Analysis

The intention-to-treat principle (see Chapter 9) is always applicable in a randomized trial, and, as discussed in later chapters, the exclusion of any patients from an analysis may have the potential to bias the results. A possible exception is exclusion based on some form of review - perhaps histological or radiological - which is carried out blind to the patients' allocated treatment and to their outcome. Pathology review is often carried out and there are inevitably a proportion of cases where discrepancies between the local and review diagnosis are apparent. Again it is useful to consider the extremes of approach. The explanatory trial might exclude those patients found ineligible on blind review, on the basis that their inclusion introduces another source of variability, if the outcome or response to treatment of the 'ineligible' patients is anticipated to be substantially different from that of the truly eligible patients. However, the pragmatic trial would have to recognize that in routine clinical practice, the great majority of patients will be treated on the basis of local diagnosis, and thus it is appropriate to quantify the impact of treatment in the type of patients who would be treated outside of the trial on the basis of local diagnosis - in other words not to exclude patients on the basis of conflict with review diagnoses. It therefore follows that central review need not be an essential component of a large-scale pragmatic trial. One exception to this rule might result from consideration of the educational value of the review. If there is an opportunity to identify and resolve problems for the future by systematically comparing local and review diagnoses, then this may be valuable. However this adds an organizational burden to the trial participants, and the impact and value of this therefore needs to be considered.

0 0

Post a comment