The story of gabapentin is not the most recent, but it is certainly the most egregious example of this practice. In 1993 gabapentin, under the trade name Neurontin, was approved by the FDA to be used, in combination with other drugs, for the treatment of seizures. Parke-Davis, the drug’s maker, then began to market it to doctors for an impressive number of maladies. A list of these conditions (included in a later Justice Department press release) included bipolar disorder, pain disorders, Lou Gehrig’s disease, attention-deficit disorder, migraine, restless-leg syndrome, and seizures—but as a first-line, single-agent treatment. The marketing of gabapentin as a single agent for seizures was particularly outrageous because the FDA had actually rejected the company’s application for this indication. The aggressive marketing initially paid off—Parke-Davis sold a lot of gabapentin. But in the end, the firm was hit with a $430 million settlement by the Justice Department.* More than 20 years later, gabapentin is still not approved for most of the indications for which it was initially promoted. It is unknown how many patients were harmed because they received a treatment that did not work, experienced adverse effects of gabapentin, or were delayed in their receipt of other, approved treatments for their condition.
::
The roots of reversal are extensive. Primarily, these roots are made of insufficient or flawed data. Often the treatments that are fed by these roots are accepted with the best intentions—doctors and drug and device developers truly believe in the treatments and think the data, or explanatory model, they have are sufficient. This chapter has explored instances when the intentions are less than admirable. Drug developers have willfully distorted evidence in order to sell medications. Guidelines recommend therapies not because they work but because the recommendations serve the interests of the people who wrote the guidelines. Therapies that are approved based on their promise are often not adequately tested. If these therapies are tested and found wanting, they often remain available. It is difficult and expensive to develop a new treatment. It takes dedicated researchers to determine whether a novel therapy is effective. A system riddled with conflicts of interest makes it nearly impossible for the process to work.
13 WHY ARE WE SO ATTRACTED TO FLAWED THERAPIES?
RECENTLY, A PHYSICIAN FRIEND and insomniac confided in us that the FDA was messing up (not actually her original choice of words). She had read about a new sleep medication, suvorexant, which works not by causing drowsiness but by decreasing wakefulness. Regulators needed to hurry up and approve that medication, she said, because she desperately needed it. Although the comment should not have been surprising, it was. First, being a physician, our friend should know that the “decreasing wakefulness” effect was almost certainly a marketing ploy, being no different from causing drowsiness and not readily assessed by science. Second, she likely also knew that approved sleep aids offer quite minimal benefits.* But even though our physician friend should know all this, she still felt like—hurry up and give me access to that pill. What was wrong with her? It just so happens to be the same thing that is wrong with all of us.
Throughout this book we have given examples of how doctors and patients are seduced by practices that do not work. The phenomenon is ubiquitous—in chapter 7 we argued that as much as 40 percent of the things doctors do are ineffective. In the last few chapters, we have tried to summarize why reversal is so common. Thus far, we have explained the cause of reversal from a scientific standpoint: reversals occur when a practice is adopted before it has a strong evidence base. The weak evidence base is often ignored because of doctors’ faith in mechanistic explanations or studies that were designed to be deceptive by industry. The perverse financial incentives among doctors and inventors often assist the process. But, despite this explanation, we have not tackled the harder problem: what is it that makes us so readily seduced? This is a subtle distinction. To some extent we have dwelled on the how, but there is also a why. Why is it that we, as doctors and patients, are so likely to adopt flawed therapies, even if we understand what makes the promise of a specific novel therapy dubious. Why was our friend so confident that the next new pill would be so great? What is it about us that makes us ready and willing (if not eager) to adopt therapies that may not help?
ACT NOW, DATA LATER
We have a problem; we need a solution. We hear the mantra every day. We need to solve this problem now. Ten minutes ago. Yesterday. It is not just in medicine but everywhere. Global warming, immigration, the federal deficit, obesity. We have to do something now. The problems cited are often serious and real and sometimes truly frightening. Obesity now contributes to one in five deaths in America. Absolutely, it is serious, and absolutely, we need to do whatever we can to reduce the burden of disease. Ridding the country (the world) of obesity cannot come a minute too soon. But what exactly do we do?
Do we promote diet and exercise? Do we concentrate only on the most obese or on those with medical problems resulting from their obesity, or do we try to shift the entire bell curve? Do we change farm subsidies? Do we tax sugar-sweetened beverages? Do we encourage personal responsibility, or corporate responsibility, or both? Do we tell people to eat at home? Does organic help? Does weight depend more on diet or exercise, fat or carbohydrates? When you get to proposing an actual intervention, things get very complicated. Knowing that obesity, HIV, cancer, heart disease, malaria, domestic violence, and suicide are problems is easy. Figuring out how to solve the problems is very hard.
At one of the hospitals where we work, a group recently launched a “stand up once an hour initiative” based on data that time spent sitting is linked to obesity and early death. This “act now” group felt that enough was enough. If you must sit, stand once an hour. Of course, that recommendation has not been tested at all, and posting “stand up” flyers (the actual intervention here) is even further removed from the actual science. It is unlikely that all of the posters and fliers printed—completely well-intentioned—changed any health outcomes. But we live in an “act now” world.
Medical reversal occurs when we act before we have data. Something seems like it should work, so we act on our assumptions. One way of satisfying our need to act, without doing so prematurely, might be to redirect our energy from “act now, data later” to “data now, act later.” This could be accomplished by doing exactly what you were going to do anyway: institute your “stand up every hour” initiative, but take 10 percent of the funds you planned to spend on the intervention and use it instead to pay for data collection. Enlist participants and randomize them to stand once an hour or not, and measure weights before and after the study. Better still, randomize medical schools to the placement of posters and fliers and follow the weights of a random sample of students, faculty, and staff before and after. This design is called a “cluster randomized trial.” Or be really ambitious and recognize that weight is just a surrogate end point. Can you test whether once-per-hour standing at work will improve quality of life or decrease the incidence of disease or mortality? Yes, it is hard to test an intervention on young people and wait decades for an outcome, so why not pick a high-risk group? Enroll patients in a heart failure clinic and randomize them to stand once an hour. Can you improve clinically important outcomes in this vulnerable population with a simple intervention?
It feels good to do something about a problem, but what is important is to do something that works. “Data now, act later” could actually prove efficacy, thus providing a real solution to the problem.
THE TECHNOLOGY EFFECT
Another reason we are so attracted to flawed therapies is the obviousness of the benefits of technology. Over the past century, technology has successfully addressed problems that, historically, seemed intractable. We now communicate and travel around the world with ease. We are freed from the discomfort of hot summer days and cold winter nights. Information is now available anytime, anywhere. Just 25 years ago, disputes about basic facts— for instance, was Kevin Bacon in Apollo 13?—could not be answered until the library (or
the video store) opened the next day. Now, there is no reason to speculate.
Because technology has addressed so many of our daily challenges, we are eager to embrace the next innovation. But, as we discussed in chapter 10, not every advance that would seem to offer obvious benefit really does. Sometimes we do not need a study to tell us whether a new technology is better than the old one; sometimes we do; and this helps explain why we are so vulnerable to reversal. When it comes to your new iPhone or television, you do not need a randomized trial to tell you that the screen is crisper, the processor is smoother, and the sound is clearer. Because of this, we come to believe that the new sleeping pill must be better than the old one, or that a prostatectomy performed by a robot must be better than one performed by a surgeon. But the human body is very different from technology. It is much more complicated and more challenging (as well as more magnificent). Its complete workings remain largely a mystery, and the effects of our interventions are often small, below the sensitivity of crude human observation. You cannot observe two surgical approaches and know which is better the way you can view two TVs and choose the superior one. And yet, the temptation to extrapolate our experiences with technology to biology is irresistible.
Whatever the reason, our tendency to believe that “newer is better” is real. Focus groups of health-care consumers reveal this bias.* In one study, not only did people believe that newer technology was usually better, but one-third also felt that medical treatments that work the best usually cost the most. Only one-third of participants recalled having a physician explain the scientific evidence that shows which care is best. Many surveyed were unclear what “evidence” even is—thinking it was the totality of their test results and exam findings and not the studies that support the benefit of an intervention.
DIRECT-TO-CONSUMER ADVERTISING
In America, it is not just our predispositions that lead us to believe in new technologies; we also have a little help in the form of direct-to-consumer (DTC) advertising, advertisements produced by drug companies and shown to potential patients. No matter how little media you consume, you are certainly familiar with the nearly ubiquitous advertisements for the newest drugs. They appear on TV, radio, and Internet pop-ups. The United States is one of only two nations that permit the DTC marketing of medical products. Besides being annoyed by some of the advertisements, many people do not see a problem with DTC advertising of prescription drugs. Admittedly, the advertisements create interest in a new, generally more expensive product, but proponents assert that the advertisements are also educational, and if a person still needs to ask his doctor for a prescription, what is the problem?
In 2005 a group of researchers nicely illustrated why we should worry about DTC advertising. They performed a randomized trial in which they enlisted standardized patients (SPs). SPs are actors who are trained to convincingly play the part of a patient. In this study, SPs were sent to visit local general practitioners and were randomized to display one of two disorders: depression or adjustment disorder. We all know what depression is, a serious mental illness that is often, appropriately, treated with medications. Adjustment disorder is on the depression spectrum; it resembles depression but is generally milder. It occurs when the normal reaction to a real life trauma—the loss of a job or a loved one—becomes excessive. Adjustment disorder occurs when life stressors overwhelm our coping ability. The treatment is generally time and counseling, not medications.
In the trial, each SP was randomized to describe depression or adjustment disorder. The SPs were also randomized to make one of three types of requests of the physicians: brand-specific; general; or none. The brand-specific request was something like “I recently saw an ad for Paxil, might it help me?” The generic request was “Could I benefit from a medication?” The results were intriguing on multiple levels. First, advertisers got what they paid for: if an SP made a brand-specific request, she was much more likely to receive the brand of medication she requested. Second, if an SP requested a medication, she was more likely to get one than if she did not. Third, the SPs who made any type of request were more likely to receive appropriate care than those who did not. Thus, being an informed patient helps us to get the best care, and being an informed patient who has been told by a drug company what to ask for helps the drug company. Among SPs presenting with symptoms of adjustment disorder, a condition usually not treated with medication, 10 percent who did not make a request left with a prescription—probably an appropriate proportion. For those who made a general or brand-specific request, the prescribing rates were 39 and 55 percent, respectively.
These results tell us just why drug companies love DTC advertising. The ads do two things. First, they “educate” the public that medications exist, and, second, they provide specific brand names. Even though a doctor is the ultimate gatekeeper, when patients bring up drugs and specific brands to doctors, they often get what they ask for. After all, there is a lot of art to medicine, and as a general rule, unless the request is completely off the mark, it pays to work with patients.
Between 1999 and 2005, DTC advertising increased from just under $1 billion to just over $4 billion. This increase tells you that DTC marketing successfully sells drugs. Does it also work to improve health? Unfortunately, we do not know. While it might seem draconian to restrict the ability of companies to advertise their products, prescription drugs are a different commodity from Caribbean vacations or new cars. You need a great deal of training to make sense of the studies that evaluate new drugs and to impartially conclude when they are best used. Laws that require motorcycle helmets, seatbelts, and contributions to the social security program exist because people do not always weigh risk appropriately. As a society, we believe that, to an extent, a role of government is to protect people from unnecessarily risky or bad decisions. DTC advertising is good for business but may not be good for health. For these reasons, DTC advertising remains banned in every country except the United States and New Zealand.
BASIC SCIENCE IN EDUCATION
“Act now, data later,” the technology effect, and advertising influence the choices of people outside the medical field as much as they do people inside it. The next couple of topics preferentially influence doctors and how they make patient-care decisions. One reason doctors are so easily swayed by the newest breakthrough in the medical sciences is that mechanistic thinking is so deeply engrained in our training. In medical school, the study of how therapies should work is much more extensive and comes before the study of whether therapies do work. In other words, we would much rather quiz medical students about whether venlafaxine inhibits postsynaptic reuptake of norepinephrine, serotonin, or both, than ask them to critically appraise the randomized trials testing whether, and to what extent, venlafaxine improves depressive symptoms. If that sounds odd—it is. Our current system is a result of how medical knowledge and education have evolved. For most of the 20th century, quizzing medical students about how things should work was all medical schools could do. There was a paucity of trials testing whether and to what extent therapies actually did work. Then the evidence-based movement began, and clinical practice shifted, but education has been slow to follow.
When we talk about how therapies should work, we are talking about the basic sciences. We are talking about the mechanisms of how things work. In the scientific tradition, there are two complementary schools of thought, reductionism and empiricism. Reductionists believe that the more we understand about the mechanism by which disease happens and therapies act, the closer we are to optimally targeting therapy and curing disease. Empiricists are concerned foremost with whether interventions work. Why should you give a drug in the angiotensin-converting enzyme-inhibitor class to a patient with heart failure? The reductionist would answer, Because it prevents ventricular remodeling (a maladaptive process). The empiricist would answer, because multiple randomized trials show that it improves outcomes. You see the distinction? The views are complementary. You have to be a reductionist to have a good
understanding of biology and to drive medical innovation. But in the end, it is the empiricist whose answer is correct. However, historically, the reductionist approach has always had priority in medical education.
Doctors who care for patients should be, and mostly are, empiricists. We are happy to use drugs that have been shown to work even though we do not quite understand their mechanisms of action. For instance, we know inhaled anesthetic gases can knock you out, but we have no idea just how they work. The training in premedical courses and in medical school, however, is heavily steeped in reductionism. We learn organic chemistry and biochemistry. We learn molecular biology. We learn anatomy. All of this understanding is important—to a certain extent. If you are planning to have a career developing new drugs, then it is very important. But the internist who writes a prescription for a blood-pressure pill does not really care a whole lot about how it works; he cares that it lowers blood pressure (a surrogate end point) and decreases cardiovascular events (a clinically important end point). Even a surgeon is most concerned with the few dozen anatomic structures she crosses paths with—she has long forgotten the thousands of tiny tendrils she dissected in anatomy class. It is for this reason that empiricism should carry the day in medical training.
Rather than being replaced, reductionism is experiencing a revival. When people celebrate the coming era of “personalized medicine,” they imply that soon our understanding of the mechanism of disease will be so robust, so sophisticated, that you will not need to empirically demonstrate the efficacy of interventions. This is unrealistically optimistic. Whether the mechanism for a therapy is the mechanics of lung inflation or the tiny binding pocket of a renegade protein, you still need a trial to ensure that what you are doing actually accomplishes the ends you envision. The rumors of empiricism’s demise have been greatly exaggerated.
Ending Medical Reversal Page 17