Hi, Sorry to trouble the list. I have a problem which I'm not sure how to resolve. I have a Cox model with 1 independent variable with 2 categories (and thus 2 survival curves on plotting survfit) How can I get an estimate of survival for each category at a particular time point, with standard error? Looking through ?cph and ?coxph, I'm not quite sure how to go about that. I would really appreciate any direction here. Thanks a million! Min-Han
Apologies, I located the function in cph (... time.inc =30, surv = TRUE) ... $surv.summary Thanks. Min-Han On Tue, 2 Nov 2004 21:06:33 -0500, Min-Han Tan <minhan.science at gmail.com> wrote:> Hi, > > Sorry to trouble the list. I have a problem which I'm not sure how to resolve. > > I have a Cox model with 1 independent variable with 2 categories (and > thus 2 survival curves on plotting survfit) > > How can I get an estimate of survival for each category at a > particular time point, with standard error? > > Looking through ?cph and ?coxph, I'm not quite sure how to go about > that. I would really appreciate any direction here. > > Thanks a million! > > Min-Han >
Min-Han Tan wrote:> Hi, > > Sorry to trouble the list. I have a problem which I'm not sure how to resolve. > > I have a Cox model with 1 independent variable with 2 categories (and > thus 2 survival curves on plotting survfit) > > How can I get an estimate of survival for each category at a > particular time point, with standard error? > > Looking through ?cph and ?coxph, I'm not quite sure how to go about > that. I would really appreciate any direction here. > > Thanks a million! > > Min-HanThe survest function in the Design package makes this easy. You can also probably do it without a lot of hassle using standard survival package routines. With survest you can easily specify covariate combinations for which to obtain predictions. -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University
On Tue, 2 Nov 2004, Min-Han Tan wrote:> Hi, > > Sorry to trouble the list. I have a problem which I'm not sure how to resolve. > > I have a Cox model with 1 independent variable with 2 categories (and > thus 2 survival curves on plotting survfit) > > How can I get an estimate of survival for each category at a > particular time point, with standard error? >Here's an example> fit <- coxph( Surv(futime,fustat)~rx,data=ovarian)> summary( survfit( fit, newdata=list(rx=c(1,2))))Call: survfit.coxph(object = fit, newdata = list(rx = c(1, 2))) time n.risk n.event survival1 survival2 59 26 1 0.952 0.973 115 25 1 0.903 0.945 156 24 1 0.855 0.917 268 23 1 0.806 0.888 329 22 1 0.758 0.858 353 21 1 0.710 0.828 365 20 1 0.663 0.797 431 17 1 0.613 0.764 464 15 1 0.560 0.727 475 14 1 0.509 0.689 563 12 1 0.454 0.648 638 11 1 0.403 0.606 -thomas