Kaplan-Meier and Cox proportional hazards regression in survival analysis: statistical standard and guideline of Life Cycle Committee

Abstract

In medical research, analyzing the time it takes for a phenomenon to occur is sometimes crucial. However, various factors can contribute to the length of survival or observation periods, and removing specific data can lead to bias results. In this paper, we discuss the Kaplan-Meier analysis and Cox proportional hazards regression model, which are the most frequently used methods in survival analysis. For the first step, we shall discuss the temporal concepts needed in survival analysis, such as cohort studies and then the basic statistical functions dealt with in survival analysis. After solidifying the concepts, methods of understanding and practical application of the Kaplan-Meier survival analysis is noted. After that, we will discuss the analysis methods for the Cox proportional hazards regression model, which includes multiple covariates. With the interpretation method of Cox proportional hazards regression result, we then discuss methods for checking the assumptions of the Cox proportional hazards regression, such as log minus log plots. Finally, we briefly explain the concept of time-dependent regression analysis. It is our aim that through this paper, readers can obtain an understanding on survival analysis and learn how to perform it.