public domain would have helped the situation immensely.

"The rosiglitazone example is one of the most glaring. We never would have even had the debate about the safety of rosiglitazone in any substantive way had Glaxo not been forced to release the data through litigation," he observes, adding, "Even now, Glaxo has not yet released all the data."

And recalling the rofecoxib (Vioxx, Merck) debacle, he observes: "Merck always said, 'We have evidence [of the increase in MI with rofecoxib] compared with naproxen but we don't know the risk compared with placebo. The truth is that they had that placebo evidence as early as 2001. But doing a systematic review at that time couldn't possibly have uncovered the problem because a miniscule amount of the data was available in the public domain. But the data existed and the drug was being sold and people were making decisions about it."

Other examples in cardiology include the smoking-cessation drug varenicline (Chantix, Pfizer) and the cholesterol-lowering medication ezetimibe (Zetia, Merck), says Krumholz. These companies could really help themselves if they made all of the data available so that accurate assessments could be made of the risk/benefit balance for these drugs, he says.

"But right now, the companies are saying, 'Yes, we have data that are relevant to the safety and efficacy of this drug in our vaults, so just trust us and trust the FDA.' There, you've only got two groups, and one has a vested interest. I'm not saying I don't trust the FDA, but it is generally understaffed and underfunded," he maintains. "We have a big problem in medicine in general. No one can replicate anyone else's results because they don't have the data."

Industry is getting pummeled, and they need a way to show the public that they are interested in competing on science.

Key to the approach suggested by Krumholz and Ross is that the teams assessing the data be independent. "If you have groups analyzing the data that are getting grants from the company, that's not really true, independent analysis. If they hire someone who finds in the product's favor, no one believes it anyway. Industry is getting pummeled, and they need a way to show the public that they are interested in competing on science and that they can stand behind their work," he observes.

Krumholz favors voluntary, rather than enforced, participation. On this point, he and Ross are slightly at odds with the Cochrane Collaboration, which is calling on governments to think about "introducing legislation that makes it a requirement to provide data from all trials to the public" and "to consider punitive measures for noncompliance."

Calls for data sharing have come from a variety of organizations, says the Cochrane Collaboration, including the World Health Organization, the US National Institutes of Health, and the UK Medical Research Council.

There Must Be a Way to Transparency

Krumholz is keen to stress, however, that his intentions are "not about bashing industry"; he and Ross are cognizant of the fact that companies have "some legitimate concerns about competitors distorting data or people mining data in ways that are inappropriate."

To this end, he believes the process they are suggesting should be confined to postapproval. "Companies need to share all their data with the FDA--but I can understand them not sharing all their data with the public in the preapproval period because of proprietary interests. The strongest case [for this] is when the drugs are being sold."

And he adds that industry is not the only culprit: "Many studies that are publicly funded are not much better. This isn't just companies holding back, it's the state of the art right now in science."

But the pharmaceutical and medical communities owe it to the people who participate in clinical trials to make all data accessible, he argues. "Patients who have consented to be in these studies are doing so because they want to be part of the evidence base, they want to contribute to science. It's somewhat unethical to ask someone to participate in a study that you never report."

He acknowledges that some of the fault lies in the publishing process. It is often difficult for journal editors to publish negative studies or trials that may be underpowered, he says. "Let's be honest, it's a business, and they want articles that people are going to want to read."

But he argues there still needs to be a way to get all of the data for a specific product into the public domain, "because in aggregate, it may provide us with some information." The PLoS journals are a good example of publications "trying to create avenues" for publishing without judging, he notes, and clinicaltrials.gov "has a reporting arm." However, there are still limitations to these initial attempts at transparency, he observes.

Domino Effect: If One Company Does This, They All Will

With the mantra in mind that "you've got to get the data out there," Krumholz and Ross set out to try to design a way of doing this that would be acceptable to all parties involved. They approached Medtronic about its rhBMP-2 product to promote bone growth, which has recently been beset by safety concerns, and the company agreed to be the first case study for the new model; Krumholz says that it "raised some great questions."

The first step in the proposed process is that an external coordinating organization--ideally an academic center--contracts with the pharmaceutical or medical-device manufacturer, which agrees to provide access to all of its patient-level clinical research data, documentation, and case report forms from clinical trials, postmarketing studies, and safety programs. "The data will be processed so it is not possible to know who the patients are," Krumholz explains.

Just the pressure of one company signing up to this puts pressure on the other companies.

Second, the coordinating organization chooses two independent intermediary groups "to add further credibility," he notes. "You let them have access to all the data, everything, and then when they are finished, the company commits to a process to release all the data to the public. Then anyone who doesn't believe these two groups can repeat the work for themselves."

Krumholz says they expect to hold a national conference in January to try to further establish the process, and they hope to release the data on the Medtronic product within six to nine months of this conference.

"Once we show we can do this successfully, and it's feasible, it's just going to be a matter of time before everyone asks, 'Why aren't all the companies doing some form of this?' Just the pressure of one company signing up to this puts pressure on the other companies," he concludes.

Krumholz chairs a cardiac scientific advisory board for UnitedHealth. Ross is a member of a scientific advisory board for FAIR Health. Krumholz and Ross are recipients of a research grant from Medtronic through Yale University.