Preventing Stroke in Atrial Fibrillation: Warfarin vs. Factor Xa Inhibitors

Pratyaksh K. Srivastava, BA1

1Harvard Medical School, Boston, MA 02115, USA
Correspondence should be addressed to P.K.S. (


The heart is a muscular pump that circulates blood to the body. It accomplishes this through an electrical network that dictates the timing and nature of its contraction. Atrial fibrillation is an electrical abnormality whereby the heart beats irregularly and too quickly [1]. One consequence of atrial fibrillation is an increased tendency to form blood clots in the upper chambers of the heart [1]. These blood clots have the ability to break off and travel to the brain, where they can cause a stroke.

Patients with atrial fibrillation are at five times greater risk for stroke than similar aged individuals in normal sinus rhythm [2]. In order to prevent strokes, patients with atrial fibrillation are frequently anticoagulated with warfarin, a Vitamin K antagonist.

The decision to begin anticoagulation is usually based on a patient’s CHADS2 (history of congestive heart failure and/or moderate or severe systolic dysfunction; hypertension; age ≥ 75; diabetes mellitus; prior stroke, transient ischemic attack, or thromboembolism) score (Table 1) [3,4]. US guidelines recommend the initiation of warfarin therapy in patients with CHADS2 scores of 2 or above, which correlates to an ischemic stroke rate of 4.2% per year (Table 2) [3,4]. 

Table 1. Calculation of CHADS2 Score.

Table 2. Ischemic Stroke Rate (%/year) by CHADS2 Score.

Currently, only 60% of patients with atrial fibrillation who meet criteria for anticoagulation take warfarin [5]. The main reasons cited for this low rate include fear of bleeding complications, difficulty dosing the medication, and the bothersome nature of frequent blood draws [5]. These associated difficulties were the impetus for trialing direct Xa inhibitors, which anticoagulate by blocking factor Xa in the coagulation cascade. Xa inhibitors were initially attractive for their ease of dosing (few drug and food interactions, standard dosing regimens) and for their convenience (no frequent blood draws) [5]. 


The first two large randomized controlled trials (RCTs) to compare Xa inhibitors to warfarin were “Rivaroxaban versus Warfarin in Nonvalvular Atrial Fibrillation” (ROCKET AF) and “Apixiban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation” (ARISTOTLE) [6,7].

ROCKET AF was a randomized, double blind, controlled trial performed between 2006 and 2010 that compared oral rivaroxaban to warfarin for the prevention of stroke or systemic embolism in atrial fibrillation patients [6]. The trial enrolled 1,178 individuals with atrial fibrillation and moderate to high risk of stroke (CHADS2 ≥2). The patients were randomized to receive once daily oral rivaroxaban or dose-adjusted warfarin. Results of the ROCKET AF study revealed a decreased incidence of stroke and systemic embolism in the rivaroxaban group when compared to the warfarin group (HR 0.79, CI 0.66–0.96). The rivaroxaban group also demonstrated a significant reduction in intracranial hemorrhage (0.5% vs. 0.7%, p=0.02) and fatal bleeding (0.2% vs. 0.5%, p=0.003) when compared to the warfarin group. 

The ARISTOTLE trial (2006–2010, n=18,201) used the Xa inhibitor apixiban and was similar to ROCKET AF in design and outcomes measured [7]. ARISTOTLE demonstrated similar findings to those of ROCKET AF, with a reduced incidence of stroke and systemic embolism in the apixiban group when
compared to the warfarin group (HR 0.79, CI 0.66–0.95), as well as reductions in major bleeding (HR 0.69, CI 0.60–0.80), hemorrhagic stroke (HR 0.51, CI 0.35–0.75), and all-cause mortality (HR 0.89, CI 0.80–0.99). Neither study reported major side effects of the drugs beyond the bleeding described above.

One limitation of ROCKET AF was that the warfarin group had a therapeutic International Normalized Ratio (INR) only 55% of the time. Previous studies have demonstrated that a therapeutic range more than 58% of the time is required for warfarin anticoagulation to be considered superior to aspirin [8]. There was also a significant difference in number of prior myocardial infarctions between the rivaroxaban and warfarin groups [6]. The warfarin group exhibited a greater number of prior myocardial infarctions, suggesting the possibility of greater underlying cardiovascular disease burden. Both studies evaluated individuals at moderate-high risk for stroke—mean CHADS2 for ROCKET AF and ARISTOTLE were 2.1 and 3.5, respectively—and so may not generalize to individuals with lower CHADS2 scores. Pharmaceutical companies sponsored both ROCKET AF and ARISTOTLE, offering a potential source for bias in result interpretation.

Many other Xa inhibitors have been evaluated since ROCKET AF and ARISTOTLE. In 2013, a Cochrane Review was released pooling the major RCTs on the topic [9]. The review included 43,084 participants from ten RCTs that compared dose-adjusted warfarin to different Xa inhibitors, including apixiban, betrixaban, darexaban, endoxaban, idraparinux, and rivaroxaban for the prevention of stroke and other embolic events in atrial fibrillation patients. The mean CHADS2 score of the study participants was 2.7 (range: 1.9–3.5), and the mean age ranged from 65 to 74 years. In the Cochrane group’s meta-analysis, it was found that Xa inhibitors significantly decreased stroke (OR 0.78, CI 0.69–0.89) and systemic embolic events (OR 0.53, CI 0.32–0.87) when compared to warfarin. They were also associated with fewer intra-cranial hemorrhages (OR 0.51, CI 0.41–0.64) and lower all-cause mortality (OR 0.88, CI .81–0.97). 


There are several limitations to the studies presented above. No studies to date compare the different Xa inhibitors directly against one another (e.g., apixiban vs. rivaroxaban), making it difficult to conclude if one Xa inhibitor is better than another [9]. Further, the majority of the large studies performed to date use either apixiban or rivaroxaban—85% of the data from the Cochrane Review was from studies that used either one of these two drugs. In fact, in sub-group analysis, the Cochrane Review found no significant reduction in stroke or systemic embolic events in any of the Xa inhibitors except apixiban and rivaroxaban. The longest follow-up time for studies included in the Cochrane analysis was 1.9 years, with some studies having follow-up of as little as 12 weeks. Finally, the studies conducted to date all include patients who both have and have not used warfarin before study enrollment, making it difficult to evaluate if prior warfarin status affects Xa inhibitor efficacy [9]. 

This last limitation was addressed in the “Endoxaban versus Warfarin in Patients with Atrial Fibrillation” (ENGAGE-AF) trial, which was a randomized, double blind trial comparing low- (30mg) and high-dose (60mg) endoxaban to dose-adjusted warfarin for the prevention of stroke or embolic events in patients with atrial fibrillation [10]. The trial enrolled 21,000 patients with atrial fibrillation and moderate to high risk of stroke (mean CHADS2 2.8) and had a mean follow-up time of 2.8 years. The study found that there was no difference in stroke/embolism rates between the high-dose endoxaban and warfarin groups in individuals with prior Vitamin K antagonist use (1.62% vs. 1.60%, p>0.05), while there was a difference in Vitamin K antagonist naïve individuals (1.49% vs. 2.12%, p=0.03). These results suggest that prior Vitamin K antagonist status may play a role in determining response to Xa inhibitors.

The clinical adoption of Xa inhibitors has been slow [11]. This can be attributed to a number of factors. Although the new Xa inhibitors are associated with a lower incidence of fatal bleeding and intracranial hemorrhage, bleeding continues to remain a feared complication due to the lack of effective Xa inhibitor reversal agents. This is especially worrisome in certain populations, such as the elderly or those with gastrointestinal ulcers, where bleeding is particularly dangerous. The current mainstay of treatment for patients who present with acute bleeding on a Xa inhibitor is supportive care with fluid resuscitation, transfusions, source identification, and drug discontinuation [12]. The use of 4-factor prothrombin complex concentrate (PCC) was shown to normalize prothrombin time in human volunteers who received rivaroxaban, although no studies to date have evaluated the use of PCC in patients with acute bleeding [12]. There are certain additional populations in which factor Xa use is contraindicated. These groups include those with prosthetic heart valves (increases risk of thrombus on the valve), pregnant women (lack of clinical data), patients with renal insufficiency (Xa inhibitors have partial renal excretion), and those at the extremes of weight (affects dosing) [12]. 

Another issue that has slowed adoption of Xa inhibitors is medication adherence. Warfarin has a half-life of around 40 hr, which allows for patients to miss occasional doses without suffering from thromboembolic complications [12]. Xa inhibitors, in comparison, have a half-life of around 12 hr. This short half-life results in the rapid decline of the drug’s antithrombotic effect if doses are missed [12]. Last, factor Xa inhibitors are expensive. Even with insurance, the copayments can be prohibitively high, increasing the likelihood of missed doses, splitting pills, and overall non-adherence [12,13].

Studies to date demonstrate either superiority or noninferiority of Xa inhibitors to warfarin for the prevention of stroke in patients with atrial fibrillation. There are limitations to the data, however. In addition, there are many downsides to factor Xa inhibitor use including the absence of an effective reversal agent, rapid decline of antithrombotic effect if doses are missed, and high cost. Taken together, these findings should prompt caution when considering the use of Xa inhibitors for the prevention of stroke in patients with atrial fibrillation


  1. Waktare JE (2002) Atrial fibrillation. Circulation 106:14-16.
  2. Wolf PA, Abbott RD, Kannel WB (1991) Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 22:983-8.
  3. Singer DE, Albers GW, Dalen JE, et al. (2008) Antithrombotic therapy in atrial fibrillation: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 133:546S-92S.
  4. Friberg L, Rosenqvist M, Lip GY (2012) Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: The Swedish atrial fibrillation cohort study. Eur Heart J 33:1500-1510.
  5. Veltkamp R, Prabhakaran S (2014) Critique of effective anticoagulation with Factor Xa Next Generation in Atrial Fibrillation trial. Stroke 45:2151-3.
  6. Patel MR, Mahaffey KW, Garg J, et al. (2011) Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 365:883-91.
  7. Granger CB, Alexander JH, McMurray JJ, et al. (2011) Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 365:981-92.
  8. Connolly SJ, Pogue J, Eikelboom J, et al. (2008) Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrilla-tion depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in therapeutic range. Circulation 118:2029-37.
  9. Bruins Slot KM, Berge E (2013) Factor Xa inhibitors versus vitamin K antagonists for preventing cerebral or systemic embolism in patients with atrial fibrillation. Cochrane Database Syst Rev 8:CD008980.
  10. Giugliano RP, Ruff CT, Braunwald E, et al. (2013) Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med 369:2093-104.
  11. Bauer KA. Pros and cons of new oral anticoagulants (2013) Hematology Am Soc Hematol Educ Program 2013:464-470.
  12. Kaatz S, Kouides PA, Garcia DA, et al. (2012) Guidance on the emergent reversal of oral thrombin and factor Xa inhibitors. Am J Hematol 87 Suppl 1:S141-5.
  13. Choudhry NK, Saya UY, Shrank WH, et al. (2012) Cost-related medication underuse: prevalence among hospitalized managed care patients. J Hosp Med 7(2):104-109.