An overview on the complement of Kaplan-Meir estimation and cumulative incidence estimation in the presence of competing risks _ simulation approach.

Valarmathi, S.; Lakshmanan, B.C.; Ponnuraja, C.

International Scientific Research Journal; 2015; 1; 61-65.     

Abstract: Researchers are concerned with the methodological problems arising in the analysis of clinical trials when competing risks are present. A competing risk is defined as an event whose occurrence precludes or changes the probability of occurrence of a main event under examination. In this setting, the appropriate estimate of the probability of failure is described by the cumulative incidence. This function is not available in all statistical software packages except very few, the complement of Kaplan–Meier estimates are often used unsuitably instead of cumulative incidence function. When competing risks are present, the appropriate estimate of the failure probabilities is the cumulative incidence rather than the complement of Kaplan–Meier estimate. This paper compares these two methods of estimating cumulative probability of cause-specific events in the present of other competing events. The simulated data with three competing events is used to demonstrate the different estimates given by one minus Kaplan-Meier (1-KM) and cumulative incidence function. Also this paper evaluates the advantages and suitability of statistical methods using the cumulative incidence estimate over the complement of Kaplan Meier estimates (1-KM) method in clinical trial time to event competing data.