Dear Statalist,
I'm using -somersd- command to compare c-index from two Cox models, one with demographics and traditional cardiovascular risk factors for predicting cardiovascular outcomes, and another one adding a new variable of interest. I've followed the pathway published from "Newson RB, Comparing the predictive powers of survival models using Harrell's C or Somers' D, The Stata Journal (201):10(3):339-358".
After -stset-, -stcox-, -estat con-, -predict hr-, generate invhr=1/hr, generate censind= 1-_d if _st==1
I've run: somersd _t invhr if _st==1, cenind(censind) tdist transf(c) following the suggestions describe above.
However, the option -cenind()- was not found. I've update the package, but it's still not working.
Any suggestion to fix this? Or an alternative approach to compare C-statistics?
Thanks,
Henrique
Henrique Turin Moreira, MD, PhD
University of São Paulo, Ribeirão Preto, Brazil
I'm using -somersd- command to compare c-index from two Cox models, one with demographics and traditional cardiovascular risk factors for predicting cardiovascular outcomes, and another one adding a new variable of interest. I've followed the pathway published from "Newson RB, Comparing the predictive powers of survival models using Harrell's C or Somers' D, The Stata Journal (201):10(3):339-358".
After -stset-, -stcox-, -estat con-, -predict hr-, generate invhr=1/hr, generate censind= 1-_d if _st==1
I've run: somersd _t invhr if _st==1, cenind(censind) tdist transf(c) following the suggestions describe above.
However, the option -cenind()- was not found. I've update the package, but it's still not working.
Any suggestion to fix this? Or an alternative approach to compare C-statistics?
Thanks,
Henrique
Henrique Turin Moreira, MD, PhD
University of São Paulo, Ribeirão Preto, Brazil
