[R] sample size for recurrent events

[Mike]

Dear All,

I'm a relative novice in R, with some experience with S-plus and quite a 
bit in SAS.  In R and S-Plus, I've primarily worked with Frank Harrell's 
Design and Hmisc libraries.  I have R 2.3.1 for windows (XP) installed.

I have a project coming up in which the outcome will be recurrent 
events.  There will be right censoring and likely to be a good stack of 
0s on the left.  In searching the R archives, I came across the gcmrec 
package, which looks like it will estimate the models I might be 
interested in. 

My question is whether anyone is aware of work done on sample size 
planning for these models?  I came across a meeting abstract for a paper 
that apparently presented some SAS code for power estimation, but I 
haven't had any luck contacting the authors.

Does anyone have any experience with this issue or pointers to where I 
might do some reading on this topic?  Of course, any thoughts on the 
approriate model would also be more than welcome.

Thanks,

Mike Babyak