by the physician. Pain is, of course, a symptom and not a sign; it cannot be measured by any physical test. The only way, in fact, that we can tell how much pain someone feels is by asking them. All the studies that document placebo analgesia are constrained by this important limitation. They may use various different techniques to gauge the level of pain, from simple yes – no questions to numerical charts on which patients are asked to indicate their current degree of suffering on a scale of zero to ten, but they all have to take the patient’s word for it. Is the placebo response, then, just a private affair, dwelling entirely inside the patient’s mind?
Some early studies appeared to suggest that this was in fact the case. Two of the most famous, both conducted in the late 1950s, were designed to investigate the effectiveness of an operation known as internal mammary ligation, which was widely practised at the time as a treatment for angina pectoris. Angina is characterised by a vice-like pain in the centre of the chest which tends to be brought on by exercise and goes away with rest. It is generally attributed to the clogging up of the coronary arteries, which reduces the amount of blood these arteries can then supply to the muscle in the heart wall. Internal mammary ligation involved, paradoxically, blocking some of the furred-up arteries completely – the rationale being that the blood would be forced to find alternative routes by sprouting new channels through the heart muscle. These new channels would be free of blockages, and so the circulation in the heart would improve – or so the surgeons hoped. Thousands of operations were carried out without any proof that the new channels did actually sprout, although thousands of patients reported that they felt much better after receiving the operation. Only when pathologists failed to detect any of the supposed new blood vessels in those who had received the operation did some doctors begin to wonder if it had the effect they thought it did.
Two groups of doctors decided to investigate by carrying out controlled trials comparing internal mammary ligation with a placebo operation.10 This ‘sham surgery’ involved cutting into the chest and exposing the arteries, but not ligating them. Much to their surprise, the doctors found that patients receiving the dummy operation showed about the same level of improvement as those receiving the real one. The reaction of the medical community was depressingly predictable: instead of being struck by the power of the placebo response, doctors quickly dropped the operation of internal mammary ligation. Nobody paused to wonder how the mere belief that one had received a proper operation could be so effective at reducing chest pain.
Of course, without a no-treatment control group, we cannot be sure that the improvement that occurred after these operations was the result of the placebo response, or whether it was simply the natural course of the disease. However, the success rates were high enough to suggest that the placebo response was playing an important role. Around three quarters of all patients reported significantly lower levels of pain, showed a great increase in their exercise tolerance, and decreased their consumption of vasodilating drugs. This is almost certainly a lot more improvement than doctors would expect in the absence of any treatment.
It is important to note, however, that the placebo response – if such it was – did not, in this case, reverse the underlying pathology. No new arteries sprouted in either the experimental group or the placebo group. True, there was an improvement in some ‘objective’ measures, such as walking distance and drug consumption, but both of these effects could have been due simply to the reduction in pain that followed both the real and the sham operations. A person who feels less pain when exercising can do more before he feels like stopping, and will consume fewer pills; there is nothing mysterious about that. These early studies, then, lend support to the view that placebos can affect symptoms but are powerless to cure disease – they may make you feel better, but they don’t make you get better.
More recent studies of treatments for angina reveal a similar pattern. The main surgical operation for angina today is the coronary-artery bypass graft, in which small sections of vein are removed from the leg and grafted onto the coronary arteries to allow the blood to bypass the blocked areas. This operation does improve survival in the rare cases when the blockage is very serious, but in the majority of less serious cases it has no effect on life expectancy. In fact, when dye is injected into the blood vessels of these patients with less severe cases of angina, it is often found that the new grafts soon become blocked themselves. Surprisingly, though, many of these patients still report a significant reduction in pain.11 Is the placebo response, then, an entirely subjective phenomenon, as some have claimed?
Beecher and other pioneers of placebo research argued otherwise, but their data was not conclusive. They claimed that patients given placebos showed objective changes, such as constricted pupils, but they could not prove that these changes were directly attributable to the placebo response because their studies did not include no-treatment control groups. The sceptic could always argue that such objective changes might have occurred anyway. One early study went some way to meeting this objection by using a patient as his own control.12 A man with a hole in his stomach was tested for the level of gastric acid after being treated with a placebo, and the results were compared with his reactions at other times when no treatment was administered. The gastric acid level fell twice as often when a placebo was used as when no agent was administered. The sceptic could still object, however, since the sample size was so small: only one patient, and only twenty-six observations.
Fortunately, more recent studies have produced much stronger evidence. The studies that looked at the effect of ultrasound on postoperative dental pain, for example, found that objective measures were also affected by the placebo response.13 Not only did those who received the fake ultrasound (while the machine was switched off) experience a reduction in pain, but in one of the studies trismus was also significantly reduced compared to the no-treatment control group. Trismus is an involuntary contraction of the jaw muscles which keeps the jaw tightly closed – and can be measured objectively. Furthermore, in both of the studies there was also a significant decrease in swelling in those receiving the fake ultrasound. Inflammation is even less of a ‘mental’ process than trismus. The placebo response is clearly not just a subjective affair.
ULCERS
The fact that trismus and inflammation can be affected by dummy treatments shows that the placebo response is not limited to subjective symptoms. Nevertheless, if these were the only objective signs that placebos could affect, it would be a hollow victory for the placebo response, falling far short of dramatic claims such as that cited at the beginning of this chapter. Inflammation and involuntary muscular spasms may well be objective signs of pathology, but they are not particularly dangerous problems. Is there any evidence that placebos can affect more serious conditions?
There is. Daniel Moerman, a medical anthropologist at the University of Michigan, has provided persuasive evidence that placebos can cure ulcers. Moerman began by gathering the results of seventy-one controlled trials of drugs for treating stomach ulcers. As usual, these studies did not include no-treatment groups, and so no individual trial provided direct evidence of a placebo response. By examining the studies together, however, Moerman was able to detect a pattern that did suggest that the placebos were having a powerful effect. Ingeniously, he compared those studies in which patients took two placebos a day to those in which patients took four placebos a day. In the first group, 33 per cent were healed, while in the second, 38 per cent were healed.14 This might not seem like a big difference, but it is statistically significant, and it has been replicated in another, more rigorous meta-analysis.15
Comparing two groups of patients taking different amounts of placebos is almost as good, scientifically speaking, as comparing a placebo group with a no-treatment group. The fact that those taking more placebos showed more improvement
