I'd like to see what the OXY number looked like for those that didn't drop out - I assume by "comparable" they meant "a bit higher than for our drug." It's very unusual to see a pain trial for a narcotic that doesn't comfortably separate from placebo.
But even producing results in the same ballpark is pretty significant. The reality is that the overall morbidity from long-term narcotics use has to be higher than that from some moderate QTc issues (although who knows if the FDA would see it that way given their QTc-philia).
Peter
(Edit - here's another study in osteoarthritis showing pain score reductions of 2.3 with oxy ER:
At end point, the OROS hydromorphone group had a mean pain relief score of 2.3 (median, 2.0) and the ER oxycodone group had a mean pain relief score of 2.3 (median, 2.3) (95% CI, -0.30 to 8).