[R] How to fit parametric survival model using counting process data

[Terry Therneau]

> I was trying to fit a parametric survival model with Weibull distribution on
> counting process type of data (NOT interval censor data), but the
> survreg(Surv(T1,T2,event)~x,data,dist="weibull") did not seem to work.
> Anyone can help me with that?

 The survreg function does not support time-dependent covariates or multiple 
events per subject, i.e., 2 of primary reasons to use (start, stop] data.  I've 
thought about it, and it is on my "someday" list.  It is a fairly difficult 
update though, so "someday" may be a loooooong way away.  (A major upgrade to 
coxme is first, to support more models and an lmer-like syntax).
 
 	Terry Therneau