Feature request: Firth correction

[carbocation]

I'm wondering if adding optional Firth correction might be possible. I ask because, although there is a coxphf package, that package does not currently support multistate models or competing hazards models like this package does. For completeness, I have also opened a parallel inquiry over on the coxphf issues to see if there is a way to instead add multistate or competing hazards models to that package. Ultimately, I think it would be helpful if there were an R package that could both handle competing hazards and perform Firth correction.

[therneau]

The Firth 'correction' is a penalized model, and in my opinion it more fits as an option in coxme. That is, it is simply a different penalty than a Gaussian. Updating coxme to handle multi state models is on my to do list (but unfortunately has been there for a long time.)