[R] Non-inferiority log-rank test

[ehud cohen]

Hi Group,

I'm looking for a method to calculate sample size for a
non-inferiority survival analysis.
Initially I've used cpower() of Hmisc, switching the roles of type-I
and type-II errors, as cpower() assumes no treatment difference as the
null hypothesis.

Now, I would like to introduce event rate difference under the
null-hypothesis, and calculate sample size accordingly. Any ideas?
(similar functionality exists in nBinomial function of gsDesign for
binomial tests)

Thanks,

Ehud