1 Biomathematics, M. D. Anderson Hospital and Tumor Institute, Houston, Texas 77025
The continuing discussion on the current prognosis for children with acute leukemia1-6 has raised at least two general points concerning the analysis and reporting of survival data. Because these points are of fundamental importance to the discussion, they require further clarification.
The first point concerns the common practice of overtly or covertly excluding a particular type of patient from the analysis.4 Patients excluded typically have unfavorable prognoses so that results extrapolated to the general population of patients often seem overly optimistic.