• Email
  • Single Page
  • Print

Your Dangerous Drugstore

Worse yet, of the thirty-two members of the 2005 panel that was charged with deciding whether the COX-2 inhibitors were safe enough to stay on the market, ten had financial connections with one of the manufacturers, according to a front-page story in The New York Times that appeared a week after the panel’s decision.15 As is often the case, these ten members with conflicts of interest were not disqualified. And as it turned out, they voted 9–1 in favor of Vioxx and Bextra. Without their votes, the panel would have recommended that these two COX-2 inhibitors be removed from the market (there would still have been enough votes to keep Celebrex). This does not prove that these nine advisers were biased, but it certainly raises the question, especially since Vioxx and Bextra were rejected by the majority of panel members with no known ties to the manufacturers. That is why FDA advisory committees should not include people with conflicts of interest, no matter how expert they may be.

The clinical trial that caused Merck to withdraw Vioxx was designed to see whether the drug could prevent the recurrence of colorectal polyps; the finding that the drug increased the risk of heart attacks and stroke was an accidental result of the trial. The company professed to be surprised. Then the CEO, Raymond Gilmartin, who claimed his wife took Vioxx right up until the drug was withdrawn, said the results were “unexpected.”16

But in fact, it could hardly have been a complete surprise. There had been signs of trouble for years.17 In 2000, a company-sponsored trial was published in The New England Journal of Medicine comparing Vioxx with over-the-counter naproxen in patients with rheumatoid arthritis.18 This was called the Vioxx Gastrointestinal Outcomes Research, or VIGOR, trial (medical researchers and their sponsors love catchy acronyms), and it was intended to show that Vioxx was easier on the stomach than naproxen (Aleve). In relieving pain, the drugs proved to be the same, but those taking Vioxx had only half the risk of serious stomach problems. Unfortunately, the study also showed at least a fourfold increase in the risk of heart attacks. The details of the cardiovascular effects were not described in the published paper, but an FDA analysis indicated that the drug was more likely to cause heart attacks or strokes than to prevent stomach ulcers. Merck tried to explain the alarming finding away by saying the difference probably showed that naproxen protects the heart, not that Vioxx harms it. But, of course, without testing that hypothesis, it was simply self-serving speculation. Moreover, within a year, other evidence came to light that Vioxx increased cardiovascular risks, and that Celebrex may do so as well.

In my book I discussed the power of the big pharmaceutical companies to sell just about anything. What Merck should have done after it got the results of the VIGOR trial was immediately launch a large enough clinical trial to investigate the cardiovascular risks as quickly as possible. Instead, a few months later, it signed Dorothy Hamill to skate its problems away. The company reportedly spent $160 million on direct-to-consumer ads for Vioxx in 2000, and continued to spend approximately $100 million a year during the next four years.19 The costs of even a very large clinical trial would almost certainly have been less than what the company spent on ads. But as expensive as they are, the costs of direct-to-consumer ads are small compared with what drug companies spend promoting their products to doctors. In a variety of expensive ways doctors became convinced—just as the public did more cheaply—that Vioxx was some sort of a breakthrough. The fact that it was no more effective than naproxen, or presumably other over-the-counter remedies, was a fact soon forgotten.

That the advisory panel recommended halting direct-to-consumer ads highlighted the hollowness of the pharmaceutical industry’s contention that such ads are “educational”—designed to encourage patients to discuss medical problems with their doctors. It should be obvious that there is nothing educational about watching Dorothy Hamill pitch a drug or about the other prescription drug ads we see on television. They are meant to persuade patients (and doctors) to use the drug. And they work—as they did spectacularly well in the case of the COX-2 inhibitors. That is why every other advanced country except New Zealand does not permit such ads. But despite the fact that senior officials of the FDA, including the last two commissioners, have repeatedly endorsed the fiction that direct-to-consumer ads are educational, the advisory panel clearly knew better. If ads were truly educational, panel members would not have recommended a moratorium on broadcasting them. 20

In a sense, Merck was doing what big drug companies often do. It was touting the benefits of its drug (such as they were), downplaying or explaining away the risks, and marketing it as though it were a medical miracle. Less explicable is the dereliction of the FDA. If Merck didn’t want to launch a study of the cardiovascular effects of Vioxx (and it had nothing to gain commercially by doing so), why didn’t the FDA insist on it? After all, the FDA’s responsibility is to ensure that prescription drugs are safe and effective, and there was clearly something that needed looking into here. The FDA should also have taken a close look at the ads, since one of its jobs is to ensure that they are accurate and balanced—which they obviously were not.

But the FDA did nothing. Later, it protested that it doesn’t have the authority to mandate additional studies once a drug is marketed, but that is sophistry. The FDA has the authority to pull drugs off the market, and that threat would have been enough to get Merck to organize a trial. Finally, in 2002, after a year of wrangling, it got Merck to add a tepid warning to the drug’s labeling information—the material in small print that comes with prescription drugs (and that few actually read). That hardly met the agency’s responsibility. The part of the FDA that approves new drugs now receives half of its funding from “user fees” paid by drug companies for each drug evaluated. These were first authorized by Congress in 1992, and in return the FDA was required to speed up its review of drugs. I warned in my book about the baleful effects of the user fees, which put the FDA on the payroll of the industry it regulates. This story brings that warning home.


The Vioxx case underlines the importance of clinical trials in deciding whether drugs work or not. Without them, doctors and patients would have to decide on the basis of personal accounts indicating whether or not a given patient seems to improve. That is an unreliable (not to mention dangerous) method. As I wrote:

The assumption that a drug works if a patient gets better does not allow for natural variations in the illness, for the placebo effect (the tendency of both doctors and patients to imagine a drug is working), for all the other times when the drug might fail, or for the possibility that another drug might have worked better.

That is why clinical trials were such an important medical advance when they were first introduced in the middle of the twentieth century, and why the FDA requires clinical trials rather than a collection of testimonials to decide whether drugs are safe and effective.

I mention this because the public hearings held by the FDA advisory panel permitted testimonials from people who said they wanted the COX-2 inhibitors left on the market because nothing else relieved their pain. The panel was reportedly much influenced by these testimonials (could that have explained the pro-Celebrex vote by panel members not on Pfizer’s payroll?), and it ultimately concluded that, although the drugs should be used far less widely than they were, they might be uniquely effective for some people. But while that is possible, there was no evidence on which to base that conjecture.

Not only are testimonials an unreliable way to judge a drug’s effectiveness, they are particularly useless when those giving the testimony are selectively chosen. According to the transcript, at least one of the patients who spoke at the hearings was brought there by Pfizer, the maker of Celebrex and Bextra. I have not seen any evidence that patients who had suffered heart attacks or strokes while on these drugs were brought to Washington to testify; nor, so far as I know, were those who had tried a COX-2 inhibitor but preferred Advil or Aleve.

The notion that testimonials are useful in evaluating drugs reflects the biases of some doctors (mainly those in clinical practice) and much of the public. Many patients like to think they are uniquely different from others, not only as persons, but biologically. However, biological variations among patients are probably far less significant than is widely believed, and would need verification, in any case. Clinicians, for their part, want as many choices of drugs as possible, even if they have little basis for choosing among them. When a drug doesn’t seem to work, they want to be able to say, “Here, try this and see if that’s better.” So it should not be surprising that testimonials played such a big part in these hearings. They were open hearings, and it would have taken a brave panel member to point out publicly that testimonials are not a good basis for making a decision about the benefits of drugs.

Drug companies are quick to exploit the view that there is great biological variability among patients; it is one of their justifications for turning out so many me-too drugs. They like to present me-too drugs as backups for one another. But there is little evidence to support the notion that if a particular drug doesn’t work for a patient, another one in the same class will. Drug companies don’t test their me-too drugs on patients who have found a similar drug ineffective, so there is no way to know for sure. But much of what may look like differences among me-too drugs probably has mainly to do with the size of doses. It may be, for example, that a double dose of Celebrex would act just like Vioxx for virtually everyone.

Even if we grant the possibility that some people may respond very differently to these drugs, does that justify accepting a greatly increased risk of heart attacks and strokes? Since those who most need pain relievers for arthritis are precisely older people who are most vulnerable to cardiovascular disease, the drugs have been estimated to have caused tens of thousands of heart attacks among the millions of people who have taken them regularly.21 That is a tremendous cost to balance against the dubious proposition that the drugs offer something unique for pain relief. It is the FDA’s job to see that the benefits of prescription drugs outweigh the risks. It seems clear to me that the agency failed to do so in this case.

The courtroom is not a good place to resolve questions like whether Vioxx caused someone’s heart attack. For one thing, when there are other possible causes of a medical condition, it’s usually impossible to know which one was responsible in any specific case. The best we can do is deal with statistical probabilities. For example, if we look at large numbers of people and find that heart attacks are more than twice as common in those who took Vioxx (as some studies have found22 ), then we can say that the drug outweighs all other risks in the average person like those in the trials. But our legal system requires juries to determine causation in particular individuals, with particular sets of risks, something that is rarely possible. It is an illusion of our tort system to think that we can, and it is one reason these kinds of cases yield such inconsistent verdicts. The best we can do is assume each person is typical of those studied in clinical trials.

John McDarby had a number of other conditions that could have affected his heart, such as diabetes and clogged arteries, and Merck made the case that his heart attack should be attributed to them, not Vioxx. This strategy had been successful in earlier cases. But McDarby’s lawyers stood the argument on its head. They made the case that precisely because of these other heart risks, he was especially vulnerable to the dangerous effects of Vioxx—in their words, “the last person who should have been on Vioxx.” That is certainly a commonsensical view, and clearly the jury saw it that way. Essentially, they were persuaded that a push is more dangerous if you’re standing on the edge of a cliff.

The punitive damages in the McDarby case were based largely on the jury’s finding that Merck had misled the FDA by not supplying it with an internal analysis the company had done of Vioxx’s risks. Merck argued that it withheld the analysis because it was flawed, but had supplied all the underlying data. Whatever the truth about McDarby’s claims, the FDA certainly had enough information to take stronger action years ago.

We have an FDA precisely because we know that drug companies, given their inherent conflict of interest, should not be left to decide on their own whether their products are safe. Caveat emptor may be a reasonable approach for many consumer products, but not for prescription drugs. But the Vioxx story underscores the extent to which the FDA has come to see itself as representing the drug companies, not the public.

Two reforms are necessary. First, the conflicts of interest that pervade the agency should be eliminated. The Prescription Drug User Fee Act (PDUFA) that authorized drug companies to pay user fees in return for quick approval of their drugs comes up for renewal in 2007. Congress should let it die, and instead appropriate adequate support from public funds for this vitally important agency. The FDA should also prohibit experts who consult for drug companies from sitting on its advisory panels. No one’s expertise is indispensable, contrary to the current practice of granting virtually automatic waivers on those grounds.

Second, the FDA should be made much more transparent. Drug companies are required to inform the agency about all the clinical trials they sponsor on drugs for which they are seeking FDA approval. But the agency does not make the studies public without permission of the sponsor, and drug companies naturally publish only the most favorable results. By allowing less favorable results to remain buried, the agency puts proprietary interests ahead of the public interest, and doctors and the public come to believe prescription drugs are better than they are. That should stop. All studies involving human subjects should be registered at inception in a publicly available, central database, and the salient results added at completion.

These reforms will require congressional legislation—an uphill battle, in view of the power of the pharmaceutical lobby in Washington. That power will need to be offset by concentrated pressure from citizen and advocacy groups with an interest in regulating prescription drugs, something that has so far been lacking. But the spotlight does belong squarely on the FDA. That, after all, is where the public’s protection should lie, not with juries after the damage is done.


The ‘Dangerous Drugstore’ November 16, 2006

  1. 15

    Gardiner Harris and Alex Berenson, “10 Voters on Panel Backing Pain Pills Had Industry Ties,” The New York Times, February 25, 2005.

  2. 16

    Quoted in Anna Wilde Mathews and Barbara Martinez, “E-Mails Suggest Merck Knew Vioxx’s Dangers at Early Stage,” The Wall Street Journal, November 1, 2004.

  3. 17

    For several accounts of the longstanding safety concerns about Vioxx, see Mathews and Martinez, “E-Mails Suggest Merck Knew Vioxx’s Dangers at Early Stage”; Eric J. Topol, “Failing the Public Health—Rofecoxib, Merck, and the FDA,” The New England Journal of Medicine, October 21, 2004; Barbara Martinez, “Vioxx Lawsuits May Focus on FDA Warning in 2001,” The Wall Street Journal, October 5, 2004.

  4. 18

    Claire Bombardier et al., “Comparison of Upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in Patients with Rheumatoid Arthritis,” The New England Journal of Medicine, November 23, 2000.

  5. 19

    The $100 million figure was widely reported; see, for example, Topol, “Failing the Public Health”; see also www.todaysseniorsnetwork.com/excessive _ads.htm.

  6. 20

    In April of this year, Pfizer resumed advertising Celebrex after a steep drop in sales during the moratorium. The ad included the following text: “Important information: Celebrex may increase the chance of a heart attack or stroke that can lead to death.”

  7. 21

    For estimates, see Eric J. Topol, “Failing the Public Health. Also, testimony of David J. Graham, associate director for science and medicine in the FDA’s Office of Drug Safety, before the Senate Finance Committee, November 18, 2004.

  8. 22

    For an overview of studies, see Peter Jüni et al, “Risk of Cardiovascular Events and Rofecoxib: Cumulative Meta-analysis,” The Lancet, December 4, 2004.

  • Email
  • Single Page
  • Print