Meta-analyses: Novel Anticoagulants Halve Intracranial Hemorrhage vs. Warfarin

Download this article's Factoids (Factoid 1Factoid 2) (PDF & PPT for Gold Subscribers)

Novel oral anticoagulants halve the risk of intracranial hemorrhage compared with warfarin in patients with nonvalvular atrial fibrillation (A-fib), according to 2 meta-analyses of major randomized trials.

In a paper published in the December 2013 issue of JAMA: Neurology, investigators led by Saurav Chatterjee, MD, of Brown University (Providence, RI), pooled data from 57,491 A-fib patients in 6 randomized trials comparing a novel oral anticoagulant (1 administering dabigatran, 2 rivaroxaban, and 3 apixaban; n = 31,830) with warfarin or aspirin (n = 25,661). The mean time in therapeutic range for warfarin was 61.2%. There was no significant heterogeneity among the trials.

No Differences Among Individual Agents

Overall, the novel anticoagulants reduced the rate of intracranial hemorrhage by more than half (OR 0.49; 95% CI 0.36-0.65). Risk was reduced by each agent, with Bayesian random-effects analysis showing no differences among the individual medications. The absolute risk was 0.52% for dagibatran (combined 110-mg and 150-mg doses), 0.78% for rivaroxaban, and 0.52% for apixaban vs. 1.24% for warfarin.

Compared with warfarin, the median number needed to treat for prevention of intracranial hemorrhage was:

  • 29.32 for 110-mg dabigatran
  • 34.53 for 150-mg dabigatran
  • 59.11 for rivaroxaban
  • 35.07 for apixaban          

In an exploratory subgroup analysis, the rates of intracranial hemorrhage were reduced by the novel anticoagulants regardless of the bleeding site (OR 0.42; 95% CI 0.18-1.00 for intracerebral, intraparenchymal, or intraventricular vs. OR 0.54; 95% CI 0.34-0.85 for epidural, subdural, or subarachnoid; P = 0.64).

The authors say that although the exact mechanism behind the lower rate of intracranial bleeding with the novel oral anticoagulants is unknown, a probable explanation is that they target a single site in the coagulation cascade compared with multiple sites with warfarin. In addition, in contrast to warfarin, the novel anticoagulants do not affct factor VIIa.

“[O]ur analysis suggests that any of the current available [novel oral anticoagulants] are reasonable choices when risk of [intracranial hemorrhage] is a consideration,” Dr. Chatterjee and colleagues write.

In an accompanying editorial, Robert G. Hart, MD, of McMaster University (Hamilton, Canada), and colleagues say “the importance of the low rate of intracranial hemorrhage associated with the [novel anticoagulants] cannot be overemphasized,” since it lowers the threshold for their use to prevent disabling ischemic strokes.

However, the editorial cautions, the conclusion that the novel anticoagulants are similar should be interpreted with care as “[s]imple comparisons of estimates of intracranial hemorrhage across trials suggest that there may be differences among agents and that the risk appears to be dose dependent.”

Second Meta-analysis Includes All 4 New Agents

In a separate meta-analysis, published online December 4, 2013, ahead of print in the Lancet, Christian T. Ruff, MD, of Brigham and Women’s Hospital (Boston, MA), and colleagues evaluated pooled data from phase 3 trials that randomized 71,683 A-fib patients to a novel anticoagulant (n = 42,411) or warfarin (n = 29,272). The trials included:

  • RE-LY (dabigatran, 110 mg or 150 mg, twice daily)
  • ROCKET AF (rivaroxaban 20 mg, once daily)
  • ARISTOTLE (apixaban 5 mg, twice daily)
  • ENGAGE AF-TIMI 48 (edoxaban [not approved in the United States], 30 mg or 60 mg, once daily)

Compared with warfarin, the novel anticoagulants reduced stroke or systemic embolism (primary composite endpoint) by 19%, driven by a 51% decrease in hemorrhagic stroke; no difference was seen in rates of ischemic stroke. The new drugs also reduced intracranial hemorrhage and mortality, with a nonsignificant decline in major bleeding, but increased GI bleeding (table 1).

Table 1. Efficacy and Safety Outcomes



95% CI

P Value

Stroke or Systemic Embolism



< 0.0001

Ischemic Stroke




Hemorrhagic Stroke



< 0.0001

All-Cause Mortality




Major Bleeding




Intracranial Hemorrhage



< 0.0001

GI Bleeding




The benefit of the novel anticoagulants in reducing the primary endpoint was consistent across all subgroups examined.

Additionally, the efficacy of low-dose regimens was similar to that of warfarin for the primary endpoint, with a trade-off of lower risk of hemorrhagic stroke for higher risk of ischemic stroke. These regimens matched warfarin in effect on GI bleeding.

Not All Agents Are the Same

In an accompanying editorial, Torben Bjerregaard Larsen, MD, of Aalborg University Hospital (Aalborg, Denmark), and Gregory Y. H. Lip, MD, of the University of Birmingham (Birmingham, United Kingdom), write that the meta-analysis “assumes that all the novel oral anticoagulant drugs are the same (which they are not) and work on the basis of a class effect or are broadly equivalent; and that the randomized trials are homogeneous, which again they are not.” Moreover, the meta-analysis “does not really answer the question of which novel oral anticoagulant is best, whether from an efficacy or a safety perspective,” they say.

Nonetheless, study coauthor Michael D. Ezekowitz, MD, PhD, of Thomas Jefferson Medical College (Philadelphia, PA), told TCTMD in an interview, the bottom line is that “there is absolute consistency among all the novel agents [in terms of] substantially lowering intracranial hemorrhage of all forms against warfarin, whether it is very well controlled or less well controlled. [Both studies] agree that all these agents are better than . . . warfarin.

“Now we’re into the . . . phase where we try to tailor our therapy to individual patients,” he continued, which requires translating the trial results into practice while recognizing that the study populations were quite selective.

Although meta-analyses clearly have limitations, he observed, they are valuable because they provide rough directions. Unfortunately, head-to-head comparisons among the agents are unlikely to be conducted, he said, because event rates with the new anticoagulants have been much lower than expected, meaning the trials would need to be enormous and would thus be cost-prohibitive.

According to the Lancet editorial, a “major dilemma of clinical management” is how to predict which newly diagnosed A-fib patients are likely to do well on vitamin K antagonists and thus perhaps reap minimal benefit from use of a novel anticoagulant. In fact, it cites a clinical algorithm, the SAMe-TT2R2 score (Apostolakis S, et al. Chest. 2013;144:1555-1563) developed to guide such decisions.  

However, Dr. Ezekowitz,  was skeptical of this strategy “because my read of the literature is that even at the highest level, time in therapeutic range does not diminish the effectiveness of the novel agents for reducing intracranial bleeds.”

Note: Dr. Ezekowitz is a faculty member of the Cardiovascular Research Foundation, which owns and operates TCTMD. 


1. Chatterjee S, Sardar P, Biondi-Zoccai G, et al. New oral anticoagulants and the risk of intracranial hemorrhage: Traditional and Bayesian meta-analysis and mixed treatment comparison of randomized trials of new oral anticoagulants in atrial fibrillation. JAMA Neurol. 2013;70:1486-1490.

2. Hart RG, Pogue J, Eikelboom JW. Direct-acting oral anticoagulants: The brain gets a break. JAMA Neurol. 2013;70:1483-1484.

3. Ruff CT, Giugliano RP, Braunwald E, et al. Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: A meta-analysis of randomised trials. Lancet. 2013;Epub ahead of print.

4. Larsen TB, Lip GYH. Warfarin or novel oral anticoagulants for atrial fibrillation? Lancet. 2013;Epub ahead of print.



Related Stories:

Meta-analyses: Novel Anticoagulants Halve Intracranial Hemorrhage vs. Warfarin

Novel oral anticoagulants halve the risk of intracranial hemorrhage compared with warfarin in patients with nonvalvular atrial fibrillation (A-fib), according to 2 meta-analyses of major randomized trials
  • Drs. Chatterjee and Hart report no relevant conflicts of interest.
  • Dr. Ruff reports serving as a consultant to and receiving honoraria from Boehringer Ingelheim, Bristol-Myers Squibb, and Daiichi Sankyo.
  • Dr. Larsen reports serving as an investigator and speaker for multiple pharmaceutical companies.
  • Dr. Lip reports serving as a consultant to multiple pharmaceutical companies.
  • Dr. Ezekowitz reports serving as co-principal investigator for the RE-LY and X-ERT trials; receiving consulting fees, lecture fees, and grant support from Aryx Therapeutics and Boehringer-Ingelhiem; consulting fees from Sanofi-Aventis; and lecture fees and grant support from Portola Pharmaceuticals.