Home » Uncategorized » Rivaroxaban or Warfarin in Stable Coronary Artery Disease – Should the COMPASS Study Lead Us Back to the Future?

Rivaroxaban or Warfarin in Stable Coronary Artery Disease – Should the COMPASS Study Lead Us Back to the Future?

Henry I. Bussey, Pharm.D.


The recently published (online) COMPASS trial [1] of rivaroxaban (brand name Xarelto) with or without aspirin in patients with stable coronary artery disease (CAD) and peripheral vascular disease has been touted as a practice-changing break-through study that supports adding rivaroxaban to the therapy of such patients.  Approximately 90% of enrolled patients had CAD. Four key questions to be addressed are listed below.  The COMPASS trial was a double blind, double dummy randomized trial that compared the three therapies: aspirin alone, aspirin + rivaroxaban 2.5 mg twice a day, and rivaroxaban 5 mg twice a day (without aspirin).  Because the 5 mg rivaroxaban group did not show clear benefit and had a higher bleeding rate than the other 2 arms, this discussion will focus only on the aspirin alone and the aspirin + 2.5 mg rivaroxaban twice a day groups.

  • Exactly how beneficial is the addition of rivaroxaban to aspirin?
    •  Answer: One stroke, heart attack, or cardiovascular death prevented for every 154 patients treated for a year.
  • What is the risk of adding rivaroxaban?
    • Answer: One additional major hemorrhage for every 166 patients treated for a year.
  • Why was a vitamin K antagonist (VKA) such as warfarin not tested in this trial ?
    • Perhaps the safety and efficacy of warfarin in coronary artery disease has been forgotten?
    • The pharmaceutical industry lacks interest in funding research with warfarin.
    • What if warfarin outcomes were superior to rivaroxaban?
  • What might be the impact of warfarin if well managed with frequent INR self-testing?
    • Projected to double the safety and efficacy of warfarin reported in previous studies (based on data from atrial fibrillation data bases).
    • Might be significantly more effective and safe than rivaroxaban.
    • Reduced time, hassle, and expense of warfarin management.

General Considerations: The COMPASS study has 3 “red flag” issues that raise concerns from a study design and execution perspective:

  • Size of study: As a general rule, the smaller the anticipated impact of a given therapy, the larger the study population will need to be in order to find a statistically significant difference.  For that reason, I am usually skeptical of the clinical benefit of any study that requires more than 10,000 patients to show a statistically significant difference. The COMPASS study enrolled 27,395 patients.
  • Composite endpoint:  Combining several clinical events into a single primary outcome (such as stroke, heart attack, and cardiovascular death) can be another indicator of trying to make a small clinical difference be statistically significant; but in this case, this is the same 3-component composite that has been used in previous post CAD studies and each component is very important.
  • Geographic and ethnic considerations:  Individuals living in different parts of the world, those living in different cultures, and those with different genetic predispositions may have different outcomes of treatment.  Therefore, one should always consider whether the patients studied are comparable to one’s own patients.  In COMPASS, patients were enrolled from 5 geographic regions around the world and each region showed a trend toward benefit of rivaroxaban + aspirin.  However, only 14% of patients came from North America sites and the impact of rivaroxaban was not statistically in this sub- of group of almost 4,000 patients.  The impact of rivaroxaban was statistically significant in 2 of the 5 regions (Eastern Europe and Asia/Pacific) which together enrolled almost 9,000 patients (32% of all patients).  One has to question whether the impact of rivaroxaban + aspirin would have been statistically significant without including the patients from Eastern Europe and Asia/Pacific area who appeared to benefit more than did those from other regions.

Exactly how beneficial was rivaroxaban?  COMPASS compared rivaroxaban 2.5 mg twice a day + 100 mg of aspirin to rivaroxaban 5 mg twice a day (and no aspirin) to aspirin 100 mg daily.  The primary outcome of stroke + heart attack + cardiovascular death was reduced by 24% in the group that got 2.5 mg of rivaroxaban + aspirin twice a day!!  In the group getting 5 mg of rivaroxaban twice a day (without aspirin), the primary outcome was reduced by 9%, but that difference was not statistically significant even though each group had more than 9,000 patients. But what does the 24% reduction with the 2.5 mg dose actually mean?  The event rates in the two groups were 5.4% with aspirin alone vs. 4.1% with rivaroxaban 2.5 mg twice a day + aspirin over a mean follow-up of 23 months.  Those event rates (over 23 months) would equal annualized event rates of approximately 2.7% with aspirin alone vs. 2.05% with rivaroxaban + aspirin; a difference of 0.65% per year [see Table 1].  That means that one would need to treat 154 patients for a year in order to prevent one patient from experiencing one of the composite events.   While preventing one stroke, heart attack, or cardiovascular death per year is certainly a valuable goal, one must consider the risks of such therapy.  The cost of the additional therapy for all 154 patients should also be factored in but the cost will not be considered here because, as far as I know, the 2.5 mg tablet is not yet available.

What is the risk of adding rivaroxaban?

The major risk of adding rivaroxaban to aspirin is bleeding.  The rate of major bleeding with aspirin was 1.9% while the rate of major bleeding with 2.5 mg of rivaroxaban + aspirin was 3.1%.  Although that is a 63% relative increase, the absolute annualized difference is only 0.6% [see Table 1].  The number needed to treat of 154 vs the number needed to harm of 166 indicate that approximately 9 additional major hemorrhages would occur for every 10 composite events prevented.

Table 1. Annualized Event Rates per 100 patients from the COMPASS Trial


Aspirin + R 2.5 mg Diff. NNT


Primary Outcome


2.05 0.65


Major hemorrhage


1.55 0.60


Primary outcome = composite of stroke, heart attack, and cardiovascular death.  NNT = number of individuals who would need to be treated for a year to prevent one event, NNH = number of individuals who would need to be treated for a year in order to cause one additional major hemorrhage.

Why was a VKA such as warfarin not included?  For the last several years, the focus of clinical literature (and pharmaceutical advertising) in the area of anticoagulation has been on the newer DOACs (Direct-acting Oral AntiCoagulants) such as rivaroxaban.  But slightly earlier data has documented the merits of warfarin alone or in combination with aspirin in patients with a recent acute CAD event.  Anand and Yusuf provided a nice review of this issue in 2003 and pointed out the importance of achieving an adequate International Normalized Ratio ( INR).[2]  Studies that targeted an INR less than 2.0 did not show benefit with VKA therapy, but benefit was seen with an INR target of 2 to 2.5 + aspirin or an INR of approximately 3 to 4 without aspirin.  The COMPASS trial, however, enrolled patients with stable CAD and I am not familiar with any recent studies that evaluated warfarin in similar patients.  The COMPASS trial was based on the results of the ATLAS trial[3] which was conducted in patients with a recent acute CAD event, so the following discussion will compare the ATLAS results with a couple of example studies of warfarin +/- aspirin in patients recent acute CAD events.  The ATLAS trial randomized 15,526 patients to rivaroxaban 2.5 mg twice a day, rivaroxaban 5 mg twice a day, or placebo; all patients received aspirin and/or other antiplatelet therapy.  The primary endpoint was stroke, heart attack, or cardiovascular death.  The WARIS II [4] and ASPECT 2 [5] trials were somewhat similar studies that randomized patients with acute CAD events to aspirin alone, aspirin + warfarin with a target INR of 2 to 2.5, or warfarin alone at an INR of approximately 3 to 4.  Based on NNT and NNH, it would appear that the warfarin regimens in WARS II and ASPECT 2 were at least comparable in safety and efficacy to rivaroxaban in the ATLAS trial. [see Table 2].

Table 2: Outcomes in Acute CAD with Rivaroxaban or Warfarin +/- Aspirin

ATLAS: Aspirin +/- Rive 2.5 BID or 5 mg BID (n = 15,526)


Aspirin + R 2.5 mg  Aspirin + R 5 mg Diff. v. 2.5/5 R NNT 2.5/5


Primary Outcome


9.1 8.8 1.6/1.9 63/53
Major Hem.


2.1 2.1 1.5


WARIS II: Aspirin +/- VKA INR 2-2.5 or VKA 3-4


Aspirin+ VKA INR 2-2.5 VKA 3 – 4 Diff vs INR 2-2.5/3-4 NNT at INR 2-2.5/3-4
Primary Outcome


3.67 4.21 1.49/0.95


Major Hem.


0.62 0.62 0.45


*ASPECT 2: Aspirin +/- VKA INR 2-2.5 or VKA 3-4

Primary Outcome

9.2 4.8 5.2 4.4/4.0 23/25
Major Hem.


2.0 1.0 1/0


CAD = coronary artery disease (defined differently in the different studies), VKA = vitamin K antagonist (such as warfarin, brand name Coumadin). R = rivaroxaban, Hem = hemorrhage, INR = International Normalized Ratio which is the blood test used to measure the intensity of the anticoagulant effect of VKAs. NNT = the estimated number of individuals who would need to be treated for one year to prevent one primary event.  NNH = the estimated number of individuals who would need to be treated for one year in order to cause one additional major hemorrhage.

*Median (rather than mean) length of follow-up of 12 months was reported in the ASPECT study but it was presumed that the mean follow-up was also 12 months.

What might be the impact of warfarin if well managed with frequent INR self-testing?  There have been two major problems with warfarin management – (1) the difficulty of keeping the INR in the optimal range and (2) the time, hassle and costs of trying to achieve optimal INR control.  Several small studies [6 – 13] have demonstrated that these two obstacles to warfarin use can be greatly reduced by combining INR self-testing with online automated management.  In our own small study of self-testing and online management [11], we were able to maintain the INR in range more than 80% of the time for a group of patients who previously had a mean time in range of 56%.  Such improved INR control was achieved (and maintained) with a minimal amount of time and effort on the part of the clinician and the patient.  Two large data sets in patients with atrial fibrillation have defined the relationship between improved INR time in range and major clinical events and mortality.[14, 15]  Based on the relationship defined in those two reports, the improved INR control with INR self-testing and online management is projected to reduce major clinical events by more than 80% and mortality by more than 40% [see Figure].  If patients with an acute CAD event can derive benefit similar to that projected for better INR control in patients with atrial fibrillation, then the VKA-related event rates in Table 2 would be reduced by more than 50% making well-managed VKA therapy superior to rivaroxaban.


Conclusion:  The addition of rivaroxaban 2.5 mg twice a day plus low dose aspirin can reduce the incidence of the composite endpoint of stroke, heart attack, and death in patients with stable CAD.  But this benefit is small (less than 1%/yr.) and off-set by a similar increased rate of major hemorrhage (also less than 1%/yr.).  How VKA (warfarin) therapy (with or without aspirin) may alter outcomes in stable CAD patients is not known.  But data from different studies (with the acknowledged hazards of cross-study comparisons) in patients with a recent acute CAD event suggest that warfarin alone or warfarin plus low dose aspirin may be at least as safe and effective as rivaroxaban plus antiplatelet therapy.  INR self-testing and online management, which has the potential to transform INR control and VKA management efficiency, may provide an avenue to improve outcomes with warfarin therapy beyond that which can be achieved with rivaroxaban and other DOACs.  There is a critical need to compare such optimal management of warfarin to the DOACs in patients with stable CAD, those with a recent acute CAD event, atrial fibrillation, and other indications.


  1. Eikelboom JW, et al. Rivaroxaban with or without aspirin in stable cardiovascular disease. N Engl J Med 2017 DOI: 10.1056/NEJMoa1709118    (http://dx.doi.org/10.1056/NEJMoa1709118).
  2. Anand SS, Yusuf S. Oral anticoagulants in patients with coronary artery disease. J Am Coll Cardiol 2003; 41:62S-69S. DOI: 10.1016/S0735-1097(02)02776-6
  3. Mega JL, et al. Rivaroxaban in patients with a recent acute coronary syndrome. N Engl J Med 2012; 366:9-19.  DOI: 10.1056/NEJMoa1112277.
  4. Hurlen M, et al. Warfarin, aspirin, or both after myocardial infarction. N Engl J Med 2002; 347:969-974.
  5. Van Es RF, et al. Aspirin and coumadin after acute coronary syndromes (the ASPECT-2 study): a randomised controlled trial. Lancet 2002; 360:109-113.
  6. O’Shea SI, et al. Direct-to-patient expert system and home INR monitoring improves control of oral anticoagulation. J Thromb Thrombolysis 2008; 26(1): 14-21.
  7. Ryan F,  Byrne S, O’Shea S, et al.  Randomized controlled trial of supervised patient self-testing of warfarin therapy using an internet-based expert system. J Thromb Haemost 2009; 7:1284-1290.
  8. Ferrando F,  Mira Y, Contreras MT, et al. Implementation of SintromacWeb(R), a new internet-based tool for oral anticoagulation therapy telecontrol: Study on system consistency and patient satisfaction. Thromb Haemost 2010; 103:1091–1101.
  9. Harper PL,  Pollock D.  Improved anticoagulant control in patients using home international normalized ratio testing and decision support provided through the Internet. Internal Medicine Journal 2011; 41:332-7.
  10. Verret L,  Couturier J, Rozon A, et al. Impact of a pharmacist-led warfarin self-management program on quality of life and anticoagulation control: a randomized trial. Pharmacotherapy 2012; 32:871-879.
  11. Bussey HI,  Bussey M, Bussey-Smith KL, et al. Evaluation of warfarin management with international normalized ratio self-testing and online remote monitoring and management plus low-dose vitamin k with genomic considerations: a pilot study. Pharmacotherapy 2013;33:1136-46.
  12. Bereznicki LR,  Jackson SL, Peterson GM, et al. Supervised patient self-testing of warfarin therapy using an online system. J Med Internet Res 2013; 15(7):e138.
  13. Harper P, et al. Evaluation of a community pharmacy anticoagulation management service utilizing point-of-care testing and online computerized decision support (Abstract ATT07). J Thromb Haemostas 2013; 11(S3):11-12.
  14. Gallagher AM, et al. Risks of stroke and mortality associated with suboptimal anticoagulation in atrial fibrillation patients. Thromb Haemost 2011; 106:968-977. doi: 10.1160/TH11-05-0353.
  15. Wan Y, et al. Anticoagulation control and prediction of adverse events in patients with atrial fibrillation: A systematic review. Circ Cardiovasc Qual Outcomes. 2008; 1:84-91. Doe: 10.1161/CIRCOUTCOMES.108.796185.

1 Comment

  1. Colucci, Vince says:


    Great points as always. As I recall, the initial ATLAS trial with riva was stopped at one of the interim analyses because of excessive bleeding in the rivaroxaban group; same with the CAD trial with apixaban.

    The same point can be made (as I usually try to do at my institution as an educational point in journal clubs, etc.) with the AFIB trials. The absolute diff in (for instance) major bleeding in the ARISTOTLE trial was about .3% in 18K patients, so cranking up the ‘N’ to extract a small, statistical difference. So 1 in 333pts over 2.5 years in a relatively healthy cohort when compared to warfarin that was at best mediocre in control to prevent one major bleed….ICH notwithstanding of course. Economically not a good deal either if you work those numbers out… I calculated about $1-1.5 to prevent that bleed. Further, in flipping that….this says that, given our major bleeding rate of ~1.5% in our ACC clinic in a population that is almost exponentially more ill and co-morbid than the ARISTOTLE or ROCKET or RELY trials, if we switched everyone to apix or riva, we accept a de facto increase of 100% bleeding. I suspect that out of the 1100 patients in my clinic…. 10% would have been eligible. The Xa inhibitors are good drugs (not great drugs) but without a way to adjust and optimize they lose signficant clinical utility.

    vince colucci

    missoula, mt



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: