What are randomised controlled trials?

It has been claimed that

In clinical research, the efficacy of a treatment A cannot be assessed purely by the observation of a single group of patients treated with A, because the natural course of the disease and other co-factors can be the reason for the improvement of the state of treated patients. [1]

The randomised controlled trial (RCT) is a research tool developed (in theory) to deal with the particular problems of variability in individuals and disease processes. The participating patients are divided into two groups, with the test group being given the substance being tested (the verum), while the control group is given a substance believed to be inactive (the placebo). The process of random allocation of people to each group is intended to equalise the distribution of unknown factors across the two groups, so that their influences on the result cancel out, leaving clear evidence of what changes are actually caused by the substance being tested. In fact there are important questions about the validity of the evidence RCTs can produce. In tests of treatments the unknown factors and how they are theoretically managed by the RCT include:

• Placebo effect: People can benefit from the mere fact of taking something for their illness (the placebo effect), so the drug has to be shown to have a greater effect than a placebo, rather than a greater effect than doing nothing. Statistical analysis is used to arrive at the results. Because participants’ responses are also affected by knowing whether they are receiving the drug or the placebo, the trial may be conducted ‘blind’, where this information is kept from them.

 
• Natural course of the disease, regression toward the mean and spontaneous remission:  Some people may get better during the period of the trial as part of the disease process or without the reason being known. Random allocation of participants to the test and the control groups is intended to equalise this in the two groups. The random allocation may be adjusted so as to allocate to each group roughly equal numbers of participants by sex, age and particular stage of illness. In a “double blind” RCT (DBRCT) the researchers do not know who is allocated to which group, so that allocation will not be affected by a natural bias in experimenters to achieve a particular result.

 
• Variation in responses: Just as different people have different specific patterns of symptoms, so the responses of participants in a trial will vary. The greater this variety of response, the harder it is to distinguish the effect of the drug, so participants may be selected for the similarity of their symptoms (homogeneity), though this limits knowledge of the broader effectiveness (generalisability) of the results.

 
• Other factors: The random allocation of participants is also used to minimise the likelihood of other factors (such as social circumstances) affecting one group more than the other.

Problems
It is important to understand that the RCT compares two sets of evidence rather than testing a theory against experimental evidence, which is the norm for the mature sciences. Furthermore, in practice, clear evidence is still not easy to obtain, and an understanding of the mechanism of action of a drug may not emerge for some time.

The RCT generates a number of questions about its theoretical assumptions when used as a test of curative interventions. For example:

• If there is so much individuality in people’s symptoms, why is a matching individuality not needed in treatment, especially given the evidence that all drugs produce side effects?
• How can it be sufficient to look at the success of a drug in improving specific symptoms, when this may mean that other consequences of treatment are overlooked?
• Since diseases are products of a number of factors, which may be unknown or peculiar to individuals, how is it sufficient to find a treatment for these effects without having a clear understanding of their relationship to the causes?
• How is it possible for medicine to be scientific if it cannot clearly distinguish between the natural progress of a disease and the curative process of a treatment?

The importance of these questions is shown by the lists of drugs that are withdrawn after a period of clinical use. Such consequences call into doubt the validity of the RCT as a mechanism for providing consistent and reliable evidence of the curative action of drugs. In contrast, it should be noted that the information gathered by homeopaths about their remedies (including those first tested in the late 18th century) is as valid now as it was when the tests were first made.

References
1. Edzard Ernst and Eckhardt G. Hahn (eds), Homoeopathy: a critical appraisal (Butterworth Heinemann, 1998), p.3.

Related pages:
Why it works

• Genetics
• Health and illness
• Homeostasis
• Placebos
• Symptoms and homeopathy

Orthodox medicine
• What are diseases?
• What is effectiveness?
• What is evidence?
• What are homogeneity and generalisability - a paradox?
• What factors affect RCTs of homeopathy?
• What is the rebound effect? 
• What are side effects?