Are placebo run ins justified?BMJ 1997; 314 doi: http://dx.doi.org/10.1136/bmj.314.7088.1191 (Published 19 April 1997) Cite this as: BMJ 1997;314:1191
- Stephen Senn, professor of pharmaceutical and health statistics ()a
- Accepted 16 January 1997
Every medical man commits that act of treachery, Mr Blake, in the course of his practice…. Every doctor in large practice finds himself, every now and then, obliged to deceive his patients.—Wilkie Collins, The Moonstone
The placebo, it is agreed, is an excellent aid in determining the specific pharmacological effects of pharmaceutical agents. Any criticisms of placebos have usually been directed at the morality of giving an inert substance to patients hoping for effective treatment.1 2 Indeed, it might be argued that the object of drug development is to make the use of placebos unethical: to find a treatment that is so effective that it will be unacceptable to withhold it in the future. The debate has focused on acute conditions (especially if serious), where the ethical stakes are highest. For chronic conditions (especially if they are less severe) the situation is different. Here the patient may have to live with the condition for many years and, in his or her own interest, be prepared to try a number of treatments–including, from time to time, no treatment at all. With informed consent, the use of placebos in randomised clinical trials may then be uncontroversial.
It is often overlooked, however, that fully informed consent and a placebo can go together only if the clinical trial is randomised.3 Where this happens, the patient's doctor may say: “If you come on this trial you will be given at random either a new treatment or a dummy pill which looks like it. Neither you nor I will know until the end of the trial which you have been given.” The doctor may even offer the patient the opportunity to read the protocol and, in any case, the trial can certainly be conducted as if the patient had read it. This is the standard of open protocol, hidden allocation4 used by R A Fisher in his famous description of the tea tasting experiment: “The subject has been told in advance of what the test will consist, namely that she will be asked to taste eight cups, that these shall be four of each kind, and that they shall be presented to her in a random order.”5 As Freedman has pointed out, concealment, rather than deception, is involved in such experiments.6
Many trials, however, are preceded by a “placebo run in,” in which all patients are given placebo. The practice is common within the pharmaceutical industry and recommended by standard texts as a means of weeding out non-compliers before randomisation,7 8 eliminating placebo responders,8 ensuring that patients are stable,9 washing out previous treatment, or simply to provide a period for baseline measurement. This is incompatible with informed consent, since a doctor is hardly likely to say: “Take this ineffective substance for the next month and record your symptoms daily in this diary.” The allocation may be hidden but the protocol will not be open. The doctor will be two thirds honest at the most: the patient may be given the truth and nothing but the truth, but the whole truth will be deliberately withheld. A clear example is provided by the physician's health study,10 11 12 in which 33 000 doctors themselves were the subjects. All were given placebo to ß-carotene in the run in, and a description of the study admits, “To the participants, it appeared that the trial had begun,”adding later, “a run-in might still be implemented if a plausible scenario were developed to describe to subjects the reasons for switching their medication after they entered trial.”11
Quite apart from the ethical objections, there is also the logical difficulty involved with double guessing. The use of the placebo run in appeals to the argument from the stupidity of others since, if placebo run ins are regarded by trialists as being excellent, trialists must assume that patients do not know about placebos: if patients do, they will suspect what is happening and undo the value of concealment. There is a further logical and ethical difficulty: if trialists are dishonest with their patients, what right have they to expect honesty in return?
The statistical case
What is also not generally appreciated is that the statistical case for a placebo run in is weak. If the object is to screen patients for entry, then not only is there disagreement as to whether this is efficient,13 14 15 but in any case it can be done just as well using an active treatment.3 (Indeed, in the physician's health study, which had a factorial treatment plan, and in which the prophylactic effects of aspirin on cardiovascular mortality and ß-carotene on cancer were studied, aspirin was given in the run in. The reasons given for giving placebo to ß-carotene in the run in are unconvincing.11) If the placebo run in is thought to provide level baseline conditions, then again this can be done just as well using the active treatment, or another treatment, or, indeed, no treatment at all: none of these devices need involve deception.
The purpose of a randomised clinical trial is not to measure the natural course of a treatment; it is to measure the causal effects of a treatment. The relevant question is not “How has the state of the patient changed over time?” but rather “Is the state of the patient different, having been treated, than it would have been otherwise?” Not only is this question the one the trial is designed to answer but it is the only one that has relevance to the doctor's actions: to his or her behaviour as healing agent. If the answer to this question depends on what has been given in the run in, how can we apply the results of clinical trials at all? It is true that if two active treatments seem equal it is difficult to interpret the result. It might be argued that comparison to a placebo run in would help. But if this is to be the placebo's job of work, then to do it properly it should be added as a third randomly allocated treatment.16 17
In the course of their practice, doctors may indeed find themselves “every now and then, obliged to deceive … patients” but, when this is done, it can at least be justified with the argument that it is in the patient's interest. No such justification can be given for the placebo run in, which, furthermore, involves an outright deception. Such deception should be eliminated from randomised clinical trials before it succeeds in bringing into disrepute that other use of placebos, as a control in the randomised section of the trial, which only requires concealment and is compatible with fully informed consent. If not, not only may the trial itself be jeopardised but trialists may eventually find, as in other cases involving consent, that the matter will be tested in the courts.18
Conflict of interest: None.