SAS code for the estimation and between-group comparison of cumulative incidence functions in competing risks survival analysis

Abstract

Abstract Competing risks extends survival data by also observing a cause of failure. Once a subject fails, it is impossible for him to subsequently fail from any other cause. A well-established tool for summarising competing risks data is the cumulative incidence estimate, which estimates the probability of a subject failing from a specific cause of interest before a given time. Comparisons of the cumulative incidence estimates between groups of subjects can be made using Gray's test. However, there is no commonly available SAS code to perform such analyses, which helps to perpetuate the mistaken use of Kaplan–Meier estimates and log-rank tests in the analysis of competing risks data. This paper presents SAS code to provide cumulative incidence estimates and Gray's tests.