Ending Medical Reversal

by Vinayak K Prasad

  The third is perhaps the most challenging case. A 72-year-old man with thyroid cancer that has spread to his lungs begins to get symptoms from his cancer. He feels “tightness” in his neck and shortness of breath. Without a doubt, he has a serious condition that, unfortunately, will eventually lead to his demise. A new medication comes on the market (vandetanib) expressly for his cancer. A lot of things make the drug attractive. For starters, it is a pill (not an intravenous treatment) that you only have to take once a day. But there are significant downsides. The side effects are so bad that about one-third of patients cannot tolerate the starting dose, and 12 percent end up having to stop taking it altogether. These numbers are based upon the patients who are fit enough to participate in a randomized trial that studied the drug. For patients outside of trials, cancer drugs often have worse side effects. Vandetanib was approved by the FDA because it slows the growth of the tumor, but it did not show any improvement in survival. (Remember bevacizumab in chapter 3.) What does all this mean? It makes for a very tough decision. The patient, with his doctor, has to balance the symptoms he is currently having against any side effects he might develop if he takes the drug, while knowing that the drug may not make him live a single day longer.

  Decision making without data is hard. It is also, often, necessary.

  STEP 5. FIND SOMEONE WITH WHOM YOU CAN WORK

  You and your doctor will need to work together to figure out what is the right treatment for you. When there are data, they might require you to ask questions that will require some legwork on your doctor’s part. When there are no data, you will need honest advice from someone you trust. It is helpful to get a sense of the doctor’s philosophy and make sure it matches your own. We all want good care, but some may want a “less is more” approach, while others prefer an aggressive, “leave no stone unturned,” approach. If it is true that no one can really change, do not expect your doctor’s philosophy to change completely to suit your own.

  One thing that you should expect from any doctor is a willingness to answer your questions. In the hurried pace of medicine today, this may seem a lot to ask. It is not. What follow are a few suggestions about how to get the most out of your exchange with your doctor. Some of these questions and statements come from our patients and have helped them get the best care out of us; others are things we wish our patients would ask more often.

  :: USE THE DOCTOR’S EXPERIENCE

  Yes, your doctor has had a lot of training, but what makes the consultation valuable is that she has cared for many people having the experience that you are having for the first time. When a therapy is offered, take advantage of this. “How long have you been using this treatment?” “Before this drug/surgery was available, what were you doing, and why is this better?” “Have you read any studies that would make you doubt this treatment?”

  :: USE THE DOCTOR’S EXPERTISE

  If you could read and understand all the studies of all the treatments you are being offered, you would not need a doctor. Asking your doctor to explain the data can be challenging, but to make the decisions as we outlined above, you need to. Ask to see the articles that your doctor is referencing. The abstracts of articles are usually easy to understand and can open the door to asking further questions. Make it clear that you are not second-guessing. Doctors are more impressed than put off when we hear patients say,

  “I don’t mean to be a problem, but this is important to me and I just want to make sure I am going to be comfortable with what I choose.”

  “I trust your judgment, but I want to make sure this is right for me.”

  “I understand that you might not have time for this now. When is a good time that I can come back? Or can I ask you more questions by phone or email?”

  :: AIM HIGH IN STEERING

  This is probably the only phrase we still remember from driver’s ed. Sometimes the decisions that lie further ahead can inform the decisions at hand. “If this treatment does not help, what will we do next?” “Why are you suggesting the treatments in this order rather than the reverse?” These questions are not only for making decisions about therapy; they apply when considering diagnostic tests and evaluations. “If this test is negative, what do we do?” (“Are we done?” “Will we have ruled out the disease?” If not, “Should we start with a different test?”) “If the test is positive, what will we do?” “Is there a proven treatment?” (If there is not, do you really want to know about the diagnosis?) “If the test is nondiagnostic, what will we do?” “How likely is it to be nondiagnostic?”

  PARTICIPATE IN RANDOMIZED TRIALS, CONDUCTED BY IMPARTIAL SPONSORS

  We end with a final plea. There are clinical trials in which you should participate. We cannot endorse any or every clinical trial because, frankly, there are some bad studies out there. If you have cancer and are interested in a phase 1 trial, you should know that the trial is overwhelmingly unlikely to benefit you. Your motivation should be pure altruism. If you have any condition and are considering a study without a control arm, you have to probe the motivation of the investigator and ask whether the trial will provide useful information—no matter how excited you are to try something new. When considering randomized trials, remember that some industry-sponsored studies have flawed designs or do not ask the best clinical question. On the other hand, do sign up for any randomized trial fully conducted and paid for by impartial sponsors. The National Institutes of Health, the Veterans Administration—these are groups that conduct trials only to answer important clinical questions. Furthermore, you would be in a randomized trial. These are trials where there is true equipoise—we do not know which treatment is better. Not only is your treatment conducted with the safety of a treatment arm and a control arm; you guarantee yourself safe and effective treatment, but in addition, your care will help other patients who will someday be in your position. When impartial sponsors conduct clinical trials, there is no reason to hesitate. The upside is large, and the downsides are minimal and carefully monitored. Ask your doctor whether he knows of any such trials that are for people like you.

  ::

  Reversal is common in medicine. You are sure to be offered therapies that are, at the very least, potentially ineffective. Thoughtful questions, asked in the right way, to a doctor who is willing (or can be encouraged) to work with you will expose the riskiest therapies. These questions will also help you to receive proven therapies when they are available and allow you to make informed decisions when they are not.

  18 BEYOND DOGMA :: WHEN RANDOMIZED TRIALS ARE UNNECESSARY

  OVER THE PRECEDING 17 CHAPTERS, we have tried to make the case that reversal is one of the most important problems in medicine today—if not the most important. The solution to the problem of reversal is to improve the evidence on which medical practice is based. The reversal of widely accepted medical practices can be costly, challenging, contentious, and even injurious. Anyone who has followed the medical news and come away confused should feel vindicated. Medical news is confusing because medical practitioners are confused about the weight of evidence that should be present before adopting a therapy. In our research, we found that that just over half (56 percent) of authors of observational studies in toptier medical journals believed their data should change medical practice. However, as we have demonstrated over the course of this book, observational studies are unreliable, and trusting them has led to notorious mis-steps in medicine.

  In chapters 14 to 16, we presented our best ideas on how to solve the problem. Currently, doing randomized trials is difficult, costly, and complicated—facts compounded by the indifference of most patients toward participating in these trials. Our solution calls for randomized controlled trials to test every treatment decision we make. Day-to-day medical care should not just have randomized trials as a garnish—that sprig of parsley next to your steak—but should have randomized trials embedded throughout. To accomplish this, we have to create a culture in which demonstrating that a treatment works is an absolute necessity prior t
o approval and widespread adoption. The default process should be that consenting patients are enrolled in randomized trials, designed by impartial sponsors. By combining a burden-of-proof philosophy with the nudge principle, we could achieve the right model: a medical system in which every patient is treated with practices that we know work and every untested practice is tested in a randomized trial. This is our vision for medicine.

  Fundamentalism in all forms, however, is bad. We recognize that in the near term at least (and in some situations for far longer), practices not supported by randomized controlled trials must exist. As clinicians, we are aware that there needs to be wiggle room—we will need to continue to make some decisions without a solid evidence base. However, we hope that this necessity, one that is now the norm, will be restricted to a few situations, such as when a patient with a rare condition is critically ill, when a patient has a truly unique problem, and during diagnostic evaluations.

  SITUATIONS THAT ARE BOTH DIRE AND RARE

  Every physician occasionally cares for a patient whose condition is truly dire. Depending on where the doctor practices, these situations might be rare or painfully common. Consider one of the common ones. Pulmonary embolism, which we discussed in chapter 16, occurs when a blood clot, typically from the legs, has traveled to the lungs. Usually this problem is accompanied by chest pain and shortness of breath and is treated relatively easily. However, if the clot is massive in size, a pulmonary embolism can strain the heart as it struggles to pump blood past the clot, and cardiac arrest can result. This situation is dire with odds of death higher than anyone would be willing to bear. This is an uncommon occurrence but, in a large hospital, not a rare one. Several strategies have been implemented to treat these patients. Some doctors will send a patient to have open-heart surgery in an attempt to remove the clot. Others will deliver a powerful clot-busting drug via an intravenous infusion. Still other doctors will choose to deliver the same drug, but by threading a catheter right up against the clot—with the hopes of avoiding the catastrophic bleeding that can result from the clot-buster. Most commonly, doctors simply provide basic anticoagulation, the usual treatment for emboli that are not imminently dangerous, allowing the body’s own clot-busting systems to do the bulk of the work.

  The last choice here—anticoagulation alone—is the best-supported strategy. The other interventions might work, but they are not yet supported by robust randomized trials, and each of them carries real risks. For these situations—dire but not particularly rare—the model we outlined in previous chapters is best. We should do only what we know is beneficial, and each of the more experimental strategies should be tested in randomized trials (using the burden of proof as a guide and the nudge principle to optimize enrollment).

  Too often, however, situations are both dire and rare. A patient with an unusual cancer, who is receiving a novel therapy, develops fevers and begins to decline. He becomes delirious. Despite the doctors’ best efforts, including broad-spectrum antibiotics and antifungals to treat infections that might be causing the fevers, the patient does not improve and the cause of his decline (and the appropriate solution) remains a mystery. A consulting doctor suggests further suppressing the patient’s already compromised immune system. Maybe this makes sense: the fevers could be a response to the novel therapy. Maybe this course is dangerous: if there is an infection, further weakening the immune system might prove deadly. There are no clinical trials to guide us. Do the doctors give it a shot?

  In these cases, when a patient’s condition is deteriorating, and when the cause is neither clear nor typical, it is reasonable to suspend the burden-of-proof principle and do what seems reasonable. Recently, a team in Philadelphia genetically engineered an immune cell to attack cancer. Patients developed unrelenting fevers and illness after the infusion of these cells. On a hunch, they tried powerful immune blockers. In these cases, the patients got better. Did the drugs help? Possibly, yes. At a minimum, no one could fault the doctors for trying something unproven in this situation.

  The tricky part about dire and rare situations is that dire and rare are both relative terms. Dying of cancer is often dire, but for the most part, doctors should not just try random drugs hoping for a cure. In fact, one of the charges of the U.S. FDA is to prevent dying people from taking a remedy that is not very likely to help them and far more likely to hurt them. Dying can often lead to desperation, and desperate people need doctors to counsel them against futile and foolish actions. At the same time, there is a part of human nature that says you have to try something for a patient who is young (another relative term) and is dying of something you do not quite understand. Physicians need to be free and willing to improvise in situations that are dire and rare. However, if that dire situation begins to become a more common one, researchers and clinicians must design a trial to test the interventions.

  The frequency of medical reversal should inspire us toward a medical system that reliably improves human health. At the same time, in certain circumstances doctors need to be free to improvise. It is difficult to watch someone suffer or die and it is all the more difficult when that death is untimely or inexplicable. In such cases, the burden of proof should be a bit flexible. However, doctors should try to find commonalities among the dire conditions and use clinical trials to discover successful interventions. Someday, the team in Philadelphia may come to realize that their predicament is not as uncommon as it once was. It may come to be more like the case of massive pulmonary emboli. When that time comes, we should start thinking about how to test what succeeded in a pinch to make sure that it is truly beneficial.

  UNIQUE CASES

  One of the greatest accomplishments in medicine has been our ability to recognize patterns. For thousands of years, doctors have noticed commonalities among the people they treat and, in the process, identified several thousand diagnoses. With each year, we improve upon this—finding parsimonious genetic or physiological causes of complex diseases. In 1981, for instance, just five men with an unusual pneumonia led doctors to suspect that a new illness was emerging. That illness, of course, was AIDS.

  And yet, despite our remarkable ability to find patterns, the complexity that is the human body in health and disease has prevented us from figuring it all out. There are certainly diseases that have yet to be named, and some of the diseases that are presently included in our textbooks will turn out to be syndromes—collections of signs and symptoms—rather than the product of precise physiological defects. Sometimes patients have truly unique conditions or, more commonly, unique constellations of problems. A patient may have a neuroendocrine tumor (a rare cancer) and also have acute kidney failure requiring dialysis. A patient with hemophilia, a bleeding disorder, may also develop an aneurysm (a fragile swelling) of an artery in the brain. These are tough cases. Pragmatic randomized trials, those that try to reflect the real-world populations, are important, but even they are unlikely to fully guide us when the situation is unique.

  When treating patients whose problems are unique, physicians need to start with what is known. How do we approach an aneurysm in a patient with hemophilia? We start by considering the surgical approach used to treat this condition in people who do not have the bleeding disorder. Then we begin to weigh in the hemophilia. Can that disease be controlled during surgery and in the postoperative period? What is the excess risk of bleeding in patients with hemophilia? We must include those issues into the risk-benefit calculations.

  This is the sort of thinking that doctors are good at. However, when we say “good at,” we mean good at thinking that way, not necessarily good at getting the conclusions right. It is impossible to know whether doctors get these difficult cases right (even on average). We can all point to a case for which the result was favorable and in which we are proud of our decision making, but we cannot really say that it was our decision making that led to the good outcome. We can imagine creative ways to test, generally, how best to take care of patients with unique cases. For example: is it better to
have the decisions made by a specialist in that disorder, or a specialist in evidence-based medicine, or a doctor with training in risk assessment? (Remember the superspecialists of chapter 14.) A randomized trial of many patients with unique problems could answer these questions but will not provide direction on how to care for an individual in a unique situation. In these cases, doctors will always need to make decisions in an unsystematic way.

  Thus, taking care of patients with unique issues is another area in which it is acceptable to suspend the burden of proof. In the years to come, randomized controlled trials will allow us to apply evidence to patients with a broader array of medical problems, but there will always be patients in unique situations, for whom doctors will need to make educated guesses.

  DIAGNOSTIC TESTING

  Throughout this book, most of the examples of reversal we discussed had to do with treating a patient. A person has a diagnosed ailment and her doctor must decide among treatment options. Some of these options might not be supported by evidence, and if one of these interventions is eventually found wanting, then we will have an example of reversal. Much of clinical medicine, however, happens before the point where treatment is prescribed. The diagnostic process occurs when a person is taken from a complaint to a diagnosis. It is an aspect of medicine that could be far more evidence-based, but this process sometimes also needs to exist outside the strictest demands of the burden of proof.