Unbiased assessment of treatment effects on disease recurrence and survival in clinical trials.

Recently there has been considerable discussion concerning the analysis of data and presentation of results from clinical trials in breast cancer. A problem stems from there being several events of interest, notably disease recurrence and death. Current methods based on 'so-called' disease-free survival are criticized on the grounds that they confuse events, such as disease recurrence and death, in the treatment comparison and consequently could be very misleading. Alternative methods of analysis, based on separate measures of treatment effect are presented and illustrated.