Dabigatran is an oral direct thrombin inhibitor, touted as non-inferior to Warfarin (though there is some debate about its safety (Yao, Abraham et al. 2016, Hernandez, Baik et al. 2018). It is now routinely prescribed for the treatment of non-valvular atrial fibrillation, venous thromboembolism (VTE) and post-surgical prophylaxis. Dabigatran blocks the last stages of the coagulation cascade, specifically the cleavage of fibrinogen into fibrin, activation of platelets, and stabilization of forming clots. Previously there were no reversal agents available for dabigatran and monitoring of the drug is considered unreliable using standard lab tests. The drug is marketed in the US as Pradaxa, manufactured by Boehringer Ingelheim who subsequently designed and marketed the only available reversal agent for dabigatran, a monoclonal antigen-binding fragment (“Fab”) Idarucizumab, approved in 2015 by the FDA. Initial data from the only clinical study of the reversal agent was released in 2015 (Pollack, Reilly et al. 2015) and this study now represents the full analysis of a cohort of patients given Idarucizumab for the reversal of dabigatran in “uncontrollable or life-threatening” bleeding.

Clinical Question

Does 5g of intravenous idarucizumab reverse the anticoagulant effect of dabigatran in patients with uncontrolled bleeding or who are about to undergo an urgent procedure?


Adults ≥ 18 years receiving dabigatran with either life-threatening bleeding “judged by the treating clinician to require rapid anticoagulant reversal” (group A) or about to undergo surgery or other invasive procedures that “could not be delayed for at least 8 hours and for which normal hemostasis was required” (group B). The study included patients from 173 hospitals (out of almost 400 participating), enrolled over 2 years time.


All patients received 5g of Idarucizumab as two 50 ml vials, each containing 2.5g no more than 15 minutes apart.




Primary: maximum percentage reversal of the anticoagulant effect of dabigatran, determined at any point from the end of the first idarucizumab infusion until 4 hours after the end of the second infusion, as determined by measurements of diluted thrombin time or the ecarin clotting time.
Group A: Extent of bleeding and hemodynamic stability.
Group B: Peri-procedural hemostasis as classified by the clinician as normal or as mildly, moderately, or severely abnormal.


Prospective non-blinded multi-center observational trial


Group A: Absence of clinical signs of active major bleeding or hemodynamic instability, or the presence of only minor bleeding (e.g. epistaxis, hematuria) that could be managed with standard supportive care.
Group B: Procedures that are elective or for which the risk of uncontrolled or unmanageable bleeding is low.

Primary Results

  • 503 patients were enrolled
    • Group A: 301 patients
    • Group B: 202 patients
  • 95% of patients were taking dabigatran for atrial fibrillation

Critical Findings

  • Primary outcome: median maximum percentage reversal within 4 hours after the administration of idarucizumab was 100% (95% CI, 100 to 100).
  • Secondary outcomes
    • Group A
      • Of 301 patients in this arm, 98 had intracranial bleeding for which cessation was not even assessed, 67 “could not be determined” and 2 stopped spontaneously.
      • Of the remaining 134 patients, the study reports a “median time to hemostasis after administration of idarucizumab” as 2.5 hours.
    • Group B
      • 197 patients underwent procedures for which peri-procedural hemostasis was assessed as normal in 184 patients (93.4%), mildly abnormal in 10 (5.1%), and moderately abnormal in 3 (1.5%).
  • Safety outcomes:
    • Rate of thrombotic events was 4.8% at 30 days and 6.8% at 90 days.
    • 30-day mortality rate was 16.4% among patients with intracranial hemorrhage, 11.1% among patients with gastrointestinal bleeding, and 12.7% among patients with bleeding at other sites.


  • Study was multicenter
  • Attempt was made to look at clinical outcomes


  • Study has no control group which fundamentally limits the ability to draw any conclusions about the clinical outcomes of this study
    • Of the 134 bleeding patients in whom they were able to assess hemostasis, less than half of the original group A, they report a “median time to hemostasis” as 2.5 hours but without a control arm we have no idea if this is any different than patients who would receive “usual” care
    • In group B, surgeons made a subjective assessment of hemostasis but this assessment was unblinded and thus susceptible to bias
    • Without a control group, reported safety outcomes and mortality are of unclear significance
  • The study is un-blinded and was controlled by the drug manufacturer Boehringer Ingelheim which creates ample room for bias
    • In group A the assessment of hemostasis was largely based off of surrogate endpoints such as additional use of blood products, use of hemodialysis, etc. so not only is it unclear how they assessed when hemostasis had been “achieved” but the clinicians making these decisions were clearly aware that the patient had received a drug they were told would reverse the patient’s anti-coagulant which certainly could have affected their decisions about treatment.
    • Similarly, in group B, surgeons making a qualitative judgement were not blinded and in fact, we can conclude that they believed the drug would work even before they made their observations since the baseline assumption was that they would not even be willing to operate on patients who don’t receive the drug.
    • The risk of bias affects all aspects of the study including patient selection. 369 sites were initiated into this trial but the 503 patients came from just 173 hospitals which amounts to only about 1.5 patients per hospital per year. We are not given any explanation as to how these patients were chosen or if there were patients who were not even considered for enrollment.
  • Study uses surrogate end-points instead of patient-relevant ones. It is incorrect to assume that improvement of laboratory values and time to hemostasis equate to better patient outcomes. These patients are usually critically-ill with potentially altered physiology therefore it is not safe to draw conclusions based only on this proxy data.

Author's Conclusions

“In emergency situations, idarucizumab rapidly, durably, and safely reversed the anticoagulant effect of dabigatran.”

Our Conclusions

Idarucizumab effectively normalizes laboratory values that are used as surrogate markers for the anticoagulant effect of dabigatran. Despite the biologic plausibility of benefit, the clinical value and safety risks of the treatment cannot be conclusively determined from this study.

Potential Impact To Current Practice

Approval by the FDA of idarucizumab in 2015 makes any decision not to use the drug (when available) problematic in clinical practice.

Bottom Line

  • The degree of benefit—or harm—of idarucizumab for the treatment of life-threatening bleeding in patients taking dabigatran is not known.
  • Not mentioned in this paper is that the initial interim analysis of this study which included 90 patients reported a median time of 11.4 hours (Pollack, Reilly et al. 2015) when patients with intracranial hemorrhage were included. As pointed out by Drs. Radecki and DeLoughery (Radecki 2017), it seems the the authors managed to report a lower value of time to hemostasis by systematically removing patients in their cohort from this analysis.
    • In subsequent correspondence about the paper, the study authors acknowledge that exclusion of patients with intracranial hemorrhage “limits the generalizability of the findings to this subgroup” (Pollack, Reilly et al. 2017).
  • The manufacturers of the drug (and the FDA who oversaw its accelerated approval process) stated separately of the decision not to include a control group that it would be “unethical to withhold a reversal agent from a patient who may benefit from it” (Pollack, Reilly et al. 2015). However this would only be true, of course, if we knew that the drug would benefit the patient—which we now cannot ascertain from this data. Since we know that normalization of laboratory values or positive outcomes in animal studies do not necessarily correlate with benefit to our patients, this rationale is simply flawed.
  • At an estimated cost of $3500 per vial and its indication for use in critically-ill patients, not performing a true randomized-controlled trial should be considered un-ethical.

Read More


  • Hernandez, I., S. H. Baik, A. Piñera and Y. Zhang (2018). “Risk of Bleeding With Dabigatran in Atrial Fibrillation.” JAMA Internal Medicine 175(1): 18-24. PMID: 25365537
  • Pollack, C. V., Jr., P. A. Reilly, R. Bernstein, R. Dubiel, J. Eikelboom, S. Glund, M. V. Huisman, E. Hylek, C. W. Kam, P. W. Kamphuisen, J. Kreuzer, J. H. Levy, F. Sellke, J. Stangier, T. Steiner, B. Wang and J. I. Weitz (2015). “Design and rationale for RE-VERSE AD: A phase 3 study of idarucizumab, a specific reversal agent for dabigatran.” Thrombosis and Haemostasis 114(1): 198-205. PMID: 26020620
  • Pollack, C. V. J., P. A. Reilly, J. Eikelboom, S. Glund, P. Verhamme, R. A. Bernstein, R. Dubiel, M. V. Huisman, E. M. Hylek, P. W. Kamphuisen, J. Kreuzer, J. H. Levy, F. W. Sellke, J. Stangier, T. Steiner, B. Wang, C.-W. Kam and J. I. Weitz (2015). “Idarucizumab for Dabigatran Reversal.” PMID: 26095746
  • Pollack, J., Charles V., P. A. Reilly and J. I. Weitz (2017). Response to Correspondance on Idarucizumab for Dabigatran Reversal — Full Cohort Analysis. New England Journal of Medicine.
  • Radecki, R. D., Thomas G. (2017). Correspondance on Idarucizumab for Dabigatran Reversal — Full Cohort Analysis. New England Journal of Medicine.
  • Yao, X., N. S. Abraham, L. R. Sangaralingham, M. F. Bellolio, R. D. McBane, N. D. Shah and P. A. Noseworthy (2016). “Effectiveness and Safety of Dabigatran, Rivaroxaban, and Apixaban Versus Warfarin in Nonvalvular Atrial Fibrillation.” PMID: 27412905