Looking at secondary efficacy outcomes (Figure 17; see also Hylek, Figure 4) from the same meta-analysis shown in the previous Figure,[17] it appears that the reduction in stroke is largely driven by a reduction in hemorrhagic stroke, and this finding of about a 50% reduction in hemorrhagic stroke versus the results with warfarin is very solid across all 4 of the NOACs listed in Figures 15 and 16.

There appeared to be a trend toward a reduction in ischemic stroke, but that was not significant in this meta-analysis. There may be some difference among the different agents, but that could also have been due to trial design differences.

With respect to myocardial infarction (MI), there was no clear signal of an increase or decrease compared with warfarin. However with respect to all-cause mortality, there was a significant reduction of about 10%, although that was probably driven by the fact that hemorrhagic stroke is so significantly decreased.