Research updates
Phase III clinical trials   page 4
Go back a page
Go forward a page
3. Randomised controlled trials Link to the Medical Research Council web site
Randomised Controlled Trials (RCTs): the Gold Standard for assessing new treatments.
Experiments that focus on people are particularly prone to bias. Bias is any factor that may favour some outcomes over others and that may produce misleading evidence. For example, if a physician selects the treatment for each of her patients, she may give a new treatment to those who are sicker and the standard treatment to those who are healthier. The new treatment may then not look as good because people who are less healthy are likely to do less well on any treatment. This type of patient selection can lead to bias. The randomised controlled trial has evolved specifically to avoid such biases. It is now accepted as the ‘gold standard’ for assessing treatments and provides the foundation of evidence-based medicine.

Controlled means that one of the study groups is given the current standard treatment. This provides a comparable reference group. Thus a new, promising treatment should be compared with the standard treatment (Figure 4). For example:

  • A new drug may be compared with the existing drug.
  • New combinations of different (already) approved drugs may be compared with standard combinations.
  • New dosages of (already) approved drugs may be compared with standard dosages.
  • Surgical techniques may be compared.
  • Effectiveness of receiving chemotherapy before and after surgery may be compared.

If there is no effective standard treatment, the control group may receive no treatment or a placebo. You will see later that giving a placebo does not necessarily mean that the patients receive no treatment.

Chart of randomisation
Figure 4. Two trial arms for treatments X and Y. Treatment X would normally be the current standard treatment for the disease.
Methods of randomisation
In a randomised trial, the choice between the treatments under investigation is made by a random process, the equivalent of tossing a coin. In this way groups of patients are formed that should only differ by the treatment assigned; any differences between other characteristics of patients arise only by chance. With a sufficiently large number of patients in each group, this chance variation is reduced to a minimum. Randomisation evens out any systematic differences between the patient groups. A number of randomisation methods are available, some of which are more appropriate for certain types of trials. The best, because it is secure and difficult to tamper with, uses a computer-based system that automatically randomises individuals based on telephone calls or faxes to a central office like the MRC Clinical Trials Unit.
There may be identifiable factors that can be expected to influence the results of treatment. These factors can be taken into account when the patients are allocated to treatment groups. For example, if age and sex were expected to influence outcome, the randomisation process could ensure treatment is allocated at random in such a way that the groups become similar with regards to the ratio of women to men and age distribution.

Without randomisation, the groups of patients may differ in some way, simply because of the systematic way in which the groups were recruited. In the above example, we would not wish one group to comprise older men and the other group younger women: we need to ensure balance. For example, the progress of patients grouped according to the site where they are treated may depend more on the environment in which they are treated than on the type of treatment they receive (e.g. availability of supportive care and specialist consultant time). It may even vary because less fit patients are more likely to be referred to specialist hospitals.

Question 3
In a trial of a new treatment for lung cancer, what factors other than the treatment could affect the outcome for the patient?
(Try to come up with at least 5)

Go back a page Go to the top of the page Go forward a page