If you’ve enjoyed my previous take-downs of evidence-based medicine but can’t let go of your attachment to the randomized controlled trial, this post is for you.
My aim is to show you practical ways you can safely and effectively exercise clinical judgment without recourse to “evidence-based” knowledge, provided you follow simple but fundamental principles of clinical care: circumspection, parsimony, and due respect to patient autonomy.
What’s more, I will make my case against RCTs using examples that EBM apologists have precisely identified as paradigmatic of this “single greatest medical advance.”
— Vinay Prasad (@VinayPrasad82) May 11, 2016
So here we go.
1) Hormone replacement therapy
Prior to the Women’s Health Initiative trial (WHI), many physicians thought hormone replacement therapy (HRT) would reduce rates of heart disease. There was a plausible pathophysiological rationale, and retrospective studies were encouraging. In fact, the WHI study was undertaken to firmly establish benefit, but the trial was stopped early because of an unexpected increase in rates of cardiovascular events and cancers in women taking HRT.
If this were 1999, Clinician “Do we really need RCT of Hormone replacement? Ugh, what a waste, it works!” Oops! https://t.co/XBn1D9evLX
— Vinay Prasad (@VinayPrasad82) May 15, 2016
But did a doctor really need to know the outcomes of the WHI trial in order to make an appropriate decision concerning post menopausal care?
Here is a hypothetical example of how one could have handled the question of HRT without any recourse to WHI knowledge:
You are now post-menopausal, and some physicians argue that the use of HRT could reduce your chance of a heart attack because it improves your cholesterol profile. Many previous studies also suggest such a benefit. However, all these studies have important limitations and besides, your cholesterol level is only one of many risk factors for heart disease. Therefore the true benefit of hormone replacement in regards to the prevention of heart disease remains unknown.
We also know that at high doses, estrogens and other hormones can cause problems, such as blood clots, and can increase one’s risk for certain cancers. Granted, HRT contains very low doses of hormones, but there is still reason to be cautious about these drugs.
While the overall effect of hormone replacement is uncertain, given that menopause is a natural and universal phenomenon, my personal stance in regards to hormone replacement is one of caution.
If your symptoms are so severe that you are having difficulty functioning in your day-to-day life, we could certainly consider treatment. Our cumulative experience with these drugs seems to indicate that if they are harmful, the risk of complications is low.
One could have offered these arguments pre-WHI and, as it turns out, these same arguments are perfectly kosher today post-WHI. If HRT was indiscriminately prescribed prior to the WHI, that was in part under the influence of the academic and public health community and their undue enthusiasm for risk factor modification.
Prior to the publication of the famous COURAGE trial, patients with stable coronary disease received untold numbers of unnecessary stents on the alleged basis that these stents might prevent heart attacks or even death. Following publication of the COURAGE trial, which showed that stenting in patients with stable coronary syndromes could be safely deferred, the number of stent procedures declined significantly.
And the mother of all questions: “Do we really need the COURAGE trial? We KNOW stents prevent MIs and death.” https://t.co/4x3aEpSb1u
— Shannon Brownlee (@ShannonBrownlee) May 15, 2016
But did an interventional cardiologist really need the COURAGE trial to start exercising self-restraint? Here’s an example of a conversation one could easily have had at any time before the 2007 publication of COURAGE:
You have symptoms of coronary disease, but the non-invasive tests that we have done do not indicate that you are at high risk of complications in the short term.
Now, we know from experience that a stent procedure could improve your symptoms quickly, but we also know from experience that any stent procedure runs some risks of complications, including heart attacks, both in the short term and in the long term.
Whether a successful stent procedure would, in your case, reduce the overall chance of a future heart attack is really unclear, but it could not possibly be something we can assume. If there is an overall reduction in the rate of future heart attacks, that reduction in risk is likely to be very small.
My opinion is that we should start with medications and see how things go.
You have symptoms of coronary disease. At age 58, you are still young and very active. Before your symptoms began, you were running up the hills of San Francisco and you wish to resume your activities as soon as safe and feasible. Although one could try to manage your symptoms with medications, my experience has been that the most expeditious and likely way to take care of your needs is to proceed with an angiogram and stent the lesion that is likely responsible for your chest pain. Whether the stent reduces your chance of a heart attack or not remains unproven. But based on what I know about you, I would recommend stenting.
Either one of these options was available to doctors prior to COURAGE, and each is still indisputably a very reasonable option post COURAGE.
If stents were overutilized—and they undoubtedly were—it is not because doctors could not reason their way toward an appropriate use of stents, but more likely because doctors and hospitals had financial incentives to prescribe them and implant them liberally.
3) Cancer therapeutics
In the 1980s and 1990s, some oncologists and many patients expressed a great deal of enthusiasm for the treatment of metastatic breast and lung cancer with bone marrow transplantation (BMT). Clinical trials eventually concluded that the treatments had no benefit. Until those results were established conclusively, however, thousands of patients pursued the futile procedures, and controversies and litigations ensued regarding who should be financially responsible for the enormous costs of an unproven therapy.
If this were 1992: Clinician “Do we really need RCT of autotx for breast cancer? Effect is so obvious!” Oops! https://t.co/XBn1D9evLX
— Vinay Prasad (@VinayPrasad82) May 15, 2016
But did physicians really need the advent of RCTs to give the proper recommendation in regards to bone marrow transplantation for solid tumors?
The uncertainty as to the benefit of BMT was evident from the get-go. The high risks of the treatment were also known. These would have been counterbalanced by a quasi certainty of a poor outcome with usual care. There is no reason to think that careful deliberation on the part of doctors and patients couldn’t have led to an appropriate use of these experimental therapies, but such “appropriate use” would clearly have been on the basis of subjective risk tolerance and subjective value judgments.
Subjective value judgments, however, were swayed by the fact that we practice in a system where the costs of treatments are born by others. Third party payment of care divorces value from price and creates financial conflicts of interests for all involved. If patients (or doctors) had to directly bear the cost for these expensive protocols, their enthusiasm for wishful experimentation would likely have been greatly reduced.
4) CAST OFF!
Now for a tough example.
Prior to the 1989 publication of the CAST randomized trial, patients who had survived a heart attack and had frequent extra heartbeats (identified on the electrocardiogram as premature ventricular complexes, or PVCs) were treated with anti-arrhythmic medications to suppress the PVCs. PVCs are a marker for an increased risk of cardiac arrest, which is an erratic heart rhythm frequently initiated by a PVC. It was hoped that suppressing the PVCs would reduce mortality rates—a plausible hypothesis. Instead, the CAST trial showed the opposite: anti-arrhythmic medications can precipitate a cardiac arrest.
If this were 1986, Clinician “Do we really need RCTs of flecainide? Come on! Definitely works!” Oops! https://t.co/XBn1D9evLX
— Vinay Prasad (@VinayPrasad82) May 15, 2016
Could one have managed to treat heart attack patients appropriately without the benefit of the CAST results? Denying the value of CAST seems like something only an inveterate contrarian like me would do. Indeed, purely preventive treatments present a challenge for the EBM skeptic: the benefit is plausible and theoretically great (avoiding death), the costs relatively modest (few side-effects), and the risks of the intervention may be unknown or underestimated. In cases like these, it seems that nothing short of a randomized control trial can provide the needed information to inform medical decisions.
Am I willing to concede the point? Not so fast.
Almost a decade before the publication of CAST, Dr. Curt Furberg had published a review of all the small short-term studies available to-date and concluded that “the findings from the reviewed trials support the explanation that control of ventricular arrhythmias does not improve prognosis (emphasis mine).” Furthermore, a subsequent CAST Pilot Study also confirmed that despite effective suppression of PVCs, a beneficial effect of anti-arrhythmics on mortality at one year could not be demonstrated. Mention was made in that paper that some of these drugs have “pro-arrhythmic” adverse effects.
Why did cardiologists persist in their enthusiasm for anti-arrhythmic therapy in the face of this sobering reality? Could it be out of an undue desire for scientific certitude? Dr. Furberg, for example, was unsatisfied with the remaining uncertainty: “the [early] trials were too small.” He therefore called for a large study to be conducted and many others agreed with him. Yet common sense should have caused one to pause and wonder if the benefit of the added information was worth the risk of obtaining it.
If a large trial is required, that means that the benefit is likely going to be small at best. Were those patients who ultimately enrolled in the CAST trial fully informed of the unimpressive extant track record of the drugs and told of the meager benefit one could realistically hope for? I sincerely doubt it. A physician genuinely concerned with the welfare of patients should have been highly guarded about the value of such a trial.
To be clear, I don’t give myself as an example of such such virtuousness and I realize that my retrospective judgment is easy to hold, now that the CAST trial has made the case against anti-arrhythmics incontrovertible. But the point remains that sound judgment can always be applied when using therapeutics for which benefit or harm is uncertain, even in the absence of large RCT data.
As I related in a previous post, the focus of medicine underwent an essential transformation in the wake of the Flexner reforms. The warnings of Francis Peabody and William Osler went unheeded: care for the patient was demoted while scientific rigor was elevated. Prior to the CAST trial, to say that anti-arrhythmics “could” save lives was scientifically correct. But being scientifically correct is not the same as being clinically right.
Mistaking the symptom for the cause
EBM is invariably promoted as a reliable alternative to the fallible clinical judgment of physicians, a fallibility that can mislead doctors on a grand scale. But, as the examples above have shown, indiscriminate use of therapeutic agents is not a failure of clinical judgment per se, but a sign that clinical judgment is impaired systematically by some factor, be it a financial conflict of interest, an abuse of professional privilege, a fixation on scientific purism, or a misguided public health attitude. To view EBM as cure for a systematic impairment in clinical judgment mistakes the symptom for the cause.
Besides, good clinical judgment can be taught and should be fostered. If medical schools are failing in that regard, their preoccupation with obtaining research grants or generating clinical revenue may be distracting them from their primary mission. EBM and guideline medicine should not be substitutes for inadequate medical education.
So friends, let go of EBM. Be not afraid. You were meant to use your intellectual abilities and your imagination, supplied by what you have learned from trusted sources and what you experience yourself, if you pay close attention to the patient. And be mindful of any inordinate quest for scientific certainty, lest you be paralyzed by EBM’s empirical skepticism.
You have pledged to serve individual patients. That means that you must strive to be prudent and circumspect at all times and, as much as possible, “particularize” the available information to the patient at hand, not “generalize” RCT findings to broad swaths of the population which is how EBM would lead you to practice.
So don’t let EBM turn your professional calling into a mindless application of rules. Instead, enjoy the freedom to do what you know is the right thing to do.