Please login or register to print this page.

News & Views

New risk scoring systems better identify AF patients for anticoagulation


While existing risk scores help identify patients with atrial fibrillation (AF) who warrant anticoagulation and monitoring, none are able to predict those who are likely to have adverse effects with warfarin and thus likely to benefit from non-vitamin K antagonist oral anticoagulation (NOAC) therapy. The ENGAGE-AF TIMI risk score addressed this issue, using data from 21,105 patients from the trial with CHADS2 ≥2.1

The score included demographics (aged 66-75 or ≥75 years, ethnicity, male sex), ejection fraction, baseline AF, prior ischaemic stroke or myocardial infarction (MI), carotid artery disease, diabetes, haemoglobin <13 g/dL and creatinine ≥110 umol/l with a maximum score of 17 (low risk 0-6; intermediate risk 7-9; high risk ≥10) for a net outcome on warfarin in VKA naïve patients.

Using this score, edoxaban led to improved net clinical outcome versus warfarin in intermediate risk (absolute risk reduction 7.7%, p<0.001) and high-risk patients (absolute risk reduction 20.8%, p=0.004), but was comparable to warfarin in low risk AF patients (absolute risk reduction 0.1%, p=0.9). Thus, the ENGAGE-AF TIMI risk score could help guide in the selection of NOAC or warfarin for treatment-naïve AF patients.


While the use of oral anticoagulant (OAC) therapy has increased over time, the key question is whether this increase has been in the right patients. New data from the GARFIELD-AF registry suggest that this is not the case, as 51% of very low-risk patients (CHA2DS2-VASc score =0) were on OAC therapy, implying additional factors influence OAC prescribing.2

To address this challenge, the Global Anticoagulant Registry in the FIELD – Atrial Fibrillation (GARFIELD-AF) score, a novel computer-generated machine learning risk model score, was developed using data from 38,984 patients enrolled in GARFIELD-AF between March 2010 and July 2015. In a presentation by Professor Keith AA Fox (University of Edinburgh, UK), the GARFIELD‑AF Score was shown to be superior to the CHA2DS2-VASc score in predicting ischaemic stroke/systemic embolism or haemorrhagic stroke/major bleeding, both in all patients and those at low risk.14 These data indicate that the GARFIELD-AF score has potential to help clinicians assess the appropriateness of OAC therapy in low risk patients. Currently, a simplified GARFIELD-AF Score, validated using data from ORBIT-AF, is under development for web-based and mobile device applications. Further information is available from


1. Fanola CL, Giugliano RP, Trevisan M et al. A novel risk prediction score in atrial fibrillation for a net clinical outcome from the ENGAGE AF-TIMI 48

randomized clinical trial. Abstract 18767. Eur Heart J 2016;37(Abstract supplement): 393–4.

2. Fox KAA. Identifying patients with atrial fibrillation and “truly low” thromboembolic risk who are poorly characterized by CHA2DS2-VASc: Superior performance of a novel machine learning tool in GARFIELD-AF. Abstract 2950. Eur Heart J 2016;37(Abstract supplement): xi–xviii.

Published on: October 19, 2016

Members Area

Log in or Register now.



Subscribe to our RSS feed


Sign up for our regular email newsletters & be the first to know about fresh articles and site updates.


    None Found


  • ArrhythmiaAlliance
  • Stars
  • Anticoagulation Europe
  • Atrial Fibrillation Association

You are not logged in

You need to be a member to print this page.
Sign up for free membership, or log in.

You are not logged in

You need to be a member to download PDF's.
Sign up for free membership, or log in.