AMPE’s post hoc spin in DME trial is as silly as some of the stuff you joked about Accounting for odds: 1) It's a post hoc - strike 1 2) It's implies a U-shaped dose response which is rare indeed (i.e. improbable) - strike 2 3) Even if it is, in fact, efficacious with a U-shaped dose response, it is much more difficult to run trials that will succeed - strike 3. PS FWIW my subgroups aren't intended, per se, to be funny. Just to make the fallacy of subgroup logic clear.