How to generate a C-index from the Cox proportional hazard regression model

[Denali]

I am developing a risk model for a binary outcome cardiovescular disase (Yes/No) using all the betas that generated from a multivariable Cox proportional hazard regression model. The nest step that I need to do to generate a C-index from the multivariable model. Does anyone know how to do it?

Below is my code:

proc phreg data = RS;
class Age_C Race BMI_C LVEF_C SBP ;
model CAD(0) = Age_C Race BMI_C LVEF_C SBP /rl;
run;

Thank you!

[sbxkoenk]

Hello,

I guess with the C-index you mean the concordance index, c, as given by PROC LOGISTIC for a binary response?

That concordance index, c, in the "Association of Predicted Probabilities and Observed Responses" table is in fact the area under the ROC curve.

From your question I understand this value is not given by PROC PHREG. Correct? (I do not know by heart.)

To get the c-index I think you can do a PROC LOGISTIC with the same target variable and as an input (only input) the estimated probabilities of PROC PHREG.

Make sure you are modelling the same target level in both PROC PHREG and PROC LOGISTIC!!

Next time it would be better to post your question in the 'Statistical Procedures' board under Analytics. You will get faster response.

Kind regards,

Koen

[OsoGris · Accepted solution]

Check the documentation for the PROC PHREG statement where you will see a CONCORDANCE option:

https://go.documentation.sas.com/doc/en/statug/15.2/statug_phreg_syntax01.htm#statug.phreg.phrprocco...

You will need a relatively current release of SAS to access the CONCORDANCE option. 

[StatDave]

Many questions on availability of statistics, like this one, are answered in the list of Frequently Asked-for Statistics (FASTats) in the Important Links section of the Statistical Procedures Community page.