I am fitting Cox regression models with time dependent
covariates using the counting process formulation. I have a dichotomous time
dependent covariate named ‘presence of risk factor XYZ’. For a
particular patient, this risk factor was identified on day A. Is the general
convention to assign a value of ‘yes’ to the time dependent
covariate on day A or day A+1? If the outcome of interest occurred on day A
would this affect the value assigned to the time dependent covariate? I’ve
looked at a few books and there seem to be varying opinions on this issue.
Thanks in advance for your help.