Conducting analyses

Data analysis and interpretation are considered in detail in Chapters 9 and 10. The responsibility for ensuring that data relating to all the trial outcome measures are fully and correctly analysed, and that appropriate conclusions are drawn from the results, rests with the trial management group and more widely with the collaborators. This is consistent with the responsibility the trial management group bears for the successful conduct of the trial and for the preparation of reports.

Interim analyses for the data monitoring and ethics committee, undertaken while the trial intake is still in progress, contain confidential results on trial outcome measures presented by treatment group. They must therefore be prepared with great care. Interim results may suggest differences between treatment groups in either direction or no differences, and results may swing from one direction to another in the early stages of a trial. If they suggest differences, they might disturb clinicians' equipoise; if they suggest no differences, they might reduce interest in the trial. Either way, they could discourage further collaboration on wholly inadequate grounds, jeopardizing patient accrual and successful completion of the trial. The data monitoring and ethics committee should be totally independent of the trial coordinating team, are not involved in the care of patients in the trial, and can therefore make independent and objective recommendations (see Section 8.10.1). The analyses should be undertaken by the statistician, and the results must not be made available to anyone else involved in the trial, including the principal investigator, those with clinical responsibilities, expert advisers, independent assessors, or collaborating centres.

It is more debatable whether other members of the coordinating team (for example the clinical trials manager) should be aware of confidential interim results. The best principle to follow is that access to the unblinded data is restricted to the minimum number of people. Clinical trials managers might find it difficult in their day-to-day contacts with clinicians if they were aware of interim findings unknown to the clinicians.

In the final analyses, all the trial outcome measures stated in the protocol should be analysed. Careful thought needs to be given to the results, which may suggest additional analyses that need to be done to clarify some details. The interpretation of the analyses and their implications need to be fully discussed by all collaborators (see Chapter 10).

From time to time, for various reasons, trials fail to recruit and have to be terminated prematurely. The results of such trials should nevertheless be analysed and should always be reported in some format. It is almost always possible to learn something from them, even if only the reasons for their failure [3,4].

0 0

Post a comment