By Michael H. Crawford, MD

Professor of Medicine, Associate Chief for Education, Division of Cardiology, University of California, San Francisco

Dr. Crawford reports no financial relationships relevant to this field of study.

SYNOPSIS: Using the Delphi method of arriving at a consensus among clinicians concerning for which patients with atrial fibrillation should oral anticoagulants be recommended, the risk of stroke, the risk of hemorrhage, and patient-specific factors emerged. Many of these factors are not included in the guidelines and should be studied further.

SOURCE: King PK, Fosnight SM, Bishop JR. Consensus clinical decision-making factors driving anticoagulation in atrial fibrillation. Am J Cardiol 2019;124:1038-1043.

Surveys have shown that up to 38% of high-risk-for-systemic-emboli patients with atrial fibrillation (AF) are not treated with anticoagulants. To explore the reasons for this, 27 practicing cardiologists, neurologists, internists, and clinical pharmacologists used the modified Delphi method to rate factors important in oral anticoagulation (OAC) decisions. The Delphi method employed three rounds of surveys conducted blindly through electronic means. A list of potential factors for initial evaluation was derived from published literature and submitted to the participants in round 1. Each participant then rated their level of agreement of each factor’s importance for decisions regarding OAC in AF. In round 2, factors with partial consensus (60-75% agreement) and full consensus (more than 75% agreement) were reconsidered and assigned a score of importance from 0-50 points. All factors with a score above the median were reconsidered in round 3 and arranged in rank order.

Sixty-six factors were identified from the literature review. In round 1, 41 failed to meet consensus, and seven new factors were added. Thus, 32 factors were considered in round 2, and 16 did not meet consensus. The 16 factors in round 3 were arranged in rank order: CHA2DS2-VASc score, ischemic stroke history, transient ischemic attack (TIA) history, any major GI bleed less than 12 weeks ago, platelets < 50,000, nontraumatic intracranial hemorrhage > 12 weeks ago, any major bleed < 12 months ago, CHADS2 score, adherence concerns, HAS-BLED score, GI bleed from a peptic ulcer < 12 weeks ago, goals of care, liver disease with international normalized ratio (INR) > 1.4, age, patient values/preferences, and overall prognosis. The authors concluded that several of these factors are not addressed in current guidelines, including hematologic indicators of bleeding risk, previous bleeding episodes by specific type, other risk factors for bleeding, and adherence concerns. Thus, more research needs to be conducted on the clinical implications of these emerging factors and the lack of consensus on other factors found in the literature.


In patients with AF, the decision on whom to prescribe OACs is difficult because the risk of major bleeding is real. Not surprisingly, many AF patients end up on no OAC or on aspirin alone, which is no longer recommended in guidelines because the risk of bleeding outweighs the small potential benefit. Also, many patients are treated with OAC doses that are below those recommended. This study uncovered another issue: the lack of consensus about which factors are important in making the decision to anticoagulate.

Out of 73 factors identified, there was consensus only on 16. Of these 16, the three highest-ranked factors all concerned protection from systemic emboli: CHA2DS2-VASc score, ischemic stroke, and TIA history. CHADS2 score was number 8 on the list, so it has clearly fallen out of favor. Six of the next 13 factors involved avoiding hemorrhage: major GI bleed within 12 weeks, platelet count < 50,000 nontraumatic intracranial hemorrhage more than 12 weeks ago, HAS-BLED score, bleeding peptic ulcer < 12 weeks ago, and liver disease with an INR higher than 1.4. The final factors involved patient-specific issues: adherence concerns, age, goals of care, patient values/preferences, and overall prognosis. For several factors that would seem important, consensus could not be reached: dual antiplatelet therapy need, cognitive impairment, frailty, systolic heart failure, anemia, and transient postoperative AF. Why there was no agreement on these factors could not be determined in this study, but as the authors concluded, these could be good research targets. Of special importance is the patient who requires dual antiplatelet therapy and has AF. Currently, several different approaches to this problem have been studied or research is ongoing. Thus, the lack of consensus here is not surprising. There are strengths to the Delphi method as applied in this study. Since all the grading was conducted online, there was no peer-to-peer influence. Also, there was a broad group of practitioners, all of whom were knowledgeable about AF and OACs. In addition, the Delphi method studies thought processes, not prescribing patterns. Still, there were weaknesses in this investigation. Participants were limited to physicians and clinical pharmacists. No nurse practitioners or physician assistants were included, yet such individuals often are involved in the decision to prescribe OACs. Also, all practitioners were in the United States, so the results may not apply internationally. Finally, with the Delphi method, the more participants included, the better the analysis, but more than 50 is considered impractical. These investigators recruited 103 clinicians: 49 did not respond, 24 declined, and 30 agreed to participate. Three dropped out during round 1, leaving 27. This probably is enough for a robust analysis, as less than 10 is considered inadequate. This study interested me because it pointed out that an emphasis on protecting patients from emboli is the most important concern. However, factors addressing bleeding risk are important, too, and many of these are not things I normally ask about. Patient-specific factors seem to be the least important to clinicians, but perhaps not to patients. In the final analysis, patients may opt out of OAC therapy because of these reasons. There is no grand formula for deciding to prescribe OACs in AF, and many clinical characteristics need to be considered beyond the simple scores.