Running a Cox PH Survival analysis - prognostic vs predictive

Question

[psyc_biostars]

I am running a Cox PH Survival analysis to determine if a biomarker is prognostic, predictive, both, or neither.

For a biomarker to be classed as prognostic, then the biomarker term must be significant. For a biomarker to be classed as predictive, the treatment:biomarker interaction term must be significant.

At the moment I'm running two models:

Prognostic model: survival (1 or 0, with 1 as death) ~ biomarker

Predictive model: survival (1 or 0, with 1 as death) ~ biomarker + treatment + biomarker:treatment

My question is - will the effect and hazard ratio for the biomarker alone (in bold) be the same in both models? If not, is it better to chose one model over another?

Answer 1

[Kevin Blighe]

No, it will be different in both models. Also, I fail to see what is the difference between prognostic and predictive here.

Your second model is the same as ~ biomarker * treatment, by the way.

Without further information about what you're aiming to do, one cannot infer which model is 'better'.

Biostar