Are you certain you've never had a placebo?

The most striking indication of the powerful placebo effect is provided by placebo surgery. Here, as in a drug trial, everything apart from the actual surgical intervention must remain the same. A patient undergoing placebo surgery must be anaesthetized, for example, and the same incisions in the skin and, if necessary, bone must be made, so that it would not be possible for two patients, one of whom had been in the clinical group and one in the placebo group, to work this out who had the non-placebo surgery by comparing scars. One famous example is that of Leonard J. Cobb, who in 1959 conducted a placebo controlled study of internal mammary arterial ligation, an operation for angina, in which knots were tied in two arteries to increase blood flow to the heart. The operation had been regarded as a spectacular success, with three quarters of all patients reporting improvements and one third being cured. But in Cobb’s trial, the patients who received incisions and nothing more reported the same improvements as those whose arteries were tied. Due to Cobb’s study the operation was dropped as a treatment for angina.

There is a somewhat obvious ethical dilemma involved in placebo surgery: it seems to violate the Hippocratic oath, which demands that a doctor must not harm his patient, and indeed placebo surgery had fallen out of favour as a research tool. However, more recently, with the advent of keyhole surgery and more advanced anaesthetics that pose a far lower risk to the patient, many (although by no means all) ethicists have come to believe that the risk of harm to the patient from a placebo operation may be justified, as it is smaller than the risk of harm to a patient of a surgical procedure which turns out to be useless, and hospital ethics committees are increasingly making the decision on a case by case basis. A modern example that was permitted was the trial of arthroscopic surgery for arthritis of the knee in 2002. In 180 patients randomly assigned to receive either the surgery itself or simply the incisions in the skin required for the surgery, those who received the surgery gained no greater benefit than those who did not. Baruch Brody was an ethicist consulted by the committee. “My first reaction was what anybody’s would be,” he says. “I said, ‘You’re going to do what? Cut people up, pretending to do surgery when you’re not.” But he eventually came round to the idea because the placebo operation could be done using a simple sedative rather than full anaesthesia. Nevertheless, there are many surgical techniques for which a placebo controlled trial would not be considered ethical.

While placebo surgery is the most obvious example, there are less dramatic but equally important ethical considerations for placebo drug trials. One such sticking point is that it is considered unethical to deny patients treatment where it is available if long term harm could result. For example, to give a patient placebo chemotherapy for cancer could result in the cancer spreading and becoming untreatable. Kirsch and Sapirstein’s aforementioned 75% figure for the placebo effect of antidepressants has been widely disputed because the trials were conducted in patients with mild to moderate depression: it would clearly be unacceptable to give a patient at risk of suicide or self-harm pharmacologically inactive antidepressants.

This is not a problem when two therapies can be used together, since both groups of patients can be given the accepted standard treatment and the placebo controlled trial can be conducted in addition. However, if the new treatment would be dangerous to administer alongside the standard treatment, the double blind, comparative trial can be used instead. Here, the placebo is replaced by the standard medication, and the effectiveness of the two drugs can be compared. Nevertheless, when conducting trials involving patients at risk of serious harm, the results must be continually monitored, so that the trial can be halted prematurely if there is evidence that one group or another has significantly improved clinical outcomes. An example of this is the originals trials of the HIV drug AZT, which were halted when it became clear that patients receiving AZT were living significantly longer and it was considered unethical to deny the drug to the control group.

Surprising as it may seem, there is a small but nevertheless significant body of opinion that suggests the powerful placebo effect may in fact be a myth. The reasoning behind this is that few studies have been conducted comparing the placebo treatment with no treatment at all, since most clinical trials are funded by drug companies trying to prove that their drug is effective beyond placebo. The placebo group is simply the control group and the reasons for improvement in the control group are irrelevant to the study. The placebo group may improve for reasons not attributable to the placebo, such as the body’s natural recovery from a disease, or they may report improvements out of a subconscious desire to please their physician. In 2001, two Danish researchers conducted a meta-analysis of 114 studies comparing placebo with no treatment and concluded that there was no evidence that objective outcomes were improved by placebo. They did note, however, that since the studies were generally small, they could not rule out a slight “powerful placebo effect”. Curiously, two out of the 15 studies included in Beecher’s original 1955 review included a “no treatment” group as well as a placebo group: when these were analysed in isolation from the other 13 trials recently, no improvement was found in the placebo group. The epidemiologists who conducted this analysis concluded: “The placebo topic seems to invite sloppy methodological thinking.”

However effective the placebo effect actually is, its effectiveness seems to be enhanced if the doctor has a relaxed, reassuring bedside manner. A study conducted at Massachusetts General Hospital on patients who were all undergoing the same operation found that those treated sympathetically by the anaesthetist required only half the amount of painkilling medication as those treated in a deliberately brusque, offhand manner and were discharged an average of two days earlier. Since one of the principal criteria used to evaluate a doctor by his (or her) patients is whether or not he has a good “bedside manner,” we may perhaps reach the somewhat surprising conclusion that a “good” doctor is one who successfully induces in his patients the placebo effect!

For more information

The Placebo Prescription
http://www.nytimes.com/library/magazine/home/20000109mag-talbot7.html

Placebo and Placebo Effects in Medicine: Historical Overview
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1297390