Page 29 - BH-2-3

Brain & Heart Digital tools for stroke and bleeding risk in AF

variables in CHA DS -VASc, resulting in a meaningful blood cells. Among these scores, HAS-BLED is the most
56
2
2
improvement in stroke prediction. 51 widely validated and commonly used score in clinical
Information derived from echocardiography also practice. However, each scoring system has its merits and
provides insights into the development of stroke. Recently, drawbacks, which will be explored in the following parts.
52
left atrial (LA) size has emerged as a useful marker to The HEMORR HAGES score was introduced in 2006
2
characterize stroke risk. The Northern Manhattan Stroke to help identify patients with AF who may benefit from
Study concluded that moderate-to-severe LA enlargement anticoagulation but require closer monitoring while on
was associated with recurrent strokes in patients with anticoagulation therapy. This score was created based on
57
and without AF. The creators of the ABCD score cited the bleeding risk factors identified in the National Registry
53
this finding when adding echocardiographic LA dilation of AF, the same registry from which the CHADS score was
2
17
as a risk factor. In a more recent study, LA deformation derived. HEMORR HAGES assigns 2 points for a history
39
2
characterized by LA strain using a three-beat method had of bleeds and 1 point for hepatic or renal disease, ethanol
a higher predictive value for ischemic stroke compared to abuse, malignancy, age >75 years, reduced platelet count
traditional CHA DS -VASc scoring, although the improved or function, uncontrolled hypertension, anemia, genetic
2
2
performance was attenuated when the global longitudinal factors, excessive fall risk, and stroke. A score of 0 or 1 is
strain was added to the model. 54 interpreted as a low risk, 2 or 3 as an intermediate risk, and
≥4 as a high risk. 57
Artificial intelligence (AI) tools incorporating these
non-discrete measures are being developed to predict Noting that HEMORR HAGES was developed on a
2
the risk of AF in patients. Although validated AI tools historical cohort a decade earlier, and had a high degree of
55
58
specifically for predicting thromboembolic stroke in overlap with risk factors for estimating stroke, Pisters et al.
patients with AF are not yet in clinical use, it is likely that introduced the HAS-BLED score in 2010. HAS-BLED was
in the coming years, AI-augmented risk prediction may developed and validated in the Euro Heart Survey on the
enhance existing validated risk scores and further inform AF cohort. In addition to several risk factors included in
clinical decision-making, especially for patients classified HEMORR HAGES, such as uncontrolled hypertension,
2
as “intermediate risk” by conventional scoring methods. abnormal renal and liver function, stroke history, and
history of prior bleeding, HAS-BLED also includes a labile
4. Scores for estimating bleeding risk in AF international normalized ratio, simultaneous use of drugs
Stroke prevention therapy in patients with AF must balance and alcohol, and qualifies older age as >65 years. Each
the risks of ischemic stroke against the increased risks of factor was assigned 1 point except for abnormal renal
bleeding with treatment. Various scoring systems to assess and liver function and drug/alcohol use, which have a
bleeding risk have been introduced. However, employing maximum of 2 points should both individual factors be
these scores can be challenging given that several present. Scores allow the classification of patients with AF
comorbid conditions and risk factors simultaneously into three risk strata, in which a score of 0 indicates low
elevate both stroke and bleeding risk, including advanced risk, 1 – 2 moderate risk, and ≥3 high risk for bleeding. 58,59
age, hypertension, renal impairment, and a history of In 2011, the ATRIA score for assessing the risk of
stroke. The overlap in variables across scores for estimating hemorrhage associated with warfarin use in patients with
both stroke and bleeding risk is illustrated in Figure 1. AF was introduced. The ATRIA score for risk of bleeding
60
Fortunately, many risk factors for bleeding are reversible or was derived from split-sample testing of a cohort of
controllable, such as alcohol use, elevated blood pressure, patients with non-valvular AF in the Kaiser Permanente
and the use of NSAIDs. Assessment of risk factors specific system of Northern California, which was also used to
to bleeding may better inform interventions to reduce derive the 2013 ATRIA score for assessing the risk of stroke
bleeding risk and recommendations on anticoagulation. in AF. ATRIA included the following factors selected by
33
Major scores for estimating bleeding risk in AF include bootstrapping: severe renal disease (3 points), anemia
HEMORR HAGES (2006), HAS-BLED (2010), ATRIA (3 points), age ≥75 years (2 points), prior bleeding (2
2
(2011), ORBIT (2015), GARFIELD-AF (2017), and DOAC points), and hypertension (1 point). Notably, patients aged
(2023) (Table 2). Most scores are designed to estimate 65 – 74 did not receive a point, hypertension was defined as
the risk of major bleeding at critical sites, including any history of hypertension rather than a specific numeric
intracranial bleeding, retroperitoneal bleeding, intraspinal BP threshold, and concomitant aspirin use was not
bleeding, and pericardial bleeding. Major bleeding can included as a risk factor for bleeding. In the ATRIA cohort,
also be defined by the specific treatments required, such this score demonstrated good discrimination performance
as the transfusion of two or more units of packed red and net reclassification improvement compared to prior

Volume 2 Issue 3 (2024) 6 doi: 10.36922/bh.3068