@drjohnm However, stat power can be fairly good in examining HTE for a continuous subject attribute w/ reasonable dispersion in the study pop, such as baseline risk, age, LDL, A1c, BP, etc. It drives my crazy when trialists bin continuous vars & lose t
This is a nice paper. The intro (alone) is pure gold for trial translation. Thx for sharing. I guess, one pushback is that I am not sure you can hold treatment related harm constant. B/c, it seems to me that that is not a constant. cc @Tufts_PACE @AndrewFo
@AndrewFoy82 @johnwmcevoy @kaulcsmc @drjohnm @MRuzieh Can U think of example in which RRR in a specific outcome variables sig by baseline risk (& not by a mechanistic or physiological modifier, or by competing risk or Rx-harms)? And BTW, stat power for
@StatModeling OK, I agree w/ the basic premise, but this is way over-stated. It depends greatly on the degree of HTE AND the nature of the interaction variable (categorical vs continuous). https://t.co/gtEviwC8jr