Survival on integrated therapies--what assumptions shall we make?

The problems inherent in using actuarial survival calculations in order to compare different methods for treating end-stage renal failure (ESRF) are emphasized. Assumptions must be made concerning the date for commencement of follow-up observation, the method for handling patients who change therapy, and the impact of antecedent therapies on the one under consideration. It is stressed that patients involved in different treatment methods do not comprise comparable groups since they are not randomly assigned but are selected by the availability and applicability of treatment modes as well as physician and patient preferences. We have proposed that interval mortality and treatment change rates are descriptions of clinical events that have happened to any defined group of patients and may be easily understood by the physician who must advise and make decisions.