NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Adam SS, McDuffie JR, Ortel TL, et al. Comparative Effectiveness of Warfarin and Newer Oral Anticoagulants for the Long-Term Prevention and Treatment of Arterial and Venous Thromboembolism [Internet]. Washington (DC): Department of Veterans Affairs; 2012 Apr.

Cover of Comparative Effectiveness of Warfarin and Newer Oral Anticoagulants for the Long-Term Prevention and Treatment of Arterial and Venous Thromboembolism

Comparative Effectiveness of Warfarin and Newer Oral Anticoagulants for the Long-Term Prevention and Treatment of Arterial and Venous Thromboembolism [Internet].

Show details


We identified eight good-quality RCTs comparing newer oral anticoagulants to conventional anticoagulant therapy with warfarin, either alone or in combination with low molecular weight heparin. Of these eight studies, five compared newer oral anticoagulants to warfarin for prevention of stroke in nonvalvular AF. Three studies compared newer oral anticoagulants with a combination of parenteral anticoagulation and warfarin for management of VTE. Overall, newer oral anticoagulants were no worse and were—for some clinical outcomes—superior to adjusted-dose warfarin. However, in the absence of head-to-head comparisons between the newer anticoagulants, our analysis may have failed to detect important differences between drug classes or between individual drugs. Comparative effects on HRQOL and patient experience were not reported. The observational literature on adverse effects is sparse, consisting only of case-reports describing bleeding and thrombotic events. The FDA has issued alerts that it is evaluating reports of serious bleeding with dabigatran, mostly in older adults or those with impaired renal function. Our main findings and the strength of evidence (SOE) for each major outcome are summarized by key question in the next section.


Key Question 1. Chronic Atrial Fibrillation

Table 10 summarizes the findings and SOE for each major outcome. In brief, newer oral anticoagulants were associated with a lower rate of all-cause mortality compared with warfarin (high SOE). Newer oral anticoagulants were also associated with fewer hemorrhagic strokes (moderate SOE). For these outcomes, we estimated the absolute risk difference to be 8 fewer deaths and 4 fewer hemorrhagic strokes for every 1000 patients treated with the newer oral anticoagulants compared with adjusted-dose warfarin over approximately 2 years of treatment. However, VTE-related mortality and ischemic stroke were not significantly lower with newer oral anticoagulants.

Table 10. Summary of the strength of evidence for KQ 1—chronic AF.

Table 10

Summary of the strength of evidence for KQ 1—chronic AF.

For dabigatran, the comparative effects on vascular outcomes were dependent, in part, on the quality of adjusted-dose warfarin treatment. While anticoagulation control in the VHA appears to be at least as good as that found in clinical trials, the ROCKET-AF study had a mean TTR that was worse than typical standards. In the RE-LY study, the advantages of dabigatran were greater at sites with poor INR control than at those with good INR control for all vascular events, nonhemorrhagic events, and mortality. Warfarin and dabigatran showed comparable outcomes in centers with good mean TTR.99

Key Question 2. Venous Thromboembolism

Table 11 summarizes the findings and SOE for each major outcome. In comparison with the chronic AF studies, there are fewer studies and patients enrolled and shorter duration of followup for this population. The summary risk ratio favored newer oral anticoagulants for all-cause mortality, VTE-related mortality, recurrent VTE, and major bleeding, but in each instance the CI included no effect. Overall, these results support the conclusion that newer anticoagulants are no worse than adjusted-dose warfarin for major clinical outcomes.

Table 11. Summary of the strength of evidence for KQ 2—venous thromboembolism.

Table 11

Summary of the strength of evidence for KQ 2—venous thromboembolism.

Key Question 3. Mechanical Heart Valves

We did not identify any published studies that compared newer oral anticoagulants to adjusted-dose warfarin in patients with mechanical heart valves. Current evidence is insufficient to estimate the relative effects of newer anticoagulants compared with warfarin for patients with mechanical heart valves.

Key Question 4. Adverse Effects

The adverse effects of newer oral anticoagulants compared with adjusted-dose warfarin were generally consistent across treatment indications. After excluding the ximelagatran studies, the summary risk ratio for discontinuation due to adverse effects was higher for newer anticoagulants, but this result was not statistically significant. The effects on bleeding rates are complex. Fatal bleeding was significantly lower for newer oral anticoagulants, an effect that was consistent across drug classes. Major bleeding was lower for newer oral anticoagulants, but this effect was not statistically significant and varied significantly across studies. In contrast, gastrointestinal bleeding was increased with newer oral anticoagulants. Gastrointestinal bleeding was significantly increased in patients treated with dabigatran and rivaroxaban compared with warfarin.99 The efflux of dabigatran by p-glycoprotein transporters into the gastrointestinal tract may be a mechanism for this finding.105 Both the clinical trial subgroup analyses and the FDA reports suggest that bleeding risk may be increased in older adults and in those with impaired renal function. Further, the differential bleeding risk may be related to the quality of warfarin anticoagulation.

Another potential adverse effect is myocardial infarction. We found no increased risk when combining results from all studies. However, for dabigatran alone, we found an elevated risk (RR = 1.35) that approached statistical significance. A separate meta-analysis, primarily of short-term trials, found a statistically significant increase in myocardial infarction or acute coronary syndrome (OR 1.33; 95% CI, 1.03 to 1.71).70 Liver dysfunction was substantially higher for ximelagatran, a drug withdrawn from the market due to this adverse effect. Elevated rates of liver dysfunction have not been seen with the other newer oral anticoagulants. The SOE was low for several outcomes because CIs included clinically important differences and there was unexplained variability in treatment effects.

Table 12Summary of findings for KQ 4—adverse effects

OutcomeStrength of EvidenceSummary
Drug discontinuation due to adverse effectsLowAcross all indications, discontinuation due to adverse effects was higher with newer oral anticoagulants (RR 1.23; 95% CI, 0.94 to 1.61), but the 95-percent CI was large and included no effect. In subgroup analysis, rates of discontinuation were higher for dabigatran compared with FXa inhibitors.
A clinically important increase in drug discontinuation compared with warfarin cannot be excluded.
Major bleedingLowAcross all indications, the risk of major bleeding was lower with newer oral anticoagulants (RR 0.86; 95% CI, 0.71 to 1.04), but the 95-percent CI was large and included no effect. A clinically important decrease in major bleeding compared with warfarin cannot be excluded. In December 2011, the FDA issued a notice that it was evaluating reports of serious bleeding with dabigatran.
Fatal bleedingModerateAcross all indications, the risk of fatal bleeding was lower with newer oral anticoagulants (RR 0.59; 95% CI, 0.46 to 0.77). Risk difference was 1 fewer death per 1000 patients.
Gastrointestinal bleedingModerateAcross all indications, the risk of gastrointestinal bleeding was increased with newer oral anticoagulants (RR 1.30; 95% CI, 1.17 to 1.49). Risk difference was 1 additional gastrointestinal bleed per 1000 patients.
Myocardial infarctionLowAcross all indications, the risk of myocardial infarction was not different with newer oral anticoagulants (RR 1.02; 95% CI, 0.76 to 1.39). In a subgroup analysis, the risk was increased with dabigatran (RR 1.35; CI, 0.99 to 1.85) compared with FXa inhibitors (RR 0.86; CI, 0.66 to 1.11); p = 0.03 for between-group comparison.
Liver dysfunctionModerateAcross all indications, the risk of liver dysfunction was not different with newer oral anticoagulants (RR 0.82; 95% CI, 0.61 to 1.11).


Clinicians have used adjusted-dose warfarin to prevent systemic emboli related to chronic AF, recurrent VTE, or mechanical heart valves for decades. The benefits and limitations of warfarin are well known. Adjusted-dose warfarin reduces the risk of stroke by 62 percent in patients with chronic AF, the most common indication for anticoagulation in veterans, compared with a 19-percent reduction with aspirin.74 The primary limitations of warfarin are the variability in anticoagulant effect together with drugñdrug and drugñfood interactions that require frequent laboratory monitoring. A recent VA multicenter trial showed that home warfarin monitoring compared with high-quality conventional monitoring did not affect stroke rate, major bleeding episodes, or mortality rates but did lead to small improvements in patient satisfaction and quality of life.23

Our review shows that the newer oral anticoagulants are a viable option for long-term anticoagulation. DTIs and FXa inhibitors have the advantage of more predictable anticoagulation, fewer drug–drug interactions, and equivalent or better mortality and vascular outcomes compared with warfarin. The data are most robust for chronic AF, with fewer studies evaluating use to prevent recurrent VTE and no studies in patients with mechanical heart valves.

The absolute benefits for clinical outcomes are small. For chronic AF, the number needed to treat compared with warfarin over a 2-year period is 132 to prevent 1 death, 260 to prevent 1 hemorrhagic stroke, and 758 to prevent 1 fatal bleeding episode. Because no studies reported effects on patient experience and HRQOL, effects on these important outcomes are unknown. A recent systematic review74 found that, for most patients, warfarin therapy does not have important negative impacts on quality of life.

Safety and Use of Newer Oral Anticoagulants in VA

For clinicians and policymakers, important questions remain. These include questions about which patients are most likely to benefit and which, if any, of the new drugs are most effective. Patients with higher bleeding risks and markedly impaired renal function were excluded from these studies. Clinicians should also consider the quality of INR monitoring available to their patients. In a prespecified subgroup analysis, Wallentin et al.99 found that the advantage of dabigatran over warfarin in terms of major bleeding rates was evident only at sites with poor-quality anticoagulation (TTR <57.1%), while rates of major bleeding were not significantly different at sites with higher quality anticoagulation. Hence, better INR controlled to similar bleeding rates between both groups. In the VHA, time in treatment exceeds this threshold, but newer oral anticoagulants could have important advantages for individual patients who have difficulty maintaining a therapeutic INR. However, since newer oral anticoagulants are dosed twice daily, compared with once daily dosing of warfarin, better outcomes would not be expected if poor medication adherence were the cause of the subtherapeutic INR. A pragmatic concern related to adherence is the FDA notification that dabigatran may lose potency if placed in pill boxes and that it should be dispensed and stored only in the original bottle or blister package.106

Although newer oral anticoagulants are associated with a lower risk of fatal bleeding compared with warfarin, this advantage may be tempered by the increased risk of gastrointestinal bleeding with dabigatran.70,84,91,99 The FDA is currently evaluating reports of high rates of serious bleeding. The reports of bleeding appear to be concentrated in older adults and those with impaired renal function. Another worrisome finding is elevated rates of myocardial infarction with dabigatran, although the strength of evidence for this finding is low. The higher myocardial infarction rate could be related to the drug specifically, to differences in the patient sample studied, or to the protective effect of warfarin on myocardial infarction.69 Alternatively, increased risk of myocardial infarction maybe due to a rebound thrombin effect after the discontinuation of dabigatran, a DTI.105 VA should carefully consider the potential benefits and harms, along with patients at higher risk for adverse effects when establishing eligibility criteria for newer oral anticoagulants.

Clinicians may wonder whether the benefits of newer oral anticoagulants observed in chronic AF will extend to those patients with mechanical heart valves. While this is possible, we caution against extrapolating these data since the INR target for patients with mechanical valves is higher and the dosing may differ. A Phase II trial is currently underway comparing three different doses of dabigatran.


The 2011 American College of Cardiology Guideline update for the management of AF was published before the studies evaluating rivaroxaban and apixaban were published. It recommends dabigatran as a useful alternative to warfarin in patients with chronic nonvalvular AF who do not have severe renal failure or advanced liver disease.9,107 This guideline also noted that patients already taking warfarin with excellent INR control may have little to gain by switching to dabigatran. The more recent American College of Chest Physicians guidelines recommend dabigatran 150 mg for prevention of stroke in AF over the use of adjusted-dose vitamin K antagonists.25 Both the nonprofit QuarterWatch and other groups have raised concern or made recommendations for dosing adjusted to age or renal function. The European Society of Cardiology recommends dabigatran at a dose of 150 mg be used in patients with a low risk of bleeding, while the lower dose of 110 mg is reserved for those with a high risk of bleeding.108 In Canada, dabigatran is approved for the prevention of stroke in AF, and dabigatran 110 mg twice daily is recommended for elderly patients 80 years of age or older or those at a high risk of bleeding.109 In the United States, the FDA has only approved the 150 mg dose and recommends a dose of 75 mg twice daily for patients with CrCl of 15 to 30 mL/min.110

Cost and Cost-Effectiveness

An important disadvantage of the newer oral anticoagulants is their higher drug acquisition costs. The cost-effectiveness of dabigatran compared with warfarin for stroke prophylaxis has been evaluated in three recent publications.93,94,111 Each of these analyses found dabigatran to be cost-effective. However, the studies varied in the factors affecting cost-effectiveness, including drug costs used in the analyses, assumptions about the adequacy of warfarin anticoagulation, and the baseline risk of bleeding or stroke. Depending on the study, cost-effectiveness increased with lower drug costs for the newer oral agents, worse INR control, and higher baseline risk of bleeding or stroke. However, none of these analyses considered the possible expansion in the pool of patients who might be offered and choose chronic anticoagulation with newer agents. An analysis of Medicare beneficiaries showed that only two-thirds of patients with chronic AF who were ideal candidates for anticoagulation were discharged on warfarin. Although an expansion in the indicated use of anticoagulation would be beneficial clinically, it would increase health care costs since these drugs have been shown to be cost-effective, not cost-saving. In an era where health systems and individuals are considering costs ever more carefully, a budget impact analysis would be useful to VA policymakers. Policymakers will have to consider how best to meet the needs of patients while considering health care value. A study by Rose et al. has made the business case for quality improvement programs to improve adjusted-dose warfarin treatment as another viable alternative.100


Our study has a number of strengths, including a protocol-driven review, a comprehensive search, careful quality assessment, and rigorous quantitative synthesis methods. Our study, and the literature, also has limitations. An important limitation is the lack of head-to-head comparisons of the newer oral anticoagulants and an inability to examine the comparative effectiveness across classes (DTIs versus FXa inhibitors) or within class. As the literature grows, subgroup analyses or a network meta-analysis that includes studies comparing warfarin with placebo or aspirin might better address this question—but this comparison was beyond the scope or our review. Based on currently available data, important differences in efficacy or frequency of adverse effects could be present but undetected. A limitation of the literature is the relatively short-term experience with these drugs. It is possible that additional adverse effects may emerge with more widespread and longer duration use.


We used the framework recommended by Robinson et al.112 to identify gaps in evidence and classify why these gaps exist (Table 13).

Table 13. Evidence gaps and future research.

Table 13

Evidence gaps and future research.

PubReader format: click here to try


Other titles in this collection

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...