ABSTRACT
Competing risks data are inherent to cancer clinical research in which failure can be classified by its types, and the information on each type of failure is as important as the overall survival probability. For instance, cause of death for cancer patients can be either death due to treatment-related toxicity (treatment-related mortality) or disease recurrence. Disease recurrence is an important event of interest as is treatment-related mortality for evaluation of efficacy and toxicity of a treatment. If disease recurrence is the event of interest and if an individual dies from treatment-related toxicity, this competing event removes the individual from being at risk for disease recurrence. Therefore, applying methods of standard survival analysis to an event of interest when a competing risk is present would lead to biased results since standard survival analysis assumes independence of events and does not take types of failure into account. In this article, we review fundamentals of cumulative incidence function, Gray test, Fine and Gray model, and Klein and Andersen model and illustrate competing risks data analysis using clinical datasets of hematopoietic stem cell transplantation. In addition, we present limitations of Fine and Gray model and Klein and Anderson model, model selection in competing risks regression analysis, and computing tools.
