Thursday, 19 July 2012

Why placebo surgery is ethical, and necessary

The research discussed in my previous post on placebo surgery for Parkinson’s disease sparked a rush of publications debating the ethics of placebo surgery, even before the results were released. One of these articles (here) gives a good overview of the ethics of placebo (sham) surgery controls, using one of the Parkinson’s trials as its case in point. The authors give both sides of the argument, but I disagree with two important points; points that lead them to conclude that placebo surgery is not necessary.

They state that in order to demonstrate the effectiveness of a surgical procedure, you only need to compare it to any non-operative therapy because the “net therapeutic advantage of an intervention [the benefit beyond that of the non-operative control group] represents a robust measure of its performance”. The “net therapeutic effect” to which they refer is the perceived therapeutic effect, and may be due to the placebo effect of surgery. It is no measure at all of the specific (true) therapeutic effect of the intervention. Pragmatically, they are correct, but in saying that this is acceptable, they are saying that doing surgery for its placebo effect is acceptable, thereby shooting themselves in the foot, logically speaking.

They underestimate the benefits that result from showing a surgical procedure to be ineffective. They display concern about the risks of placebo surgery in trial subjects. Fair enough, but ethically, the risks of the sham surgery need to be outweighed by the potential benefits of the research. They discount the harmful effects of real, but ineffective operations that continue to be performed, and the potential benefits of removing such procedures. On my scales, I would stack the harms from any single sham surgery trial, against the harms that would have resulted from all the extra operations that were avoided due to the results of that trial. Surgery for Parkinson's disease (that they use as their example) is a perfect example; there goes the other foot.

The standards of ethics versus the standards of science
Similar arguments against placebo surgery are echoed in another article on the same topic (here), which highlights the conflict between the standards of ethics (to minimise harm, and have genuine informed consent, free from ‘therapeutic misinterpretation’) and the standards of methodology (which dictate that the most rigorous methods should be used in order to produce the least biased estimate of the true effect). The standards of ethics make it easy to argue against placebo surgery (it adds risk). However, the standards of methodology dictate that you should use placebo surgery where a placebo effect might be expected, because it is the best way to allow for that effect. Obviously, there is also some conflict between the ethics relating to an individual, and those relating to the 'greater good' for society.

Do we always need to use placebo surgery in trials?
I am not suggesting that placebo surgery trials are necessary for all procedures. It depends on many factors but basically, you need placebo surgery when you expect a placebo effect. For example, you don’t need a placebo surgery trial for ‘hard’ objective outcomes like death, because the placebo effect is unlikely to affect one’s risk of dying. A placebo trial was not necessary to show that prostatectomy did not significantly reduce mortality (here). You do, however, need a placebo trial when the outcomes are ‘soft’ (subjective: pain, for example). Tests of surgical procedures for back pain, knee pain, headaches, neuralgia, tennis elbow, etc., can never really adjust for the placebo effect of surgery unless they are compared to placebo surgery.

In summary
Placebo surgery trials are required (and are ethical) when testing some surgical procedures, particularly when a placebo effect is possible. They are necessary because:
1. They are more scientific, not just because of the placebo effect of surgery. They also allow proper blinding in order to reduce the bias of the investigators, particularly those assessing the outcome.
2. The low risks of placebo operations on a relative few trial participants are outweighed by the risks of continuing to perform surgery that is not effective.
3. Practice change (improved patient care resulting from decreasing unnecessary surgery) will be more likely to occur as a result of evidence from placebo controlled trials, compared to any other trial design.


  1. How do you deal with the Declaration of Helsinki (section B, paragraph 24), that states that subjects are to be permitted to change treatment arms or withdraw from the study at any time. It's hard to withdraw once a permanent wound is made - sham or not.

    1. Subjects are always able to withdraw at any time, but they are not permitted to go back in time. Your argument would negate any trial that involves an intervention at any time point, like taking a pill.
      If you are making the point that there is irreversible risk involved with a stab wound, that is addressed in the post with respect to informed consent and balancing risks and benefits, and individual vs society etc.

  2. Delighted to see this post! I’ve been arguing for placebo surgery for years. Will return later to read more carefully and probably ask a couple questions.

    1. Thanks Paul, I am also compiling reports on a separate page ("Placebo surgery") listing all previous studies of sham surgery - will be posting number 3 soon.