[Design, management and handling of a randomized trial]

Before a newly developed treatment starts being used in practice it must pass through three clinical trial steps, called phase I, II, and III. The subject of this presentation are the phase III clinical trials. Their goal is to compare the treatment under study to the currently standard treatment (either to show the superiority of the new treatment, either to show it is equivalent in terms of efficacy with the standard one). The only way of ensuring a valid comparison is to perform a randomised clinical trial, meaning that for each patient the treatment is assigned randomly by a mechanism unknown by the investigator(s). Each clinical trial must start with the creation of a protocol. The protocol is a document that describes in details all the aspects of performing the trial. Normally, once the first patient is registered into the trial, the protocol shouldn't be modified any more. That's why it must foresee every problem that might show up during the trial, which makes it a difficult task. A crucial question which has to be answered before the trial begins is the 'sample size'--the number of patients needed. This number is based on several parameter. A bad choice of one of these parameters can completely compromise a trial. All data handling and administrative tasks are usually handled by data managers. They have a crucial role in collecting and validating all the data, and the quality of the whole clinical trial is closely related to the quality of their work. To help them in their tasks, a suitable computer system can be of invaluable help. One of the last steps of the trial is the statistical analysis itself. The statistical tests to be used must be predefined in the protocol and depend mainly on the end points used for assessing the efficacy. To avoid all bias the analyses must be done using the 'intent to treat' principle. Many problems can show up during the trial: ineligible patients or lost to follow up, protocol violations, etc. A well written and well respected protocol should help reducing this list to a minimum. Finally, all this work will only be useful if the results are widely presented in a clear and pertinent way. If the medical community is not convinced by the results, all those resources would have been wasted.