Getting more out of survival data by using the hazard function.

Abstract

Information about the survival experience for a group of patients is almost exclusively conveyed using plots of the survival function. However, the hazard function provides information about the survival experience that is not readily evident from inspection of the survival function. The hazard function tracks the instantaneous failure rate over time among the surviving patients and can be readily estimated with available software. We illustrate how these estimates can be used in conjunction with estimates of the survival function to glean clinically relevant information from survival data.