cox regression -one variable =100% of outcome

[jb26]

Hi,

hope someone can help....

I'm doing a stepwise forward Cox regression analysis examining the time taken to reach a decision (in particular the differences between those that make an early decision vs late -these are dichotomised by the optimal time taken to make this decision). I have several independent variables (all dichotomised). The problem is one of these variables predicts 100% of those making the early decision (and this is not an error and makes sense) =STATA excludes it from the model. IS there anyway I can manipulate the data or STATA command to include this variable?

Appreciate any help!!!!


j