Rivaroxaban for Stroke Prevention in Atrial Fibrillation

A Critical Review of the ROCKET AF Trial

Jeremy S Paikin; Joshua J Manolakos; John W Eikelboom


Expert Rev Cardiovasc Ther. 2012;10(8):965-972. 

In This Article

ROCKET AF Trial Design

The ROCKET AF study was a blinded, double-dummy, randomized-controlled trial in which 14,262 patients with nonvalvular AF at increased risk for stroke were assigned to rivaroxaban 20 mg daily (or 15 mg daily in those with creatinine clearance [CrCl] 30–49 ml/min) or dose-adjusted warfarin (target INR: 2–3).


The main objective of the ROCKET AF trial was to demonstrate noninferiority of rivaroxaban compared with warfarin for the prevention of stroke or systemic embolism in patients with AF.


The ROCKET AF trial recruited patients from 1178 centers in 45 countries. Patients were eligible for inclusion if they had electrocardiographic evidence of AF within 30 days prior to randomization and either a history of stroke/transient ischemic attack (TIA) or systemic embolism or at least two of the following additional risk factors for stroke: heart failure or a left ventricular ejection fraction of 35% or less, hypertension, age ≥75 years or the presence of diabetes mellitus. By excluding AF patients without a previous stroke/TIA or systemic embolism who did not have at least two other risk factors for stroke, and by limiting inclusion of patients with two other stroke risk factors to 10% of the cohort at each site (other patients without a history of stroke/TIA or systemic embolism had to have at least three risk factors), the ROCKET AF trial included a patient population at higher risk of stroke than those included in other new oral anticoagulant trials in AF.[6,7] A higher risk population increases study power but is less representative of the broader AF population that is expected to benefit from anticoagulant treatment.

The types of patients excluded from participation in the ROCKET AF trial were those at increased risk of bleeding (e.g., disabling stroke within 3 months or any stroke within 14 days of randomization, active bleeding, CrCl <30 ml/min, taking aspirin at doses exceeding 100 mg/day or dual antiplatelet therapy), those with another indication for anticoagulant therapy (e.g., prosthetic heart valve) and patients receiving treatment or potentially eligible for treatment with drugs that may interact with rivaroxaban (e.g., patients with HIV infection).


The rivaroxaban dosing regimen evaluated in the ROCKET AF trial was selected on the basis of results of Phase II dose-finding studies in VTE treatment, which showed that long-term treatment with q.d. and b.i.d. rivaroxaban doses in the range of 20–60 mg/day, were associated with similar efficacy and a dose-dependent effect on bleeding. Thus the lowest effective dose of rivaroxaban (20 mg daily) was chosen for evaluation in the ROCKET AF trial.[16] Unlike the RE-LY trial that used an open design,[6] the ROCKET AF study employed a blinded design to reduce the potential for differential use of cointerventions based on the knowledge of treatment allocation and for biased ascertainment and reporting of outcomes. To preserve the integrity of the blinding, investigators used a point-of-care device to generate encrypted INR values, either real or sham depending on the treatment to which the patients were assigned.


The primary efficacy outcome was the composite of stroke and systemic embolism. Stroke was defined as an acute, focal neurologic deficit of presumed cerebrovascular etiology that persisted beyond 24 h. If the deficit lasted for less than 24 h, it was defined as a TIA. Brain imaging was encouraged for all suspected strokes and was performed in 82.1% of patients reported to have had a stroke. Systemic embolism was defined as acute vascular insufficiency associated with clinical or radiographic evidence of arterial occlusion (not associated with another likely cause). The primary safety outcome was the composite of major and clinically relevant nonmajor bleeding. Major bleeding was defined as clinically overt bleeding associated with any of the following: fatal outcome, involvement of a critical anatomic site (intracranial, spinal, ocular, pericardial, articular, retroperitoneal or intramuscular with compartment syndrome), fall in hemoglobin concentration ≥2 g/dl, transfusion of ≥2 units of whole blood or packed red cells, or permanent disability. Critically relevant nonmajor bleeding was defined as bleeding that did not meet the criteria for major bleeding, but necessitated medical treatment, unplanned medical attention or temporary interruption of the study drug. The protocol specified the need for follow-up visits at 1, 2 and 4 weeks then monthly thereafter for INR measurement and surveillance of primary safety and efficacy outcomes.


The primary statistical analysis employed a Cox model using treatment as a covariate. The noninferiority margin of 1.46 for the primary outcome was derived from a meta-analysis by Hart et al. of studies comparing VKA with placebo/non-VKA control for stroke prevention in AF,[2] and was selected with the intent of preserving at least 50% of the benefit of VKAs for stroke prevention. Based on expected event rates of 2.3% per 100 patient years in patients treated with warfarin and a warfarin drop-out rate of 14% per year, it was estimated that 14,000 patients would be needed to have 95% power to demonstrate noninferiority.

Analytic Approach

The protocol specified that the primary analysis for the noninferiority of rivaroxaban as compared with warfarin for the prevention of stroke or systemic embolism would be tested using a per-protocol analysis. The per-protocol population was defined as all patients who received at least one dose of a study drug, did not have a major protocol violation (e.g., permanent discontinuation of randomly assigned treatment prior to a primary outcome event) and were followed for events while receiving a study drug or within 2 days after discontinuation. If noninferiority was met, the protocol specified that the superiority would also be tested in the same per-protocol population, but also including patients who had a major protocol violation (this is described as the safety population). Bleeding outcomes and other efficacy outcomes were also analyzed using the safety population. Secondary analyses for noninferiority and superiority were performed in the intention-to-treat (ITT) population, which included all patients who underwent randomization and were followed for events to the end of the study irrespective of whether they continued on study treatment.