A Simple Approach to Assess Group Differences in Estimated Baseline Survivor Functions from Cox Proportional Hazards Models

Daniel A. Powers, University of Texas at Austin

We develop a heuristic approach to generate the expected events that would occur under a Cox proportional hazards model and illustrate how the results can be used to test for group differences in the resulting risk-adjusted survivor distributions. The estimated baseline survivor functions from a Cox proportional hazards models fit separately for two groups are used to construct the expected number of events occurring at each event time, the expected number of interval-censored observations, and the adjusted risk set corresponding to the expected decrements due to events and censoring for each group. Differences in the resulting expected survival distributions can be tested using log-rank and generalized Wilcoxon tests. This method should also prove useful for making other kinds of comparisons of adjusted life tables.