1. Is the proposed treatment safe and therefore unlikely to bring harm to the patient ?

- no treatment is devoid of side effects and this must not be used to withold treatment

2. For the sake of a clinical trial can a Rx be ethically witheld from a patient. In other words can there be a control group on placebo treatment ?

- mastectomy or not in Ca breast

- antihypertensives vs control in raised BP

- analgesics in pain

If compared with 'standard' treatment, what is the standard? eg. morphine in postop pain, beta blockers in angina.

3. Can patients be randomly allocated to either treatment group ?

- are exclusions going to alter the randomisation and therefore, will we end up with different groups where no comparison can be made ?

- rigid definition of patient population before starting so that any patient can receive any treatment. This may involve restricting the age limits, health limits, sex and so on.

Randomisation ensures that local factors do not influence judgment as to effectiveness of treatment.

4. Is it necessary to obtain the patient's consent ?

- some forms of treatment are more invasive than others in producing the same end result (until proven that the invasive treatment is better than the non-invasive) this exposes the patient to a greater risk, eg. epidural opioids vs systemic opioids for postop pain, CAVG for angina vs. medical treatment, SWIFT trial.

- any discomfort that the patient may have, eg. repeated blood samples, lumbar punctures etc. must be explained.

N.B. Patients must now give written consent before entering a clinical trial.

5. It is necessary to get ethical committee approval. The Royal college guidelines in the UK demand this approval. It will ensure that the guidelines presented by the MRC and the Declaration of Helsinki are observed, i.e. with reference to proper scientific observations, qualified practitioners conducting the trial, and so on.

6. There must be no bias on the part of the observer for or against one form of treatment. This usually necessitates the trial being double-blind, i.e. where neither the patient nor the doctor know which treatment is being undertaken. Obviously, there must be a randomisation code available should an untoward reaction occur, or, of course, for subsequent analysis of results. Some treatments, such as surgery cannot be carried out in this way. Assessment of results by an impartial observer is necessary, who, hopefully does not know what treatment the patient has received.