how to predict 5-year risk using cox model

[owen_h]

Dear fellow stata users,

I have a question regarding how to predict 5-year risk using the cox model. For example, let's say we follow all patients for ~8 years for mortality (y1=death), and we want to predict the 5-year mortality risk (y2=death5, whith is death censored at year 5). How should i do this? Should i use cox regression to model y1 (death) and get the baseline survival function at year 5, or should i model y2 (death5)?

Thanks!
Owen

[Steve Samuels]

See Message # 14 of http://www.statalist.org/forums/foru...=1399502458306