Dear Users,

I am looking at the impact of 4 different treatments on recovery (yes/no) using data from a trial. The study was stratified across clinics so I would need to account for the 15 strata.

I considered running a binomial regression but, due to possible small cell sizes, I would prefer to use the MH test; however, I am not sure mhodds can deal with 2xj tables.

Also, ideally, I would like to estimate relative risk rather than odds ratio.

I'd be grateful for any suggestion.

Thanks,
Laura