ROC curve analysis for cox regression model

[hatem ali]

How to do ROC curve analysis and a graph for cox regression model on stata 11?

it can be done for logistic regression analysis using command:lroc
but it doesn't work for a cox hazard regression model?
is there a way for that?how to do it?

[Clyde Schechter]

See -estat concordance- in -help stcox postestimation-.

[hatem ali]

Thanks for getting back to me
So I did estat concordance and it gave me Harel c and somer d which I understand are the postitive predictive value of the model.

How can I graph and plot the positive predictive value?? and how to estimate the area under the curve?

Thanks for your help

[Clyde Schechter]

Harrell's C is the area under the curve, it is not the positive predictive value of the model.

But I don't know of a way to get a plot in this context.

[hatem ali]

and how to get sensitivity ,specificity and PPV for the whole cox hazard regression model?

[Clyde Schechter]

I don't know what you mean by that. The Cox regression model does not predict any dichotomous outcome. It is a model of time to events. The Harrell C statistic (which is also the area under an ROC curve) for a Cox regression model relates to the accuracy of its predictions about whose "death" precedes whose.