"Medical reversal" occurs when new data demonstrate that a commonly-accepted practice is not beneficial or even harmful to patients. is a world-leading expert in this phenomenon. In this "Doc to Doc" interview, I spoke to him about the trends in medical reversals, their underpinnings, and how they can be ended in the future. His book is titled "."
A transcript of the interview follows.
Perry Wilson, MD: Medical reversal is the phenomenon where a long-established medical practice is changed due to new, emerging evidence. One of the leading experts on medical reversal is Vinay Prasad. He's an assistant professor of medicine at Oregon Health & Science University, and in conjunction with co-author Adam Cifu, he has written . Dr. Prasad, thanks for joining me on Doc to Doc.
Vinay Prasad, MD: Thanks so much for having me.
Wilson: The concept of medical reversal I think is easy to understand, but can you give us a few examples to sort of orient us here? What are the recent examples of medical reversal that come to your mind when you think about this phenomenon?
Prasad: Yeah, so I would say a medical reversal is something that doctors had been doing for years, often even decades, that have been extolled by professional guidelines, professional societies, that we may have even been reimbursed for or incentivized to do that later were found to be of no benefit or maybe even harmful to patients.
Some of the kind of key examples I think about are, one, the glycemic targets in the intensive care unit. Even when I was a resident doing training, we really chased tight glycemic control in the medical ICU, but of course, just a few years later, a randomized control trial NICE-SUGAR came out showing that that actually led to net harm without benefit.
One of the other things I think about that comes to mind rather quickly is the use of steroid injection for spinal stenosis, so epidural steroid injection. Now a couple years ago, there was an outbreak of fungal meningitis in compounding pharmacies, and people rightly faulted the compounding pharmacies for failing to follow proper hygienic policies. But one of the things that was under-discussed was the fact, "Well, why are we doing so many epidural steroid injections?" We were doing it because we thought that it would decrease pain over the course of weeks to perhaps even months in patients, and there were a lot of uncontrolled studies and anecdotal reports that that was true. But there are also well done sham-controlled studies showing that that benefit is, in fact, probably nothing more than a placebo effect.
Then the last example I'll give you is kind of an intensely controversial topic, but that's the use of routine coronary stents for chronic stable angina. I think a huge sector of stenting revenue comes from stenting stable angina. Even surveys of patients as recently as two years ago show that 85% of patients having it done believe it's done to lower the rate of myocardial infarction or to improve longevity. Of course, the well done, randomized controlled trial COURAGE shows it does neither of those two things. We can have a longer discussion about the possible symptom benefit, but I will point out that it's never been tested against sham-control, and there are many of us who question even that symptomatic benefit.
Wilson: So certainly as I thought about these medical reversals, controversy comes to mind, and in many of the cases that you cite in your book, some of the ones you just mentioned. Another example might be PSA testing, routine PSA screening for prostate cancer. There has been a shift in thinking. There's been new data. There are new guidelines that suggest that some of these things are not appropriate to be done, and yet there seems to be an intense backlash against that. Where do you see that coming from? Why does that exist?
Prasad: I will say that's a great question and we have to admit that medicine, as an institution, is a bit like a battleship. We don't turn on a dime. So new information, even strong new information, doesn't lead to change in practice. Now you may wonder, "What is the deep cause of this?" I guess I've thought about that a bit and I think it really is two things. One, doctors are generally good people. We are really acting with the best intentions of patients in our mind. We've done things for years because we truly believe they make our patients better off.
You combine that psychological reward, doing something for someone, you believe it makes them better off with a little bit, in many cases, of financial reimbursement for doing that. I think together that's a very potent psychological stimulus, and it's very hard to break that sort of almost like an addiction. You become addicted to doing something. You believe it works and you're reimbursed for it. I think that's a lot of medical reversal, but not the full story. It's a bit about financial incentives, but it's not totally about financial incentives.
Wilson: There are certain things that do result in direct reimbursement. Spinal injections might be an example, whereas in contrast, you brought up intensive glycemic control in the ICU as sort of a counterexample where there's probably not a direct profit motive for the physicians engaging in that practice there. Do you see a difference in the adoption of new medical standards that divides along lines that are attributable to financial interest?
Prasad: That's a good question. I guess there's two related questions. One is, are we more likely to adopt things if they have a direct financial benefit? The second thing, are we less likely to de-adopt them if they're later contradicted? The answer is, I think, there is some evidence that the way in which we adopt practices does depend on the amount of skin doctors have in the game, regrettably. I know we wish it were a perfect system based solely on a clean understanding of the evidence, but there is evidence, to some degree that that's not the case.
In terms of why we de-adopt things, why we do some things faster than others, I think there just hasn't been enough empirical work in that field, and I'm interested in doing more and I hope others do more as well. But I do appreciate your point that not every reversal is a reversal that puts money in doctors' pockets, and I don't want to make it seem like that is the driving force. A lot of these reversals are simply things that are very intuitive. We believe they work. We have strong faith in the pathophysiology.
In some cases, doctors may not be the beneficiaries of these procedures, but other third parties may be -- such as the biopharmaceutical industry -- and they may be exerting influence on different parts of the medical process like the writing of guidelines or perhaps even through patient advocacy groups.
So, there's different levels and complexities to all of these examples. There's no one-size-fits-all pattern for why this happens, except for one thing. The one thing that is one-size-fits-all is where do reversals come from, which is we adopted something in biomedicine based on a promise that it would work without definitive studies that it does work. Those studies are typically not even ongoing at the time of adoption. These were things that were incredibly seductive. They had strong pathophysiologic rationale, but they didn't have, in most cases, robust empirical, randomized data showing that they worked.
Wilson: I was going to get into that a bit. Many of these reversals happen after a large randomized trial. You bring up the issue of flecainide, which was an antiarrhythmic agent, which when you talk about biologic plausibility, the use of flecainide after MI, where presumably an arrhythmia-induced heart attack was the sequela you were trying to avoid, flecainide dramatically reduces the rate of arrhythmia. All the biological plausibility in the world is there suggesting that this drug should reduce the rate of death after MI, and yet when the definitive trial was done, a trial that you point out many questioned even the ethics of doing in the first place. But when the definitive trial was done, quite the opposite was shown. Is it the standard that these large, randomized trials are redefining what we think or is it smaller ancillary studies that sort of build up to a new understanding?
Prasad: Yeah, so I think that's a fantastic question. The history of the CAS trial really revealed sort of the highlights of reversal. It came into prominence for good reason. It made perfect biological sense. PVCs, the R-on-T phenomenon, those were implicated at sudden cardiac death. This is a drug that suppresses those aberrant beats. This is a drug that should lower early mortality after myocardial infarction. As you point out, cardiologists, in many cases, were reluctant to enroll patients on the CAS study saying that there was no equipoise. It is unethical not to give these patients access to medication, and of course, that turned out to be a study where the drug actually caused harm.
So, I will say there are some takeaway lessons. There are many things in modern medicine that there are practitioners who say, "It would be unethical to study this rigorously." I believe we have to really take those claims with a grain of salt. The history of medicine is littered with the smartest, most compassionate, thoughtful investigators being reluctant to test things they thought they knew were true and only to find that when they were tested by courageous investigators, they were contradicted.
I think the lesson of reversal is we need robust, large-scale, pragmatic, randomized control trials. That should be the rule of biomedicine and not the exception. And that anecdotal, case-based, retrospective, uncontrolled studies, while those things have a role, they're hypothesis generating. They shouldn't be used for definitive understanding of the efficacy of particularly novel treatments.
Wilson: I need to ask you, then, in your opinion, what's a therapy that is happening now that is ripe for randomized trial? Where would you like to see that money spent, for example?
Prasad: There are many things and I'll give you one that's in my wheelhouse so I keep a close eye on it. There's something we're doing in oncology right now, which is offered through commercial testing services like Guardant360 or Foundation Medicine, and that is for patients with solid tumors that have failed all conventional therapies, patients who are undergoing next-generation sequencing, whole exome sequencing, and they're taking drugs for mutations they have in a unregulated fashion that might work, but we don't know that they work for sure.
We've had one randomized trial in this space of personalized cancer medicine. It's called SHIVA study. It was published in Lancet Oncology. That was a negative study. But meanwhile, we've had over 100,000 patients in the United States with cancer who've undergone next-generation sequencing, some of which may have taken off-label drugs. I have called for in several publications, but the most noted of which is a piece I wrote in Nature called "The Precision Oncology Illusion," for randomized, controlled trials of this practice. I believe that they have a huge value. They will answer a clinical question that is actually happening with tremendous cost, financial cost, as well as a human cost, the toxicity of off-label medications that may not work. There are a number of sort of more nuanced reasons why this trial is really needed, desperately needed at this juncture.
Wilson: You've elevated randomized trials to a position of extreme authority in these cases, and a lot of people would say, "Randomized trials are great, but they are far from perfect. The patients who get into a randomized trial or who choose to join a randomized trial rarely resemble the patients we see in everyday practice." There are a lot of biases that can creep in. There are potentially financial incentives that are playing on the doctors and the sponsors of the trials. To people who are skeptical of randomized trials in general, is a single randomized trial, for example, good enough to change a long-standing practice?
Prasad: Okay, that's a fantastic question. Let me just make a few points. One is I think you have to separate some of the problems of randomized trials that are intrinsic to randomization from those that are man-made problems, basically. They're problems that are intrinsic to randomized trials, and they're problems that are man made.
I like to give the analogy of just because United Airlines is doing a bad job of customer service doesn't mean that airplanes should be thrown away. So when it comes to only looking at healthy people, that's a man-made problem. We don't have to enroll just healthy people on randomized trials. We can have something called pragmatic randomized trials with very broad inclusion criteria, trials that actually do look at older individuals, patients that look like the patients you and I may see in our clinic. We can do that. That's been done before. It can be done. That's a man-made problem.
The other problem is the cost of randomized trials is often cited. It's been said that enrolling a single person is as much as $30,000. That's a man-made problem and in The Netherlands in a randomized trial called TASTE, they've lowered the cost to $50 per person. But there are some problems of randomization that are intrinsic and the biggest of which is you don't need a randomized trial for something that has a tremendous effect size.
The classic example is you don't need a randomized trial for wearing a parachute when you jump out of an airplane. That's an intervention that has a 99.9999% improvement in overall mortality in a very short period of time. No one has ever called for a randomized trial of the parachute, and similarly, no one has called for a randomized trial of if I got hit by a bus and my femur were exploded out of my leg, and I had a compound fracture, no one has said, "You need a randomized trial of putting it back in my leg or just leaving it out and letting nature take its course."
So there are, in fact, things that we won't be testing in randomized trials, that aren't necessary to test. But what is necessary to test in randomized trials are biomedical interventions for complex or multifaceted pathways with modest to marginal effect sizes. I hate to say that's the majority of biomedicine. Most of what we do has a modest to marginal effect size, and the paper that proves that is a paper called "Empirical Evaluation of Large Effect Size," by John Ioannidis in JAMA. Okay, so that's the bit about randomized trials.
There's some problems that are man made. We should be clear about those and then there's some places that randomized trials, you're absolutely right. They aren't necessary and they aren't needed. There are some parachute practices in biomedicine. They're just few and far between.
Your second point is can a single randomized trial really contradict a huge body of evidence, like these reversals we've found? I would say that in many cases and we're always taught that randomized trials build upon each other, and we should be doing systematic reviews and meta-analyses and that's absolutely correct. But in the cases of the reversals we've outlined in our book, these are really practices that the totality of the evidence that came before the randomized trial is very fragmented, lacking, uncontrolled, unreliable evidence. The randomized trial is almost the only light in the sea of darkness. It's the only reliable study where before it came no reliable studies, so we are fairly confident these are true reversals, the 146 we give in the appendix of the book.
Wilson: I want to talk about the patient perspective on all of this. What comes to mind in my practice, I'm a nephrologist, I treat a lot of hypertension, is a patient comes in, they've been on atenolol for years, and all of a sudden one day they come to me and I say, "You know what? We're going to switch that drug because there's new evidence to suggest that this drug might not be helping you avoid the outcomes we're trying to avoid like heart attack." Patients may lose trust in that situation. How do you address that with patients when you say, "We're treating things differently now?"
Prasad: Yeah, and I think so many doctors have the experience of having done something with good faith, that we really kind of cajoled or extolled our patients to do that later turned out not to be true. For me, one of the things I think of a lot is the use of extended-release niacin to boost HDL and lower triglycerides, an intervention that in the AIM-HIGH study was found to be no better when added on top of statin therapy. Yet, that was something that I had been doing prior to that randomized trial.
I think you make an excellent point. It is very tough as a clinician to go in there and say, "Look, I was wrong about something I was doing. I wasn't wrong because I was ill motivated. I was wrong because everyone was wrong about this." But I think we have to be honest about it. We do get things wrong.
The second thing I have to say is that we should be careful not to use these examples as reasons to dismiss all of Western medicine or all of biomedicine. There are a lot of people out there right now who are willing to seize upon the imperfections of medicine as reason for why we're wrong about everything, why vaccines may cause autism, or all sorts of conspiracy theories.
We have to be careful to say, "Look, we have gotten this wrong, but yet, medicine is still the best way of moving forward about issues of health." I completely sympathize with your predicament. I've been there. It isn't easy and we're walking a fine line.
Wilson: Well, Dr. Prasad, it's been a fascinating conversation. Thank you for tracking these reversals for us, and we are looking forward to seeing much more from you. The book, once again, is called Ending Medical Reversal: Improving Outcomes, Saving Lives, available at any place fine literature is sold and please join us again sometime.
Prasad: Thank you. Thanks for having me.
, is an assistant professor of medicine at the Yale School of Medicine. He is a 51˶ reviewer, and in addition to his video analyses, he authors a blog, . You can follow .