The situation prevailing before, during, and after surgery is one usually filled with grave anxiety and stress. It would be surprising if, in this charged atmosphere, surgery did not have a powerful placebo action, in addition to what it may or may not accomplish physiologically.
—Henry Beecher, “Surgery as Placebo,” 1961
As a tennis and soccer player, I’ve had my share of, dare I say, successful surgeries. Over the years, I’ve peered at images of my torn ACL, meniscus, and rotator cuff, and listened carefully to surgeons explaining their plan of action. While the post-op pain was excruciating, especially for the rotator cuff, which involved the removal of bone spurs, I trusted my surgeons along with the generous use of ice, drugs, and physical therapy to get me back to the courts and pitch. And they did. So I must admit that I was one of the skeptics who found the results of the Can Shoulder Arthroscopy Work? (CSAW) clinical trial hard to believe.
CSAW, a randomized placebo-controlled clinical trial of subacromial shoulder impingement, compared the arthroscopic removal of bone spurs to sham surgery (arthroscopy only) or no treatment. Conducted in the United Kingdom, CSAW was a large multisite trial at thirty-two hospitals including fifty-one surgeons and three hundred patients with shoulder pain who were eligible for the surgery. This was a large study, so if there was a clinical benefit of the surgery over a placebo, this study had the statistical power to show it. Quite surprisingly, CSAW found that arthroscopic shoulder decompression, a hugely popular surgery, was no better than the placebo. Perhaps I should not have been surprised since this was not the first or even last time a popular surgery failed to demonstrate efficacy beyond that of a sham control.
Four months earlier, a severe heart attack had left him a jobless semi-invalid. His breathing was labored, and his almost constant chest pain became severe on the slightest exertion. Two days after surgery he was out of bed, walking around his room. Pain had magically vanished; his breathing was improved. On the fifth day he went home and shortly thereafter he resumed work.
—J. D. Ratcliff, “New Surgery for Ailing Hearts,” 1957
You might think that surgery and surgical procedures are the one medical intervention immune to the challenge of the placebo effect, but you would be mistaken. This epigraph is from a Reader’s Digest story that boosted the popularity of a surgery called mammary artery ligation. The article chronicled the remarkable recovery of a sixty-one-year-old carpenter who received the surgery in Italy. In the early 1950s, angina pectoris (severe chest pain) was routinely treated by the bilateral ligation of the internal mammary artery, which is the internal thoracic artery that supplies blood to the chest wall and breast. This effective procedure was thought to work by increasing blood flow to the heart by collateral circulation.
While the surgery was widely viewed as successful in humans, in dogs it was a bust.1 This failure to replicate the benefits of the surgery in an animal model inspired a 1959 double-blinded clinical trial to compare the ligation to sham surgery. The trial was carried out with seventeen patients, and outcome measures were obtained three to fifteen months postsurgery. All the patients were operated on, but only eight of them had verum ligation. Amazingly, during the first six months, about half the patients in each group (five out of eight for the ligated, and five out of nine for the nonligated sham) reported improvement in their symptoms, and two patients who were not ligated showed improvement in exercise tolerance.2
A follow-up study in 1960 used a similar procedure, comparing ligation to sham surgery, and found that improvement was similar among the patients who underwent ligation and those who had just the skin incision.3 With these epic failures, the treatment was abandoned. Later, Beecher compared previously published ligation clinical trials.4 He organized the trials based on the biases of the investigators and noted that studies conducted by “enthusiasts” tended to find positive results for mammary artery ligation; in the words of one enthusiastic researcher, “Practically all of our patients have had complete relief of the anginal pain.” In contrast, studies conducted by “sceptics” found that few patients had complete pain relief.5
Despite Beecher decrying as immoral and unethical the continuation of “casual or unplanned new surgical procedures—procedures which may encompass no more than a placebo effect—when these procedures are costly of time and money, and dangerous to health or life,” the use of placebo controls in surgical trials would be slow to catch on.6 Still, placebo-controlled surgical trials continued to trickle in. For example, there was the pacemaker trial for obstructive hypertrophic cardiomyopathy that found no significant difference in perceived symptoms between the verum and sham surgery groups.7
Around the same time as the pacemaker trial, a trial of fetal tissue transplants in patients with advanced Parkinson’s disease caused an ethical maelstrom. Although patients had for decades already been openly receiving the intracerebral transplantations, the results were mixed. The trial, conducted between 1995 and 1999, was designed to demonstrate the superior benefits of the surgery by randomizing patients to receive surgical fetal brain implants or a sham equivalent of the surgical procedure.8 But from the start, ethicists raised concerns about the integrity of the human body and how the subjects would be able provide informed consent without there being therapeutic misconception. Some ethicists made the case that analysis of the individual patient’s pre- and postoperative condition compared to patients who received standard treatments would suffice. Despite the resistance, the trial proceeded, and patients in both the verum and sham groups were given anesthesia, had holes drilled into both sides of their skulls, and were prescribed immunosuppressive drugs. In the sham condition, the dura, a thick membrane surrounding the brain, was not penetrated. Disturbingly, a year after the surgeries, no statistically significant difference in clinical benefit was found between the verum and sham surgery groups.
Sham surgery is, by its very nature, an invasive placebo control involving the use of anesthesia and incisions. In the fetal transplantation studies, ethicists argued that the risks to those who received the sham intervention outweighed any potential scientific or societal benefit. Possible complications from anesthesia include nerve injury, allergic reactions, nausea, vomiting, and malignant hyperthermia. Incisions put the patient at risk for increased pain, bleeding, and/or infection. Given the risks to patients, some ethicists have contended that sham surgery is intrinsically or presumptively unethical, and violates the ethical imperative to “do no harm.”
In addition to the ethical challenges, executing a sham surgery can be costly and engage many clinical confederates to provide a convincing scene in the surgical theater. In sham surgeries, confederates must replicate audiovisual (e.g., dialogue between surgeons and technicians) and olfactory (e.g., the distinct smell of the polymer used in vertebroplasty) cues. In addition, props such as suction devices, catheters, infusion pumps, and even dark goggles to literally blind the patients can be required.
Logistically, surgery is expensive, and it is unusual for funding bodies to be motivated to fund a placebo-controlled clinical trial unless it is to test the safety of a new marketable device. Given the propensity for clinical trials with low enrollment to drag on past their funding period, and be abandoned or unpublished, the possibility of risk exposure to subjects and financial loss with no guaranteed benefit to society is a constant concern.9
A landmark study in 2002 reinvigorated the ethical debate and ushered in the modern era of placebo-controlled surgery trials. J. Bruce Moseley and colleagues randomized 180 patients with knee osteoarthritis to either arthroscopic debridement (a minor surgery to remove damaged bits of bone and cartilage), arthroscopic lavage (a minor surgery using saline to wash out the excess fluid or debris in the joint), or sham surgery. The outcomes were assessed over a twenty-four-month period. Cited by subsequent academic articles over twenty-five hundred times, the New England Journal of Medicine report of this trial would capture the attention of modern surgeons and health care quality experts. Once again, in a sham-controlled trial of an expensive and widely popular surgery, the researchers found that “at no point did either of the intervention groups report less pain or better function than the placebo group.”10 More recently, a meta-analysis comparing arthroscopic surgery to nonsurgical interventions like exercise, physical therapy, or steroid injection suggested that arthroscopic surgery is not more effective than the less invasive alternatives.11
Another widely publicized series of sham-controlled surgical trials centered around renal denervation for the treatment of resistant hypertension. Resistant hypertension is characterized by a failure to respond to three or more antihypertensive drugs. Surprisingly, there are nontrivial placebo effects in patients with resistant and nonresistant hypertension.12 In renal denervation, nerves in the wall of the renal artery are ablated by radiofrequency pulses or an ultrasound, causing a reduction in sympathetic neuronal activity and decrease in blood pressure. The early trials of the Symplicity Catheter System, Symplicity HTN-1, 2009, and HTN-2, 2010, were single-armed trials with promising results. In Symplicity HTN-3, however, a sham-controlled trial performed in 2017, no difference was observed between the blood pressure reduction from verum denervation, −14 mmHg, and the sham control, −11mmHg, six months after the surgeries. Undaunted by these disappointing results, the manufacturers of the Symplicity denervation device continued to perfect the target, number of ablation points, and selection of participants to see if these factors influenced the benefit of the treatment over a placebo. Notwithstanding these technical improvements, the benefit of renal denervation is still only modestly better than a placebo.13
Yet if either surgery or its sham equivalent ameliorate a patient’s symptoms, when neither wins, should we take both off the table?
There is a large body of evidence supporting what was long suspected by Beecher and others: that some surgeries, steeped as they are in ritual, are potent inducers of placebo effects. A 2014 review of placebo-controlled surgery trials found improvement in patients randomized to a placebo in 74 percent of the trials, and no difference between surgery and a placebo in 15 percent of the trials.14 In 49 percent of the trials, the surgery had a greater effect than a placebo, but the magnitude of this difference was small.
If the last sixty-plus years are viewed with a societal lens, one might argue that discrediting popular surgeries like mammary artery ligation and more recently renal denervation as shams and removing them from the surgeon’s arsenal saved thousands of people from undergoing a useless surgery.15 But the question of whether a surgery works or not isn’t a simple open-and-shut case. Yes, many popular surgeries, when tested in sham-controlled trials, proved to be no better than a placebo. Yet if either surgery or its sham equivalent ameliorate a patient’s symptoms, when neither wins, should we take both off the table? Shouldn’t we offer at least one of these approaches to the patient? With 150,000 patients in the United Kingdom and 700,000 in the United States still undergoing knee arthroscopies annually, perhaps at least some of these surgeries, effective or not, are still on the table.16