Page 30 - BH-2-3

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

Figure 1. Common variables used in scores for estimating the risk of stroke and bleeding in atrial fibrillation (AF). Shades of blue indicate that the variable
is included in stroke risk scores, such as CHADS2 (2001), Framingham (2003), CHA2DS2-VASc (2010), ATRIA (2013), ABC (2016), GARFIELD-AF
(2017), or IMRS-VASc (2019). Shades of red indicate that the variable is included in bleeding risk scores, such as HEMORR HAGES (2006), HAS-BLED
2
(2010), ATRIA (2011), ORBIT (2015), GARFIELD-AF (2017), or DOAC (2023). The omission of a variable from a score is denoted as a blank box with
a slash. Note: Please refer to Tables 1 and 2 for a full list of variables for each score, including the specific definitions of each variable. Source: Icons were
created using BioRender.
Abbreviations: INR: International normalized ratio; TIA: Transient ischemic attack.

scores, including HEMORR HAGES. However, in several each. The ORBIT score classified patients as having low (0
60
2
validation cohorts, ATRIA demonstrated a poorer ability – 2), medium (3), and high risk for bleeding (≥4). ORBIT
to predict intracranial bleeding compared with HAS- was validated using an external clinical trial population
BLED. 61,62 from the ROCKET-AF trial, which randomized patients
Noting the limitations of scores such as HAS-BLED with non-valvular AF to rivaroxaban or warfarin. 63,64
and HEMORR HAGES, which were formulated from a ORBIT demonstrated similar discrimination but improved
2
limited number of incidents and where HEMORR HAGES calibration compared to HAS-BLED and ATRIA scores.
2
demonstrated variable reliability across external validation Various validation studies have compared the
groups, O’Brien et al. proposed the ORBIT score in accuracy of these bleeding risk scores. HAS-BLED and
63
2015, intending to improve bleeding risk prediction using HEMORR HAGES were validated in a “real-world”
2
only readily available clinical variables. The ORBIT score cohort of patients with AF from Denmark, where they
was derived from the prospective Outcomes Registry for demonstrated similar performance in predicting major
Better Informed Treatment of AF, which enrolled patients bleeding events, although HAS-BLED was preferred
with AF from 176 sites across the US. The five strongest due to its greater simplicity. Additional studies further
25
predictors of major bleeding events were identified using a supported HAS-BLED’s superior performance. HAS-BLED
backward selection approach and formed the basis of a risk was also found to have better predictive ability than ATRIA
score with the same acronym as the prospective cohort. The in a cohort of AF outpatients from an anticoagulation
acronym ORBIT stands for older (age ≥75 years); reduced clinic. When all four scores were evaluated using data
62
hemoglobin, hematocrit, or history of anemia; bleeding from the AMADEUS trial, HAS-BLED was the best
history; insufficient kidney function; and treatment with predictor of clinically relevant bleeding, defined as major
an antiplatelet agent. Each of these factors was assigned 1 bleeding or any non-major clinically relevant bleeding. 61,65
point except reduced hemoglobin, hematocrit, history of Consequently, HAS-BLED has become the predominant
anemia, and bleeding history, which are assigned 2 points score used today to assess bleeding risk in patients with AF.

Volume 2 Issue 3 (2024) 7 doi: 10.36922/bh.3068