Estimated median survival from Cox analysis

["Hunsicker, Lawrence" <lawrence-hunsicker@uiowa.edu>]

Hi, folks:

I am trying to predict the estimated median survival for a group of patients with a defined set of values for SOME of the covariates from the cox model.  I know that I can use the coxph version of survfit to calculate the entire curve, but really all I need is the estimated median survival.  A couple of questions:

1.  Am I correct that survfit.coxph will estimate the survival curve from the mean values of all of the covariates that are not explicitly set in the “newdata” data set? 

2.  Is there any way to get survfit.coxph to print out for each of the specified covariate sets only the time to median survival, and its confidence intervals?  (I am assuming that what one gets is the estimated survivals at each of the event times, since this appears to be what plot(survfit(model)) plots.)

Many thanks in advance for any help that you can give me.

Larry Hunsicker