st: Using the predictnl command following a model containing restricted cubic splines and time-dependent effects

["Turnbull, Alison E." <aturnbul@jhsph.edu>]

Hello, 

I'm working with a model developed using the stpm2 package from SCC.  
The baseline hazard has 4 degrees of freedom with knots at time = 3, 7 and 21. 

Variable x1 is a continuous variable with a time-dependent effect and a knot at time=21. 
Variables x2 is a dichotomous variable with a time-dependent effect and a knot at time= 21.

The 4 derived spline variables for the baseline log cumulative hazard are 
_rcs1
_rcs2
_rcs3
_rcs4

And the spline variables for the time-dependent effects of x1 and x2 are
_rcs_x11
_rcs_x12
and 
_rcs_x21
_rcs_x22

I would like calculate the hazard ratios and associated 95% CIs for individuals with x1=1 relative to those with x1=0 at time = 2, 7, 10 and 30 as well as the hazard ratios for individuals with varying levels of x2 at the same time points.  I realize that these hazard ratios are non-linear functions of the model parameters  and will require use of the predictnl command but I am struggling to figure out the right equation.  

I would be greatly appreciative if someone could write out the command for one of these desired hazard ratios to help get me started.  

Thank you, 
alison

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