It makes perfect sense to consider Avonex, Rebif, Betaseron, and Extavia as a single group for this kind of analysis.
I don't know what the objective of "this kind of analysis" is. Those drugs have different shares by line and severity, so if we want to describe a "typical" Gilenya patient then a finer breakdown would be meaningful.
..half of Gilenya’s US scripts are coming from patients who were not on another drug. To a first-order approximation that is good enough for our purposes, these patient starts are being taken from other drugs in proportion to what those drugs’ share of the first-line setting was prior to Gilenya’s launch.
Again, I don't know what "our purposes" are, but that's a bit too simplistic. Why do you only consider the first line setting? What about restarts? What if there is pent-up demand in failed patients (patients who give up taking their drugs because they don't see enough benefit for the hassle that they require)? Wouldn't a new launch source disproportionately from that segment first until it's depleted? Kind of what people are expecting for Telaprevir?