st: -TVC- option in the -stcox- command

[Pete de Blank <pdeblank@gmail.com>]

Dear Stata-users,

I am trying to understand how to use and interpret the -tvc- option in
Stata’s  -stcox- command.

I have a data set of drug and infection information about subjects
during inpatient hospitalizations, with daily exposure information on
antibiotic classes of drugs.  I would like to explore whether these
drug classes are risk factors for a specific infection by determining
the hazard ratio associated with exposure to these drug classes.

Since exposure to these drugs varies with time, I have been using the
-tvc- option for my survival analysis.  My -stset- command is
    Stset hospitalday, failure(infection) id(id)

I include all drug classes that have a p value >0.2 on log rank test,
according to the command
    Sts test drugclass1, strata(hospitalsite)

And my initial model is:
    Stcox drugclass1 drugclass2 drugclass3 drugclass4 drugclass5,
tvc(drugclass1 drugclass2 drugclass3 drugclass4 drugclass5)
strata(hospitalsite)

I have three questions:
(1)  How do you choose the best multiplier for the time-varying
covariates (with -texp(exp)-)?  How do you know if ln(t) is better
than t?

(2)  Under what criteria can non-significant covariates be eliminated
from the model?  If the non-time-varying p-value is >0.05?  Only if
both the time-varying and non-time-varying covariates have
p-values>0.05?  If the time-varying portion (but not the
non-time-varying portion) of a covariate has a p-value > 0.05, can I
eliminate just that covariate in the tvc option?

(3)  Finally, how do you conceptualize the interpretation of the
results?  Do I think of this as an interaction with time?

Thank you for any and all help or suggestions,
Pete

*
*   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/