0
$\begingroup$

What is the difference in interpretation of the effect estimates in a Cox model stratified by sex vs. with a random statement for sex? Thanks.

$\endgroup$
  • $\begingroup$ Could you please explain what you mean by “a random statement for sex”? Does that mean including sex as a covariate instead of stratification, omitting sex completely from the model, or something else? $\endgroup$ – EdM Apr 9 at 11:34
0
$\begingroup$

In a stratified Cox model, the baseline hazards functions from the different strata are unrelated.

In contrast, using a random effect implies that the baseline hazard functions are proportional to each other.

A detailed comparison of these two modelling strategies are provided in

  • Glidden, D. V. & Vittinghoff, E. (2004). Modelling clustered survival data from multicentre clinical trials. Statistics in Medicine 23, 369–388.
  • Munda, M. & Legrand, C. (2014a). Adjusting for centre heterogeneity in multicentre clinical trials with a time-to-event outcome. Pharmaceutical Statistics 13, 145–152.

That being said, gender is not a random effect.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.