A new parametric model for survival data with long-term survivors

We develop a new parametric model using the three-parameter Burr XII distribution for the analysis of survival data with long-term survivors, which includes the previous Weibull mixture model as a special case. The new model is applied to the analysis of a set of leukaemia, data for which previous attempts in the literature using traditional parametric models were unsatisfactory due to lack of fit. It is shown that the new model improves the fit to the leukaemia data significantly and is thus capable of providing more credible answers to a variety of statistical inference problems that are of interest to medical researchers and practitioners. Copyright (C) 2004 John Wiley Sons, Ltd.