considered ethical.
An integral part of the consent process is providing the individual concerned with full information pertaining to the trial including the potential benefits (if any) and risks. This information may be provided in a number of ways which will depend on the context, but will usually comprise written information as well as a verbal explanation of what is involved. The language of both the written and verbal components has to be chosen carefully and should be phrased in non‐technical terms whenever possible. Clearly, the consent process must provide full details of the intervention options and should explain what the objectives of the trial are and that the interventions will be allocated using a chance mechanism. If participating in the trial involves extra medical examinations or requires more (possibly invasive) samples to be taken from the patient than would be routine for the condition concerned in standard practice, then this needs to be made clear.
Also, it is a requisite of the process that the individual should not be pressurised into giving consent so that, for example, adequate time should be allowed by the investigating team for individuals to decide and, often in the case of patients, to discuss the situation with relatives or friends. It should also be explained that even if consent is given, the subject is free to withdraw that consent at any time without compromising in any way the quality of their subsequent care.
Regulatory requirements usually insist that the informed consent process is fully documented, and the consenting individual signs a consent form which is then appropriately witnessed. This part of the process will also need to be adapted to the particular circumstances. For example, in a trial concerning paediatric patients the consent process will be directed at the parent or legal guardian of the child who then consents on the child’s behalf. Other situations where the consent process may have to be carefully considered may be in accident and emergency situations, patients who are mentally compromised, have psychiatric conditions or who are frail and elderly.
It is important for the investigating team to be, and remain, fully conversant with local, national and even international regulations pertaining to the consent process. Although for international studies, it may be sufficient to write in the protocol, quite simply, that patients must give fully informed written and signed consent, and participating clinicians must also conform to local requirements.
2.5 Choice of interventions
The particular interventions to be compared in a clinical trial will be determined by the choice of the specific research question concerned. We shall see in Chapter 9 that the size of a clinical trial, that is the number of subjects that need to be recruited, depends critically on the effect size. In loose terms, the effect size is the anticipated difference between (say two) interventions. If the effect size is small – trials have to be large, whereas if it is large – trials may be small. Now, setting all other aspects aside, small trials are preferable to large. They require fewer patients and therefore can be completed in a timely manner. Following on from this, large effects are preferable to small! Of course, before the trial is conducted we do not know the actual size of the effect (indeed that is what we are trying to determine from the trial itself once completed) but we can choose to make comparisons between interventions which at the onset are as different in their anticipated effect as reasonably possible. Thus, the way in which the comparator (often termed control and test) interventions are chosen is critical.
2.5.1 Standard or control treatment
In the development of a new or alternative approach to therapy, it is important to compare this with the current standard for the disease or condition in question if at all possible. When no effective standard or alternative therapy exists, the new therapy may be compared against ‘no‐treatment’ or a placebo control.
Example 2.3 Placebo controlled trial – advanced hepatocellular carcinoma
In a randomised controlled trial conducted by Chow, Tai, Tan, et al. (2002) in advanced liver cancer, patients were randomised to receive either placebo (P) or tamoxifen (T). Tamoxifen, although not anticipated to influence survival outcome, was often used to palliate symptoms in such patients although no randomised trial had been conducted to measure its effectiveness, and this provided the rationale for the use of a placebo.
However, placebo controls and no‐treatment controls are not possible in many circumstances. In which case, the control group may comprise those receiving current best practice. For example, in patients receiving surgery for the primary treatment of head and neck cancer followed by best supportive care, a randomised controlled trial may be assessing the value of adding postoperative chemotherapy to this standard (best supportive care only) approach. In this case, the control group are those who receive the current standard of no adjuvant treatment, whilst the test group receive chemotherapy in addition. In some situations, the no intervention control may be a watch‐and‐wait policy, whilst the comparator group receives an immediate intervention.
Example 2.4 Schizophrenia and schizophreniform disorder
In the trial described by Kahn, Fleischhacker, Boter, et al. (2008), four second‐generation antipsychotic drugs were tested against haloperidol: a first‐generation standard in first‐episode schizophrenia. Thus, although second‐generation drugs had been in use for a decade, their clinical effectiveness compared with those of the first‐generation was still debated.
2.5.2 Test treatment
If there are only minor differences in the actual type of interventions to be compared in a randomised trial, then outcome differences are likely to be small and only a very large‐scale trial would detect any differences in efficacy even if they were truly present. What is more, even if such small differences are demonstrated by a clinical trial, they may have little clinical or research consequence. Thus, it is best (within the realms of practicability and safety considerations) if the treatment options are as different as possible so that clinically important differences may therefore be potentially established. For example, if a randomised trial is planned to test against placebo (dose d = 0) a drug (at dose d), then d should be taken as high as possible. If in such a trial no effect is demonstrated, then one may be reasonably confident that the drug is not efficacious. On the other hand, if a lower dose d/2 (say) had been chosen to compare with placebo, then a ‘no difference’ outcome may be a result of the dose chosen being too low rather than the drug being truly inactive. Thus, another key question for the design team in this context is to decide how high the dose should be whilst balancing potential efficacy against toxicity and safety issues. One may imagine the different scenarios if one were choosing the dose to include for a new analgesic if the patients are young (usually healthy) individuals as compared to establishing the value of a drug suitable for those who are terminally ill.
An extreme example of a major difference in the interventions studied, which one may imagine would have been difficult to justify, is that of the hip fracture prevention trial of Example 1.11 where the authors state the following:
In homes allocated to usual care (control group), the nominated study coordinator received brief information (10 minutes) about and demonstration of the hip protector, and two hip protectors were provided for demonstration purposes.
The intervention (intervention group) consisted of structured education of staff, who then taught
