This Figure shows the efficacy outcomes across the spectrum of the components of the primary composite endpoint: for strokes of all types – hemorrhagic, ischemic, and unknown – rivaroxaban was non-inferior to warfarin.  Of note is that there were no differences in number of in ischemic strokes per 100 patient-years;  this was probably a factor in keeping the confidence intervals of the ITT and per-protocol analyses wide enough that the results were only significant for non-inferiority, not superiority. Why there was this lack of difference in ischemic stroke (which was also true for the third NOAC, apixaban, discussed below, but not for dabigatran) is not clear.

 Reiffel JA. Am J Med 2013; 126: 00-00.