st: Checking proportional hazards assumption with ordinal and categorical and time-dependent variables in the model

[Sanam P <sanamp25@yahoo.com>]

Dear All

For checking proportional hazard assumption in Cox regression model, one of the ways is to use:

stcox age ndrugtx treat site age_site, nohr tvc(age ndrugtx treat site) texp(ln(_t))

Can any one please guide me how this code will change if there is ordinal or categorical variables with more than two categories in the model?

I understand that when running cox model with categorical variables we can use some thing like below code adding i infront of the categorical variables:

xi:stcox i.agegroup ndrugtx i.treat site age-site

However I am not sure how the code for testing the PH assumption should change.

My other question is that if there is a time dependent variable in the model, something like blood pressure measures that are updated during time, I understand that we will not need to check the assumption for this variable but will that affect this test for other variables in anyway? my understanding is that it should not be included in the tvc part of the code?

Many thanks in advanced,
Shell


      

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/