August 25, 2012Circulation Arrhythmia and Electrophysiology78 citationsOpen Access
Assessing the Risk of Bleeding in Patients With Atrial Fibrillation
View Full Paper
Key Result
The HAS-BLED score significantly improved net reclassification for predicting bleeding events compared to other risk scores, though all scores had modest predictive ability (c-statistic ≈0.6).
Study Design
Type
Cohort
Multicenter
Yes
Structured PICO
Does the HAS-BLED score improve bleeding risk prediction compared to other bleeding risk scores in patients with nonvalvular atrial fibrillation?
P
Population
Patients diagnosed with nonvalvular atrial fibrillation in a 4-hospital institution between 2000 and 2010.
I
Intervention
HAS-BLED score for bleeding risk estimation
C
Comparator
Other older bleeding risk scores and the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) score
O
Outcome
Bleeding events (assessed via c-statistic and net reclassification improvement)safety
The HAS-BLED score provides significantly improved reclassification for predicting bleeding events in patients with atrial fibrillation compared to older scores and the ATRIA score.
Abstract
BACKGROUND: Management decisions for thromboprophylaxis in atrial fibrillation need to balance the risk of stroke against serious hemorrhage. The objective of the present analysis is to compare the Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalized ratio, Elderly (>65 years), Drugs/alcohol concomitantly (HAS-BLED) score against other older bleeding risk scores and the new Anticoagulation and Risk Factors in Atrial Fibrillation score in an atrial fibrillation cohort. METHODS AND RESULTS: Patients diagnosed with nonvalvular atrial fibrillation in a 4-hospital institution between 2000 and 2010 were identified. Independent risk factors of bleeding were investigated using Cox regression. The predictive value of several bleeding risk schema was assessed using the c-statistic and net reclassification improvement. Oral anticoagulation use was highest in moderate-risk patients (59.8%) but only slightly more than high-risk (50.1%) and low-risk (46.4%) patients. Those at higher bleeding risk (HAS-BLED ≥ 3) were also at highest risk of stroke/thromboembolism or stroke/thromboembolism/death, as well as bleeding and all-cause mortality. On multivariable analysis, independent predictors of bleeding were age ≥ 75 years and age ≥ 65 years, alcohol excess, anemia, and heart failure. All risk scores had only modest predictive ability for bleeding, whether on vitamin K antagonist or not (c-statistic ≈0.6). When the HAS-BLED score was compared with other bleeding risk scores, the net reclassification improvement was significantly improved against all other scores tested. CONCLUSIONS: Current oral anticoagulation prescribing patterns would suggest that bleeding risk estimation by clinicians is poor and that oral anticoagulation prescribing does not reflect bleeding risk per se. The HAS-BLED score performs well in relation to predicting bleeding events compared with older bleeding scores and the Anticoagulation and Risk Factors in Atrial Fibrillation score, with significantly improved reclassification using HAS-BLED compared with all other bleeding risk scores tested.
Cite This Study
Lip et al. (2012) conducted a cohort in Nonvalvular atrial fibrillation. HAS-BLED score vs. Older bleeding risk scores and the Anticoagulation and Risk Factors in Atrial Fibrillation score was evaluated on Bleeding events. The HAS-BLED score significantly improved net reclassification for predicting bleeding events compared to other risk scores, though all scores had modest predictive ability (c-statistic ≈0.6).