For an editorial in the Journal of the American Medical Association, Arthur Slutsky, a well-respected critical-care physician, used the subtitle “The Seduction of Physiology.” What did he mean? Slutsky understood that doctors like interventions that cause improvements in vital measures like blood pressure and oxygenation. We know these numbers are important, are objective, and correspond to improvements in physiology that we understand. It is reasonable to assume that if we can improve these measures in critically ill patients, we can improve survival. Therein lies the false seduction of physiology. Many times, we have found that interventions that improve these measures do nothing to improve the survival of the patient. Improving physiological measures is comforting, but it does not always affect outcomes. None of this is to say that paying attention to physiology or designing therapies that improve physiological measures is worthless—it is not. But any such intervention must be rigorously tested, rather than blindly followed. In the next chapter we propose a system of medical education that does a better job of balancing the importance of empiricism and reductionism.
ABANDONING A PAYCHECK
A colleague recently joked that flawed therapies have paid for many a beach house. His statement is cynical and hyperbolic but not without truth. In chapter 8, we wrote about how tenaciously interventional cardiologists have clung to the practice of stenting open coronary lesions in people with stable angina—a $12 billion a year industry—even after the COURAGE trial showed it was ineffective. Sure, interventional cardiologists have witnessed cases in which these stents have benefited patients, but they are also the group most likely to have their income negatively impacted by the findings of COURAGE.
Although doctors—ourselves included—like to believe that no amount of financial remuneration could sway their judgment, hundreds of studies have shown that financial conflicts of interest do bias physicians. When you hear of conflict of interest, you might think of doctors doing consulting work. A psychiatrist is paid by Merck to give talks on depression and then goes on to emphasize the benefits of Merck drugs. This certainly is a conflict, but it is one that our field has been pretty aggressive in exposing. We wish we had a nickel for every “COI form” we have completed before giving talks or publishing papers. However, there are far more insidious types of conflict of interest. Any time a physician recommends a procedure or treatment that will increase his take-home pay, there is at least the potential for conflict of interest. Orthopedists may not always see the downside of joint replacement as clearly as nonorthopedists; oncologists tend to err on the side of giving chemotherapy; and general internists who own X-ray machines tend to order more X-rays. A recent study tracked urology specialists’ groups that bought a machine that could deliver radiation to the prostate. Are you surprised that groups that purchased the machine tended to give more radiation?
It is very hard to accept evidence that something you have done for patients, something that you truly believed was beneficial, is not useful. The evidence is even harder to accept when you have been well compensated for your work. Because of this, acceptance of medical reversals is never easy and opposition to them is usually passionate.
::
Human beings are creative, optimistic, and hopeful. These are traits that have served our species well. At the same time, these traits make us susceptible to medical reversals. Our creative minds figure out ways that therapies may work—what intracellular pathways they would affect, what principles of physiology they satisfy. Our optimistic side feels better, even when interventions are equivalent to the placebo. In our hopefulness we continue to believe that tomorrow will be better than today. The enduring belief is that the new and more expensive must be better than the old and economical. In many aspects of life, this disposition is surely beneficial—we make more friends, solve problems, and enjoy ourselves—but when it comes to deciding how to plan our health-care systems and how to manage our own medical care, these traits can lead to erroneous conclusions. In the next chapters we argue that although the roots of reversal are vast, from the scientific method to human psychology, a few simple rules can curb its growth and even prune it back.
PART 4 BEYOND REVERSAL
14 MEDICAL EDUCATION :: A VERY GOOD PLACE TO START
OVER THE COURSE of this book, we have sought to answer the questions, how often are accepted practices overturned (reversed, not replaced)? and, why does reversal occur? We have presented the case that reversal is ubiquitous in medical practice and its causes are diverse. These include financial bias, investigator bias, and unbridled (and unjustified) optimism with regard to novel treatments on the part of all the players in medicine: doctors, scientists, industry, fledgling biotech companies, advocacy groups, the media, and patients. Our next step is to move from theory to practice. If we accept that reversal is common, harmful, and not a necessary part of medical progress, can we find a way to decrease its prevalence?
With such diverse causes, it will be a challenge to end, or even reduce, reversal. We need to find a bottleneck, a narrow canyon, where all the sources can be ambushed. Thankfully, we have such a place. Because the proximate cause of reversal is adopting therapies based on insufficient evidence, improving evidence will reduce the prevalence of medical reversal.
How do we improve the evidence on which the practice of medicine is based? In theory, it seems pretty simple. For all new medical practices, screening tests, diagnostic tests, pills, novel devices, procedures, and surgeries, well-done randomized trials must show improvement in real outcomes (such as quality of life or mortality) before those practices gain FDA approval or are widely used. Instituting this strategy would ensure that all new practices really work and might even check the rising costs of medicine by no longer spending money on ineffective innovations. We also need to address the existing therapies that lack a strong evidence base. To do this, we could enumerate all existing, unproved medical practices and rank them by cost and frequency of use. Then, prioritizing the common and costly ones, test each with a randomized trial. We would then abandon whatever does not work. John McKinlay, whom we spoke of earlier, was truly prescient when he suggested similar solutions 35 years ago.
If only it were really that simple. To implement such a solution, the entire health-care industry would require a new ethic. This ethic would demand that the burden of proof, the task of proving the utility of every new therapy (and the older, unproven ones), rest on those who stand to benefit from instituting the therapy. The implementation of this strategy would be challenged by a long list of reasonable concerns. How would we stimulate (or legislate) the adoption of a new ethic? Would requiring more solid evidence before adopting new therapies slow innovation? What would count as solid evidence? Certainly not every randomized controlled trial is a good one. What sorts of end points should randomized trials examine? And why should we not just throw out all unproven medicine in one fell swoop?
We try to answer these concerns in the coming chapters. But first, in this chapter and the next, we suggest some specific reforms that we think would go a long way toward changing the culture of medicine. We start where doctors start, in medical school, and then move to where many doctors are trained and where much of the medical innovation occurs, in academic medical sciences.
These chapters are quite different from the preceding ones. Up to this point we have taken great care to support everything we have said with evidence. When we pointed out a reversal, we outlined how the therapy was proved to be ineffective and explained why, in retrospect, doctors had mistakenly accepted the therapy. These chapters are, by necessity, more speculative. We cannot say for sure that a certain type of medical training, or a certain standard for promotion of academic faculty, increases the prevalence of reversal. Furthermore, we certainly cannot know that the reforms we suggest would reduce the frequency of reversal in the future. But we explain why we believe certain features of medical education and the structure of academic medical centers predispose our field to accepting therapies that are later
overturned. We then go on to suggest changes that we think will be effective but that we hope will be tested. We would obviously be the last people to suggest that an intervention be accepted just because it makes sense.
WHAT MAKES A GOOD PHYSICIAN?
When we consider how doctors should be trained, we need to think first about the sort of doctor we want to produce. We need to get very specific here—we need to specify who the we is when we ask, What type of doctors do we want? Medical schools are interested in graduating fine physicians and academic leaders (which often means doctors who are most productive in research). All other things being equal, they prefer to turn out doctors who generously support the alumni association. Residency directors have a shorter horizon when considering the medical school product. A resident who works hard, furthers the hospital’s clinical mission, gains admission to a prestigious fellowship, and does not cause problems is the ideal doctor for a training program. Hospital systems and medical groups are looking for “rainmakers,” those doctors who can bring in many grants or well-insured patients, ideally for highly reimbursed treatments. Payers, on the other hand, are hoping for doctors who are cost-conscious, ordering few tests and making fewer referrals. We could, of course, go on to talk about the desires of plaintiffs’ lawyers, defense lawyers, and pharmaceutical companies.
For our purposes, let us consider the sort of doctors that we, as future patients, want. Of course there is no one type of doctor that every person wants. The doctors we desire are as diverse as we are. Some of us want a doctor who will give us full autonomy in our decision making, while others want a doctor who will strongly guide us. We have worked with doctors who are warm and friendly and others who seem distant and detached. We have known doctors who refer to their patients by first names and always ask about patients’ families, travels, and holidays and others who expressly steer clear of these perceived niceties. All these doctors have had loyal cadres of patients.
What then can we say about the doctors we want our medical schools to produce? Tomorrow’s doctor probably needs to master three areas while in medical school: the doctor-patient relationship, systems-based practice, and practice-based learning. First, the doctor-patient relationship. While no doctor can be everything to everybody, it is not asking too much that our schools train their graduates to be most things to most patients. They must interpret a patient’s needs and values and respond in a productive, therapeutic manner. They need to understand how to communicate with all types of people: those who want autonomy and those who do not; those who want a hand-holder and those who want a more distant, though empathic, guide; those who look like the doctor and those who do not. To develop this kind of physician requires schools to provide extensive mentored clinical experience.
Not only do our doctors need to be experts in working with patients; they need to work within the growing health-care team in a facile manner. Decades ago doctors collaborated only with nurses (and often in less-than-collegial ways). Today, every hospitalization requires that the attending physician work collaboratively not only with nurses but with subspecialty teams, physical therapists, social workers, case managers, and many other affiliated health-care workers. In today’s lingo, this is referred to as systems-based practice.
Lastly, we want doctors to know their stuff. They cannot be satisfied with “knowing their stuff” when they graduate, since most of the specific “stuff” they learned will be obsolete before they complete their training. Tomorrow’s doctors need to master the skills necessary to practice from a strong and current evidence base throughout their careers. This means continuing to learn from their patients and from the medical literature until the day they retire. We call this practice-based learning. Doctors need to interpret data so that they know which interventions are well supported and which are not, and then be able to present this information to their patients. Medical school graduates who enter careers in research must be motivated to always question untested dogma. Physician scientists need to think not only like scientists but like clinicians as well.
RETHINKING THE FOUNDATIONS OF MEDICAL EDUCATION
Although we train many extraordinary doctors, the frequency of reversal in medicine tells us that, at present, our system of medical education is not completely successful. Medical education stands on two pillars, constructed in the early 20th century. The first pillar is premedical (college) course work. Many of these courses, such as organic chemistry, physics, and calculus, are the subjects that were important in the early 1900s. (These courses were probably deemed important at that time because they were the only subjects with any bearing on the study of medicine.) The second pillar is the four-year medical-school curriculum: years one and two spent studying science and years three and four spent seeing patients. This was the ground-breaking approach to structuring medical education put forth by Abraham Flexner in the early 1900s.
Today, both of these pillars of training are being questioned. Critics argue that the college classes students take to gain acceptance to medical school are of little (organic chemistry) or no (physics and calculus) use to the practicing physician. These courses not only do little to prepare students for their careers as doctors; they are not even necessary to understand the first two years (the science portion) of medical school. Furthermore, the first two years of studying science in medical school have been criticized as being of little value to the vast majority of physicians.
If we accept these critiques, premedical and medical education should be reformed for the sake of efficiency alone. However, we have an even deeper concern about the structure of medical education. It is not hard to see that the years spent studying how the body’s small parts work may actually be harmful. These classes indoctrinate students with a belief in the primacy of the foundational medical sciences over the clinical ones. As we noted in chapter 13, we currently train students to be reductionists rather than empiricists. Students come to believe that a medical therapy works because of its mechanism. This is not an absolute truth. The mechanism is one way a therapy might work. If the same model explains future data, then it is probably the way this therapy does work. One can only be certain that a given therapy actually works when the therapy is shown to work in randomized controlled trials.
Science is the theory that underlies medicine. Science (as well as experience) is crucial for generating hypotheses to be tested. Clinical trials are real-world data points. Let us consider the example of cholesterol-lowering medications in the treatment of high cholesterol. The hypothesis suggested that because high cholesterol is associated with heart attacks, lowering cholesterol would prevent heart attacks. Trials using statin-type cholesterol-lowering drugs were designed to test this hypothesis. The trials proved that using statins prevents heart attacks. The trials did not prove the success of the mechanism (in this case, that lowering cholesterol prevents heart attacks). It will take more trials, with other types of medications, to prove the mechanism. This has not yet been done. In many other cases, the hypothesis is actually disproved, leaving us with neither a mechanism nor a viable treatment. Recall the CAST trial from chapter 1. In that example the mechanism that was proposed was that decreasing the frequency of premature ventricular contractions would save lives. When this was tested, we found that the hypothesis was incorrect and the treatment based on it was actually harmful. Proponents of scientific models are sometimes so confident that they regard trials as a nuisance.
Thus, the problem in the organization of medical education is a problem of priority. It is not that basic science is unnecessary to physicians and that medical students should never learn or think about it. Absolutely not. Without a rich understanding of the foundational sciences, medical progress would cease. But basic science is not the first thing a practicing clinician should learn. The primacy of the basic sciences is the reason that cardiologists could not accept the finding that niacin did not save lives. It is why radiologists could not accept that vertebroplasty did not help back pain. It is the reason orthopedists cou
ld not accept that repairing torn menisci did not help knee pain. They thought, “How can an empirical study contradict the mechanism?” The reality is that the human body is so complicated, and our understanding of it so superficial, that what we believe should work often does not.
REBUILDING MEDICAL EDUCATION FROM THE GROUND UP
What, then, should a medical education that is designed to produce better clinicians, less prone to advocating ineffective treatments, look like? This training would start before medical school. A student committed to becoming a doctor should be expected to arrive at medical school already knowing the basics of biochemistry and physiology (the chemical and molecular basis of all life). An understanding of basic anatomy would also be expected. Anatomic studies have traditionally been reserved for the rarified (and formaldehyde-infused) air of medical schools, based on the idea that the privilege of dissecting the human cadaver should be reserved for physicians in training. In today’s world of lifelike models and three-dimensional computer imaging, there is no reason that college students could not take part in high-quality anatomy courses. A strong argument could be made that understanding how our bodies work, on both a microscopic and a macroscopic level, is not out of place in a 21st-century college education. This preliminary course work (biochemistry, physiology, anatomy) is certainly teachable in undergraduate and postbaccalaureate programs. It should take the place of the currently required courses (physics, calculus) that were considered critical preparation for medical school around the time of the First World War.
Ending Medical Reversal Page 18