Surgery is not what it used to be. Prior to the discovery of general anaesthesia (or antibiotics), surgical procedures in the early 19th century were mostly performed as a last resort – and reported by those who experienced them to be a ‘terror that surpasses all description’, with conscious patients, excruciating pain and considerable mortality.1 Jump to the 1850s, and the work of dentist William Morton culminated in a public demonstration of how general anaesthesia facilitates surgery (the removal of a jaw tumour) without movement or complaint from the patient. Even so, the use of chloroform anaesthetics was associated with one death for every 3000 surgeries performed. Fast forward to the 21st century, and general anaesthesia is now a safe, routine, medical procedure. Research across a broad range of surgical cases suggests that, for fit and healthy patients, the risk of anaesthetic-related death is very low. Currently, anaesthetic-related mortality falls between 0.4–1 death recorded for every 100,000 procedures performed, with most postoperative deaths associated with the primary disease.2 3
There are implications that come with this safer access to surgical intervention. Medical comorbidities that would have previously posed an insurmountable obstacle to surgery are no longer an absolute barrier, where there are appropriate staff and equipment for the level of presenting complexity. Surgery is also no longer confined to lifesaving, last resort procedures, and can be performed for a range of chronic conditions, minor complaints and cosmetic purposes. In 2017, there were 2.2 million admissions to Australian hospitals involving elective surgery, as compared to 340,000 emergency presentations requiring surgery, the most common of which was the humble appendicectomy.4As the range of possible surgical interventions continues to grow, the need for an evidence-based treatment choice is essential.
Placebo surgery, or sham surgery, can be used to demonstrate an effect of surgical interventions in randomised controlled trials (RCTs), isolating the actual intervention from any unintentional, beneficial effects from other aspects of the operating theatre – anaesthetic, skin incisions or perioperative care. Placebo surgical procedures have an unsurprisingly strong placebo effect – invasive placebos are associated with a stronger perceived effect than non-invasive placebos, as is the decisive diagnosis and approach to treatment, which is usually associated with surgical treatment. Over the years, the results of placebo controlled RCTs have shed significant light on contemporary practice. A recent systematic review of placebo surgical trials found that of the 53 studies identified, there was no difference in the outcomes for the surgical and placebo cohorts in 27 cases (51 per cent).5 It was also noted that most of the identified trials investigated surgical treatments for conditions that were not life-threatening, but negatively affected daily function and quality of life (GORD, Parkinson’s disease, obesity, chronic pain).
The use of placebo-controlled RCTs is hardly a novel concept. In 1939, there was reported improvement of angina pectoris in a single patient who underwent an internal mammary artery ligation. It was 20 years later that this treatment, which has no clear physiological basis, was found to be no more effective than a placebo procedure of a single skin incision.6 Since that landmark trial, there have been several other notable placebo trials investigating the benefits of invasive procedures and, at times, disproving popular treatments. In 1981, 30 patients with Meniere’s disease refractory to medical treatment participated on a double-blind controlled study investigating the efficacy of an endolymphatic sac-mastoid shunt. The ‘placebo’ surgery in this study was a regular mastoidectomy, and participants recorded their symptomatology three months prior to and twelve months following surgery. Interestingly, at the time of the study, it was found that while there were minor differences in recorded symptom improvement between the two groups, both intervention and placebo groups improved significantly.7 Following this, a subsequent study examined the longitudinal effects of the intervention three years later and there was no difference in outcomes reported between the intervention and placebo groups.8
While various peri-operative aspects of gynaecological surgery have been subject to placebo-controlled trials, including the use of chewing gum for stimulating postoperative bowel activity, pre-emptive port site local anaesthetic and the use of ginger as a prophylactic antiemetic, placebo-controlled RCTs for gynaecological surgical interventions are limited. In 1994, Sutton et al9 assessed the efficacy of laser laparoscopic surgery for treatment of pelvic pain associated with endometriosis. The recruited 63 women underwent a laparoscopy and were allocated to either the intervention group (laser) or a placebo group, with possible removal of fluid from the Pouch of Douglas for visual diagnostic purposes. Both groups were followed up three and six months after surgery, and after the second review, patients were made aware of their group allocation and laser laparoscopy was offered to the placebo group. This study found that there was statistically significant pain improvement in the intervention group as compared to the placebo group. Interestingly, it was noted that the rate of pain improvement was less than previous retrospective cohort studies of laser laparoscopy had reported. It was also reported that the placebo effect of non-interventional surgery on pain, which was evident in improvement for both groups at three months, was no longer affecting the participants by the extended six month follow up. This initial improvement in pain in the placebo group may be partly attributable to the removal of peritoneal fluid for diagnostic purposes that resulted in an unintended short-term beneficial effect. There were several significant limitations of the study noted, including the single measure for patient symptoms (pain scale), exclusion of grade IV endometriosis cases and the lack of histological diagnosis for any patients in the trial.
In 2004, Abbott et al undertook a placebo-controlled RCT of 39 women with histologically proven endometriosis.10 This study compared immediate laparoscopic excision of endometriosis with an initial diagnostic laparoscopy, where both arms of the study received surgical excision of any remaining pathology six months later. Outcomes were measured as changes from baseline pain scores, quality-of-life assessments and sexual activity questionnaire scores. This study found a reduction in pain scores for all participants at six months, regardless of their randomisation, with a significantly symptomatic improvement across all measured outcomes for the surgical intervention group (80 per cent), as compared to the placebo group (32 per cent). The authors noted that, within a placebo model, it is expected that participants may be more conservative in rating their symptoms when they know they may have had interventional surgery, and this may account, in part, for improved scores for the placebo group. The authors also noted that the placebo response was similar to the findings of Sutton et al,11 but the nonresponse rate in the surgical group (20 per cent) was less than in the Sutton et al study (38 per cent). This may have been due to the disease distribution of the participants, with stage 1 disease the most common diagnosis in the prior study, but a minority of cases in the Abbott et al study. The authors ultimately concluded that surgical intervention can improve quality of life for patients with endometriosis, but also that pelvic pain is not always secondary to endometriosis, when it is present.
Overall, placebo-controlled trials have clearly been shown to benefit surgical practice. Even when providing unexpected results, trials may not necessarily deem an intervention as not therapeutic, but it may identify another unintended mechanism affecting the results. There is a level of trust required in undertaking surgical procedures, and this, along with the invasive nature of surgery, is likely to heighten the placebo effect when evaluating effective procedures. While these trials are invaluable, unlike medical placebo controlled RCTs, all surgery comes with a risk profile, even when minimally invasive, and adverse outcomes of surgical procedures typically last longer (wound infection, adhesion, herniation) than those of medications. In response to growing ethical concerns around placebo-controlled surgical trials, Savulescu et al12 have proposed ‘essential criteria’ which, when fulfilled, would assist to ensure that surgical placebo-controlled trials are ethically sound. These criteria include: current uncertainty regarding the efficacy of a surgical procedure, preliminary evidence that either the procedure will result in significant improvement in outcomes or suspicion for a placebo effect, minimisation of any risks and no deception in implementation of the trial, and finally, that the area of investigation is clinically important with findings that can significantly impact on clinical practice.13 Meeting these expectations, placebo-controlled trials can safely continue to inform surgical practice by supporting effective procedures that can be proved to enhance outcomes, while identifying ineffective interventions and reducing the risk of unnecessary harm to patients.