Imaginary double blind crossover trial

Suppose our imaginary substance, which we wish to test for its effectiveness in the treatment of rheumatoid arthritis originates in the ocean. Perhaps we might call our substance ‘Seapower’.

We must first indicate some of the main properties whit we believe our substance to have. Let us say it is taken in the form of a pill and that it is a slow-acting preparation taking, perhaps, six weeks on average before beneficial results become evident. After this the patient’s condition improves steadily and the final condition lasts for several months at least without any further treatment. We are instructed to undertake a double blind crossover trial to see if the substance ‘Seapower’ can demonstrate its effectiveness in an objective clinical way.

The principle upon which our trial results will be judged involves a comparison between the change in the condition the patients whilst on the ‘active substance’ (‘Seapower’) and that whilst taking placebo. If there is no significant change between these two sessions of treatment, even if the patient on ‘Seapower’ is markedly better at the end of the trial, we shall be forced to consider that this product is ineffective. If there is a significant difference between the active and placebo conditions of the patients, then we can consider that ‘Seapower’ is effective. We shall now carry out the trial and see why this procedure is completely unsatisfactory for our purpose.

It should be emphasized that these criticisms do not apply to a properly designed double blind trial without any crossover.

Setting up the trial

First of all we must collect a number of suitable patients. We will try to pick those with a reasonable variation in degree and duration of suffering with rheumatoid arthritis to give a wide scope for our assessment. Let us say that we are lucky and manage to group together thirty patients (who seem to be of suitable temperament) within a manageable area.

Sufficient ‘Seapower’ pills, and similar-looking placebo pills, are prepared to allow each patient to have, say, eight weeks on each (this period is based on the average period of six weeks which it takes our substance to become effective). By allowing the extra two weeks, those patients who react more slowly will have a chance to benefit before being taken off the treatment. We prepare our active and placebo pills in such a manner that the only way of identifying which ones patients are receiving is by a code system. The code system is locked away.

We now decide on the various factors relating to patient condition which we will measure and set up the paperwork for this system. One more decision we make is to allow all our trial patients to be weaned off any other therapy they may be on over a period of three weeks preceding the starting date îf our trial. The patients are allowed to take pain-killing tablets but not those based on salicylates (aspirin, etc).

*29/48/5*