C-index and ploting survival curves after multiple imputation

[Sugi Jeyaganth]

Dear Statalist,
After i impute my data set. I run the stepwise cox regression analysis.I
need help for to calculate the
1) c-index for my final model
2) pooled survival curve for different levels of covariate.
3)how can i find 95% CI and p-value for two nested Cox models
of a Harrell C statistics comparison.



I am very much appreciate any guidance.

Thanks
Sugi
************below is my stata code ***************

stset fw, f(dead) id(MRN)
gen Int=ln(_t)


*only two binary variable has missing
*imputing this varibles
ice o.mgmt_m o.adjuvan Int _d , ///
gen(m_) saving(im, replace) ///
m(5) seed(4324)

use "im.dta", clear

xi: mim: stepwise, pr(0.10): stcox i.sex i.mgmt_m i.his ///
i.age i.chemo i.kps i.regimen ///
i.loca i.ppro i.adjuvan i.exin

*final model
xi:mim , storebv: stcox i.adjuvan i.mgmt_m i.chemo i.his
mim: predict pb

*c-index
mim, cat(combine) est(r(area)) se(r(se)) : roctab pb _t _d
too few variables specified
an error occurred when statsby executed roctab

*curve
stcurve, survival at1(chemo=0) at2(chemo=1)
last estimates not found

*compare two model c index

xi:mim , storebv: stcox i.adjuvan i.mgmt_m i.chemo
xi:mim , storebv: stcox i.adjuvan i.mgmt_m i.chemo i.his i.ppro

[Steve Samuels]

The Sandbox forum is for trying out Latex and features of the the Advanced Forum Editor ( "A" in the three buttons on the upper-right of the edit window). Post in the General Forum. For readability, put all code and results between CODE delimiters-use the # button in the advanced editor. Also it is considered a good habit on Statalist to post with user-names that contain full real names, first and last. You can re-register with your full name via the Contact Us button at the bottom right of this page.

[Sugi Jeyaganth]

Thanks very much for your reply , i will re-register it.