A doctor stands over you and feeds a long wire up the inside of your body, from your groin into your heart. You are conscious and comfortable, if not necessarily calm.

For nearly half a century, cardiologists have been performing angioplasty, or percutaneous coronary intervention (PCI). The point is to open up arteries that have become clogged and hardened by years of life—to reverse and improve the symptoms of heart disease. Today the procedure is done around a million times each year in the United States alone. In the span of 45 minutes or so, the looming cardiologist watches a screen as real-time X-ray images show the wire going up to the heart, traversing vessels to reach the coronary arteries. The cardiologist then feeds a small balloon over that wire and inflates it, forcing open the blood vessels at the point of narrowing. The vessels can be held open with a metal tube that expands over the balloon, known as a stent. Immediately, the vessels look better on the X-ray images. The narrow area is wide open, and blood is flowing freely.

“When we tell a patient, look, we ‘fixed’ you, this has an immense positive effect,” said John Mandrola, a cardiac electrophysiologist in Louisville, Kentucky.

He and many others have watched as their patients tend to report less chest pain, more energy, better stamina, and “all sorts of benefits.” As the Mayo Clinic tells readers of its site, “Angioplasty is used to treat a type of heart disease known as atherosclerosis. Atherosclerosis is the slow buildup of fatty plaques in your heart’s blood vessels. Your doctor might suggest angioplasty as a treatment option when medications or lifestyle changes aren’t enough to improve your heart health, or if you have a heart attack, worsening chest pain (angina), or other symptoms.”

“Your doctor” may be wrong in doing so, though, according to a controversial study published last week that, if widely accepted, has the potential to change the course of medicine. Data in The Lancet showed that among people with severe blockage of the coronary arteries, the procedure did not improve angina—the reason for nearly 500,000 PCI procedures worldwide every year—or ability to exercise on a treadmill. Now after 40 years, millions of procedures, and billions of dollars, doctors are questioning whether the common procedure is, in most nonemergency cases, doing much less good than previously believed, if any.

But many patients and doctors swear it works. Mandrola describes a typical case: A patient is told they should have this cardiac catheterization because of chest pain and a cardiac-stress test that suggests disease. Patients come to understand from friends and the internet that if a blockage is found, the doctor will “fix it.”

“Blockages are deadly and must be fixed, goes the thinking,” Mandrola explained. “So they do the PCI, and they bring the family into the lab or show them pictures of the blockage. Everyone is happy. Doctors, nurses, patient, and family. That whole scenario creates a whirl of placebo effect. And the patient feels better.”

“I’ve been saying for many years that we don’t know if patients feel better from stents, or if they feel better because patients always feel better when we do an invasive procedure,” said Rita Redberg, a professor of medicine at the University of California, San Francisco. “That’s how the mind words.”

Knowing that this procedure’s effect is based heavily in placebo, it would seem, will diminish its effect. Reading this article may cause people to have more chest pain.

If stents to open clogged coronary arteries indeed don’t help people in nonemergency situations, this would be among the highest-stakes disappointments in medicine to date, when more than anything else, people around the world die of heart disease. It usually involves blood vessels getting progressively clogged up and then totally blocked. This causes the heart muscle to die, and so the person dies.

“This is a hugely disruptive study,” said Mandrola. “The implications are huge. Billions of dollars have been spent, and many hundreds of thousands of patients have been exposed to the risks of PCI, without any documented benefit.”

Redberg went even further. “I think this has shown definitively that there was no benefit on exercise time, no benefit for angina, no benefit for functional status—it was pretty definitively negative.”

So are there any reasons—outside of an acute heart attack—to do PCI?

“I can’t think of one,” she said. “Why have a person undergo the risk of the procedure?”

Mayo Clinic makes these risks clear: The abridged version includes blood clots, heart attack (the wire can displace some of the plaque inside the wall of the artery and actually cause a heart attack), coronary-artery damage (recall that there is a wire and force-inducing balloon being inserted into the three-millimeter vessels on the surface of the heart), abnormal heart rhythms, kidney failure, and stroke. These are rare outcomes, but they become increasingly relevant as the benefits of the procedure become less clear.

* * *

Really, how could angioplasty not be effective?

The problem highlights the disconnect between how the world should be and how it is. This disease seemed so straightforward. It required no genius to suggest that the blockage of vessels could be fixed by opening up the vessels. In much the same way that a plumber removes a ball of hair and a shower is instantly new again, a cardiologist might remove years of living hard and restore the flow of blood to the heart.

The first and larger shock to doctors on this subject came in 2007, when results from the landmark COURAGE Trial appeared in the The New England Journal of Medicine. Researchers had compared PCI with medications alone and found that the procedure, surprisingly, did not reduce a person’s risk of subsequent heart attacks. It turned out that the heart was not as simple as a toilet.

Still, that study did find the procedure provided some degree of relief from angina (the chest pain associated with heart disease), though even that effect waned after three years. Despite the modest benefits, and high cost of the disease, many cardiologists continued to do the procedure. “It was already in the culture,” said Redberg. “We’ve been doing PCI for so long.”

Doctors believe it works, or patients want it, and they may leave bad reviews for doctors who don’t do the procedure. Or worse, doctors might fear getting sued if they don’t do a procedure that’s still considered a standard of care.

The new Lancet study also “teaches clinicians the crucial value of placebo effect—and its ugly cousin, the nocebo effect,” said Mandrola, referring to the concept wherein a belief that something will bring about a negative effect results in an actual negative effect. “When we frighten people about their disease or possible side effects of a drug, this, too, has immense negative effects.”

Because the placebo effect is such a powerful force in medicine, researchers in the Lancet study had to devise a way for patients not to know if they had gotten a stent or not. That meant using a “sham procedure,” in which patients with severe disease would go into the operating room, have an incision made, and have a wire inserted—but there would no ballooning, and no stenting. The entire procedure was an act, and the patient was kept in the dark as to whether they actually had a metal cage in their artery or not.

“The methods are as beautiful as I have seen. So, so, so good,” said Mandrola. The pictures in The Lancet show severe blockages among these patients’ coronary arteries. These were not the sort that many doctors would hesitate to treat with PCI. When I showed the images to Mandrola, he said, “No one—I mean no one—in the United States doesn’t fix those sorts of blockages in patients who are still having chest pain despite taking three medications.”

But the doctors in the study didn’t; they sent the patients home with ghost stents. And lo and behold, there was no statistical difference in the amount of time people lasted on a treadmill, or in quality-of-life scores.

“This study will begin to change the mind-set of cardiologists and patients that focal blockages need to ‘be fixed.’” said Mandrola. “Instead, these findings help doctors and patients understand that coronary artery disease is a diffuse systemic disease. A focal blockage is just one manifestation of a larger disease.”

* * *

Many cardiologists have come to see coronary artery disease in that way over the years, as the benefits of angioplasty have fallen away one by one. Still, even among those that agree with this general model, not everyone was convinced by this study.

“The study has a lot of warts,” said the cardiologist Eric Topol, the director of the Scripps Translational Science Institute. He was Twitter-critical of the study in the minutes after the The New York Times sent out a push alert last week with the headline: “‘Unbelievable’: Heart Stents Fail to Ease Chest Pain.”

“I wouldn’t have started with ‘Unbelievable!’” Topol said. “I laughed when I saw that. Yeah, it is unbelievable—that such a small study would lead people to talk about changing guidelines.” (An editorial published in The Lancet by Redberg and David Brown of the Washington University School of Medicine said, “All cardiology guidelines should be revised.”)

“I’m not trying to defend PCI in stable patients. It’s overused,” Topol continued. “I don’t do stenting, but I did for many years, and I saw many patients have marked relief of their angina. So this study is inconsistent with clinical experience, but my biggest concern is the flawed design. This trial is very tiny—200 people. That means beta error.”

Beta error is an increasingly common refrain in scientific criticism, noting that a study did not have enough subjects, and so poses a risk of being falsely negative. If you only flip a coin twice, you have a one-in-four chance of concluding that coins never land heads-up. Flip it 10 million times and your finger gets very tired and also the chance of beta error is effectively zero.

“This is such an important issue that if you’re going to study it right, my contention is you need to do it right, with a larger sample. Maybe if it had 600 or 1,000 people,” Topol said.“What the study suggests is really a poor comprehension of clinical-trial methodology. You can engineer a negative result of anything if you design the trial that way. Beta error is the most common error in clinical research, and the title of this paper could’ve been ‘A Trial to Highlight the Beta Error Problem.’”

Mandrola, too, had caveats. “This one small study should inspire confirmatory studies,” he said. “Confirmation is key.” He also noted that the study enrolled patients who had low levels of angina and were low-risk in other ways, and this means we can’t generalize the findings to higher-risk groups.

Redberg had no caveats. I asked her about the sample size, and she said, “I have no qualms about considering this definitive.”

* * *

So what stops medicine from changing overnight?

This raises the fundamental question of what the burden of proof needs to be to change practice. Guidelines are slow to change, and even once guidelines are issued, doctors are slow to change what they do. What does change things is when payers stop reimbursing hospitals for procedures that aren’t supported by evidence. What does that process look like going forward?

“That’s a sensitive question,” said Redberg. “When the American College of Cardiology came out with their latest guidelines on PCI for patients with no symptoms, they changed the wording from ‘inappropriate’ to ‘rarely appropriate,’ because their concern was that payers would stop paying for something that was deemed inappropriate. I would say it is not appropriate to do PCI for stable angina. There is no benefit, and there are risks, so I would think that the American College of Cardiology would change the guidelines now, based on this study.”

So it’s up to the American College of Cardiology?

“Well, payers don’t have to read guidelines,” she said—though often guidelines do turn the tide on behavior. They have in other cases of procedures that turned out not to work and that we stopped doing, like transmyocardial revascularization and renal denervation, and may soon affect the widespread practice of ablation for the abnormal heart rhythm known as atrial fibrillation. It has been widely adopted over decades and costs tens of thousands of dollars, yet has not proven to improve health or prolong life.

“I think going forward we should demand that there is data for these sorts of procedures before they get widely adopted,” said Redberg.

“People can nitpick the study as much as they want, but burden of proof should be on the procedure to show an advantage,” said Haider Warraich, a fellow in cardiovascular disease at Duke University. This is a growing sentiment in medicine—that to minimize the exorbitant costs of health care, we must move to a system where it doesn’t take 40 years of doing a procedure before we figure out it’s not worthwhile.

The challenge only stands to grow as new, expensive technologies play ever-larger roles in medicine. Does fee-for-service make it impossible to change the system for adopting new procedures? Will they always be assumed worthwhile until proven worthless?

“Look, fee-for-service means you get paid just as much to do something totally unnecessary, even harmful, as you do for a lifesaving procedure,” said Redberg. “And patients are not in a position to distinguish between those.”

She points to the role of the FDA, then, in approving medical devices for doing these procedures. Under the current administration, Redberg said, the FDA “wants to approve things faster and faster, requiring less data, and so allowing riskier procedures—instead of waiting for the studies to know that something is safe and effective.”

Here the epidemic of heart disease comes back to politics. When people argue that the “free market” will eradicate bad medical practices—from worthless drugs to ineffective devices—they may be right. But that process of eradicating bad health care from the market once it has been introduced can take decades, as the case of cardiac stents seems to be showing. And the harm and waste incurred in the process are never undone.