Snowball in a Blizzard

by Steven Hatch

  One concern was that all of these studies were either observational (the researchers just followed women who chose to take hormones and compared them to those who didn’t, but weren’t randomized into a treatment versus placebo trial) or case control (where women who suffered heart attacks or broken hips were then asked whether they had taken hormone replacement, and then their hormone use was compared to healthy controls). Still, the gathering pile of evidence seemed ever more convincing, so by the late 1980s the vast majority of physicians had become sufficiently sold on the many benefits of hormone replacement therapy. Eventually, hormone replacement therapy became part of national primary care guidelines.

  Why, then, would a pharmaceutical company wish to interfere with a juggernaut like hormone replacement therapy by performing a costly randomized trial to prove what everyone knew to be true in the first place? The answer lies in the federal legal structure concerning drug approval and drug marketing. Although physicians have a fairly wide latitude in prescribing drugs in whatever way they see fit, drug makers are given approval to market their drugs only with very specific indications. Suppose that a company seeks approval by the FDA for its antihypertensive drug, and following approval a group of neurologists use the drug experimentally to treat migraine headaches and discover it is quite effective. The trials are well designed, and they publish their results in respected journals. Despite this unanticipated additional beneficial use of the drug, the company cannot advertise this use unless it returns to the FDA and submits documents seeking approval for the new indication of treating migraine.

  Most companies don’t bother, as additional approvals for marketing are costly and time-consuming. Usually, they are content with the medical system finding “off-label” uses for their approved drugs, informally touting these soft indications through a variety of means. In some cases the off-label use becomes more important than the use for which it was approved, and drug companies try to massage those situations to their benefit.* Most often, however, the off-label market isn’t sufficiently lucrative to justify the expense of seeking additional approvals for drugs that are already available.

  The most notorious example of the abuse of the off-label system involves the marketing of the drug Neurontin, which was informally pushed as a panacea for any number of ills, including bipolar disorder, chronic pain, alcohol withdrawal, and migraine headaches, among other problems. According to a 2009 article from Bloomberg, US federal prosecutors asserted that $2.12 of $2.27 billion—or 94 percent—of Neurontin’s revenues came from off-label use. Neurontin’s initial FDA indication in 1994 was as an adjunct (that is, not even first-line) treatment for seizures; in 2002 a second approval was given for treatment of a condition called post-herpetic neuralgia, which is the sometimes severe pain that can follow an episode of shingles.

  Hormone replacement therapies, however, represented an exception, because an FDA-approved indication for prevention of heart disease could take a drug with a big market and turn it into one with a huge market. Thus the Wyeth-sponsored HERS Study (Heart and Estrogen/Progestin Replacement Study) was born. Wyeth initially sought FDA approval without having to perform such a study, but various patients’ rights advocates, such as Cynthia Pearson of the National Women’s Health Network quoted at the beginning of the chapter, objected. “You couldn’t approve a drug for healthy men without a randomized clinical trial. Even aspirin had to have a randomized controlled trial with healthy men,” Pearson told New York Times writer Gina Kolata in 2002. By contrast, some doctors were concerned about whether it was even ethical to undertake such a study given how strongly the evidence suggested hormone replacement saved lives. Since we know it works, the reasoning went, why would we give placebos to thousands of women and knowingly consign some of them to an early death?

  Because Wyeth’s goal was FDA approval for an expanded use of hormone therapy, the HERS research didn’t try to demonstrate every purported benefit that had been reported over the past two decades, but was narrowly focused on a question that could be easily defined and answered in a reasonably short time frame. The goal was to demonstrate that hormones were useful in so-called secondary prevention of heart disease—that is, they wanted to show that the estrogen/progesterone combination, when given to women who were already known to have heart disease, would prevent further heart-related complications.

  The HERS investigators ran a trial similar to the kind that I discussed in the previous chapter with respect to statins, enrolling under 3,000 women who had known heart disease and following them for four years. In effect, the investigators were proposing that the estrogen/progesterone combination might be as beneficial as statins had been shown to be. Assuming that trial was successful, it was only a short step to prove that hormone replacement could be used for preventing disease in healthy women (what is known as “primary prevention”; secondary prevention decreases the complications of people who already have a disease*) and should be indicated for preventing heart disease in all menopausal women. This represented an enormous, and as yet not fully tapped, market.

  The successful statin trials I discussed in the previous chapter were secondary prevention studies, although there is research on primary prevention with statins as well.

  The results were published in 1998, and they were not encouraging. HERS failed to show any benefit for women: almost exactly the same number of women had major cardiac events (fatal or nonfatal) whether they took hormones or not. Yet although the trial was clearly bad news for Wyeth and the other makers of estrogen and progesterone, it did not completely alter the notion that hormone replacement should remain the standard of care. Even in the conclusion to the study, the authors suggest that although this indicated that women who have heart disease should not start taking hormones, if they already happen to be taking hormones for other reasons—that is, the benefits that had been attributed to them through observational trials—then they should probably be continued. There were, after all, many such benefits that had been observed.

  But HERS was not the final word, for, in addition to it, the National Institutes of Health sponsored a trial known as the Women’s Health Initiative. WHI was much a more ambitious study. The researchers were interested in evaluating not only hormones but also diet, as well as the effects of calcium and vitamins, and they were looking at many outcomes besides heart disease, such as fractures, breast cancer, and endometrial cancers. Compared to HERS, the WHI hormone study enrolled ten times the number of women, with the goal of following them for up to twelve years. Moreover, WHI enrolled women whether or not they had heart disease and whether or not they had hysterectomies (in order to avoid problems in interpreting the data, women with hysterectomies were excluded from HERS). The sheer size of the research project could have been underwritten only by the federal government. It was a mammoth undertaking.

  The trial was originally designed to last until 2005, but by 2002 a steady accumulation of data suggested that not only were hormones not beneficial but they were harmful. Such harms included an increased risk of breast cancer, heart disease, stroke, and blood clots; we’ll look at the magnitude of those harms below. In the summer of that year, the researchers made a decision to halt the trial, sending a letter to the participants advising them to discontinue taking the estrogen/progesterone combination.† The letter was issued in early July, and the bottom fell out for hormone replacement therapy. An estimated 6 million women had been taking hormones before the announcement, and although nobody knows precisely how many women abruptly halted the medications, it was almost surely a substantial proportion of that number.

  Only part of the WHI hormone trial, which had involved 16,000 women who had not had hysterectomies and were taking the estrogen/progesterone combination, was stopped in 2002. The remaining 11,000 women who did have hysterectomies and were taking estrogen alone continued to be followed, until that arm of the trial was halted in 2004 for similar reasons.

  The consequent media frenzy was predictable if lamentable: hormone replac
ement was portrayed to be the equivalent of ingesting poison. “It was like an abrupt hit in the solar plexus—with a message that was loud and clear: If you value your life, don’t even be in the same room as a bottle of hormones,” noted doctor Steven Goldstein, a board member of the North American Menopause Society.

  One hesitates to use the word “hysteria” to describe the news coverage, but the majority of the stories dealing with the sudden end to the WHI trial had failed to make explicit the magnitude of the threat. Hormones had been shown to be of no benefit, that much was clear. But how great was the harm? The behavior of tens of thousands of women and their doctors indicated that they believed taking hormones sentenced a woman to endometrial cancer or a heart attack. Many news stories encouraged this view of matters. The bottom line was simply to stop. Even though the HERS trial should have given everyone reason to become more skeptical of the elixir-like properties of hormones, the termination of the WHI trial caused an overnight shift in the emotional attitude about them. What was once healthy had become not only useless but menacing, even terrifying.

  Yet the harms posed by prolonged hormone use were in fact relatively small. Compared to an equal number of women who were not taking replacement therapy, if 10,000 women took the combination of hormones for one year, 41 additional women would develop complications (8 cases of breast cancer, 7 heart attacks, 8 strokes, and 18 blood clots). Now this is just one year and those numbers add up over time, but there were also quantifiable benefits as well: for the cost of those problems, there would also be 6 fewer cases of colon cancer and 5 fewer broken hips.

  The more nuanced picture, then, was a portrait of uncertainty, allowing clinicians to quantify the risk of a medical approach that certainly didn’t live up to its billing, but whose problems were sufficiently minor that one could make a case for its use, particularly in women experiencing debilitating symptoms in the setting of menopause. That would definitely represent a much smaller group of patients than the market that Wyeth and other drug manufacturers had envisioned only years earlier, but in the wake of the news stories that swept the country in the summer of 2002, the notion that these drugs might still have a place was difficult to imagine. Slowly, however, a reconsideration of their role has occurred, and some patients are having conversations with their doctors about the risks versus benefits of hormone replacement. These are not completely safe drugs, but neither are they without use in particular instances.

  The story of hormone replacement therapy is one of weak effects: like statins, thousands of people needed to be followed for years before one could definitively state whether they were helpful (and, unlike statins, they weren’t). The answers could be found only through statistics using large samples. Most researchers would have guessed that there was a small benefit to hormones. That turned out not to be true, but the harms that were observed were sufficiently small that the researchers shouldn’t have been completely surprised.

  Yet the change in public perception of hormone replacement therapy pre- and post-WHI couldn’t have been more fundamental; one day the sun rose in the east, the next in the west. What could arguably be inferred from this is that because we underappreciate uncertainty, we are left with having to conceive of modern medicine as an either/or proposition of amazingly lifesaving or irredeemably toxic. So when evidence filters in that we might need to reconsider the magnitude of a treatment’s value, we don’t have a conceptual system that allows for a graded adjustment. Instead, we can only flip the treatment from one category to the other.

  The fact that hormones had been associated with all that is good in medicine and then were almost immediately associated with its polar opposite indicates that they were wildly oversold. But why was this? Public health officials and doctors could have provided caveats about the kind of correlation-based research that had been performed, and mostly they didn’t. It’s not that scientists and statisticians weren’t aware of the correlation/causation problems in the 1970s and 1980s, when hormone replacement therapy really began to take off. Many drugs are sent to trials and ultimately fail, so skepticism of hormone replacement should have been the norm, yet so many doctors were taken completely by surprise.

  There are many reasons for this, but I would suggest that one critical factor may have been that everyone wanted hormones to be successful, so the warning signs were largely ignored. What that means is that as we near the center of the spectrum of certainty, where we become less and less confident in the unimpeachable nature of the evidence, our own hopes for a treatment—or our fear of a particular disease—can have a dramatic impact on our approach to that treatment or disease, even if that means we embrace a treatment or a technology that ultimately ends up doing more harm than good. Instead of being thought of as drugs that may or may not have benefits that had not as yet been proven through trials, hormones carried an emotional resonance for women who wanted to demonstrate their commitment to their health, and their doctors shared in this enthusiasm.

  Recall the language that one pundit used to describe the cultural importance of screening mammograms: “screening is what responsible and health-conscious women do . . . those are commendable and powerful virtues.” Indeed they are. But such charged rhetoric can make it harder to appraise research in as levelheaded a manner as possible and can lead us to think we are more certain than in fact we should be. The tension inherent in the correlation/causation conundrum leaves us, doctor and patient alike, susceptible to overinterpreting the data, and it makes us captives of our own desires.

  One well-known cognitive bias from which a doctor—or for that matter anyone—can suffer is called anchoring. Anchoring means that people become so invested in their diagnosis that they actively dismiss information that should lead to a reconsideration of that diagnosis, tossing aside clues that in retrospect make perfectly clear that the original diagnosis should have been reconsidered. Ultimately, the correlation/causation problem teaches us to be aware of where we find ourselves on the spectrum of certainty and, as we near its middle, that we should be conscious of our own motivations, as those motivations may cloud our judgment. We should be especially alert of our tendency to anchor when we encounter such fuzzy data. And anyone is capable of doing this: doctors have no special corner on the market of self-awareness of their biases, or lack thereof.

  How can nonspecialists raise their antennae when hearing news about some study linking, say, dark chocolate consumption with longer life? To consider that question, we first have to think about how mass media tends to portray medical research. I’ll try to provide some specific answers in the conclusion, but, in the meantime, it is worth investigating how our health news copes with uncertainty.

  8

  “HEALTH WATCH”: HYPE, HYSTERIA, AND THE MEDIA’S OVERCONFIDENT MARCH OF PROGRESS

  To deal with the broader problem of anecdotes, what you need is a framework that tells you which anecdotes are almost surely wrong.

  —PAUL KRUGMAN

  The previous two chapters have looked explicitly at how uncertainty and our understanding of a treatment’s effectiveness are inextricably and irrevocably linked. Before that, we looked at uncertainty’s impact on diagnosis, as well as how uncertainty can make experts square off against one another, leaving not only laypeople but even ordinary practicing doctors scratching their heads, trying to understand what the right approach should be to some problem.

  Yet whether the uncertainty involves the latest technology allowing a surgeon in San Francisco to use a robot to take out a patient’s gallbladder in Longmont, Colorado, or a cutting-edge drug that may revolutionize the treatment of spinal cord injuries, or a revised guideline on the treatment of emphysema, the vast majority of people learn of these advances through some form of mainstream media, whether the New York Times, network television news, National Public Radio, or links supplied by their friends on Facebook. (This is even true for most doctors, where all but the most specialized of specialists—who stay up to date at conferences and in journals—get their n
ews from the same sources as most other people.) We swim in an endless stream of media, able to sample from dozens of stories about health each day. In this final stop on our tour of uncertainty, we’ll take a cursory look at how the media shapes our understanding of medicine, often by downplaying uncertainty’s importance. We’ll think about how it adjusts our perceptions of risk, lulling us into a false sense of security about things that are genuinely (if only subtly) dangerous in our lives, but encouraging us to become preoccupied with threats unlikely to ever do us harm.

  As we complete this survey of uncertainty, we’ll finally consider in a formal way how all this awareness of uncertainty can be enabling, rather than making us feel depressed about how deeply flawed some of our medical practices truly can be. But, before we get there, we must understand how media can, even if only inadvertently, manipulate us while trying to inform us—and how we actively participate in that process. To start, we’ll consider how the media chose to portray a singularly remarkable event when some German doctors made an announcement that shook the world of infectious disease to its core.

  The Berlin Patient

  December 15, 2010, was just another day for HIV-infected patients and the people who care for them. Across the world, more than 6,000 people became infected, and although nearly all of these people would not realize it that particular day, they had joined the growing ranks of people around the world living with HIV, which numbered some 30 million. That same day, at the other end of the HIV continuum, nearly the same number (about 5,000), who had been infected for years, succumbed to the disease.