Arthroscopic Knee Surgery Failed to Beat Placebo in Chronic Meniscus Tears

John Mandrola begins with a common clinical path: chronic knee pain, an MRI or x-ray, an anatomic abnormality in the meniscus, and a recommendation for arthroscopic partial meniscectomy. The meniscus is the knee’s shock absorber; the procedure Mandrola described is done through an arthroscope inserted into the joint, with the surgeon shaving or cleaning up the damaged tissue. He is careful not to call that “non-invasive.” It is not open knee surgery, but “anytime we put things into the body,” he said, it is invasive. “The definition of small surgery is surgery on somebody else.”

The clinical uncertainty is not whether people with knee pain sometimes improve after surgery. They do. The question is why. Mandrola lays out three possibilities that are often confounded in medicine: the condition may improve through natural history; it may improve because the surgery had a specific therapeutic effect; or it may improve through expectation and care — the placebo effect — alongside other nonspecific features of treatment. The study that prompted his essay mattered to him because it tried to separate those causes in a domain where medicine often does not: surgery.

In the meniscus study, 146 patients with chronic tears — not acute traumatic tears — underwent arthroscopy. Every patient had the tube inserted into the knee. The surgeon looked inside the joint and confirmed a meniscal problem. Only then were patients randomized. One group received the actual meniscal procedure: shaving or cleaning up the meniscus. The other group received what is commonly called sham surgery: the outward conduct of the operation without the meniscal intervention. Staff asked for instruments and behaved as though the procedure were being performed, but no meniscal procedure was done.

The design matters because both groups had scars, sedation, and the clinical ritual. Mandrola explained that patients are lightly sedated and given a nerve block; in comparable procedures, patients may wear headphones and listen to music. Follow-up was done by people who did not know which treatment patients had received. In placebo-controlled surgeryA trial design in which the control group receives a credible imitation of an intervention, so expectation and clinical ritual can be separated from the intervention’s specific effect., patients can also be asked whether they can guess their treatment assignment; when blinding works, they cannot.

That design creates an ethical tension. Russ Roberts pressed the point: if a patient “needs surgery,” is it acceptable to enter the joint and then knowingly withhold the therapeutic act? Mandrola’s reply was to put the ethical comparison elsewhere. If 700,000 arthroscopic partial meniscectomies are performed annually in the United States — the number he cited from the paper’s introduction — then the ethical question is not only whether it is permissible to randomize roughly 146 patients to placebo surgery. It is also whether it is permissible to keep performing hundreds of thousands of operations without knowing whether they outperform placebo.

Roberts drew out the asymmetry in ordinary moral intuition. People tend to see the 700,000 operations as well-intentioned attempts to help, while the placebo operation feels like deliberately “gouging” a knee in the service of knowledge. But that intuition is clouded by presumed motivation. The ordinary procedures also involve incentives and harms; the study exists because good intentions do not establish efficacy.

Mandrola also noted that the arthroscope itself is partly diagnostic: the surgeon can look into the joint and might find something else. In cardiology, he said, trialists have used creative designs to handle similar ethical concerns. He described the ORBITA study as involving patients with single-vessel coronary disease and severe coronary lesions causing angina. One group received a stent; the other underwent a pressure-wire procedure without the stent. Patients were told they could have the stent procedure after six weeks, while remaining blinded during the comparison period. Because patients in Britain often wait more than six weeks for the procedure anyway, Mandrola said, the design made it possible to learn from a placebo-controlled intervention without permanently denying treatment.

The meniscus results were straightforward in the dimension that mattered most: the pain-response curves improved similarly in both groups. The actual surgery was no better than the placebo procedure. Mandrola emphasized the limitation as well as the finding. Without a no-surgery group, the study does not quantify the true placebo effect. It shows that a procedure done at very large scale was no better than a procedure that preserved the incision, scope, sedation, and expectation but not the meniscal procedure itself.

Roberts then questioned the term “sham.” If the knife does not add benefit, he suggested, perhaps the repair itself is the sham. Mandrola said a group at Imperial College London has pushed against the word “sham” for this reason, preferring “placebo procedure.” If the surgery does not work, “the whole thing’s a sham.”

There was also a possible downstream harm signal in the 10-year follow-up to the FIDELITY study, which Roberts identified by name. Roberts read from the study results: more patients in the surgery arm progressed to high tibial osteotomy or total knee replacement, 12% versus 4%. He said he assumed that would be significant for that study size, but acknowledged the excerpt he saw did not say so. If it were significant, he suggested, it could mean the procedure itself weakened the knee and made replacement more likely. Mandrola was cautious: “12% versus 4%” were small numbers. But he accepted the broader point. Surgery has immediate risks — infection, anesthesia complications, bleeding — and if the specific procedure does not outperform placebo, those risks matter. The harms may also include possible downstream consequences, not merely the burden and cost of the initial operation.

The meniscus study is not only a verdict on a procedure. For John Mandrola, it is also a lesson in how clinicians should understand the act of caring. Placebo-controlled surgery studies teach which procedures are “placebo resistant,” but they also teach physicians how expectations, words, and relationships affect symptoms. The placebo effect is not simply an embarrassment for medical science; it is a force clinicians can either use carefully or ignore.

Mandrola defines the placebo effect as improvement associated with expectation. A colored pill or a large capsule tends to carry more expectation than a small white tablet. Surgery is an even stronger expectation signal than a pill. The nocebo effectThe counterpart to placebo: symptoms or harm produced or amplified by the expectation that something will be harmful. is the reverse: when the expectation of harm makes a patient feel worse.

His central example was statins. In Mandrola’s account, blinded trials — where patients do not know whether they are taking a statin or a placebo — show “absolutely no difference in side effects.” In observational settings, where patients know they are taking a statin, clinicians see large amounts of muscle pain and ill health. Cyclists tell him their legs are not as good on statins because they know statins can affect muscles. Mandrola characterized that as “mostly all nocebo effect.”

The SAMSON trial, which he described as one of the best trials he knows, made the point more elegantly. Imperial College London researchers took 100 patients who said they were statin-intolerant. Each patient cycled through months in which they took a statin, a visually identical placebo tablet, or no tablet at all. An app recorded how they felt each day. The best months were the months with no tablets. Patients felt “identically bad” on the statin and the statin placebo. In Mandrola’s telling, there was a clear ill effect associated with taking a pill, but it did not matter whether the pill contained the statin chemical or only looked like a statin.

Russ Roberts supplied a vivid nocebo anecdote from something he had read: a construction worker sees a long nail driven through his boot and screams in extreme pain, only for clinicians to remove the shoe and discover that the nail had passed between his toes without piercing the skin. Roberts called it “the ultimate nocebo.” The point was not that pain is imaginary. It was that pain is mediated by the brain, and the brain’s expectations can generate real suffering.

Mandrola connected that to clinical language. He once titled a lecture “Words Can Harm and Words Can Heal.” Clinicians, in his view, should learn from placebo procedures not that patients are gullible, but that confidence, reassurance, and careful framing are part of treatment. A doctor can say: we are sorry you are in pain; we understand; the pain will get better; surgery has not been shown to be better than placebo surgery; we will support you through it.

Roberts tested the logic. If the benefit comes from the ritual of surgery, does the patient need the placebo operation — the incision, the apparent intervention — in order to improve? Mandrola said no. This is where the physician comes in. He told a personal story: he had a labral tear of the hip, could not run and could barely cycle. A surgeon told him that the tear would get better, that it would take time, and that labral surgery would not really help. That advice was enough. But it depended on prior investigators having done the hard studies that made the conservative counsel credible.

Mandrola does not turn that into a blanket anti-surgery rule. Roberts explicitly cautioned that the program was not giving medical advice, and Mandrola agreed. His own labral tear got better on its own; that does not mean every labral tear does. The argument is narrower: for many chronic conditions, especially those with plausible natural recovery, a physician’s job is not simply to intervene but to help patients wait safely.

The role of words also appeared in the examples of “bone on bone” knees and “widow maker” arteries. Mandrola sees many cardiology patients who have been told by surgeons that they have bone-on-bone knees. Some report that their knees do not hurt, yet they are considering surgery because the phrase itself has taken hold. Roberts said the phrase produced an immediate fear response in him: “bone on bone” sounds terrifying. Mandrola answered that cardiology has its own equivalents. “Widow maker” makes a patient want a stent. A positive stress test is described as having “failed” the test. Such language creates an environment in which intervention feels more necessary before the patient has fully understood the trade-offs.

At the dentist, Roberts said, he still reacts to the phrase “I’m going to give you a shot” because of childhood experience with thick, painful needles, even though modern dental injections are often relatively painless. When a dentist says, “This is going to pinch and it’ll hurt,” Roberts said, the second clause does harm by directing the brain to search for pain. Mandrola said empirical studies show exactly that. He tries not to say, as he did when younger, that local anesthetic will feel like a bee sting. Bees “hurt like hell.” Better to reduce the expectation of pain, not plant it.

The risk of a conservative medical philosophy is that patients may overlearn it. Russ Roberts made that problem personal. He had leg pain and knew enough about placebo effects and medical reversalA finding that a medical practice accepted on weaker evidence does not hold up when tested more rigorously. to feel intellectually secure in waiting. But he also admitted an emotional aversion to procedures, shots, and surgery. What looks like hard-headed skepticism may partly be fear dressed up as principle.

John Mandrola recognized himself in that description. He said he is “much more afraid of intervention” than of many conditions. But he drew a distinction between watchful conservatism and denial. Some conditions are reasonable to observe under careful surveillance, with a physician who knows the disease and partners with the patient. Other conditions are hazardous enough that wise counsel should say: this is not one to slow-roll.

For many chronic conditions, however — leg pain, numbness and tingling in a foot or hand, even atrial fibrillation — Mandrola believes a conservative approach is often appropriate. He invoked Voltaire: “The best physician humors the patient until nature heals the patient.” Mandrola calls it the Voltaire approach. It is, in his view, underused.

His personal atrial fibrillation story was intended to show the point without denying exceptions. Mandrola is a heart rhythm specialist, and when he developed atrial fibrillation himself, the most frightening part was the prospect of undergoing a procedure he performs routinely on others. He chose to watch and wait, using medicines for a while. The condition improved. He gave a second example of a cyclist and engineer with atrial fibrillation who tracked episodes on a calendar: red boxes for AFib, green squares when clear. In February the calendar was full of red; by July it was green. If the patient had undergone a procedure in February, Mandrola said, the improvement would have been attributed to procedural success, even though the condition improved on its own.

That is the recurring epistemic trap. Clinicians observe patients getting better after intervention and naturally credit the intervention. Patients are grateful. Physicians feel effective. The disease’s natural history disappears from view.

Roberts asked for the numerator and denominator behind medical reversal. If the reversal examples are vivid and numerous, are they representative? Or do they risk making the skeptic forget that many treatments do work?

Mandrola’s answer was emphatic: many things work in medicine, and it has “never been a better time to be a doctor or a patient.” Congestive heart failure care, he said, has been transformed. Patients who once died within a year or two may now be cared for over 20 years. Cancer treatment has also changed many cancers into chronic diseases. He does not want medical conservatism to become pessimism or nihilism.

The problem is not progress. The problem is the pathway by which interventions become accepted. Mandrola believes there are too many reversals because clinicians are too willing to be “bamboozled” by observational, non-random evidence and by their own hubris. His preference is a culture in which procedures and devices are tested in proper trials before widespread acceptance.

He acknowledged the standard objection: first-generation devices and first-generation surgeries may not show their eventual value. Procedures iterate; technique improves. If every new approach is judged too early, medicine might block useful innovation. Mandrola accepts that this is a real concern, but he still wants more skepticism toward non-random evidence and more willingness to randomize.

Denmark was his model. He said that when walking through a Danish hospital, one may find a third or more of patients enrolled in some randomized trial — not necessarily a dramatic experimental treatment, but comparisons of ordinary care choices, such as different ICU cooling temperatures or different saline solutions. Patients are being treated, but the system is constantly studying what works. Mandrola thinks medicine needs more of that culture, especially because he sees much of modern medicine as being on a plateau where large gains are harder to achieve. When marginal improvements are difficult, it becomes more important to test before accepting.

The meniscus evidence does not instantly change practice. Russ Roberts suspected that orthopedic surgeons did not read the 10-year FIDELITY follow-up and stop doing the operation across the board. John Mandrola agreed the translation from trial to clinic is difficult. Trials are highly selective: one may screen 1,000 patients and enroll 140. That selectivity is necessary for answering a precise question, but real patients arrive in many varieties. A surgeon can reasonably argue that a trial applies to this selected group and not to every meniscal injury or knee problem.

That caution is legitimate, but it sits beside a professional bias toward action. When one’s livelihood and professional identity are built around intervention, it is easy to conclude that this patient is different. Surgeons see patients shortly after the operation, when many are grateful and improving. They may not see them 10 years later after a knee replacement; and if they do, the later deterioration can be attributed to genetics or the underlying disease. A plausible clinical story can keep the intervention in use even after a trial has narrowed the case for it.

Mandrola sees the same forces in cardiology. In the context he described — stents or procedures to fix certain coronary blockages — he said cardiologists have known for 15 to 20 years that doing a stent or fixing the blockage does not reduce heart attacks or make someone live longer. Yet walking away from such procedures is difficult. It is not only financial. It is vocational. The physician’s career is organized around fixing things. Vinay Prasad, whom Mandrola referenced, calls it a “double whammy”: clinicians get paid and feel good for doing the fix.

That is why placebo-controlled trials have moral importance beyond their narrow endpoints. They interrupt the story clinicians tell themselves. They show that patients get better for reasons other than the intervention, and that what the physician does may not work as well as the physician thinks.

Roberts framed the broader human impulse as action bias. When something in the body is identified as dangerous — even if it may be slow-growing, even if it may never affect quality of life — many patients want it removed or fixed. He mentioned prostate cancer and other cancers as examples where the idea of leaving something in the body can be intolerable.

Mandrola answered with his own screening choices. He does not allow his doctor to do a PSA test because he does not want to begin the cascade: the number leads to worry, worry to biopsy, biopsy to further decisions. He likened it to “If You Give a Mouse a Cookie.” Better, for him, not to start. He said the same about colonoscopies: based on a study that he interprets as showing no change in longevity, he personally does not do them and has no regrets. He did not present that as a universal screening rule; he described it as his own decision within his conservative approach.

Roberts added that he tells his doctor not to do PSA testing, then sometimes receives results anyway because it is treated as routine. For him, the cruelty is not only cost. It is the imposition of knowledge he deliberately chose not to acquire, because a bad score would begin a cascade.

Mandrola’s broader point is that tests and labels can start chains of intervention. A phrase such as “bone on bone,” an image, a PSA value, or a “failed” stress test may move a patient toward action before the benefits, risks, and uncertainty have been carefully weighed.

Russ Roberts asked John Mandrola to define “medical conservative,” the position Mandrola, Adam Cifu, Vinay Prasad, and Andrew Foy had argued for in a 2019 essay. Mandrola said his views had not softened; if anything, he has become more conservative as medicine has plateaued and the pressure to do more has increased. He also said the view is not as lonely as it may appear. Through his cardiology podcast, he hears from young clinicians around the world who say privately that the work has influenced them, even if they cannot say as much to their professors. He describes it as a quiet, below-the-surface effect.

The intellectual analogy he chose was Thomas Sowell’s distinction between constrained and unconstrained visions. Medical conservatives, in Mandrola’s description, hold a constrained view of what medicine can do. Unconstrained physicians — “enthusiasts,” as he put it — believe that if they do enough, they can make everything better. The medical conservative appreciates medicine’s progress, but recognizes that many interventions have limited benefit, some should be treated more conservatively, and profit or industry incentives can diverge from patients’ interests even when industry also enables real progress. The core demand is that interventions be shown effective in proper trials before they are widely accepted.

Roberts extended the Sowell analogy through Adam Smith’s “man of system” and Sowell’s line that there are no solutions, only trade-offs. In medicine, the unconstrained view imagines the body as fixable if clinicians keep expanding the frontier. That is sometimes approximately true. The problem is assuming it is true for the particular patient, condition, and intervention in front of the doctor today. Bodies have trade-offs. Interventions bring pain, misdiagnosis, type I and type II errors, side effects, anesthesia risk, downstream procedures, and unintended consequences.

The most difficult case is not the knee operation. It is the moment when a patient is dying and the family asks what can be done next. Roberts described an emergency physician’s experience: families see one intervention work for a while, then the condition worsens, and they ask where the next switch is — the next drug, surgery, or device that can turn the body back on. In 2026, Roberts said, it feels unacceptable that there may be nothing left to do.

Mandrola pushed back on that phrase. There is always something to do, but the goal may change. His wife, Stacy, is a retired hospice and palliative care physician, and that gives him a particular perspective. If a life-prolonging therapy no longer makes sense, care does not end. The goal may shift toward comfort, home, family, and maximizing the time that remains.

That shift is hard precisely because medicine trains clinicians to imagine one more technical move: one more surgery, one more pacemaker lead, one more stent. The conservative act is not abandonment. It is counseling the patient and family that the goal of care has changed.