Statistical power estimation for a time-dependent covariate in Cox (proportional hazards) regression

This setting is a standard time dependent covariate analysis using Cox regression. Cancer patients receive repeated testing to detect ctDNA in the blood. We will estimate an association between ctDNA positivity and progression-free survival. A power analysis is needed to determine if 1) logistically maximum sample size makes the study feasible (HR = 2) and for grant application. My own literature search failed to uncover papers for power analysis in this setting. Power analysis for landmark analysis is a potential fall back. Links to papers or R-code for conducting power analysis in this setting would be appreciated.

1 Like