Overdosed America (7 page)

Finally, the most important measure of safety is the overall frequency of serious side effects—including but not limited to gastrointestinal side effects. For the full 12 months of the study, the people in the CLASS study who took Celebrex
experienced 11 percent more serious complications
(in all body systems combined) than the people who took the older and less expensive anti-inflammatory drugs. This difference did not reach statistical significance but certainly is significant in countering Pharmacia’s claim that Celebrex is better than older NSAIDs because it’s safer.

These findings contributed to the FDA’s decision to send one of its rare Warning Letters to the CEO of Pharmacia in February 2001.
The letter cites repeated unsubstantiated marketing claims
that Celebrex is the preferred NSAID for people taking a blood thinner and that it is safe and effective for the treatment of acute pain—a use for which it is not approved—and points out that Pharmacia’s marketing material fails to warn of the possibility of serious GI complications caused by the drug. The Warning Letter concludes by saying:

Your promotional activities described above raise significant health and safety concerns in that they minimize crucial risk information and promote Celebrex for unapproved new uses. In two previous untitled letters dated October 6, 1999, and April 6, 2000, we objected to your dissemination of promotional materials for Celebrex that . . . contained unsubstantiated comparative claims, and lacked fair balance. Based upon your written assurances that this violative promotion of Celebrex had been stopped, we considered these matters closed. Despite our prior written notification, and notwithstanding your assurances, Pharmacia has continued to engage in false or misleading promotion of Celebrex
.

Also included in the Warning Letter was the requirement that Pharmacia send out the “Dear Healthcare Provider” letter that had landed on my desk. Of course, the letter sent out by the manufacturer was not quite as specific as the FDA’s Warning Letter. Few doctors, even if they had bothered to wade through the difficult language, had the time or inclination to find out the story behind the letter. As a result, doctors continued to prescribe Celebrex for their patients based on the scientific evidence published in JAMA, scientific evidence that I now understood was more than just biased in the manufacturer’s favor. It was incomplete and presented an inaccurate picture of the so-called safety advantage of Celebrex over other far less expensive NSAIDs. The NEJM review article didn’t even mention the Warning Letter issued to Pharmacia or the mandated “Dear Health Care Provider” letter, which had been sent six and four months, respectively, before the review article was published.

Two months after the CLASS study appeared in JAMA, the results of the Vioxx Gastrointestinal Outcomes Research study, VIGOR, were published in the November 23, 2000, issue of the
New England Journal of Medicine.
This Phase 4 study, which included over 8000 people with rheumatoid arthritis, compared the risk of serious gastrointestinal problems in people treated with Vioxx against those treated with naproxen. Treatment with Vioxx, the article concluded, “resulted in significantly fewer clinically important [major] upper gastrointestinal events than did treatment with naproxen.” The case for Vioxx appeared to be even stronger than the case for Celebrex.

But when I read the NEJM article carefully I noticed that, as with the Pravachol and stroke study described in the last chapter, the people included in the VIGOR study were very different from the vast majority of people for whom doctors prescribe NSAIDs. More than half of the people in VIGOR were taking steroids, such as prednisone, in addition to an NSAID for their arthritis. Only a small percentage of the people for whom Vioxx is prescribed in the general population, however, take steroids at the same time. This turns out to be an important detail, because buried in the text of the article is the finding that among the people in the study who were
not
taking steroids, the reduction in the risk of serious gastrointestinal complications was not large enough to be statistically significant. This may have been the most important finding in the study, but it remained hidden in plain sight and is still virtually unknown.

Next I looked at the data in the FDA files comparing the risk of serious cardiovascular complications in the people who took Vioxx with the risk of those who took naproxen. The original research plan for the VIGOR study had acknowledged the possibility that naproxen users might develop fewer serious cardiovascular complications than those who took Vioxx. The reasoning was that the COX-1 inhibiting activity of naproxen makes platelets less “sticky” (like aspirin, but not to the same extent), potentially decreasing the risk of unwanted blood clots—a property not shared by Vioxx, which is a selective COX-2 inhibitor. Because of this possibility, the research plan called for serious
“cardiovascular thrombotic or embolic”
complications
*
to be examined by an independent committee to make sure that the study results were accurate and unbiased.

It turns out that the increased number of serious cardiovascular complications in Vioxx users more than offset the highly touted GI benefit of this drug. In the VIGOR study the people who took Vioxx experienced 21 fewer serious GI complications than those who took naproxen, but they experienced
27
more
serious cardiovascular complications
. These were the results that the November 2000 NEJM article on the VIGOR study acknowledged had been assessed, but did not report.

The more closely I scrutinized these data on the cardiovascular complications in the VIGOR study, the more I learned about how data can be manipulated to color the “scientific evidence” that is so trusted by doctors and the public. The November 2000 NEJM article focused most of its discussion about cardiovascular risk on heart attacks alone, although heart attacks had not been identified in the research plan as a specific cardiovascular complication to be monitored by itself. Considering heart attacks alone, however, allowed the authors to claim that Vioxx significantly increases risk only in people who should have been taking aspirin because of a previous cardiovascular problem (but had been prevented from doing so by the design of the study). The NEJM review article reported the frequency of a broader but still incomplete list of cardiovascular complications (
increased risk of heart attacks, strokes, and sudden death
), once again minimizing the extent of the problem.

However, when all of the serious cardiovascular complications documented in the manufacturer’s own data are included, the picture that emerges is different from and much more troubling than the picture presented in either of the NEJM articles. Overall, the people in the VIGOR study who took Vioxx were 2.4 times more likely than those who took naproxen to experience a serious cardiovascular complication. The
statistical significance of this finding (
p
= .0016)
means that there are less than two chances out of a thousand that this increase in the risk of developing serious cardiovascular complications is simply due to chance. In absolute terms, for every 100 people treated with Vioxx instead of naproxen for one year there was
one additional serious cardiovascular complication
. Indeed when the FDA statistical reviewer independently analyzed all serious cardiovascular events from the study, she found that the risk of cardiovascular complications was more than double in the people who took Vioxx, and that the likelihood that this difference was simply due to chance was 1 in 10,000 (
p
= .0001).

Even for people without a history of cardiovascular problems (the people for whom the NEJM VIGOR article reported no significant increase in the risk of heart attacks) taking Vioxx instead of naproxen almost doubled the risk of developing a serious cardiovascular complication (
1.9 times the risk,
p
= .041
). An FDA reviewer commented that the greater risk of cardiovascular problems in the people who took Vioxx “could lead one to conclude that naproxen . . .
would be the
preferred
drug”
(FDA’s underscore).

Both of the NEJM articles warned that, because of an increased risk of heart attack or cardiovascular complications, patients with a history of cardiovascular disease taking Vioxx should also take prophylactic low-dose aspirin. Neither of the articles, however, gave doctors any idea of the magnitude of this increased risk. The results of the VIGOR study show that for every 100 people with a history of cardiovascular disease treated with Vioxx instead of naproxen there were
between
seven and 11 additional serious cardiovascular complications each year
.
*

No mention was made in either of the NEJM articles about the single most important finding of the VIGOR study. Overall, including GI, cardiovascular, and all other serious complications, the people who took Vioxx
had 21 percent more “serious adverse events,”
complications that usually lead to hospitalization or death, than did the people who took naproxen (
p
= .013). This translates to an absolute risk of 2.5 more serious complications for every 100 people who took Vioxx instead of naproxen each year. Something is very wrong with a system that leads patients to demand, and doctors to prescribe, a drug that provides no better relief and causes significantly more serious side effects. (Vioxx
might
be the drug of choice for a patient with arthritis who is taking steroids, who requires an anti-inflammatory drug, and who has no history of cardiovascular disease but a history of GI bleeding or complications from other NSAIDs. But even for this patient, there might be other, older drugs that would be safer to try first.) Adding insult to injury, Vioxx costs $100 to $134 a month, compared with $18.19 a month for prescription naproxen, or $7.50 for over-the-counter naproxen.
*

Just one month after publication of the NEJM review article that dismissed the increased cardiovascular risk of Vioxx as a “play of chance” and failed to mention the significantly increased risk of serious complications in people who took Vioxx, the FDA sent a
warning letter to Merck
citing it for marketing that was “false, lacking in fair balance, or otherwise misleading.” The letter pointed specifically to Merck’s “promotional campaign for Vioxx that minimizes the potentially serious cardiovascular findings that were observed in the . . . VIGOR study.” “Simply incomprehensible” is what the FDA’s letter says about a press release issued by Merck on May 22, 2001, titled, “Merck Confirms Cardiovascular Safety Profile of Vioxx.” The warning letter adds: “Your misrepresentation of the safety profile for Vioxx is particularly troublesome because we have previously, in an untitled letter, objected to promotional materials for Vioxx that also misrepresented Vioxx’s safety profile.”

Surely authors who are expert enough to have written the “Drug Therapy” review article about Celebrex and Vioxx in the
New England Journal of Medicine
were aware (or should have been aware) of the manufacturers’ unprocessed data from the VIGOR and CLASS studies—even if the manufacturers hadn’t volunteered these data on their own. The data had been available on the FDA’s website for 6 months prior to the publication of the article. And surely they had an obligation to inform their readers just how much the FDA’s interpretation of the data from these two studies differed from the articles published in our two most respected and influential medical journals.

Rather than presenting a balanced view of the scientific evidence, the NEJM review article repeats the “unsubstantiated claim” that Celebrex causes significantly fewer serious gastrointestinal complications than other NSAIDs—completely ignoring the FDA’s Warning Letter to Pharmacia and the “Dear Healthcare Provider” letter mandated by the FDA. Likewise, the NEJM review article did exactly what the FDA’s Warning Letter cited Merck for doing: minimizing the cardiovascular risk and the unfavorable safety profile of Vioxx.

The FDA reviewers’ reports posted on the Internet provide an unusual opportunity to compare the unprocessed data from the manufacturer-sponsored CLASS and VIGOR studies with the medical journal articles and “medical knowledge” that they subsequently became. Unlike the articles about statins and strokes that I investigated, this went beyond spin.

One of the most surprising (and frightening) parts of this story is that even though all this information was (and is) publicly available, that knowledge has done little to correct doctors’ and patients’ beliefs that these drugs are superior. How had we gotten to a place where expensive new drugs could become “blockbusters” when there was little scientific evidence to justify their routine use? And why hadn’t the FDA spoken up more publicly about the misleading journal articles?

At a 2002 faculty medical ethics seminar at Harvard Medical School, I had the opportunity to ask Dr. Janet Woodcock, Director of the FDA’s Center for Drug Evaluation and Research, why the FDA had not intervened in JAMA’s publication of the Celebrex study. I pointed out that the FDA was aware that publication of the CLASS article in JAMA would lead to greatly increased use of Celebrex under false pretenses. Dr. Woodcock said the FDA could not “constrain communication” in a scientific journal, and that this was “a First Amendment right of commercial speech issue.” Then Dr. Marcia Angell, former editor of the
New England Journal of Medicine,
asked Dr. Woodcock, “Do you ever write a letter to the editor if you feel that an article has misrepresented the results of a study that the FDA has reviewed?” Dr. Woodcock said that this had been done in the past, but the FDA did not do so in this case, adding, “I don’t know why.”

The pressure from my patients to prescribe Celebrex and Vioxx did not let up, intruding into alliances that had been built up over many years. I tried to explain that these drugs offered no better relief than the older, less expensive anti-inflammatory drugs. I actually started to enjoy the challenge of trying to refocus my patients’ attention back onto their underlying issues and medical problems, trying to reengage them in the search for constructive solutions. I did my best to help them understand that their beliefs about these drugs were being masterfully manipulated by the drug companies’ multipronged marketing efforts, and that these efforts were being driven far more by the goal of improving the drug companies’ sales than improving patients’ health or comfort.