Treatment effects in competing-risks analysis of prostate cancer data

Green and Byar (1980, Bulletin Cancer, Paris, 67, 477-488) have analysed some data from a randomised clinical trial comparing treatment for patients with prostate cancer in stages 3 and 4. These authors assessed the effects of treatment on survival using an exponential regression model with treatment by covariate interactions. Clinical considerations led to conclusions being drawn about the relationship between treatment and different causes of death. This paper presents a more direct approach for assessing treatment effect based on a model which recognizes the competing-risks nature of the problem. Methods for choosing the optimal treatment for particular patients, one of the aims of the Green and Byar work, are also discussed.