43 is a reasonable number to start with (it gives you some power wrt sample variance, but you have to pay attention to randomization). It might give you some i

43 is a reasonable number to start with (it gives you some power wrt sample variance, but you have to pay attention to randomization). It might give you some information if the intervention were really, really bad for patients or really really good. (So it could be used to establish possible harms). But definitely not useful farther than that except to justify larger studies. I'm still hoping we catch a break one of these days.