Most drugs never reach FDA approval. Bayer's Sunharbo, first of a new class, just did, to reduce repeat strokes. The bit I'd check in the trial data is bleeding. The mechanism is "thought to" stop bad clots without raising that risk, and that's the whole point of the class.

It goes past bleeding. This excerpt gives no trial size, no comparator and no reduction in recurrent strokes. "Significant advance" is the experts' verdict, not an effect size. The 20% is baseline risk, not the drug's effect.