Riposte to Guest Commentaries on 'Problems associated with randomized controlled clinical trials in breast cancer.

Abstract

This paper addresses the objections of Professor M. Baum and Mr W. J. Cunliffe to my thesis that the randomized controlled clinical trial is a poor tool for the investigation of the treatment of breast cancer, argued in a discussion paper entitled 'Problems associated with randomized controlled clinical trials in breast cancer' (A.E. Johnson, 1998, Journal of Evaluation in Clinical Practice 4, 119-126). The objections range from those that have a philosophical basis, through those founded on differing concepts of the classification of primary tumours and the nature of the metastatic tumour, to those that question the reliability and usefulness of the clinical evaluation of response to treatment in terms of histological grade and rate of tumour shrinkage. An alternative approach to research through primary systemic therapy with selection of treatment according to predicted tumour behaviour was severely criticized, both on the preceding grounds and because it was assumed that the alternative to randomization is management by anecdote. These objections are examined and evidence in support of reliable and useful clinical measurement of response is presented in some detail. The problems associated with randomization as a technique for the evaluation of treatments, when the intrinsic variability of tumours is very large without the intervention of treatment, remain unsolved.