Modelling survival data with a cured fraction using frailty models

Cure models have historically been utilized to analyse time-to-event data w ith a cured fraction. We consider the use of frailty models as an alternati ve approach to modelling such data. An attractive feature of the models is the allowance for heterogeneity in risk among those individuals experiencin g the event of interest in addition to the incorporation of a cured compone nt. Utilizing maximum likelihood techniques, we fit models to data concerni ng the recurrence of leukaemia among patients receiving autologous transpla ntation treatment. The analysis suggests that the gamma frailty mixture mod el and the compound Poisson improve on the fit of the leukaemia data as com pared to the standard cure model. Copyright (C) 2001 John Wiley & Sons, Ltd .