Our study illustrated that a surgical placebo for arthroscopic lavage could be designed that was acceptable to a sizeable proportion of people across the range of stakeholder groups (although a few held strong views against the use of a surgical placebo under any circumstances), but conducting such a trial in practice would face considerable feasibility issues. Our study also illustrated well the opposing and often strongly held opinions that are held about placebos in surgery, the ethical issues that underpins this controversy, and the challenges of mounting such a trial that exist even when ethics committee approval has been granted.
Issues of ethical and scientific acceptability of a placebo design in surgery
There was widespread acceptance in our study that further investigation was required into the effectiveness of arthroscopic lavage. Whether that investigation should or should not include a placebo-controlled design generated much wider discussion.
Commentators agree that the ethical principles appropriate to all clinical research must be satisfied as a minimum when considering a placebo-controlled design. These principles are that the study must: a) have scientific merit, b) be acceptable to participants in terms of the risk-to-benefit ratio of participation and c) respect the autonomy of participants by enabling them to determine whether they should participate . In the qualitative components of our study participants raised and discussed these factors for the case of arthroscopic lavage. The scientific merit of the proposed study and the need for informed consent were readily accepted. Whether the proposed study design provided an acceptable balance of risks and potential benefits for participants generated much wider debate.
The discussion of an acceptable risk-to-benefit ratio in a placebo-controlled study is not straightforward. As Horng and Miller  argue, the risks must be considered in the context of alternative study designs to answer the research question - could an evaluation of arthroscopic lavage be conducted without the use of a placebo control and without compromising scientific rigour? The subjective measurement of the primary outcome in our study (patient reported pain) was recognised to be prone to bias in an open trial design (ie, when people would know which intervention they had been randomised to) , and it was considered impossible to maintain an open trial sufficiently long for any placebo effect to have dissipated, and as such the inclusion of a placebo control was deemed to maximise scientific rigour.
An additional consideration in the risk-to-benefit ratio for participants is the nature of the proposed placebo - how "risky" the proposed placebo is perceived to be. A placebo must be able to mimic the intervention under evaluation, but minimise the risks to those who might take part in the trial. Edwards et al suggest that perceptions of acceptability of placebo are likely to vary depending on the nature of the placebo in question . In our study, agreement of the choice of placebo surgical procedure proved easier than the method of anaesthesia. The surgical approach chosen (three small skin incisions) was both a satisfactory mimic and had low intrusiveness and thus required little debate. However, the use of general anaesthesia, while an excellent mimic, was more intrusive and as such generated much greater discussion, and was the factor that caused most discussion with local decision-makers when seeking formal approval to conduct the pilot. It is interesting to note, however, that the content experts (ie, the anaesthetists) contended that the use of a general anaesthetic would be safer (because this is the technique which they used routinely and with which they have the greatest experience) than a supposedly less intrusive alternative, such as a form of analgeso-sedative regimen, with which they were less familiar. The wider literature supports this, suggesting that the success of a procedure is directly related to the number of procedures undertaken by that individual .
Further evidence of the controversial nature of the placebo design was the need to go to an appeal before the pilot trial received ethics committee approval, despite evidence of support from a range of surgeons, anaesthetists and potential participants. In response to the ethical debate raised by the Moseley trial , the American Medical Association produced a set of principles under which a placebo-controlled trial in surgery would be considered ethical . These principles outlined: that surgical "placebo" controls should be used only when no other trial design will yield the requisite data; that particular attention must be paid to the informed consent process when enrolling participants in such trials; that the use of surgical "placebo" controls may be justified when an existing, accepted surgical procedure is being tested for efficacy (but that it was not justified when testing the effectiveness of an innovative surgical technique that represents only a minor modification of an existing, accepted surgical procedure); and that when a new surgical procedure is developed with the prospect of treating a condition for which no known surgical therapy exists, using surgical "placebo" controls may be justified, but must be evaluated in light of whether the current standard of care includes a non-surgical treatment and the benefits, risks and side-effects of that treatment. Our experience suggests that these principles remain relevant; but, for a placebo-controlled trial to be conducted successfully, it is clear that it must not only be an ethically and scientifically acceptable course of action but must also be a feasible course of action.
Issues around feasibility of a placebo-controlled trial in surgery
Randomised controlled trials in surgery are well-known to be difficult to design and often suffer from recruitment problems [30, 31] and adding a placebo component to the design adds to this complexity. Our study also faced a number of practical hurdles before it commenced recruitment and it proved impossible to surmount all of these in one of the two pilot centres within the four-month timeframe of the pilot. We found that: a) stakeholders in each trial centre needed to be fully briefed and any ethical and practical concerns resolved prior to trial commencement; b) that arrangements needed to be put in place to cover the costs of the placebo before the trial could go ahead; and that c) appropriate indemnity arrangements needed to have been instituted. The crucial importance of local stakeholders and gatekeepers has been outlined by a number of authors and the need to develop explicit recruitment and communication strategies identified [32, 33].
Our study confirmed that recruitment to a surgical placebo-controlled trial is achievable - nine of 13 eligible patients approached agreed to join the trial - and the study also demonstrated that blinding of participants receiving surgery was successfully maintained. However, two of the six allocated to surgery subsequently withdrew before surgery citing concerns about the possibility of receiving placebo surgery and risks associated with hospitalisation as reasons. This raises questions about the potential influence of a placebo arm on retention rates in surgical trials.
This range of feasibility issues is likely to be faced by anyone considering a placebo-controlled surgical trial, and a checklist of appropriate issues that trialists should consider is presented in Figure 2.
Strengths and weaknesses of the study
Surgical placebos are controversial and this is one of the few studies to have explored empirically the attitudes and perceptions of stakeholders on this important issue. In addition it sought to reflect the perspectives of a wide range of stakeholders including surgeons, anaesthetists, potential participants and ethics committee chairs. It also ensured UK-wide coverage of opinions through the professional surveys (although response rates were, like those for other surveys among health professionals, quite low), and this provides reassurance that the results are reflective of the current range of opinion on the issue of surgical placebos in general. Similarly the involvement of multiple centres in the research was a strength, as previous studies of placebos have often been conducted in a single centre setting. This was one of the key criticisms of the Moseley trial as it only involved a single surgeon in a single centre.
We recognise, however, that our study concentrated on a relatively minor surgical procedure and that the results may have been different if we had been trying to design a placebo for a more invasive procedure which would have required a larger surgical incision or more complex anaesthetic or a higher risk of major complications. We anticipate that in those circumstances consensus on both the design and acceptability of the placebo would have been harder to achieve and that recruitment to the study may have been lower. In addition, the number of patients recruited to the pilot study was small; limiting the conclusions we can draw from their responses. However, we are confident that the range of issues which require to be considered when planning a placebo-controlled trial in surgery were encountered in this study.