Exploratory data analysis and the use of the hazard function for interpreting survival data: an investigator's primer.

This report discusses how one can use the hazard function to gain important insights on the patterns of failure in clinical studies when the principal endpoint is a time metric. These new insights may help gain increased understanding into the pathogenesis of a chronic disease and how it is affected by treatment intervention. The qualitative behavior of the hazard function can reveal whether mortality is increasing, decreasing, or is constant over time. Simple graphic plots are all that is necessary to show characteristic failure patterns. These informal procedures are in the spirit of carrying out exploratory analyses on the data. This report discusses the organization of clinical data using a "branch and leaf" plot, outlines the calculation of the hazard function and life table, and uses examples from lung cancer and uveal melanoma to illustrate calculations and ways of interpreting hazard functions.