ESC Congress FA eBook 2016 - 12

CLINICAL TRIAL HIGHLIGHTS

Figure 1. ICH Greater Following Warfarin Treatment Compared with Rivaroxaban
On treatment + PS match

Intention to treat + PS match
2.0 %

2.0 %

1.5 %

1.0 %

1.0 %
0.5 %

0.5 %

HR, 0.77 (CI, 0.39-1.49)

HR, 0.67 (CI, 0.43-1.03)
0%

0%
0

Years

Rivaroxaban

1.5

2.5

Years

Rivaroxaban

Warfarin

1.5

2.5

Warfarin

Multivariable HR, 0.63 (95% CI, 0.40-0.99)
Reproduced with permission from L Friberg, MD.

There was no difference in any major bleeding
between rivaroxaban and warfarin (adjusted HR, 0.89;
95% CI, 0.73 to 1.10). There was also no difference in urogenital bleeding (adjusted HR, 0.90; 95% CI, 0.56 to 1.44),
or other bleeding (adjusted HR, 0.80; 95% CI, 0.56 to
1.16) regardless of whether the data were analyzed based
on ITT, estimated on treatment, or using a multivariable
analysis on top of the ITT data. There was also no significant increase in gastrointestinal bleeding with rivaroxaban (multivariable HR, 1.16; 95% CI, 0.83 to 1.62).
After multivariable adjustment, there was a significant 37% lower risk for intracranial hemorrhage among
patients in the ITT group treated with rivaroxaban compared with those treated with warfarin (adjusted HR,
0.63; 95% CI, 0.40 to 0.99). The difference observed in the
on-treatment group was not significant (Figure 1).
Prof Friberg concluded there is no difference in bleeding risk between rivaroxaban and extremely well-managed warfarin, except for intracranial bleeds, which are
significantly higher with warfarin.

New GARFIELD Risk Score
Improves Characterization of
Low Risk Patients Compared
With CHA2DS2-VASc
Written by Toni Rizzo

Identification of patients with atrial fibrillation (AF)
who have true low risk for thromboembolic events and
should not receive anticoagulation therapy is an ongoing
challenge. Programs such as the Global Anticoagulant
Registry in the FIELD-Atrial Fibrillation (GARFIELD-AF)

October 2016

have shown that anticoagulant use has increased over
time [Camm AJ et al. Heart. 2016]. The critical question, according to Keith AA Fox, MB, ChB, University
of Edinburgh, Edinburgh, United Kingdom, is whether
patients at low risk are correctly characterized.
With current tools, the risk/benefit balance of anticoagulation is poorly defined in low risk patients. Using
current strategies to define risk, 28% of high risk patients
do not receive anticoagulation therapy, whereas 51% of
very low risk patients do [Camm AJ et al. Heart. 2016].
Factors beyond those in current risk scores appear to
influence prescribing decisions for anticoagulation.
The aim of this study was to compare the performance
of CHA2DS2-VASc with a novel computer-generated
(machine learning) risk model, developed using contemporary data from the GARFIELD-AF registry.
A coalescent regression model that allows joint modeling of all outcomes was developed to determine study
patients' thromboembolic risk. The model was based
on 38,984 patients in GARFIELD-AF from 2010 to 2015,
using outcomes for all-cause mortality, ischemic stroke/
systemic embolism (SE), and hemorrhagic stroke/major
bleed that occurred within 1 year of enrollment in the
registry. The performance of the model was compared
with CHA2DS2-VASc in all patients and in those with low
stroke risk. The results were externally validated with
ORBIT-AF, an independent registry. C statistics (area
under the Receiver Operating Characteristic curve) and
P values were used to assess performance of the new risk
score. A C statistic of 0.5 indicates a 50% chance of the
result being true positive or false positive.
Low-risk patients (CHA2DS2-VASc 0-1 for men and
1-2 for women) made up 20.2% of the total cohort. The
C statistics for the relationship between true positives

medicom-publishers.com/mcr

http://www.medicom-publishers.com/mcr

Table of Contents for the Digital Edition of ESC Congress FA eBook 2016