presides in counsel, let him beware how he opens his own inclination too much, in that which he propoundeth; for else counsellors will but take the wind of him, and instead of giving free counsel, sing him a song of placebo.
In the eighteenth century, placebo entered the medical lexicon as a term for fake remedies. When the physician thought nothing was wrong with a patient, he might give him a bread pill or some other innocuous substance just to keep him happy. This way, the patient would at least be spared the danger of taking a real treatment when nothing was wrong with him. In 1807, the American President Thomas Jefferson wrote in his diary that one of the most successful physicians he had ever known had assured him that ‘he used more bread pills, drops of coloured water and powders of hickory ash than all other medicines put together’. Jefferson added that he considered this practice ‘a pious fraud’ – a phrase which nicely captures both aspects of the original use of the term placebo. Just like a prayer for the dead sung by monks who never knew the deceased, a bread pill was both deceptive and consoling, a white lie to cover up a nasty fact.
All this changed after World War II. The studies conducted by Beecher, Gold, Lasagna and other elite medical researchers revolutionised the way doctors thought about placebos. By the mid-1950s, the medical profession was beginning to think that handing out placebos might not be such a fraudulent practice after all. Experiments had shown that inactive substances could induce similar effects to those of caffeine and alcohol when people were fooled into thinking that the innocuous liquids they were given contained coffee or wine. Perhaps equally powerful effects could be produced by the bread pills given out by physicians. Perhaps placebos could really heal people.
PLACEBOS BECOME RESPECTABLE
In 1955, Beecher summed up the new view of placebos in an influential article published in the Journal of the American Medical Association.5 Entitled ‘The Powerful Placebo’, the article claimed that placebos could ‘produce gross physical change’, including ‘objective changes at the end organ which may exceed those attributable to potent pharmacological action’. Placebos, in other words, had real effects on real bodies. No longer were sugar pills to be dismissed as a harmless but ineffectual sop given to please hypochondriacs and desperate people. The placebo effect was born.
Beecher’s article has been enormously influential. Fifty years after publication, it is still regularly cited in almost every scientific paper on the placebo effect, and even those that do not cite it directly usually repeat its claims without acknowledgement, or refer to papers that do cite it. It is unlikely, however, that all of those who refer to it have bothered to read it; if they had done, they might not be quite so enthusiastic. For one thing, the range of conditions that Beecher claims can be affected by placebos is not very extensive. Most of the studies he reports concern the effects of placebos on various forms of pain – postoperative pain, pain from angina and headache. The only other medical problems mentioned are cough, common cold, seasickness and anxiety, and for each of these Beecher mentions only one study. More importantly, none of the studies he refers to provides any real evidence at all for the existence of a placebo effect.
The reason why these early studies provide no evidence of a placebo effect is that, with one exception, all failed to include a control group who received no treatment. In any group of people suffering from a particular condition, some will get better without any medical help. To provide convincing evidence of a placebo effect, you would have to show that those receiving the placebo did significantly better than those who received no treatment at all. Yet almost all the studies cited by Beecher merely show that some of those who received a placebo – a third, on average – felt better afterwards. Without a no-treatment control group to compare it with, this figure is meaningless. The improvement shown by those who received the placebo might well have occurred anyway, even if they had received no placebo. The one study that did include a no-treatment group found no difference between it and the placebo group.
The authors of the original studies did not think they had found any evidence of placebo effects. A few of them even said so explicitly. They realised that the improvement shown by those receiving placebos could easily be accounted for by the natural course of the disease (spontaneous remission, random fluctuation of symptoms and so on) and various other factors such as additional treatments. In one of the studies, for example, 35 per cent of the patients with mild colds felt better within two days of taking a placebo (or six days after the start of their cold). The authors of the study pointed out that many patients with a mild cold get better within six days even if they receive no medical treatment at all. Beecher, however, ignored this remark, and attributed all the improvement shown by these patients to the fact that they had taken a placebo.
Besides spontaneous remission, other factors also play a part in the improvement shown by patients receiving placebos. In one study cited by Beecher, for example, patients with a variety of conditions were treated for anxiety and tension. After four months, during which they were given two-week courses of an anti-anxiety drug called mephensin and placebo alternately, between 20 and 30 per cent of them improved. However, another 10–20 per cent of patients deteriorated. The authors of the study subtracted the deterioration rate from the improvement rate and reported, correctly, a net improvement of around 10 per cent. Considering the rather long observation period of sixteen weeks, this seems quite a low figure, especially when one considers that the patients had taken an active drug for eight of the sixteen weeks. Some of them may also have received other medical support outside the context of the study. Yet Beecher failed to take the deterioration rate into account, and claimed that the study showed a placebo effect of 30 per cent.
In fact, Beecher misquoted ten of the fifteen trials he cited – including one which he had co-authored himself.6 His cavalier attitude in reporting these studies is paradoxical, since his underlying objective in marshalling evidence for the placebo effect was to persuade medical researchers to be more rigorous in their approach to evaluating new treatments. Before World War II, the evaluation of new therapies was largely determined by the personal judgement of distinguished doctors. Beecher was a leading figure in the movement to reform this long tradition. Along with Gold and others, he argued vigorously that medical treatments could best be tested by a new method: the randomised, placebo-controlled clinical trial.
THE LONG BIRTH OF THE CLINICAL TRIAL
As to different methods of treatment, it is possible for us to assure ourselves of the superiority of one or another … by enquiring if the greater number of individuals have been cured by one means than another. Here it is necessary to count. And it is, in great part at least, because hitherto this method has not at all, or rarely been employed, that the science of therapeutics is so uncertain.
PIERRE LOUIS, Essay on Clinical Instruction (1834)
In some ways, the clinical trial was not new. Something very similar is described in as venerable a text as the Old Testament. The first chapter of the Book of Daniel describes how Nebuchadnezzar, King of Babylon, offered his own food to some of the most noble Israelites he had taken prisoner after capturing the city of Jerusalem. Daniel refused to eat the foreign food, since it did not conform to the Jewish dietary laws. Nebuchadnezzar’s chief eunuch was sympathetic, but warned Daniel that his own head would be in danger if the King saw him looking thinner in the face than the other Israelites. At this, Daniel turned to the guard and made a request: ‘Please allow your servants a ten-day trial, during which we are given only vegetables to eat and water to drink. You can then compare our looks with those of the boys who eat the King’s food.’ The guard agreed, and after ten days Daniel, and the friends who had accompanied him on his vegetarian diet, looked in better shape than those who had eaten at the royal table. This is not exactly a clinical trial – for one thing, it concerns a dietary regime rather than a medical treatment – but the basic idea of comparing an experimental group with a control group is there.
Historians have unearthed various other ancient progenitors of the modern clinical trial. In the thirteenth century, the King of Sicily, Frederick II (1272–1337), is reported to have studied the effects of exercise on digestion