The cummulative incidence estimate versus the Kaplan Meier method in medical research

The Kaplan Meier (KM) method is very popular in medical research, but many researchers are not aware of its deficiencies when handling studies that have multiple failure types and/or competing risks. The Cumulative Incidence (CI) estimate, has been around for some time yet is rarely used, would be an excellent choice to implement when competing risks occur. Unlike the KM method, which only estimates the survival probability based on the non-censoring group, the CI estimates each failure type separately. To illustrate the superior performance of CI estimate in theory and practice, the CI estimate and the KM method will be compared and contrasted in simulations and with clinical solid tumor data. Baseline progression free survival is accurately estimated for use in further analysis at the Children's Oncology Group