Overdosed America (5 page)

I came away with a whole new understanding of how behind-the-scenes financial relationships between the drug companies and the academic experts (who write the articles in medical journals that are then received as “scientific evidence”) can neutralize potential criticism. I was starting to understand that these issues were all of a piece: the changes in my own practice, the disordered priorities, and the growing commercial influence in clinical research and medicine. I went back to my day-to-day practice with the sense of having lost my innocence and with a new wariness about the research findings presented in the medical journals.

In June 2001 another article reporting the results of
a study about stroke prevention
appeared, this time in the
Journal of the American Medical Association
(JAMA). The stated purpose of the study was to examine the association between ischemic strokes and the three commonly measured forms of cholesterol—HDL (good), LDL (bad), and total cholesterol. (Ischemic strokes, the most common kind, are caused by loss of blood supply to an area of the brain, resulting in cell death and irreversible brain injury.) By examining hospital records, the study identified cases of ischemic stroke among racially and ethnically diverse residents of Harlem. Each case was then matched with two similar control patients from the same community who had not suffered a stroke.

The cases and controls were then compared for significant differences in cholesterol levels, as well as other lifestyle, medical, and demographic factors, to identify risk factors associated with ischemic stroke. The study found that the people with low levels of HDL (good) cholesterol were at increased risk of suffering a stroke. On the basis of this finding, the authors recommended checking HDL cholesterol levels routinely and suggested that people with low levels consider treatment with a statin drug to increase their “good” cholesterol. From a quick read of the abstract, this recommendation made sense, but as with the Pravachol and stroke article, as I carefully went through the details I realized that the data led to a very different conclusion. And as in the case of the Pravachol article, the direction of the spin would lead doctors to prescribe more statins. I took out my pencil again.

Not mentioned in the article’s abstract, and mentioned only once in passing in the text, were the unexpected findings that the lower (that is, what we think of as healthier) the total and LDL (bad) cholesterol, the
greater
was the risk of stroke. (More on cholesterol later, but generally total cholesterol and more specifically LDL cholesterol play a role in blocking arteries, and HDL cholesterol partially counteracts this effect.) Buried within the tables included in this article were statistics showing that lower levels of total cholesterol and lower levels of LDL cholesterol were both significantly correlated with a
higher
risk of stroke (
p
< .001 and
p
= .04, respectively). As I read on, I was completely baffled by the authors’ statement that “we found no relation between total cholesterol levels and stroke risk.” How could they say there was no relationship between total cholesterol and stroke when their own data showed that the odds were greater than 1000 to 1 that lower total cholesterol levels were associated with a higher risk of stroke? A
follow-up letter to the editor
of JAMA from a doctor employed by the U.S. Department of Health and Human Services (expressing his own views) pointed out that the authors had “neglected to discuss these findings,” but once the cows are out of the barn such a letter has very little effect in correcting the misperception created by the original article.

The article got even stranger when it argued that its data supported the use of statins to prevent strokes in patients with low HDL cholesterol levels.
Statins raise HDL cholesterol
only half as much as the article found would be necessary to significantly reduce the risk of stroke. But statins lower total and LDL cholesterol at least three times more on a percentage basis,
far more than enough to significantly
increase
the risk of stroke
, according to the data from the study. Nevertheless, the article concluded that treatment of low HDL cholesterol with statin drugs could significantly decrease the risk of stroke—ignoring its own findings that the overall effect on cholesterol would be associated with increased risk of stroke.

I started to wonder why the article focused on cholesterol at all. The study found that other factors were just as significant as low HDL cholesterol in increasing the risk of stroke: untreated blood pressure, lack of exercise, cigarette smoking, heavy drinking, not graduating from high school, and being uninsured or on Medicaid. In fact, the authors of this article had
used data from the same case-control study
in an article published in 1998 to show that even light to moderate physical activity reduced the risk of stroke in the same people by 61 percent and that heavy exercise reduced the risk of stroke by 77 percent. The benefit of exercise documented by these authors certainly overshadows the 19 percent reduction in stroke associated with an increase in HDL of 5 mg/dL
*
—almost twice as much as is achievable with statins. Curiously, the authors’ earlier findings about the important role of exercise were not even mentioned in the current article. And, though the authors did cite the
earlier NEJM article about Pravachol
and stroke, they failed to mention that that article found no relationship between low HDL levels and increased risk of stroke.

It was hard for me to believe that the
Journal of the American Medical Association
would have published an article that had strayed so far from its own data and the medical literature, and so far from recommending what seemed like the best approach to helping patients avoid strokes. Thinking that it would probably not be a good idea to go back to the same research expert to review the article, I asked a well-respected cardiologist who had published more than 50 papers to look over my critique and make sure I had it right. He saw no problems with my analysis.

The JAMA study focusing on the increased risk of stroke in people with low HDL cholesterol levels was done in a community with many more risk factors and health problems than most. The
life expectancy of a black man in Harlem
is only 60 years, less than the life expectancy of a man in Bangladesh. So why did this article ignore the much more powerful antistroke effects of positive lifestyle changes and blood pressure control? And why did it focus on raising HDL cholesterol with statins, while ignoring its own findings that the more powerful effect of statins on total and LDL cholesterol would increase, not decrease, the risk of stroke?

At the time the study was published, there were two potential changes coming that could have provided commercial incentive to do so.
Pfizer had a new “HDL-elevator” drug
that was well along in the pipeline of drug approval, and already being tested in clinical trials. It was also becoming clear that senior citizens would soon get some kind of assistance with prescription drug coverage from the federal government. Lower-income minority communities like Harlem represent relatively unpenetrated markets for expensive drugs with purported widespread benefits. A generous prescription drug benefit would make statin drugs affordable to many residents in this community who could not previously afford them. It seemed sinful to me to be recommending that more than $1000 per patient be spent each year—in a community with a 34 percent poverty rate—on treating people with statins with at best shaky scientific justification, while ignoring proven interventions such as increasing participation in exercise, smoking cessation programs, nutritional counseling, blood pressure control, and other medical outreach. These could have a much stronger impact not only on stroke reduction but on overall health and quality of life.

These two articles about using statins to prevent strokes, appearing in the two most influential American medical journals, seemed to approach stroke not as a human tragedy but as a commercial opportunity. Both appeared to spin research results to provide “scientific evidence” that justified more use of expensive drugs. The article in the NEJM presented findings from a study done on an unrepresentative group of patients, findings that were not even statistically significant. The article in JAMA ignored its own data showing that treatment with a statin drug was more likely to increase the risk of stroke. Both articles focused almost exclusively on drug therapy rather than inexpensive lifestyle changes that have been shown to be far more effective.

Why was all of this so important to me? For the elderly, a debilitating stroke is one of the worst possible fates. Both of these articles seemed off-key, particularly as I reflected back on Mrs. Rose’s suffering and what I hoped to accomplish as a doctor. Remembering my own medical training and watching the medical students whom I was teaching struggle to learn how to base their care on the scientific evidence in the medical journals, I could see that doctors’ trust in the literature was being skillfully exploited by commercial interests. But my colleagues and students were skeptical when I tried to show them that the material they were trying so hard to keep up with could not always be trusted.

I had always thought of myself as a disciplined mainstream practitioner grounded in medical science, caring, and common sense. But I was losing my faith in the knowledge that guides medical practice, and there was no going back.

In retrospect I can see that these two articles on stroke reduction, for all their flaws, had at least presented enough data to allow the commercial sleight of hand to be uncovered by careful analysis. I was soon to learn that this is not always the case.

In April 2001
I noticed on my desk what looked like a piece of routine junk mail from Pharmacia, the manufacturer of Celebrex. When I opened it, the first thing I saw was the following headline, in large capital letters: “IMPORTANT CORRECTION OF DRUG INFORMATION.” This was clearly not ordinary junk mail.

The letter continued:

Dear Healthcare Provider,

This letter is being sent to you at the request of the U.S. Food and Drug Administration. The FDA’s Division of Drug Marketing, Advertising, and Communications has notified Pharmacia Corporation that it considered . . . promotional statements and actions by or on behalf of Pharmacia [concerning Celebrex] to be false or misleading and therefore in violation of the Federal Food, Drug, and Cosmetic Act.

I recalled the unusually forthright
editorial about Celebrex and Vioxx
that I had read in the
Journal of the American Medical Association
about a year and a half earlier, when these drugs were first approved. Since then, the marketing blitz for these drugs—both to doctors and to the public—had been unprecedented.

By the time the “Dear Healthcare Provider” letter arrived, most people considered Celebrex to be a breakthrough anti-inflammatory drug. Though the advertising was doing a spectacular job of making the drug seem “better” than the older nonsteroidal anti-inflammatory drugs (NSAIDs), its single claim to superiority rested on its lower risk of causing stomach problems. Nonetheless, Celebrex was selling like hotcakes, racking up well
over $3 billion in sales
in its first two years on the market. The drug was being widely prescribed for every kind of ache and pain. Rarely did a day go by when I didn’t hear the word “Celebrex” from a patient, a fellow doctor, or an ad on TV. The new drug was now accounting for
one-third of all arthritis drug sales
in the United States.

I read on, but the letter was vague. It mentioned several routine warnings, including a potentially dangerous drug interaction with the blood thinner coumadin and allergic reactions, and then went on to say that the FDA objected to marketing that “promoted Celebrex for unapproved uses, and made unsubstantiated comparative claims.” This was followed by several warnings in bold print, including the following admonition:

Serious gastrointestinal toxicity such as bleeding, ulceration or perforation of the stomach, small intestine, or large intestine, can occur . . . in patients treated with NSAIDs, including Celebrex
.

I read the letter several times, trying to understand exactly what the important message was that the FDA had instructed Pharmacia to get out to all health care providers. It seemed to be saying that Celebrex had just about the same risk of serious GI complications as other anti-inflammatory drugs. But this didn’t jibe with the results of the large Celecoxib Long-Term Arthritis Safety Study that had been published in JAMA in the fall of 2000. This study, informally known as CLASS, had made a strong case that Celebrex was safer. I wondered if new evidence had become available to the FDA in the interim. Suspecting that this was not the last I would hear of the matter, I filed the letter away.

Four months after receiving the letter from Pharmacia, I noticed an article in the influential Drug Therapy section of the NEJM,
“The Coxibs, Selective Inhibitors of Cyclooxygenase-2.”
It was a review article primarily about Celebrex and Vioxx, both members of this new class of arthritis drugs. (Review articles are designed to provide practicing doctors with expert and up-to-date overviews of important and evolving therapies.) While studies published about each of these drugs in the fall of 2000 had concluded that both were safer than the older alternatives, there were still some outstanding questions, and I assumed that this review article would answer them. Perhaps, I thought, the article would also explain the FDA-mandated letter about Celebrex.

It did nothing of the sort. The article started with the basic science: Celebrex and Vioxx were the first two selective cyclooxygenase-2 (COX-2) inhibitors approved by the FDA. COX-2 molecules participate in the inflammatory process within joints that causes the symptoms of arthritis. COX-1 molecules, on the other hand, protect the lining of the stomach from developing erosions and ulcers. The older nonselective NSAIDs (pronounced “en-sades”), such as ibuprofen (Motrin, Advil), naproxen (Aleve), and diclofenac (Voltaren), reduce inflammation by blocking the activity of COX-2 inside joints. The problem is that they also inhibit the activity of COX-1, which can lead to irritation of the lining of the stomach. The theoretical advantage of the new selective COX-2 inhibitors is that they can block inflammation without causing irritation of the stomach lining.