Review of Biostatistical Methods and Interpretation of Hyperthermia Clinical Trials

Chapter 21 has reviewed topics connected with the design requirements for hyperthermia clinical trials. The object of this chapter is to review some of the statistical methodology required to assist in the interpretation of such trials. Emphasis is given to estimation of measures of the relative effect of the treatments under comparison in a (randomised) trial together with associated confidence intervals. We describe how these can be calculated for situations where it is relevant to compare two proportions and we describe in some detail how this is adapted when comparisons have to be adjusted for variables which describe prognostic characteristics of the patients. We point out that it is usually more efficient (statistically) to record, for example, the exact time that a complete remission has occurred following commencement of treatment, rather than noting its presence or absence at a particular fixed time point. This approach permits survival time comparisons to be made between patient groups. We also use specific clinical examples to focus the discussion.