← Back
NeuroTrials.ai
Neurology Clinical Trial Database

ROCKET AF

Rivaroxaban versus Warfarin in Nonvalvular Atrial Fibrillation

Share Share Copied! Compare

Year of Publication: 2011

Authors: Manesh R. Patel, M.D., Kenneth W. Mahaffey, ..., Robert M. Califf

Journal: The New England Journal of Medicine

Citation: N Engl J Med 2011;365:883-91.

Link: https://doi.org/10.1056/NEJMoa1009638

PDF: https://www.nejm.org/doi/pdf/10.1056/NEJMoa1009638

Clinical Question

In patients with nonvalvular atrial fibrillation at increased risk for stroke, is once-daily rivaroxaban noninferior to dose-adjusted warfarin for the prevention of stroke or systemic embolism?

Bottom Line

In patients with atrial fibrillation at moderate-to-high risk for stroke, once-daily rivaroxaban was noninferior to warfarin for the prevention of stroke or systemic embolism. There was no significant difference in major bleeding, although rivaroxaban was associated with significantly lower rates of intracranial and fatal bleeding but higher rates of gastrointestinal bleeding.

Major Points

  • ROCKET AF was the third pivotal DOAC trial in AF after RE-LY (dabigatran, 2009) and ARISTOTLE (apixaban, 2011), and the only one to use a once-daily factor Xa inhibitor — offering a simpler regimen than twice-daily alternatives.
  • Enrolled the highest-risk AF population of any DOAC trial: mean CHADS2 score 3.5 (vs 2.1 in RE-LY, 2.1 in ARISTOTLE), with 55% having prior stroke/TIA and 63% having heart failure — making results most applicable to secondary prevention.
  • Primary per-protocol analysis: rivaroxaban was noninferior to warfarin for stroke or systemic embolism (1.7% vs 2.2%/yr; HR 0.79, 95% CI 0.66–0.96; p<0.001 for noninferiority). Pre-specified noninferiority margin was HR 1.46.
  • ITT superiority analysis was not significant (2.1% vs 2.4%/yr; HR 0.88, 95% CI 0.74–1.03; p=0.12) — a key distinction from ARISTOTLE (superior) and RE-LY 150mg (superior). ROCKET AF only met noninferiority.
  • Rivaroxaban significantly reduced intracranial hemorrhage (0.5% vs 0.7%/yr, HR 0.67, p=0.02) and fatal bleeding (0.2% vs 0.5%/yr, HR 0.50, p=0.003) — consistent across all DOACs vs warfarin.
  • Major bleeding rates were similar overall (3.6% vs 3.4%/yr, HR 1.04, p=0.58), but GI bleeding was significantly higher with rivaroxaban (3.2% vs 2.2%, p<0.001) — a class effect shared with dabigatran 150mg but not apixaban.
  • Double-dummy design: patients received either real rivaroxaban + sham INR monitoring, or real warfarin + sham tablets — the most rigorous blinding of any DOAC trial. INR values in rivaroxaban arm were generated by a point-of-care device programmed to display sham values.
  • Built-in renal dose adjustment: 15mg daily for CrCl 30–49 mL/min (vs 20mg standard) — the first DOAC trial to prospectively incorporate renal dosing in the main protocol. 21% of enrolled patients received the reduced dose.
  • Controversial warfarin TTR of 55% (lowest among DOAC trials: RE-LY 64%, ARISTOTLE 62%, ENGAGE AF-TIMI 48 65%) — critics argue the warfarin arm underperformed, inflating rivaroxaban's apparent benefit. Pre-specified center-level TTR analysis showed consistent results.
  • Event-driven design: continued until 405 primary events accrued, leading to variable follow-up (median 707 days). This contrasts with time-driven designs (ARISTOTLE) and affects exposure duration calculations.

Design

Study Type: Multicenter, randomized, double-blind, double-dummy, event-driven, noninferiority trial.

Randomization: 1

Blinding: Double-blind, double-dummy.

Enrollment Period: December 18, 2006, through June 17, 2009.

Follow-up Duration: Median of 707 days.

Centers: 1178

Countries: 45 countries

Sample Size: 14264

Analysis: The primary analysis was a per-protocol, as-treated population. An intention-to-treat analysis was also performed.

Inclusion Criteria

  • Nonvalvular atrial fibrillation (paroxysmal, persistent, or permanent) documented by ECG or rhythm strip.
  • History of prior stroke, TIA, or systemic embolism — OR ≥2 of: heart failure (or LVEF ≤35%), hypertension, age ≥75 years, diabetes mellitus (effectively CHADS2 ≥2).
  • Notably required higher stroke risk than other DOAC trials — RE-LY and ARISTOTLE allowed CHADS1 score ≥1, whereas ROCKET AF required ≥2 risk factors or prior cerebrovascular event.
  • Stable anticoagulation-eligible patients — no planned cardioversion or ablation within 30 days.

Exclusion Criteria

  • Hemodynamically significant mitral stenosis or mechanical prosthetic heart valve (valvular AF).
  • Active internal bleeding or history of hemorrhagic diathesis.
  • Intracranial hemorrhage, intracranial neoplasm, arteriovenous malformation, or aneurysm.
  • Significant liver disease (hepatitis B/C with ALT >3× ULN) or Child-Pugh class B or C cirrhosis.
  • Creatinine clearance <30 mL/min at screening (CrCl 30–49 eligible for reduced dose).
  • Uncontrolled hypertension (systolic >180 mmHg or diastolic >100 mmHg) at screening.
  • Concomitant use of strong CYP3A4 and P-glycoprotein inhibitors (e.g., ketoconazole, ritonavir) or inducers (e.g., rifampin, phenytoin).
  • Active endocarditis, recent major surgery (<30 days), or planned invasive procedure during trial.
  • Known hypersensitivity to rivaroxaban or its excipients.
  • Stroke within 14 days or severe disabling stroke (mRS ≥4) within 3 months of screening.

Baseline Characteristics

CharacteristicControlActive
Median Age (IQR) - yr73 (65-78)73 (65-78)
Female sex - no. (%)2832 (39.7)2831 (39.7)
Median Body-mass index (IQR)28.1 (25.1-31.8)28.3 (25.2-32.1)
Type of atrial fibrillation (Persistent) - no. (%)5762 (80.8)5786 (81.1)
Mean CHADS2 score (±SD)3.46±0.953.48±0.94
Previous stroke, systemic embolism, or TIA - no. (%)3895 (54.6)3916 (54.9)
Congestive heart failure - no. (%)4441 (62.3)4467 (62.6)
Hypertension - no. (%)6474 (90.8)6436 (90.3)
Diabetes mellitus - no. (%)2817 (39.5)2878 (40.4)
Previous myocardial infarction - no. (%)1286 (18.0)1182 (16.6)

Arms

FieldControlRivaroxaban
InterventionDose-adjusted warfarin targeting INR 2.0–3.0, managed by local investigators using point-of-care INR testing. Sham rivaroxaban tablets provided for double-dummy blinding. INR monitored at least monthly with dose adjustments per local practice. Mean TTR was 55% (median center-level TTR 58%) — lower than RE-LY (64%) and ARISTOTLE (62%). Patients also received sham rivaroxaban-matched placebo capsules daily.Rivaroxaban 20 mg once daily with the evening meal (to optimize absorption), or 15 mg once daily for patients with CrCl 30–49 mL/min (21% of enrolled). Direct factor Xa inhibitor with predictable pharmacokinetics — no routine monitoring needed. Half-life 5–9 hours (younger) to 11–13 hours (elderly). Sham INR device provided sham values to maintain blinding. 2.6× bioavailability increase with food — hence the requirement to take with evening meal.
DurationMedian of 590 days (event-driven trial)Median of 590 days (event-driven trial)

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
The composite of stroke (ischemic or hemorrhagic) and systemic embolism, analyzed in the per-protocol, as-treated population.Primary2.2% per year (241 events)1.7% per year (188 events)0.79<0.001 for noninferiority
Primary endpoint in the intention-to-treat populationSecondary2.4% per year (306 events)2.1% per year (269 events)HR 0.88 (95% CI, 0.74 to 1.03)0.12 for superiority
Myocardial infarction (as-treated safety population)Secondary1.1% per year (126 events)0.9% per year (101 events)HR 0.81 (95% CI, 0.63 to 1.06)0.12
All-cause mortality (as-treated safety population)Secondary2.2% per year (250 deaths)1.9% per year (208 deaths)HR 0.85 (95% CI, 0.70 to 1.02)0.07
Major and nonmajor clinically relevant bleedingAdverse14.5% per year14.9% per yearHR 1.03 (95% CI, 0.96 to 1.11)0.44
Major bleedingAdverse3.4% per year3.6% per yearHR 1.04 (95% CI, 0.90 to 1.20)0.58
Intracranial hemorrhageAdverse0.7% per year0.5% per yearHR 0.67 (95% CI, 0.47 to 0.93)0.02
Fatal bleedingAdverse0.5% per year0.2% per yearHR 0.50 (95% CI, 0.31 to 0.79)0.003

Criticisms

  • TTR of 55% was the lowest among major DOAC trials (RE-LY 64%, ARISTOTLE 62%, ENGAGE AF 65%) — well-managed warfarin patients may have comparable outcomes to rivaroxaban, limiting generalizability to settings with good anticoagulation clinics.
  • Failed to demonstrate superiority in the ITT analysis (HR 0.88, p=0.12) — unlike ARISTOTLE (superior) and RE-LY 150mg (superior for stroke). Only met noninferiority, making it the weakest evidence among DOACs.
  • Significantly higher GI bleeding (3.2% vs 2.2%, p<0.001) — a clinically meaningful trade-off, especially in elderly patients where GI bleeding is a major cause of DOAC discontinuation.
  • Once-daily dosing may create trough vulnerability: rivaroxaban's 5–13h half-life means subtherapeutic levels before the next dose, potentially explaining residual stroke events. ARISTOTLE and RE-LY used BID dosing with more stable drug levels.
  • Highest-risk population (CHADS2 3.5) limits applicability to lower-risk AF patients — cannot extrapolate noninferiority to CHADS2 1–2 populations where treatment decisions are most uncertain.
  • Industry-sponsored (Bayer/Janssen) with industry involvement in design, data collection, and analysis — though an independent academic steering committee oversaw the trial.
  • Post-randomization event rate spike after drug discontinuation: patients stopping rivaroxaban had more stroke events in the transition-to-warfarin period, suggesting a 'rebound' phenomenon when abruptly stopping without bridging.
  • Sham INR monitoring was novel but raised ethical concerns — patients in the rivaroxaban arm received fake INR values, delaying identification of any coagulopathy from other causes.
  • No direct comparison with other DOACs — all DOAC-vs-warfarin trials used warfarin as the common comparator, making cross-trial DOAC comparisons indirect and subject to population differences.

Funding

Bayer HealthCare AG and Janssen Research & Development (Johnson & Johnson)

Based on: ROCKET AF (The New England Journal of Medicine, 2011)

Authors: Manesh R. Patel, M.D., Kenneth W. Mahaffey, ..., Robert M. Califf

Citation: N Engl J Med 2011;365:883-91.

Content summarized and formatted by NeuroTrials.ai.