Abstract
Aim: Oral anticoagulants (OACs) reduce the risk of ischemic stroke but may increase the risk of major bleeding in patients with non-valvular atrial fibrillation (NVAF). Various risk scores, such as HAS-BLED, ATRIA, ORBIT, and DOAC, have been proposed to assess the risk of major bleeding in patients with NVAF receiving OACs. However, limited data are available regarding bleeding risk stratification in Japanese patients with NVAF.
Methods: Of the 16,098 NVAF patients from the J-RISK AF study, the combined data of the five major AF registries in Japan (J-RHYTHM Registry, Fushimi AF Registry, Shinken Database, Keio interhospital Cardiovascular Studies, and Hokuriku-Plus AF Registry), we analyzed 11,539 patients receiving OACs (median age, 71 years old; women, 29.6%; median CHA2DS2-VASc score, 3).
Results: During the 2-year follow-up period, major bleeding occurred in 274 patients (1.3% per patient-year). In a multivariate Cox proportional hazards analysis, an advanced age, hypertension (systolic blood pressure ≥ 150 mmHg), bleeding history, anemia, thrombocytopenia, and concomitant antiplatelet agents were significantly associated with a higher incidence of major bleeding. We developed a novel risk stratification system, HED-[EPA]2-B3 score, which had a better predictive performance for major bleeding (C-statistics 0.67, [95% confidence interval, 0.63-0.70]) than the HAS-BLED (0.64, [0.60-0.67], P for difference 0.02) and ATRIA (0.63, [0.60-0.66], P for difference ＜0.01) scores. Furthermore, it was non-significantly higher than the ORBIT (0.65, [0.62-0.68], P for difference 0.07) and DOAC (0.65, [0.62-0.68], P for difference 0.17) scores.
Conclusion: Our novel risk stratification system, the HED-[EPA]2-B3 score, may be useful for identifying Japanese patients receiving OACs at a risk of major bleeding.
＊Present address: Section of Cardiovascular Medicine, Department of Internal Medicine, Yale University School of Medicine, Center for Outcomes Research and Evaluation, Yale-New Haven Hospital, New Haven, Connecticut, USA
Abbreviations: AF = atrial fibrillation, OAC = oral anticoagulant, DOAC = direct oral anticoagulant, NVAF = non-valvular atrial fibrillation, PT-INR = prothrombin time international normalized ratio, SBP = systolic blood pressure, TIA = transient ischemic attack
Introduction
Stroke prevention is an important issue in the management of patients with atrial fibrillation (AF) in clinical practice. Patients with AF are increasingly treated with oral anticoagulants (OACs), which have been reported to reduce the risk of thromboembolism and all-cause mortality1, 2). However, OACs may increase the risk of major bleeding in patients with AF. The average annual incidence of fatal and major bleeding during warfarin therapy was reported to be 0.6% and 3.0%, respectively, which is approximately 5 times that expected without warfarin therapy3). Once major bleeding occurs, the case fatality rate is high. A sub-analysis of the ARISTOTLE trial reported that mortality within 30 days occurred in 11.0% and 15.4% of the patients who experienced major bleeding while receiving apixaban and warfarin, respectively4). Major bleeding is associated with an increased risk of subsequent all-cause mortality and stroke/systemic embolism (SE) in the long term among Japanese patients with AF5).
Various bleeding risk scores have been proposed for patients with AF taking warfarin as an OAC. Most bleeding scores, such as the “uncontrolled Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile prothrombin time international normalized ratios [PT-INR], Elderly, and Drug/alcohol concomitantly” (HAS-BLED)6) and anemia, renal disease, age, prior bleeding, and hypertension (ATRIA)7) scores, were developed and derived from studies conducted in the warfarin era. However, direct oral anticoagulants (DOACs) are being used increasingly frequently for stroke prevention. Therefore, the older age, reduced hemoglobin or hematocrit, bleeding history, insufficient kidney function, and treatment with antiplatelets (ORBIT) score8) is derived from data in AF patients taking either warfarin or a DOAC and has been proposed as an alternative score in the DOAC era. Most recently, the DOAC score9) was developed as a novel bleeding risk prediction tool for patients with AF taking DOACs.
However, limited data are available on the clinical value of various bleeding risk scores among Japanese patients with AF receiving warfarin or DOACs in contemporary clinical practice. Furthermore, there have been no studies regarding the differences in bleeding risk between Japanese and Western populations.
Aim
The present study determined the risk factors for major bleeding and developed a novel scoring scheme for the occurrence of major bleeding in a large-scale Japanese cohort of non-valvular AF (NVAF) patients administered OACs.
Methods
Study Population
The detailed study design, patient enrollment, definition of the measurements, and baseline clinical characteristics of the J-RISK AF Research have been previously described10). We pooled data from five major AF prospective registries in Japan: the J-RHYTHM (Japanese Rhythm Management Trial for Atrial Fibrillation) Registry (n=7,937), Fushimi AF Registry (n=3,749), Shinken Database (n=2,957), Keio Interhospital Cardiovascular Studies (n=783), and Hokuriku-Plus AF Registry (n=1,492). Data from each registry were collected and pooled in March 2016, and those from the Keio Interhospital Cardiovascular Studies were updated in April 2018. Of the 16,918 enrolled patients, 820 patients with valvular AF, 81 with a lack of event data for major bleeding, and 4,476 not receiving OACs at baseline were excluded. After excluding two patients with missing information regarding the type of OAC, analyses were performed on 11,539 patients.
The patient flow diagram is shown in Fig.1. The data of 4,476 patients not receiving OACs at baseline are presented as a reference only in Fig.2A and Supplemental Table 3. To balance the follow-up period among the registries, event data and follow-up periods exceeding 730 days were excluded from the analysis. We collected demographic data of the patients at baseline and follow-up data at two years.
Supplemental Table 3.Clinical backgrounds of patients receiving warfarin vs. DOAC at baseline
|
Warfarin |
DOAC |
P value
|
| 9,977 (86.5) |
1,562 (13.5) |
| Female |
2,974 (29.8) |
439 (28.1) |
.18 |
| Age, years |
71 (64-78) |
70 (62-76) |
＜.01 |
| ＜65 years |
2,594 (26.0) |
498 (31.9) |
＜.01 |
| 65-74 years |
3,613 (36.2) |
562 (36.0) |
|
| ≥ 75 years |
3,770 (37.8) |
502 (32.1) |
|
| Body weight, kg |
62.0 (54.0-70.0) |
63.0 (54.4-72.0) |
＜.01 |
| Systolic blood pressure, mmHg |
125 (114-136) |
126 (116-138) |
＜.01 |
| Diastolic blood pressure, mmHg |
72 (65-80) |
74 (67-82) |
＜.01 |
| Paroxysmal type |
3,573 (35.8) |
840 (53.8) |
＜.01 |
| CHADS2 score
|
2 (1-3) |
1 (1-2) |
＜.01 |
| CHA2DS2-VASc score
|
3 (2-4) |
2 (1-4) |
＜.01 |
| HAS-BLED score |
1 (1-2) |
1 (0-1) |
＜.01 |
| ATRIA score |
1 (1-3) |
1 (0-3) |
＜.01 |
| ORBIT score |
1 (0-2) |
1 (0-2) |
＜.01 |
| DOAC score |
5 (3-7) |
4 (2-6) |
＜.01 |
| Prior stroke/TIA |
1,817 (18.2) |
177 (11.3) |
＜.01 |
| Heart failure |
2,964 (29.7) |
296 (19.0) |
＜.01 |
| Hypertension |
6,153 (61.7) |
942 (60.3) |
.30 |
| Diabetes mellitus |
2,135 (21.4) |
310 (19.9) |
.17 |
| Coronary artery disease |
1,245 (12.5) |
158 (10.1) |
＜.01 |
| eGFR, mL/min/1.73 m2
|
61.6 (50.0-73.3) |
63.6 (53.7-74.2) |
＜.01 |
| History of bleeding |
408 (4.1) |
23 (1.5) |
＜.01 |
| Hemoglobin, g/dl |
13.6 (12.5-14.8) |
13.8 (12.6-15.1) |
＜.01 |
| Platelet, x104 /μl
|
19.0 (15.8-22.7) |
20.1 (16.6-24.1) |
＜.01 |
| Antiplatelets |
2,363 (23.7) |
190 (12.2) |
＜.01 |
| Aspirin |
2,019 (20.2) |
147 (9.4) |
＜.01 |
Continuous variables are presented as median (interquartile range) for non-normal distribution.
Categorical variables are presented as numbers (%).
eGFR: estimated glomerular filtration rate, TIA: transient ischemic attack.
The primary endpoint of the analysis was the incidence of major bleeding events. The definition of major bleeding varies among registries but is in accordance with the criteria of the International Society on Thrombosis and Hemostasis (ISTH), which consists of the following: (i) reduction in hemoglobin level of at least 2 g/dl; (ii) transfusion of at least 2 units of blood; and (iii) symptomatic bleeding in a critical area or organ (intracranial, intraocular, intraspinal, intraarticular, intramuscular with compartment syndrome, pericardial, retroperitoneal). Major bleeding was defined as (i) intracranial hemorrhage, gastrointestinal hemorrhage, and other hemorrhages requiring hospitalization in the J-RHYTHM Registry; (ii) according to the ISTH criteria in the Fushimi AF Registry and the Keio Interhospital Cardiovascular Studies; (iii) bleeding that required emergency hospitalization, including intracranial and extracranial hemorrhage in the Shinken Database; and (iv) intracranial hemorrhage, including hemorrhagic stroke, hemorrhagic events requiring transfusion, and hemorrhagic events with reduction of hemoglobin by ＞2 g/dl, in the Hokuriku-Plus AF Registry.
Valvular AF was defined as AF with rheumatic mitral stenosis or prosthetic heart valve. The OACs included warfarin and DOACs (dabigatran, rivaroxaban, apixaban, and edoxaban). Japanese treatment guidelines during this study period recommended different target PT-INRs for patients taking warfarin: 1.6-2.6 for elderly patients (≥ 70 years old) and 2.0-3.0 for younger patients (＜70 years old)11). Antiplatelet agents included aspirin and thienopyridines.
We categorized the patients according to the risk of major bleeding using the HAS-BLED score6), ATRIA score7), ORBIT score8), and DOAC score9) (age, creatinine clearance, underweight, stroke/transient ischemic attack/embolism history, diabetes, hypertension, antiplatelet use, nonsteroidal anti-inflammatory use, liver disease, and bleeding history), as originally described, except for the factor ‘L’ (labile PT-INR) of the HAS-BLED score. For the labile PT-INR criterion, we defined the value of PT-INR as ＞2.6 for elderly patients (≥ 70 years old) or 3.0 for younger patients (＜70 years old) at baseline according to the Japanese AF guidelines11), because the values of PT-INR were collected at the time of enrollment. Furthermore, for warfarin control, we defined appropriate as a PT-INR within the target range, overdose as that over the target range, and underdose as that under the target range at baseline.
Ethics
This study was approved by the Ethics Committees of Hirosaki University Graduate School of Medicine (2015-117, 2017-1051), National Cerebral and Cardiovascular Center (M27-092-4), National Hospital Organization Kyoto Medical Center (15-101), Cardiovascular Institute (279), Kanazawa University Graduate School of Medical Science (2035-1, 2460-1), and Keio University School of Medicine (20120029) and was performed within Ethics Committee-approved research protocols at other institutes.
Statistical Analyses
Categorical variables were presented as numbers and percentages. Continuous variables are presented as the mean and standard deviation for normally distributed data or median and interquartile range for non-normal distribution. Categorical variables were compared using the chi-square test, and continuous variables were compared using the independent samples t-test for normally distributed data or the Mann-Whitney U-test for non-normal distribution. Crude event rates are presented as percentages per patient-year.
We performed univariable Cox proportional hazards models after confirming proportional hazards and constructed multivariable Cox proportional hazards models to determine the variables associated with the incidence of major bleeding. All variables were evaluated for collinearity. Candidate variables for Model 1 were chosen from the significant variables in the univariate Cox proportional hazards models. The variables for Model 2 were the significant variables included in Model 1. Multiple imputation methods were used for missing data (6.3% of the total data). Multiple imputation methods using the chained equation algorithm were conducted based on 10 replications, assuming missing random mechanisms. The imputed variables are shown in Supplemental Table 1. A score was assigned to each significant risk factor based on the coefficient calculated using logarithmically transformed hazard ratios (HRs). We assessed model performance by examining calibration and discrimination at two years of follow-up in the J-RISK AF cohort. Calibration was evaluated by plotting the event rates of major bleeding according to the HED-[EPA]2-B3 score and the calibration plot in the J-RISK AF cohort. Confidence intervals (CIs) of event rates were calculated by bootstrap resampling of 1,000 replicates. The modified Hosmer-Lemeshow test was performed to determine the goodness-of-fit of the HED-[EPA]2-B3 score. To check the risk of overfitting in the model performance, we created a training dataset and a test dataset with a 5:5 ratio. After fitting a Cox model with scores from the present risk stratification scheme to the training set and creating the inverse hazard rate variables, we estimated the C-statistics in the test dataset for the model fit to the training set using the bootstrap method with 100 iterations. Discrimination was evaluated using C-statistics for a comparison with previous risk stratification schemes, such as HAS-BLED, ATRIA, ORBIT, and DOAC scores. In addition, a sensitivity analysis was performed on patients divided into subgroups consisting of those taking warfarin and those taking DOACs.
Supplemental Table 1.Imputed variables in the study
| Variables |
| Body weight |
| Systolic blood pressure |
| Diastolic blood pressure |
| Type of AF |
| Prior stroke/TIA |
| Heart failure |
| Hypertension |
| Diabetes mellitus |
| Coronary artery disease |
| History of bleeding |
| Hemoglobin |
| Platelet |
| eGFR |
| Liver function |
| Prothrombin time-INR |
| Antiplatelets |
AF: atrial fibrillation, eGFR: estimated glomerular filtration rate, INR: international normalized ratio, TIA: transient ischemic attack.
We performed a Cox proportional hazards analysis for variables associated with the incidence of major bleeding by using the Fine and Gray model to account for the competing risk of death. Significance was set at a two-sided p-value ＜0.05. Analyses were performed using the JMP software program, version 13.2.0 (SAS Institute, Cary, NC, USA), and R version 4.1.0 for Windows.
Results
Of 11,539 NVAF patients, 274 patients experienced major bleeding during the 2-year follow-up period (1.3% per patient-year), 238 patients were taking warfarin (1.4% per patient-year), and 36 patients were taking DOACs (1.3% per patient-year) (p=0.697). Intracranial hemorrhage occurred in 93 of 274 patients with major bleeding (33.9%) (0.4% per patient-year).
The demographics of the patients at baseline, categorized by the occurrence of major bleeding during the follow-up period, are shown in Table 1. NVAF patients with major bleeding were significantly older and had a lower body weight, higher systolic blood pressure (SBP), lower hemoglobin level, and lower platelet count than those without major bleeding. They were significantly more likely to have a history of stroke/transient ischemic attack (TIA), heart failure, hypertension, coronary artery disease, chronic kidney disease, and bleeding history than those without major bleeding. The baseline stroke risk scores (CHADS2 and CHA2DS2-VASc) and bleeding risk scores (HAS-BLED, ATRIA, ORBIT, and DOAC) were significantly higher in patients with major bleeding.
Table 1.Patient characteristics, medication and previous therapy at baseline.
|
Overall |
Major bleeding (+) |
Major bleeding (-) |
P value
|
| 11,539 (100.0) |
274 (2.4) |
11,265 (97.6) |
| Female |
3,413 (29.6) |
70 (25.6) |
3,343 (29.7) |
.13 |
| Age, years |
71 (64-78) |
74 (68-80) |
71 (64-78) |
＜.01 |
| ＜65 years |
3,092 (26.8) |
40 (14.6) |
3,052 (27.1) |
＜.01 |
| 65-74 years |
4,175 (36.2) |
97 (35.4) |
4,078 (36.2) |
|
| ≥ 75 years |
4,272 (37.0) |
137 (50.0) |
4,135 (36.7) |
|
| Body weight, kg |
62.0 (54.0-70.0) |
60.0 (51.4-67.6) |
62.0 (54.0-70.0) |
.03 |
| ＜60 kg |
4,757 (41.2) |
136 (49.6) |
4,621 (41.0) |
＜.01 |
| Systolic blood pressure, mmHg |
125 (114-136) |
128 (116-140) |
125 (114-136) |
＜.01 |
| ≥ 150 mmHg |
975 (8.5) |
38 (13.9) |
937 (8.3) |
＜.01 |
| Diastolic blood pressure, mmHg |
72 (66-80) |
72 (66-80) |
72 (66-80) |
.71 |
| Paroxysmal type |
4,413 (38.2) |
91 (33.2) |
4,322 (38.4) |
.08 |
| CHADS2 score
|
2 (1-3) |
2 (1-3) |
2 (1-3) |
＜.01 |
| CHA2DS2-VASc score
|
3 (2-4) |
3 (2-5) |
3 (2-4) |
＜.01 |
| HAS-BLED score |
1 (1-2) |
2 (1-2) |
1 (1-2) |
＜.01 |
| ATRIA score |
1 (1-3) |
3 (1-3) |
1 (1-3) |
＜.01 |
| ORBIT score |
1 (0-2) |
2 (1-3) |
1 (0-2) |
＜.01 |
| DOAC score |
5 (2-7) |
6 (4-8) |
5 (2-7) |
＜.01 |
| Prior stroke/TIA |
1,775 (15.4) |
57 (20.8) |
1,718 (15.3) |
.02 |
| Heart failure |
3,260 (28.3) |
101 (36.9) |
3,159 (28.0) |
＜.01 |
| Hypertension |
7,095 (61.5) |
194 (70.8) |
6,901 (61.3) |
.01 |
| Diabetes mellitus |
2,445 (21.2) |
72 (26.3) |
2,373 (21.1) |
.04 |
| Coronary artery disease |
1,403 (12.2) |
52 (19.0) |
1,351 (12.0) |
＜.01 |
| Chronic kidney disease |
5,264 (45.6) |
156 (56.9) |
5,108 (45.3) |
＜.01 |
| eGFR, mL/min/1.73 m2
|
61.8 (50.5-73.5) |
57.2 (46.6-70.4) |
61.9 (50.7-73.5) |
＜.01 |
| ≥ 90 mL/min/1.73 m2
|
777 (6.7) |
13 (4.7) |
764 (6.8) |
＜.01 |
| 60-89 mL/min/1.73 m2
|
5,498 (47.6) |
105 (38.3) |
5,393 (47.9) |
|
| 45-59 mL/min/1.73 m2
|
3,414 (29.6) |
96 (35.0) |
3,318 (29.5) |
|
| 30-44 mL/min/1.73 m2
|
1,334 (11.6) |
40 (14.6) |
1,294 (11.5) |
|
| 15-29 mL/min/1.73 m2
|
390 (3.4) |
12 (4.4) |
378 (3.4) |
|
| ＜15 mL/min/1.73 m2
|
126 (1.1) |
8 (2.9) |
118 (1.1) |
|
| Bleeding history |
431 (3.7) |
40 (14.6) |
391 (3.5) |
＜.01 |
| Hemoglobin, g/dl |
13.8 (12.5-14.8) |
13.1 (11.6-14.4) |
13.8 (12.5-14.8) |
＜.01 |
| ＜ 11 g/dl |
926 (8.0) |
51 (18.6) |
875 (7.8) |
＜.01 |
| Platelet, x104 /μl
|
19.1 (15.8-22.8) |
18.3 (15.1-22.5) |
19.1 (15.8-22.8) |
.06 |
| ＜10 x104 /μl
|
253 (2.2) |
14 (5.1) |
239 (2.1) |
＜.01 |
| PT-INR |
1.80 (1.53-2.15) |
1.90 (1.60-2.24) |
1.80 (1.53-2.15) |
.01 |
| Warfarin control |
|
|
|
.03 |
| Overdose |
513 (4.5) |
21 (7.7) |
492 (4.4) |
|
| Appropriate |
5,270 (45.7) |
134 (48.9) |
5,136 (45.6) |
|
| Underdose |
4,217 (36.6) |
83 (30.3) |
4,134 (36.7) |
|
| OAC |
11,539 (100.0) |
274 (100.0) |
11,265 (100.0) |
.85 |
| Warfarin |
9,977 (86.5) |
238 (86.9) |
9,739 (86.5) |
|
| DOAC |
1,562 (13.5) |
36 (13.1) |
1,526 (13.6) |
|
| Antiplatelets |
2,553 (22.1) |
88 (32.1) |
2,465 (21.9) |
＜.01 |
| Aspirin |
2,166 (18.8) |
69 (25.2) |
2,097 (18.6) |
＜.01 |
| DAPT |
264 (2.3) |
5 (1.8) |
259 (2.3) |
.59 |
Continuous variables are presented as medians (interquartile ranges) for non-normally distributed data. Categorical variables are presented as number (%).
eGFR, estimated glomerular filtration rate; PT-INR, prothrombin time international normalized ratio; OAC, oral anticoagulants; DOAC, direct oral anticoagulants; TIA, transient ischemic attack; DAPT, dual antiplatelet therapy.
The type of OACs administered was comparable between patients with NVAF with and without major bleeding at baseline. Concomitant use of antiplatelet agents, predominantly aspirin, was found more often in patients with major bleeding. The PT-INR at baseline was significantly higher in the patients with major bleeding. Warfarin was appropriately controlled in 45.5% of patients, but NVAF patients with major bleeding were more often controlled at overdoses than at underdoses of warfarin.
Factors Associated with Incidence of Major Bleeding and Score Assignments
We performed a Cox proportional hazards analysis to identify the variables associated with the incidence of major bleeding (Table 2). On a univariate analysis, age 65-74 and ≥ 75 years old, body weight ＜60 kg, SBP ≥ 150 mmHg, history of stroke/TIA, heart failure, hypertension, diabetes mellitus, coronary artery disease, bleeding history, hemoglobin ＜11 g/dl, platelet ＜10×104/µl, an estimated glomerular filtration rate (eGFR) ＜15 mL/min/1.73 m2, and concomitant use of antiplatelets were significant. In a multivariable model, which included factors that were significantly associated in the univariable model (Model 1), we indicated that age 65-74 and ≥ 75 years old, SBP ≥ 150 mmHg, a bleeding history, hemoglobin level ＜11 g/dl, platelet ＜10×104/µl, and concomitant use of antiplatelets were significantly associated with a higher incidence of major bleeding. Multivariable Model 2 was constructed by excluding non-significant variables in Model 1. No collinearity was found for any of the variables used in the multivariate models (variance inflation factor ＜10).
Table 2.Cox proportional hazards analysis for variables associated with incidence of major bleeding
|
Univariable |
Multivariable: Model 1 |
Multivariable: Model 2 |
| HR |
95% CI |
HR |
95% CI |
HR |
95% CI |
Coefficient |
Points |
| Age |
| ＜65 years |
Reference |
Reference |
Reference |
| 65-74 years |
1.79 |
1.24, 2.58 |
1.57 |
1.08, 2.28 |
1.66 |
1.15, 2.40 |
0.51 |
1 |
| ≥ 75 years |
2.52 |
1.78, 3.59 |
1.79 |
1.23, 2.60 |
2.00 |
1.31, 3.06 |
0.69 |
2 |
| Body weight ＜60 kg |
1.44 |
1.14, 1.83 |
1.11 |
0.86, 1.43 |
|
|
|
|
| Systolic blood pressure ≥ 150 mmHg |
1.78 |
1.26, 2.50 |
1.65 |
1.16, 2.34 |
1.74 |
1.23, 2.45 |
0.55 |
1 |
| History of stroke/TIA |
1.48 |
1.10, 1.98 |
1.13 |
0.83, 1.52 |
|
|
|
|
| Heart failure |
1.51 |
1.18, 1.93 |
1.24 |
0.96, 1.60 |
|
|
|
|
| Hypertension |
1.52 |
1.17, 1.98 |
1.30 |
0.99, 1.70 |
|
|
|
|
| Diabetes mellitus |
1.34 |
1.02, 1.75 |
1.16 |
0.88, 1.53 |
|
|
|
|
| Coronary artery disease |
1.73 |
1.28, 2.34 |
1.17 |
0.83, 1.65 |
|
|
|
|
| Bleeding history |
4.65 |
3.32, 6.50 |
3.97 |
2.82, 5.59 |
4.18 |
2.98, 5.86 |
1.43 |
3 |
| Hemoglobin ＜11 g/dl |
2.85 |
2.10, 3.87 |
1.94 |
1.40, 2.71 |
2.13 |
1.55, 2.92 |
0.76 |
2 |
| Platelet ＜10 x104 /μl
|
2.54 |
1.49, 4.35 |
2.10 |
1.22, 3.63 |
2.13 |
1.24, 3.66 |
0.76 |
2 |
| eGFR ＜15 mL/min/1.73 m2
|
3.09 |
1.53, 6.25 |
1.32 |
0.63, 2.75 |
|
|
|
|
| Warfarin (vs. DOAC) |
1.06 |
0.75, 1.51 |
|
|
|
|
|
|
| Antiplatelets |
1.67 |
1.30, 2.16 |
1.38 |
1.04, 1.84 |
1.54 |
1.19, 1.99 |
0.43 |
1 |
CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; DOAC, direct oral anticoagulant; TIA, transient ischemic attack.
Based on this multivariable model’s regression coefficients, we propose a new score (“HED-[EPA]2-B3 score”): hypertension (SBP ≥ 150 mmHg), elderly (age 65-74 years old), and drugs (concomitant use of antiplatelets) are assigned 1 point each: elderly (age ≥ 75 years old), platelet ＜10×104/µl, and anemia (hemoglobin ＜11 g/dl) are 2 points; bleeding history is 3 points, resulting in a risk scheme with a possible range of 0 to 11 points.
Model Performance
Fig.2A shows the crude incidences of major bleeding stratified by HED-[EPA]2-B3 score in patients receiving OACs, as well as in those not receiving OACs as a reference (patient backgrounds are briefly presented in Supplemental Table 2). The observed major bleeding rates increased with increasing risk score in the entire cohort with OACs (Fig.2A), and patients were divided into subgroups consisting of those receiving warfarin or DOACs (Fig.2B). The baseline characteristics of patients receiving warfarin or DOACs are shown in Supplemental Table 3; patients receiving warfarin were older and had more co-morbidities and thus higher stroke and bleeding risk scores.
Supplemental Table 2.Clinical backgrounds of patients not receiving OAC at baseline
|
Overall |
| 4,476 (100.0) |
| Female |
1,486 (33.2) |
| Age, years |
69 (60-78) |
| ＜65 years |
1,633 (36.5) |
| 65-74 years |
1,324 (29.6) |
| ≥ 75 years |
1,519 (33.9) |
| Body weight, kg |
61.7 (52.5-70.3) |
| Systolic blood pressure, mmHg |
125 (114-138) |
| Diastolic blood pressure, mmHg |
73 (65-80) |
| Paroxysmal type |
1,580 (35.3) |
| CHADS2 score
|
1 (0-2) |
| CHA2DS2-VASc score
|
2 (1-4) |
| HAS-BLED score |
1 (1-2) |
| ATRIA score |
1 (0-3) |
| ORBIT score |
1 (0-2) |
| DOAC score |
4 (2-7) |
| Prior stroke/TIA |
391 (8.7) |
| Heart failure |
671 (15.0) |
| Hypertension |
2,263 (50.6) |
| Diabetes mellitus |
707 (15.8) |
| Coronary artery disease |
510 (11.4) |
| eGFR, mL/min/1.73 m2
|
64.6 (52.1-77.0) |
| History of bleeding |
127 (2.9) |
| Hemoglobin, g/dl |
13.6 (12.3-14.8) |
| Platelet, x104 /μl
|
20.3 (16.7-24.3) |
| Antiplatelets |
1,549 (34.7) |
| Aspirin |
1,395 (31.3) |
Continuous variables are presented as median (interquartile range) for non-normal distribution.
Categorical variables are presented as numbers (%).
eGFR: estimated glomerular filtration rate, TIA: transient ischemic attack.
Event rates calculated using the bootstrap method were similar to the observed crude incidence rates (Supplemental Table 4). Fig.3 shows the crude incidences of major bleeding stratified by HED-[EPA]2-B3 score in patients ≥ 75 years old (Fig.3A) and those with an eGFR ＜60 mL/min/1.73 m2 (Fig.3B). Calibration plots comparing the major bleeding rates predicted by our model and those observed at two years of follow-up are shown in Fig.4. In all patients (Fig.4A), the predicted event rates by the HED-[EPA]2-B3 score were a good fit for the actual event rates (modified Hosmer-Lemeshow test; P=0.965). In addition, in the patients divided into subgroups of those receiving warfarin or DOACs (Fig.4B), the HED-[EPA]2-B3 score showed good calibration for major bleeding rates (modified Hosmer-Lemeshow test; P=0.876 and P=0.998, respectively).
Supplemental Table 4.Incidence rates (% per patient-year) calculated with the bootstrap method
| HED-[EPA]2-B3 score
|
Incidence rate |
95% CI |
| 0 |
0.48 |
0.27, 0.74 |
| 1 |
0.82 |
0.61, 1.06 |
| 2 |
1.38 |
1.09, 1.66 |
| 3 |
1.69 |
1.15, 2.22 |
| 4 |
2.34 |
1.50, 3.24 |
| 5 |
4.45 |
2.63, 6.29 |
| 6 |
5.46 |
2.42, 9.85 |
| 7 |
11.51 |
4.44, 21.35 |
| ≥ 8 |
15.40 |
4.61, 31.2 |
The C-statistics for major bleeding events are presented in Table 3. The HED-[EPA]2-B3 score had significantly higher C-statistics (0.67, [95% CI, 0.63-0.70]) than the HAS-BLED (0.64, [95% CI, 0.60-0.67], P for difference 0.02) and ATRIA (0.63, [95% CI, 0.60-0.66], P for difference ＜0.01) scores, but it was non-significantly higher than the ORBIT (0.65, [95% CI, 0.62-0.68], P for difference 0.07) and DOAC (0.65, [95% CI, 0.62-0.68], P for difference 0.17) scores. In a sensitivity analysis, the HED-[EPA]2-B3 score also showed the best discrimination in patients taking warfarin, as well as those taking DOACs, with regard to the C-statistic (Table 3).
Table 3. Comparison of bleeding risk schemes
| Risk scheme |
C-Statistics |
95% CI |
P value
|
| Overall |
| HED-[EPA]2-B3 score
|
0.67 |
0.63, 0.70 |
|
| HAS-BLED score |
0.64 |
0.60, 0.67 |
.02 |
| ATRIA score |
0.63 |
0.60, 0.66 |
＜.01 |
| ORBIT score |
0.65 |
0.62, 0.68 |
.07 |
| DOAC score |
0.65 |
0.62, 0.68 |
.17 |
| Warfarin |
| HED-[EPA]2-B3 score
|
0.67 |
0.63, 0.70 |
|
| HAS-BLED score |
0.64 |
0.61, 0.67 |
.02 |
| ATRIA score |
0.63 |
0.60, 0.67 |
＜.01 |
| ORBIT score |
0.65 |
0.62, 0.69 |
.07 |
| DOAC score |
0.66 |
0.62, 0.69 |
.28 |
| DOAC |
| HED-[EPA]2-B3 score
|
0.64 |
0.55, 0.73 |
|
| HAS-BLED score |
0.62 |
0.53, 0.70 |
.53 |
| ATRIA score |
0.62 |
0.52, 0.71 |
.48 |
| ORBIT score |
0.63 |
0.54, 0.72 |
.79 |
| DOAC score |
0.61 |
0.52, 0.69 |
.39 |
CI: confidence interval, DOAC: direct oral anticoagulant
We performed a cross-validation procedure to check for overfitting of our model. The C-statistic for its model was 0.65 [95% CI, 0.61-0.70] (Warfarin: 0.65 [0.60-0.71], DOAC: 0.66 [0.55-0.78]), which was comparable to the C-statistic of the original dataset (0.67, [0.63-0.70]).
We performed an additional analysis to account for the competing risk of death (Supplemental Table 5) and confirmed that the results were consistent.
Supplemental Table 5.Cox proportional hazards analysis for variables associated with incidence of major bleeding (Fine and Gray model to account for the competing risk of death)
|
Univariable |
Multivariable: Model 1 |
Multivariable: Model 2 |
| HR |
95% CI |
HR |
95% CI |
HR |
95% CI |
| Age |
| ＜65 years |
Reference |
|
Reference |
|
Reference |
|
| 65-74 years |
1.71 |
1.18, 2.48 |
1.53 |
1.05, 2.23 |
1.57 |
1.08, 2.28 |
| ≥ 75 years |
2.39 |
1.68, 3.41 |
1.75 |
1.19, 2.57 |
1.86 |
1.29, 2.70 |
| Body weight ＜60 kg |
1.38 |
1.09, 1.76 |
1.07 |
0.82, 1.40 |
|
|
| Systolic blood pressure ≥ 150 mmHg |
1.75 |
1.23, 2.48 |
1.64 |
1.15, 2.34 |
1.62 |
1.13, 2.31 |
| History of stroke/TIA |
1.46 |
1.08, 1.97 |
1.12 |
0.82, 1.54 |
|
|
| Heart failure |
1.53 |
1.19, 1.97 |
1.29 |
0.99, 1.68 |
|
|
| Hypertension |
1.52 |
1.16, 1.98 |
1.32 |
1.01, 1.72 |
1.32 |
1.01,1.72 |
| Diabetes mellitus |
1.31 |
0.99, 1.73 |
|
|
|
|
| Coronary artery disease |
1.69 |
1.24, 2.31 |
1.19 |
0.85, 1.66 |
|
|
| Bleeding history |
4.58 |
3.24, 6.47 |
3.94 |
2.76, 5.64 |
4.09 |
2.87, 5.83 |
| Hemoglobin ＜11 g/dl |
2.64 |
1.92, 3.62 |
1.84 |
1.29, 2.62 |
1.99 |
1.43, 2.78 |
| Platelet ＜10 x104 /μl
|
2.25 |
1.26, 4.00 |
1.86 |
1.04, 3.33 |
1.95 |
1.10, 3.48 |
| eGFR ＜15 mL/min/1.73 m2
|
2.80 |
1.32, 5.93 |
1.26 |
0.56, 2.82 |
|
|
| Warfarin (vs. DOAC) |
1.11 |
0.78, 1.58 |
|
|
|
|
| Antiplatelets |
1.60 |
1.23, 2.08 |
1.34 |
1.01, 1.78 |
1.46 |
1.12, 1.89 |
CI: confidence interval, eGFR: estimated glomerular filtration rate, HR: hazard ratio, DOAC: direct oral anticoagulants, TIA: transient ischemic attack.
Discussion
The principal findings of this study are as follows: (i) the incidence rate of major bleeding was 1.3% per patient-year in AF patients receiving OAC; (ii) on a multivariable model, an age 65-74 and ≥ 75 years old, SBP ≥ 150 mmHg, bleeding history, hemoglobin ＜11 g/dl, platelet count ＜10×104/µl, and concomitant use of antiplatelets were significantly associated with a higher incidence of major bleeding; and (iii) the novel HED-[EPA]2-B3 score had a predictive performance better than or comparable to previous bleeding risk scores, such as the HAS-BLED, ATRIA, ORBIT, and DOAC scores.
In global phase III randomized clinical trials of DOACs, the annualized rate of major bleeding was approximately 3% in AF patients taking warfarin and ranged from 2.1% to 3.6% in those taking DOACs2). In a sub-analysis of these trials of DOACs in Japanese patients with AF, the incidence rate of major bleeding ranged from 3.3%/year to 4.0%/year in those taking warfarin, and from 1.2%/year to 5.5%/year in those taking DOACs12-15). In previous observational studies, the incidence rates of major bleeding were 1.56% per patient-year in the Euro Heart Survey on Atrial Fibrillation6), and 1.9% per patient-year in the Swedish Atrial Fibrillation cohort study16), which was comparable to the present study.
Clinical Factors Associated with Major Bleeding
Advanced age was one of the most consistent factors associated with major bleeding in previous studies and was included in all bleeding scores, such as the HAS-BLED, ATRIA, ORBIT, and DOAC scores. The ATRIA and ORBIT scores included an age cutoff of 75 years old, whereas the HAS-BLED score and our score included an age cutoff of 65 years old. In the present analysis, the adjusted HR in patients ≥ 75 years old was modestly higher than that in patients 65-74 years old, but scores of those ≥ 75 years old were assigned 2 points, based on their coefficients rounded to the nearest integer.
Concerning hypertension, uncontrolled hypertension, but no history of hypertension, was included in the HAS-BLED and HED-[EPA]2-B3 scores, whereas the ATRIA and DOAC scores included a history of hypertension. ‘Uncontrolled hypertension’ was defined as an SBP ＞160 mmHg in the HAS-BLED score and ≥ 150 mmHg in our score. A cutoff SBP value of 150 mmHg was adopted based on our previous studies in Japan17, 18).
Concomitant antiplatelet use was significantly associated with an increased incidence of major bleeding in the present study and was included in the HAS-BLED, ORBIT, DOAC, and HED-[EPA]2-B3 scores. The AFIRE trial reported that rivaroxaban monotherapy was non-inferior to the combination therapy of rivaroxaban and antiplatelets for thromboembolism and was superior for major bleeding in AF patients with stable coronary artery disease19). Physicians are encouraged to withhold concomitant antiplatelet therapy, considering thromboembolic and bleeding risk factors.
Anemia, defined as a hemoglobin level of ＜11 g/dl, was significantly associated with major bleeding in this study, which is consistent with previous scores. The ATRIA and ORBIT scores include anemia; however, the definition of anemia differs for each score. In the present score, we determined a cutoff hemoglobin level of ＜11 g/dl, based on moderate or severe anemia, using the World Health Organization classification of anemia.
Thrombocytopenia, defined as a platelet count ＜10×104/µl, was significantly associated with an increased incidence of major bleeding. This result is consistent with that of a previous study from Korea, which showed that lower platelet counts were associated with an increased risk of bleeding events20). Although a sub-analysis of the J-RHYTHM Registry reported that the association between low platelet count (＜10×104/µl) and major bleeding was not significant (P=0.073) in a multivariable model21), the present pooled analysis, which included the J-RHYTHM Registry, found a significant association. The sub-analysis of the AFIRE trial also identified thrombocytopenia as a bleeding risk factor in Japanese patients with AF and stable coronary disease22). However, data regarding the association between thrombocytopenia and bleeding risk are lacking, and it has not been included in the previous bleeding risk scores.
All previous scores included a reduced renal function as an independent risk factor, whereas our score did not. This may be due to the presence of confounding factors, such as advanced age or anemia. It is also likely that patients with severe renal dysfunction did not receive OACs and thus were not included in the study population. The kidney function may dynamically change and worsen with age and comorbidities23).
Comparison with Previous Scores
The components of the HED-[EPA]2-B3 score overlap with those of other scores, such as the HAS-BLED, ATRIA, ORBIT, and DOAC scores. Our findings suggest that the HED-[EPA]2-B3 score outperformed the previous scores; the C-statistics of the HED-[EPA]2-B3 score (0.67, 0.67, and 0.64 in the total cohort, warfarin subgroup, and DOAC subgroup, respectively) indicated relatively good performance. O’Brien et al. found that the C-statistics were 0.67 in the ORBIT-AF cohort and 0.62 in the ROCKET-AF cohort, respectively8). In addition, they reported that the C-statistics of the HAS-BLED and ATRIA scores were approximately 0.6 in both the ORBIT-AF and ROCKET-AF cohorts. Furthermore, Lip et al. reported that the C-statistics for ATRIA, HAS-BLED, and ORBIT scores were 0.59, 0.58, and 0.61, respectively, using data from a Danish nationwide database24). The most recent DOAC score showed a good performance in the development cohort (0.73) and was further validated in the external cohort (0.67)9). However, it did not have better predictive performance in our Japanese cohort. Risk factors for major bleeding may vary according to ethnicity; therefore, the optimal risk score may differ among races, which warrants external validation, especially in Asian cohorts.
The HED-[EPA]2-B3 score may be a useful predictive score for major bleeding, which includes well-known and easily available risk factors for major bleeding and elements relevant to all patients taking OACs in the contemporary population of AF in Japan. The performance of this score appears to be acceptable for patients of advanced age or those with renal dysfunction. Given that patients with an annual incidence of major bleeding ＞2%–3%/year are at ‘high risk,’ patients with an HED-[EPA]2-B3 score ≥ 4 in the present cohort were considered at high risk for major bleeding. Physicians are encouraged to modify reversible factors such as anemia, blood pressure, and unnecessary concomitant antiplatelets in daily practice.
Limitations
Several limitations associated with the present study warrant mention. The present results were not prospectively validated using external data. Medications and indications for OACs were selected at the discretion of the attending physician. We also had no data on the therapeutic range for individual patients taking warfarin. We collected prescription data at baseline; however, we did not know the date of interruption or resumption of drugs, including OAC. The number of patients receiving DOACs was relatively small, which limits the utility of this score in the contemporary DOAC era and warrants further investigation. In addition, we do not know the exact type of OACs used at the time of major bleeding. The definition of major bleeding varied among the five registries, which may have affected the results. Bleeding risk is dynamic and may change with age and incident risk factors or by mitigation of modifiable risk factors25, 26). Finally, this study involved Japanese patients with AF; therefore, the study results may not be generalizable to other populations.
Conclusion
We demonstrated that advanced age, uncontrolled hypertension, history of bleeding, anemia, thrombocytopenia, and concomitant use of antiplatelets were significantly associated with an increased incidence of major bleeding. Our novel risk stratification system, the HED-[EPA]2-B3 score, may be useful for identifying Japanese patients with OACs at a risk of major bleeding.
Funding
This research was supported by the Practical Research Project for Lifestyle related Diseases including Cardiovascular Diseases and Diabetes Mellitus from the Japan Agency for Medical Research and Development, AMED (19ek0210082h0003).
Conflicts of Interest
Dr Akao received research funding from Bayer, and Daiichi-Sankyo, and Speakers’ Bureau/Honorarium from Pfizer, Bristol-Myers Squibb, Boehringer Ingelheim, Bayer and Daiichi Sankyo; Dr. Tomita received research funding from Boehringer Ingelheim, Bayer, Daiichi-Sankyo, and Pfizer, and Speakers’ Bureau/Honorarium from Boehringer Ingelheim, Bayer, Daiichi-Sankyo, and Bristol-Myers Squibb; Dr. Kodani received remuneration from Daiichi-Sankyo; Dr. Suzuki received Speakers’ Bureau/Honorarium from Daiichi-Sankyo and Bristol-Myers Squibb; Dr. Hayashi received Speakers’ Bureau/Honorarium from Bayer, Daiichi-Sankyo, and Bristol-Myers Squibb; Dr. Sawano received lecture fees from Boehringer Ingelheim, Bristol-Myers Squibb, Astellas Pharma, Sanofi, and research funding from Takeda Pharmaceutical; Dr. Goya received Speakers’ Bureau/Honorarium from Daiichi-Sankyo, Abbott, and Japan Life Line; Dr. Yamashita received research funding from Daiichi-Sankyo and Bristol-Myers Squibb, and Speakers’ Bureau/Honorarium from Daiichi-Sankyo, Bristol-Myers Squibb, Bayer, Ono Pharmaceutical, Boehringer Ingelheim, Novartis, Otsuka Pharmaceutical and Toa Eiyo; Dr. Isobe received research funding from Medtronic, Abbott, Boston Scientific, BIOTRONIK, Japan Life Line, Terumo, NIPRO, DVx, Active Medical, TORAY INDUSTRIES, KANEKA MEDIX, Johnson & Johnson, and Boehringer Ingelheim; Dr. Toyoda received Speakers’ Bureau/Honorarium from Daiichi-Sankyo, Otsuka Pharmaceutical, Bayer, Bristol-Myers-Squibb, and Novartis; Dr. Okamura received Speakers’ Bureau/Honorarium from Bayer and Daiichi-Sankyo. The remaining authors declare no relevant conflicts of interest.
References
- 1) Hart RG, Pearce LA and Aguilar MI: Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med, 2007; 146: 857-867
- 2) Ruff CT, Giugliano RP, Braunwald E, Hoffman EB, Deenadayalu N, Ezekowitz MD, Camm AJ, Weitz JI, Lewis BS, Parkhomenko A, Yamashita T and Antman EM: Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: a meta-analysis of randomised trials. Lancet, 2014; 383: 955-962
- 3) Landefeld CS and Beyth RJ: Anticoagulant-related bleeding: clinical epidemiology, prediction, and prevention. Am J Med, 1993; 95: 315-328
- 4) Hylek EM, Held C, Alexander JH, Lopes RD, De Caterina R, Wojdyla DM, Huber K, Jansky P, Steg PG, Hanna M, Thomas L, Wallentin L and Granger CB: Major bleeding in patients with atrial fibrillation receiving apixaban or warfarin: The ARISTOTLE Trial (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation): Predictors, Characteristics, and Clinical Outcomes. J Am Coll Cardiol, 2014; 63: 2141-2147
- 5) Ogawa H, An Y, Ishigami K, Ikeda S, Doi K, Hamatani Y, Fujino A, Ishii M, Iguchi M, Masunaga N, Esato M, Tsuji H, Wada H, Hasegawa K, Abe M, Lip GYH and Akao M: Long-term clinical outcomes after major bleeding in patients with atrial fibrillation: The Fushimi AF Registry. Eur Heart J Qual Care Clin Outcomes, 2020; 7: 163-171
- 6) Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ and Lip GY: A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest, 2010; 138: 1093-1100
- 7) Fang MC, Go AS, Chang Y, Borowsky LH, Pomernacki NK, Udaltsova N and Singer DE: A new risk scheme to predict warfarin-associated hemorrhaging: The ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) Study. J Am Coll Cardiol, 2011; 58: 395-401
- 8) O’Brien EC, Simon DN, Thomas LE, Hylek EM, Gersh BJ, Ansell JE, Kowey PR, Mahaffey KW, Chang P, Fonarow GC, Pencina MJ, Piccini JP and Peterson ED: The ORBIT bleeding score: a simple bedside score to assess bleeding risk in atrial fibrillation. Eur Heart J, 2015; 36: 3258-3264
- 9) Aggarwal R, Ruff CT, Virdone S, Perreault S, Kakkar AK, Palazzolo MG, Dorais M, Kayani G, Singer DE, Secemsky E, Piccini J, Tahir UA, Shen C and Yeh RW: Development and validation of the DOACs Score: A novel bleeding risk prediction tool for patients with atrial fibrillation on direct-acting oral anticoagulants. Circulation, 2023; 148: 936-946
- 10) Okumura K, Tomita H, Nakai M, Kodani E, Akao M, Suzuki S, Hayashi K, Sawano M, Goya M, Yamashita T, Fukuda K, Ogawa H, Tsuda T, Isobe M, Toyoda K, Miyamoto Y, Miyata H, Okamura T and Sasahara Y: Risk factors associated with ischemic stroke in Japanese patients with nonvalvular atrial fibrillation. JAMA Netw Open, 2020; 3: e202881
- 11) Group JJW: Guidelines for Pharmacotherapy of Atrial Fibrillation (JCS 2013). Circ J, 2014; 78: 1997-2021
- 12) Hori M, Connolly SJ, Ezekowitz MD, Reilly PA, Yusuf S and Wallentin L: Efficacy and safety of dabigatran vs. warfarin in patients with atrial fibrillation—sub-analysis in Japanese population in RE-LY trial. Circ J, 2011; 75: 800-805
- 13) Hori M, Matsumoto M, Tanahashi N, Momomura S, Uchiyama S, Goto S, Izumi T, Koretsune Y, Kajikawa M, Kato M, Ueda H, Iwamoto K and Tajiri M: Rivaroxaban vs. warfarin in Japanese patients with atrial fibrillation - The J-ROCKET AF study. Circ J, 2012; 76: 2104-2111
- 14) Ogawa S, Shinohara Y and Kanmuri K: Safety and efficacy of the oral direct factor Xa inhibitor apixaban in Japanese patients with non-valvular atrial fibrillation. -The ARISTOTLE-J study. Circ J, 2011; 75: 1852-1859
- 15) Shimada YJ, Yamashita T, Koretsune Y, Kimura T, Abe K, Sasaki S, Mercuri M, Ruff CT and Giugliano RP: Effects of Regional Differences in Asia on Efficacy and Safety of Edoxaban Compared With Warfarin--Insights From the ENGAGE AF-TIMI 48 Trial. Circ J, 2015; 79: 2560-2567
- 16) Friberg L, Rosenqvist M and Lip GY: Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J, 2012; 33: 1500-1510
- 17) Ishii M, Ogawa H, Unoki T, An Y, Iguchi M, Masunaga N, Esato M, Chun YH, Tsuji H, Wada H, Hasegawa K, Abe M and Akao M: Relationship of hypertension and systolic blood pressure with the risk of stroke or bleeding in patients with atrial fibrillation: The Fushimi AF Registry. Am J Hypertens, 2017; 30: 1073-1082
- 18) Kodani E, Tomita H, Nakai M, Akao M, Suzuki S, Hayashi K, Sawano M, Goya M, Yamashita T, Fukuda K, Ogawa H, Tsuda T, Isobe M, Toyoda K, Miyamoto Y, Miyata H, Okamura T, Sasahara Y and Okumura K: Impact of baseline blood pressure on adverse outcomes in Japanese patients with non-valvular atrial fibrillation: the J-RISK AF. Eur Heart J Open, 2022; 2: oeac081
- 19) Yasuda S, Kaikita K, Akao M, Ako J, Matoba T, Nakamura M, Miyauchi K, Hagiwara N, Kimura K, Hirayama A, Matsui K and Ogawa H: Antithrombotic therapy for atrial fibrillation with stable coronary disease. N Engl J Med, 2019; 381: 1103-1113
- 20) Park J, Cha MJ, Choi YJ, Lee E, Moon I, Kwak S, Kwon S, Yang S, Lee S, Choi EK and Oh S: Prognostic efficacy of platelet count in patients with nonvalvular atrial fibrillation. Heart Rhythm, 2019; 16: 197-203
- 21) Kodani E, Inoue H, Atarashi H, Okumura K, Yamashita T and Origasa H: Impact of hemoglobin concentration and platelet count on outcomes of patients with non-valvular atrial fibrillation: A subanalysis of the J-RHYTHM Registry. Int J Cardiol, 2020; 302: 81-87
- 22) Iijima R, Tokue M, Nakamura M, Yasuda S, Kaikita K, Akao M, Ako J, Matoba T, Miyauchi K, Hagiwara N, Kimura K, Hirayama A, Matsui K and Ogawa H: Thrombocytopenia as a bleeding risk factor in atrial fibrillation and coronary artery disease: Insights from the AFIRE Study. J Am Heart Assoc, 2023; 12: e031096
- 23) Fordyce CB, Hellkamp AS, Lokhnygina Y, Lindner SM, Piccini JP, Becker RC, Berkowitz SD, Breithardt G, Fox KA, Mahaffey KW, Nessel CC, Singer DE and Patel MR: On-treatment outcomes in patients with worsening renal function with rivaroxaban compared with warfarin: Insights from ROCKET AF. Circulation, 2016; 134: 37-47
- 24) Lip GYH, Skjoth F, Nielsen PB, Kjaeldgaard JN and Larsen TB: The HAS-BLED, ATRIA, and ORBIT Bleeding Scores in atrial fibrillation patients using non-vitamin K antagonist oral anticoagulants. Am J Med, 2018; 131: 574.e513-574.e527
- 25) Chao TF, Lip GYH, Lin YJ, Chang SL, Lo LW, Hu YF, Tuan TC, Liao JN, Chung FP, Chen TJ and Chen SA: Incident risk factors and major bleeding in patients with atrial fibrillation treated with oral anticoagulants: A comparison of baseline, follow-up and Delta HAS-BLED scores with an approach focused on modifiable bleeding risk factors. Thromb Haemost, 2018; 118: 768-777
- 26) Esteve-Pastor MA, Rivera-Caravaca JM, Shantsila A, Roldan V, Lip GYH and Marin F: Assessing bleeding risk in atrial fibrillation patients: Comparing a bleeding risk score based only on modifiable bleeding risk factors against the HAS-BLED score. The AMADEUS Trial. Thromb Haemost, 2017; 117: 2261-2266
