I am modeling rates of people given drug $X$ or drug $Y$, working out probability of getting drug.
The drug $X$ and drug $Y$ points have been matched on exposure time (person years) to specific drug.
In this case, is there a need to include an offset term or should it now be treated as count data?