[R] Any examples of a frailty model actually used for prediction ?

[Koffijberg, H.]

Hi everyone,

I'm looking for any examples of useful frailty models, in particular any situation in which a cox proportional hazards model with frailty outperforms a regular cox proportional hazards model with respect to prediction of the time to event (or the X-year risk of an event). I have defined my own gamma-frailty cox PH model in R but on my simulated data sample it does not predict any better than a regular cox model. In fact the predictions of the frailty model are often worse, *even* when I purposely add gamma-distributed frailty to the simulated sample (as I have previously posted). Can anyone help me, e.g. by pointing me to some R code in which the performance of models with and without frailty component are assessed ? Also any examples in which the actual estimated individual frailty values are used for predictions are very much welcome.

Thanks.
Erik Koffijberg.

=====================================================
Julius Center for Health Sciences and Primary Care
University Medical Center Utrecht Str. 6.131
P.O.Box 85500, 3508 GA Utrecht, The Netherlands
Tel:	+31 30 250 3013,
Fax: 	+31 30 250 5480
E-mail: H.Koffijberg at umcutrecht.nl