The design of ROCKET AF is shown in this Figure.[97][115] ROCKET AF enrolled patients with nonvalvular AFib at moderate-to-high risk of stroke, with the following CHADS₂ risk factors: the patients had to have prior stroke, transient ischemic attack (TIA), or non-CNS embolus, or at least 2 other CHADS₂ risk factors.  If subjects had only 2 risk factors, but no prior stroke, TIA, or embolism, they were excluded -- in other words, the ROCKET AF population had a high CHADS₂ risk score.

Subjects were then randomized to rivaroxaban (20 mg/day or 15 mg/day for patients with CrCl 30–50 mL/min) or warfarin, in a double-blind, double-dummy fashion for the patients and the physicians in the trial, with adjudication of results also double-blind.  The primary efficacy endpoint was the composite of stroke and non-CNS embolism, and the primary safety endpoint was a composite of major and non-major (but clinically relevant) bleeding. 

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