Using Cox-regression to identify predictors for cardiovascular mortality: Model assumptions or not?

Question

I have used cause-specific Cox regression to identify predictors of cardiovascular mortality. When doing this kind of analysis, are there any model assumptions to test for? Thanks in advance!

Answer 1

I got these post from a quick search post 1, post 2 and post 3.

Otherwise look at text book like Regression Modeling Strategies, Survival AnalysisA Self-Learning Text, or Dynamic Regression Models for Survival Data.

As far as I remember, you have to check that the linear predictor is correctly specified (dependents between outcome and the co-variate is linear or some appropriate transform), the proportional hazard assumption is not violated, and maybe make some calibration. You can also check for variance inflation and influential measures as in typical glms.