Summary of evidence-based guideline update: Prevention of stroke in nonvalvular atrial fibrillation: Report of the Guideline Development Subcommittee of the American Academy of Neurology

Objective: To update the 1998 American Academy of Neurology practice parameter on stroke prevention in nonvalvular atrial fibrillation (NVAF). How often do various technologies identify previously undetected NVAF?Which therapies reduce ischemic stroke risk with the least risk of hemorrhage, including intracranial hemorrhage? The complete guideline on which this summary is based is available as an online data supplement to this article. Methods: Systematic literature review; modified Delphi process recommendation formulation. Major conclusions: In patients with recent cryptogenic stroke, cardiac rhythm monitoring probably detects occult NVAF. In patients with NVAF, dabigatran, rivaroxaban, and apixaban are probably at least as effective as warfarin in preventing stroke and have a lower risk of intracranial hemorrhage. Triflusal plus acenocoumarol is likely more effective than acenocoumarol alone in reducing stroke risk. Clopidogrel plus aspirin is probably less effective than warfarin in preventing stroke and has a lower risk of intracranial bleeding. Clopidogrel plus aspirin as compared with aspirin alone probably reduces stroke risk but increases the risk of major hemorrhage. Apixaban is likely more effective than aspirin for decreasing stroke risk and has a bleeding risk similar to that of aspirin. Major recommendations: Clinicians might obtain outpatient cardiac rhythm studies in patients with cryptogenic stroke to identify patients with occult NVAF (Level C) and should routinely offer anticoagulation to patients with NVAF and a history of TIA/stroke (Level B). Specific patient considerations will inform anticoagulant selection in patients with NVAF judged to need anticoagulation. Neurology® 2014;82:716–724 GLOSSARY AAN 5 American Academy of Neurology; AF 5 atrial fibrillation; CI 5 confidence interval; CKD 5 chronic kidney disease; CrCl 5 creatinine clearance; GI 5 gastrointestinal; HR 5 hazard ratio; INR 5 international normalized ratio; NVAF 5 nonvalvular atrial fibrillation; RRR 5 relative risk reduction; RR 5 relative risk. The prevalence of atrial fibrillation (AF) in the United States was estimated to be 3.03 million persons in 2005 and is strongly associated with increasing age. Because AF is a major risk factor for cardioembolic stroke, there is an urgent need to develop strategies for identification of AF and prevention of cardioembolic stroke at all ages. The ischemic stroke rate among patients with AF averages 5% yearly but varies greatly depending on individual clinical characteristics such as age, sex, race/ethnicity, and associated stroke risk factors. History of stroke or TIA identifies those patients with a high stroke risk, averaging 10% yearly. This evidence-based guideline updates a 1998 American Academy of Neurology (AAN) practice parameter on stroke prevention in nonvalvular atrial fibrillation (NVAF). The complete guideline on which this summary is based is available as an online data supplement on the Neurology® Web site at www. neurology.org. This updated guideline reviews the evidence published since 1998 with regard to the detection of NVAF in patients with stroke and new therapies for the prevention of stroke in patients with NVAF, with a focus on 2 questions: 1) For patients with cryptogenic stroke, how often do various technologies identify previously undetected NVAF? 2) For patients with NVAF, which therapies that include From the Department of Neurology (A.C.), SUNY Upstate Medical University, Syracuse, NY; the Department of Stroke and Neurocritical Care (S.R.M.), Hospital of the University of Pennsylvania and the Pennsylvania Hospital, Philadelphia; the Stroke Program (S.C.), Wayne State University School of Medicine, Detroit, MI; the Department of Neurology (C.S.K.), Boston University School of Medicine and Boston Medical Center, Boston, MA; and the Department of Neurology (G.G.), University of Kansas Medical Center, Kansas City, KS. Approved by the Guideline Development Subcommittee on January 12, 2013; by the Practice Committee on April 29, 2013; and by the AANI Board of Directors on October 29, 2013. Go to Neurology.org for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. 716 © 2014 American Academy of Neurology antithrombotic medication, as compared with no therapy or with another therapy, reduce stroke risk and severity with the least risk of hemorrhage? DESCRIPTION OF THE ANALYTIC PROCESS This guideline was developed in accordance with the processes described in the AAN guideline development process manuals (2004, 2011). The panel searched MEDLINE, EMBASE, Cochrane, and Web of Science using appropriate search terms to locate relevant articles published between 1998 and March 2013. The search was restricted to peer-reviewed articles on human subjects written in English. The panel synthesized the evidence and developed conclusions using a modified form of the Grading of Recommendations Assessment, Development and Evaluation process. Evidence synthesis tables are available in appendix e-6 of the complete guideline. The panel formulated practice recommendations on the basis of the strength of evidence systematically reviewed and other factors, including axiomatic principles of care, the magnitude of anticipated health benefits relative to harms, financial burden, availability of interventions, and patient preferences. The panel assigned levels of obligation (A, B, C, U) to the recommendations using a modified Delphi process. ANALYSIS OF EVIDENCE For patients with cryptogenic stroke, how often do various technologies identify previously undetected NVAF? Two Class II and 15 Class III studies were identified that address this question. Figure 1 lists these studies and the associated monitoring techniques and durations involved. Studies were downgraded 1 level if they failed to provide data on a cryptogenic stroke cohort, because some of the patients in noncryptogenic cohorts had known NVAF. The most common technique used to identify NVAF in these studies was Holter monitoring, followed by serial EKG, event loop recorders, inpatient continuous EKG telemetry, outpatient transtelephonic EKG monitoring, and mobile cardiac outpatient telemetry. Several studies described a stepwise approach for NVAF screening that used serial EKGs and Holter monitoring. Monitoring duration varied from 24 hours to 30 days. The proportion of patients identified with NVAF (figure 1) ranged from 0% to 23%. Some of these estimates of NVAF incidence include very brief (e.g., ,30 seconds) episodes, and the future risk of cardioembolic stroke in this setting is uncertain. The average detection rate of all studies was 10.7% (95% confidence interval [CI] 7.9%–14.3%) (weighted average calculated using a random effects model). A meta-regression of studies using continuous monitoring techniques identified a significant increase in NVAF detection with longer monitoring duration (p , 0.0000). Conclusions. In patients with recent cryptogenic stroke, cardiac rhythm monitoring probably detects previously unidentified NVAF at a rate ranging from 0% to 23% (weighted average of 10.7% [95% CI 7.9%–14.3%]) (2 Class II studies, 15 Class III Figure 1 Proportion of patients with ischemic stroke identified with nonvalvular atrial fibrillation, by study Studies sorted by monitoring duration. CI 5 confidence interval; ELR 5 event loop recorder; HM 5 Holter monitoring; inptTele 5 continuous inpatient telemetry; MCOT 5 mobile cardiac outpatient telemetry; phoneEKG 5 outpatient transtelephonic EKG monitoring; sEKG 5 serial EKG; sELR 5 serial event loop recordings. Neurology 82 February 25, 2014 717 studies). The detection rate is probably related to the duration of monitoring. For patients with NVAF, which therapies that include antithrombotic medication, as compared with no therapy or with another therapy, reduce stroke risk and severity with the least risk of hemorrhage? Warfarin, influence of international normalized ratio level. Since the publication of the 1998 practice parameter, 2 Class II studies have evaluated the relationship between international normalized ratio (INR) level at the time of stroke presentation and stroke severity and mortality. Both studies demonstrated that an INR of less than 2 as compared with an INR greater than 2 was associated with an increased risk of disabling stroke (odds ratio 1.9 [95% CI 1.1–3.4]) or death (hazard ratio [HR] for death at 30 days 3.4 [95% CI 1.1–10.1]). Conclusion. In patients with NVAF, anticoagulation that results in an INR of 2.0–3.0 likely reduces the frequency and severity of ischemic stroke as compared with anticoagulation resulting in lower INR levels (2 Class II studies). Antithrombotics compared with warfarin or its derivatives. Our search strategy identified 6 randomized studies (5 Class I studies, 1 Class II study) comparing various antithrombotic regimens with warfarin or its derivatives in patients with NVAF. All studies employed masked or adjudicated outcome assessment. Antithrombotic regimens studied were dabigatran, rivaroxaban, apixaban, fluindione plus aspirin, clopidogrel plus aspirin, and triflusal plus acenocoumarol. Dabigatran is a direct thrombin inhibitor. Rivaroxaban and apixaban are factor Xa inhibitors. Dabigatran, rivaroxaban, and apixaban are administered in fixed doses and do not require regular blood coagulation monitoring. Antithrombotic reversal agents for these drugs are unavailable. Triflusal is an antiplatelet drug structurally related to aspirin that is used in Europe, Latin America, and Southeast Asia (see appendix e-9 of the complete guideline for the relevant countries). Acenocoumarol, a coumarin derivative, is used mostly in European countries. Fluindione is a vitamin K antagonist used in France. Figure 2 summarizes the effects (relative risk reductions [RRRs]) of these antithrombotic regimens as compared with dose-adjusted warfarin for the outcomes o