Overdosed America (12 page)

The Nurses’ Health Study was the source of the data for the 1997 NEJM article showing a
lower mortality rate among women taking HRT
. This benefit of HRT lasted for up to five years after women stopped taking the therapy, but after five years they had a 16 percent
increase
in their death rate compared with women who had never taken hormones. These contradictory findings deserved a lot more attention than they got at the time. Why did the nurses experience a lower death rate while they were taking hormones, and for five years after they stopped? And why did they then go on to experience a significantly higher death rate than women who had never been on HRT?

As the study published in the NEJM concluded, the lower mortality rate among nurses taking hormones could have been due to the “survival benefit” of the hormones. In other words, women had a lower mortality rate because they took hormones. The reverse could also have been true: the women who were already going to have a lower mortality rate might have been more likely to take hormones. In other words, perhaps it was their greater propensity toward health or the absence of disease that led them to take the hormones, and not the reverse. The third possibility is that the women who took the hormones believed that they were doing something that would protect their health and that this placebo effect played a role in keeping them healthier.

The Nurses’ Health Study researchers statistically adjusted their results for many potentially confounding factors: body weight, cigarette smoking, high blood pressure, high cholesterol, early heart attack in a parent, history of breast cancer in mother or sister, previous use of birth control pills, number of children, age of onset of period, diet, alcohol use, multivitamin use, vitamin E use, aspirin use, and regular exercise. Despite this meticulous statistical caution, however, the most elegant aspect of the Nurses’ Health Study may also have been its tragic flaw. The study purposely included only registered nurses to minimize the impact of differences in “socioeconomic status and access to health care.” But the researchers may have trusted this assumption too much.

It turns out that several important confounding factors had not been adequately taken into account, precisely the kinds of variables that the researchers thought that their study design—by including only registered nurses—had eliminated. Women who chose to take hormones tended to be
wealthier and better educated, were twice as likely to be white
than black, and
received more preventative health care
.
Dr. Elizabeth Barrett-Connor
, a professor at the University of California, San Diego, points out that while the observational studies of HRT were going on, doctors across the country knew, from Premarin product labeling (included in the
Physicians’ Desk Reference
), that women with a history of heart disease, high blood pressure, or diabetes should not be treated with estrogen. This would certainly contribute to the spurious finding in the observational studies that women who took hormones had less heart disease. Of course they did: Not only were they the type of women who were going to be healthier; the women taking hormones had already been screened to exclude those with heart disease, high blood pressure, and diabetes. The observational studies that found that Premarin decreased the risk of heart disease had not, according to Dr. Barrett-Connor, adequately adjusted for this.

In 1998 the results of the
first randomized controlled clinical trial of HRT
were published. The Heart and Estrogen/Progesterone Replacement Study (HERS), published in JAMA, had been designed to determine whether HRT decreased the risk of recurrent heart disease in women who already had heart problems (so-called secondary prevention). The results came as a huge surprise. This manufacturer-sponsored study showed that despite significantly lowering LDL (bad) cholesterol and raising HDL (good) cholesterol, HRT
increased
women’s risk of heart disease by 50 percent in the first year. And over the four years of the study, treatment with hormones provided no reduction in the risk of developing cardiovascular disease. In fact, this study showed that the overall death rate was not lower in the women who took hormones; it was slightly higher.

Why had Wyeth-Ayerst, the manufacturer of Premarin, agreed to do this study, taking a chance that the results would come out against the company’s interests? The answer is simple: the insatiable quest to expand its market for hormones. In 1990,
Wyeth-Ayerst had requested that the FDA approve
Premarin for the prevention of heart disease in postmenopausal women, notwithstanding the lack of evidence from randomized controlled trials documenting such a benefit. Cynthia Pearson, of the nonprofit, independent National Women’s Health Network, pointed out in an FDA hearing that the evidence supporting this claim was weak: “You couldn’t approve a drug for healthy men without a randomized clinical trial.
Even aspirin
[to prevent heart disease] had to have a randomized controlled trial with healthy men.”

Ms. Pearson’s argument—that the standard for the gander ought to apply to the goose—prevailed.
The FDA ruled that a randomized controlled trial was necessary
to justify the claim that HRT decreased a woman’s risk of heart disease. Wyeth-Ayerst agreed to perform the requisite study, confident that the results would come out in their favor. Eight years later, when the results became available, we learned that combined HRT (estrogen plus progestin) does not prevent heart disease.

The truth about HRT came out very slowly and was difficult for most doctors to accept. Even after the article appeared in JAMA in 2000 showing the
8 percent per year increase in the risk of breast cancer in women
taking combined hormone therapy, most experts continued to recommend, and most doctors continued to prescribe, routine hormone replacement for postmenopausal women. In 2001, Premarin was
still the third most frequently prescribed drug
in the United States.

In July of 2002, everyone found out that it was not Mother Nature but the American College of Physicians and the
American College of Obstetrics and Gynecology
that had made the big mistake. The newspapers were full of articles about the
government-sponsored Women’s Health Initiative study
, which had been specifically designed to determine whether routine HRT was beneficial for postmenopausal women. The 16,000 women in the study had been randomly assigned to take either combined (estrogen and progestin) HRT or a placebo. The study had been scheduled to run through 2005, but the women received letters instructing them to stop taking the study medication because the risk (increase in breast cancer, heart attack, stroke, and blood clots) associated with combined HRT had been found to be significantly greater than the benefit (decrease in hip fractures and colon cancers). The study found a statistically significant increase (15 percent) in the overall frequency of adverse events in the women taking HRT compared with the women who took a placebo. This translated into about one adverse event for every 100 women who took hormones for five years. The data and safety monitoring board of the study determined that the increased frequency of complications in the women who had taken HRT instead of a placebo had “crossed the designated boundary . . . of a finding of overall harm,” and that it was no longer ethical to continue the study knowing that the women taking HRT were going to be harmed more than helped.

Two more defeats for routine combined HRT (other than for treatment of menopausal and postmenopausal symptoms) followed soon thereafter. In May 2003 more results from the Women’s Health Initiative were published showing that combined
HRT not only did not prevent Alzheimer’s disease
, but actually doubled the risk of developing dementia (primarily Alzheimer’s disease) in women aged 65 and older, causing about 1 additional case of dementia for every 100 women treated with HRT for five years.

Just three months later came the final nail in the coffin, when the results of probably the biggest study of all time, the Million Women Study, were published in
The Lancet.
One million women in the United Kingdom completed questionnaires about their personal health, sociodemographic information, whether they had reached menopause, and whether they had ever taken or were currently taking hormones. Over the following four years, the local cancer registries reported when women participating in the study developed or died from breast cancer. The results showed that women who were currently taking hormones had a 66 percent higher chance of getting breast cancer (30 percent for those taking only estrogen, and 100 percent for those taking both estrogen and progestin) than the women who were not taking hormones. The women taking hormones were also significantly more likely to die of breast cancer than the women not taking hormones. To put the risk into perspective, the researchers calculated that there had been about 20,000 extra cases of breast cancer caused by HRT in the United Kingdom over the previous 10 years. Based on the difference in population size alone, even at the same rate of hormone use, there would have been an extra 94,000 cases of breast cancer in the United States in the previous 10 years as a result of women’s taking HRT. The total number of American women who developed breast cancer because of taking HRT was likely to be much higher than this, though, because
women in the United States were about four times more likely
to take HRT than women in the United Kingdom.

There is yet another disturbing finding from the
Million Women Study
: Starting in the early 1980s, progestins were added to estrogen to reduce the risk of uterine cancer in women who had not had a hysterectomy. The risk of uterine cancer was reduced to near zero, but nobody evaluated the overall effect of adding progestin to routine HRT. The Million Women Study confirmed that the taking of estrogen without progestin caused 10 extra uterine cancers and 5 extra breast cancers, a total of 15 extra cancers per 1000 women over 10 years. Adding progestin did, in fact, eliminate the risk of uterine cancer but caused an extra 19 breast cancers per 1000 women over 10 years. In other words, the problem of uterine cancer was “solved” by adding a drug that
increased
a woman’s overall risk of getting other types of cancer.

The estrogen-only part of the Women’s Health Initiative Study was ended prematurely, in February 2004. Researchers concluded that after almost seven years, the
women taking estrogen
had more strokes and fewer broken hips than the women taking placebos. The most important finding was that there was no overall benefit to taking estrogen and that, therefore, it “should not be recommended for chronic disease prevention in postmenopausal women.”

Twenty million
American women have taken HRT not only to relieve symptoms such as hot flashes and vaginal dryness but also believing that hormones would protect their hearts, decrease Alzheimer’s and Parkinson’s disease, prevent tooth loss and diabetes, strengthen their bones, preserve sexual function and urinary continence, improve the quality of their lives, and increase their longevity. The women who took HRT had access to the best care that American medicine had to offer: Compared with the population at large, they were more likely to have graduated from college, were wealthier, and were more likely to have received preventive health care. Despite this, they unwittingly exposed themselves to increased risks of breast cancer, heart attack, stroke, Alzheimer’s disease, and blood clots.

In the end, the 26 percent increase in the risk of breast cancer caused by HRT found in the Women’s Health Initiative study was deemed by the National Institutes of Health “too high a price to pay, even if there were a heart benefit.” This risk had been known a decade earlier, when the American College of Physicians made its recommendation that all women without unusual risks should consider taking HRT, but was not of enough concern at the time to invoke the principle “First do no harm.” Far more than 100,000 unnecessary cases of breast cancer resulted.

The important lessons from this chapter of American medical history will be lost if we simply attribute the debacle of routine HRT for healthy postmenopausal women to the vagaries of medical progress. Failure to understand how this mistake occurred commits us (doctors and patients alike) to being naively swept up by each new cycle of exaggerated claims about the effectiveness and safety of ever more expensive medical therapies—that is, until even newer “medical knowledge” is produced that supports even costlier drugs. The fundamental lesson to be learned from the HRT debacle is that therapeutic decisions must be based on solid and unbiased scientific evidence. The trend, however, is going in exactly the opposite direction. The ideals and personal goals that had brought me to medicine and then to family practice now called me to investigate, full-time, just how the fundamental mission of American medicine was being undermined—and how we might begin to fix it.

In 1982, the
National Governors Association sponsored a conference in New Orleans to explore innovative ways for states to control their unsustainably increasing health care costs. I presented research that I had done as a Robert Wood Johnson Fellow showing the benefit of offering inner city of Cleveland families covered by Medicaid the option of enrolling in a health maintenance organization that would give them access to private primary care doctors. For families who opted to join the HMO, hospital admissions and emergency room visits had plummeted, immunization rates and well-child care visits had improved, and costs had gone down. All the pilot programs being presented at this conference were variations on the same theme, and all came to the same conclusion: establishing a relationship between each Medicaid patient and a primary care doctor responsible for providing and coordinating all medical care improved the quality of care and, at the same time, decreased costs.