“My immediate reaction was: ‘yes, of course I’ll do it’, because that’s the sort of person I am,” Brennan tells me. “But I was concerned that I might end up in the ‘do nothing’ category, as I was already at my wits’ end.”
To Brennan’s relief she was assigned for surgery. In January 2014, she was shown to a ward at the Nuffield Orthopaedic Centre, where she changed into a hospital gown and was given a general anaesthetic. Inside the operating theatre, Carr made two tiny incisions in Brennan’s right shoulder: one at the back for the endoscope, a thin tube fitted with a light and camera that relays pictures to a monitor, and another at the side of the shoulder to allow entry for the tool with a rotating burr used to shave bone. The procedure took around 30 minutes.
When Brennan awoke in the recovery room, she discovered she was unable to talk, an upsetting side effect of an anaesthesia that lasts for a day. Her arm was painful, heavily bandaged and in a sling. Her son drove her home. After a few days she was allowed to remove the sling and gently use her arm. When she returned to the hospital for a consultation a month later she was downbeat. “I said to the nurse: I’m still so sore. This is a disaster.”
Then, as if she were slowly recovering from an illness, the pain began to subside and her mobility returned. After six months, she was able to do some light yoga and Pilates again. After a year she had gone from “a nine to a two” on the one-to-10 scale, which is the most common tool used by doctors to assess pain and requires the patient to put a number to their level of discomfort. “To all intents and purposes I was cured,” she says.
Humans have operated on each other for thousands of years. Ancient evidence of trepanation – drilling holes into the skull to excise evil spirits or treat disease – has been discovered in places as distant and diverse as South America, Africa and the Far East. Bloodletting, which involved cutting veins to purge bad “humours” from the body, was practised by the Greeks and Romans, and endured in some parts of Europe as recently as the early 20th century.
Modern surgery evolved on the battlefield during the Middle Ages. Operations were crude and often life-saving. Amputating an injured soldier’s leg allowed him to survive; no clinical trial was needed to show that it worked. This approach set the culture for surgery that has largely endured: young doctors learn from older ones, improvising as they go along. Unlike with new drugs, which are tested against placebos and require approval by regulatory authorities, new operations become common simply because surgeons believe they work.
Andy Carr, who qualified as a doctor in Bristol in 1982, had a moment of clarity early in his career that this system was flawed. He was in his late 20s and working for a senior surgeon.
“He was a wonderful, charming man, with silver hair, horn-rim spectacles, grey suit. He looked impeccable,” Carr told me. “One day I went to see one of his patients whose operation had not gone well. She had told the surgeon it had gone fine, though, and when I asked her why she replied: ‘I did not want to upset him. He is such a nice man.’ I knew then that we needed a more objective way to measure success.”
To help achieve this, Carr and his colleagues developed the “Oxford Scores”, a 12-point questionnaire where the patient evaluates how well an operation has gone, which have been widely adopted in orthopaedic surgery. But Carr was still troubled. Even if doctors and patients believed a procedure had worked, there should be scientific evidence to support it. For some operations, including hip replacements and cataract surgery, the results were clear and measurable: the patient was able to climb out of the bath unaided, or to see clearly. But when the complaint being treated was more subjective, such as pain or stiffness, proof of efficacy was usually absent.
Shoulder decompression surgery was especially of concern to Carr. Typically, he would shave off 5mm of bone during an operation. But when he removed more bone, or less, the results were similar. “I became increasingly worried that I did not understand what was going on,” he says.
Would the patients have got better without surgery? Was it possible that there was a placebo effect? It is well known that the more invasive the placebo – and surgery is very invasive – the stronger the impact.
With other researchers at Oxford, Carr scoured medical journal databases going back decades, looking for studies that tested surgical procedures against placebos. They found only 53 trials that fitted their criteria. In about half of these, including one in the New England Journal of Medicine in 2002 that assessed keyhole surgery for arthritis of the knee, the authors had concluded that surgery and the placebo surgery delivered similar outcomes. “That was a really powerful justification for saying we should not be doing these types of operations,” Carr says.
The main reason so few surgical studies had been conducted was concern over ethics. Giving a patient a placebo pill does not expose them to risk; performing sham surgery does. In discussions about conducting a sub-acromial decompression trial, Carr faced opposition. “Some people said: ‘Andy, you can’t do this. It’s unethical. It’s almost criminal.'” But when he spoke to prominent ethicists at the university, they argued that it was unethical not to test surgical procedures of questionable merit. After all, if an operation offered no benefit, people were being unnecessarily put at risk of surgery, and many millions of pounds that could have been spent on more effective treatments were being wasted.
And so, decades after the first shoulder decompression surgery was performed, Carr began to recruit doctors for his trial: 51 surgeons at 32 hospitals in the UK signed up for the study, which began in September 2012.
Soon after her follow-up visit to the hospital Carol Brennan forgot about the trial. It felt so much like she had experienced proper surgery – the post-operative pain, the loss of voice, the sling – and she had got better, so she assumed she had not been in the placebo group.
Three years later, in 2017, she bumped into Carr at a social occasion in their village. By then, the trial was over, and Carr was allowed to reveal to her the truth. An endoscope had been inserted in her shoulder, allowing him to see inside, but that was the extent of the operation. She had received sham surgery.
“I was flabbergasted,” Brennan says. “I started questioning what had happened in my brain.”
The results of Carr’s study were published in The Lancet in November that year. The groups that received genuine surgery (90 patients) and placebo surgery (94 patients) reported substantial improvement six months and one year after the operation. This suggested that the “treatment effect” of the surgery was not due to the removal of bone and soft tissue.
The third group of 90 patients, who had received no treatment – not even physiotherapy or pain relief – also reported feeling much better, if not by quite the same amount.
This small difference in improvement between the surgical and non-surgical groups, Carr and his co-authors wrote, could be due to a number of factors, including a “surgical placebo effect” of the procedure, and the rest and physiotherapy treatment that prescribed to patients following the operation. The most plausible conclusion, they wrote, was that shoulder decompression surgery “does not provide patients with a clinically important benefit”.
The reaction of shoulder surgeons to Carr’s study was “shock”, he says. Imagine performing an operation for years, believing you were healing people, only to be told you had been wasting your time. There was also denial. In Germany, where 92,000 shoulder decompression operations are performed a year, seven professional surgical associations issued a joint statement criticising the design and conclusions of the British trial, which it says have “no consequences” for the health system there.
Carr’s findings were backed up in June when the British Medical Journal published a 10-year study of shoulder decompression surgery in Finland. Again, the conclusion was that the operation was ineffective.
To Ian Harris, professor of orthopaedic surgery at the University of New South Wales, in Australia, the results of the shoulder trials were no surprise. In 2016, he published a book, Surgery, the Ultimate Placebo, in which he made the case that “for many complaints and conditions, the real benefit from surgery is lower and the risks are higher than you or your surgeon think”.
Why, then, would someone’s condition improve after receiving surgery that had no scientific impact? The most obvious cause is what Harris calls the “natural history” of the complaint: what happens when it is left alone. Many conditions are self-limiting thanks to the body’s tendency to heal itself.
Another reason is what is known as “regression to the mean”: if a variable is extreme on first measurement, it will tend to be closer to the average on the second instance. Take back pain. Most people will experience a sore back, of varying severity, at different points in their lives. If, when testing a new treatment, you choose from the population patients that currently have pain, the chances are good that their pain will have eased in a month or two, regardless of whether the procedure was effective or not.
The patient may also have improved because of other treatments they received at the same time as the surgery. Finally, they may not, in fact, have got better, but they – or their doctor – just believed they did.
Even when faced with solid evidence that a surgical procedure doesn’t work, such as the 2002 knee study in the US, it can take many years before the number of operations begins to drop. But with shoulder decompression surgery, in the UK at least, the decline should be faster, thanks in part to Carr’s study. Last year NHS England announced it was taking steps to reduce 17 routine procedures that were “ineffective or risky”, including shoulder decompressions and knee surgery for arthritis.
Carr believes that other common operations also need proper investigation. Harris says spine fusions for back pain is a good place to start. “Unfortunately doctors have a biased view: they tend to favour their own interventions, he says. “Surgeons don’t ask often enough: what happens if I don’t operate? You have to say: what’s the probability of the procedure making the patient better. If it’s 50 per cent yes and 50 per cent no, then you should not go for it.” Patients should also take responsibility for questioning their doctors about the evidence that surgery is better than not operating, he says.
When I ask Carol Brennan if she could explain why her pain disappeared after placebo surgery, she is at a loss. “I cannot believe that I was ever on the road to improvement or that I would have got better without the surgery,” she says. “The surgery seemed to trigger something.”
Is it then fair, as some have argued, to continue with the operations simply to get the placebo effect?
Andy Carr is adamant this should not be an option. “The risk and costs are too high to do an operation just for placebo.” In the vast majority of cases of shoulder pain, a combination of pain relief, physiotherapy and rest will work. And surgeons can compensate for the placebo effect in other ways. Carr says: “It’s about bedside manner and giving people time and realistic expectations.”