TR-05545; No of Pages 6 Thrombosis Research xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Thrombosis Research journal homepage: www.elsevier.com/locate/thromres
Regular Article
The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience☆ Geoffrey D. Barnes a,⁎, Xiaokui Gu a, Brian Haymart a, Eva Kline-Rogers a, Steve Almany b, Jay Kozlowski c, Dennis Besley d, Gregory D. Krol e, James B. Froehlich a, Scott Kaatz f a
Cardiovascular Center, University of Michigan Health System, Ann Arbor, MI William Beaumont Hospital, Troy, MI Huron Valley Sinai Hospital, Commerce, MI d West Michigan Heart, Grand Rapids, MI e Henry Ford Hospital, Detroit, MI f Hurley Medical Center, Flint, MI b c
a r t i c l e
i n f o
Article history: Received 11 December 2013 Received in revised form 22 April 2014 Accepted 20 May 2014 Available online xxxx Keywords: Anticoagulants Atrial Fibrillation Risk Factors Stroke Warfarin
a b s t r a c t Introduction: Guidelines recommend the assessment of stroke and bleeding risk before initiating warfarin anticoagulation in patients with atrial fibrillation. Many of the elements used to predict stroke also overlap with bleeding risk in atrial fibrillation patients and it is tempting to use stroke risk scores to efficiently estimate bleeding risk. Comparison of stroke risk scores to bleeding risk scores to predict bleeding has not been thoroughly assessed. Methods: 2600 patients followed at seven anticoagulation clinics were followed from October 2009-May 2013. Five risk models (CHADS2, CHA2DS2-VASc, HEMORR2HAGES, HAS-BLED and ATRIA) were retrospectively applied to each patient. The primary outcome was the first major bleeding event. Area under the ROC curves were compared with C statistic and net reclassification improvement (NRI) analysis was performed. Results: 110 patients experienced a major bleeding event in 2581.6 patient-years (4.5%/year). Mean follow up was 1.0 ± 0.8 years. All of the formal bleeding risk scores had a modest predictive value for first major bleeding events (C statistic 0.66-0.69), performing better than CHADS2 and CHA2DS2-VASc scores (C statistic difference 0.10 - 0.16). NRI analysis demonstrated a 52-69% and 47-64% improvement of the formal bleeding risk scores over the CHADS2 score and CHA2DS2-VASc score, respectively. Conclusions: The CHADS2 and CHA2DS2-VASc scores did not perform as well as formal bleeding risk scores for prediction of major bleeding in non-valvular atrial fibrillation patients treated with warfarin. All three bleeding risk scores (HAS-BLED, ATRIA and HEMORR2HAGES) performed moderately well. © 2014 Elsevier Ltd. All rights reserved.
Introduction Atrial fibrillation (AF) is known to significantly increase the risk of stroke and thromboembolism, leading to significant morbidity and mortality. For over 40 years, warfarin has been used to reduce the risk
Abbreviations: ACS, Anticoagulation clinics; AF, Atrial Fibrillation; BCBSM/BCN, Blue Cross Blue Shield of Michigan/Blue Care Network; MAQI2, Michigan Anticoagulation Quality Improvement Initiative; NRI, Net Reclassification Improvement; ROC, Receiver Operator Curve; TTR, Time in Therapeutic Range. ☆ Part of this data was presented as an oral abstract at the XXIV Congress of the International Society of Thrombosis and Haemostasis on July 1, 2013 in Amsterdam, Netherlands. ⁎ Corresponding author at: University of Michigan Cardiovascular Center, 1500 E Medical Center Dr – SPC 5853, Ann Arbor, MI 48109-5853. Tel.: + 1 734 998 5909; fax: + 1 734 998 9939. E-mail address: [email protected] (G.D. Barnes).
by up to 60% [1]. However, the beneficial effects of warfarin must be balanced with the increased risk of major bleeding [2]. Guidelines recommend taking into account both thromboembolic and bleeding risk when considering stroke prevention therapy for AF [3]. Prediction of thromboembolic risk is predominately performed using the CHADS2 or CHA2DS2-VASc risk scoring systems [4,5]. Multiple scoring systems have been proposed to predict the risk of major bleeding in AF populations, including the HEMORR2HAGES, HAS-BLED and ATRIA scores [6–8]. Choosing the most effective scoring system has been the topic of debate in the literature [9–13]. Additionally, use of bleeding risk assessment tools, especially the HAS-BLED score, has been endorsed by several guidelines for management of AF [14–16]. With overlap of some of the risks factors for stoke or bleeding in the risk prediction rules, the CHADS2 scoring system has been associated with an increased risk of both stroke and bleeding in patients with AF [17–19]. In this study, we retrospectively compared the CHADS2 and
http://dx.doi.org/10.1016/j.thromres.2014.05.034 0049-3848/© 2014 Elsevier Ltd. All rights reserved.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
2
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
CHA2DS2-VASc stroke scores to the HEMORR2HAGES, HAS-BLED and ATRIA bleeding scoring systems for predicting major bleeding in a contemporary, “real-world” population of non-valvular AF patients treated with warfarin.
Methods MAQI2 Collaborative The Michigan Anticoagulation Quality Improvement Initiative (MAQI 2 ) is a Blue Cross Blue Shield of Michigan/Blue Care Network (BCBSM/BCN) sponsored continuous quality improvement consortium. Full details of the MAQI2 consortium have previously been published [20,21]. Briefly, the aim of the MAQI2 collaborative is to describe anticoagulation care across the state, identify best practices and procedures associated with best outcomes, and to partner in quality improvement projects. The collaborative was formed in 2008 with initial patient enrollment starting in October 2009. As a BCBSM/BCN-funded quality improvement collaborative, selected anticoagulation clinics (ACS) in the state of Michigan were invited to participate. Participating clinics are provided reimbursement by BCBSM/BCN to cover the costs of participation in MAQI2, including data abstraction and associated quality improvement efforts. All data abstractors undergo training and each center undergoes regular audits to ensure high quality data collection. Data collected is verified by random audits to ensure agreement with pre-defined data element definitions, including the primary outcomes for this study.
Patient Selection Between October 2009 and December 2012, 2600 new patients with non-valvular AF were enrolled in MAQI2. Patients were identified at the time of ACS enrollment in this inception cohort and clinical data was abstracted from the individual ACS database and hospital or group medical records.
Risk Stratification The CHADS2, CHA2DS2-VASc, HEMORR2HAGES, HAS-BLED and ATRIA scoring systems were retrospectively calculated and patients were grouped into low, intermediate and high risk as noted in Table 1 [4–8,10,19,22]. Presence of an aortic plaque was not available and not included in the calculation of vascular disease for the CHA2DS2-VASc score. Genetic factors associated with increased risk for bleeding were not available and not included in the calculation of the HEMORR2HAGES score. Labile INR was not included in the HAS-BLED score as this data was not available at the time of warfarin initiation. Time in the therapeutic range (TTR) was calculated using the Rosendaal method [23]. Study Endpoints The primary endpoint was occurrence of first major bleeding event. Major bleeding was defined according to the International Society of Thrombosis and Haemostasis consensus, which includes fatal bleeding, bleeding into a critical organ, overt bleeding requiring transfusion of 2 + units of red blood cells or an overt bleed causing a hemoglobin fall of 2+ g/dL [24]. Statistical Analysis Because of censored data, the association of patient demographics and comorbidities with the bleeding outcome was evaluated by using Cox regression analysis. We determined the cumulative incidence of first major bleeding in all patients. Patients were stratified according to the various risk scoring systems. Kaplan-Meier analysis was performed to determine the likelihood of a major bleeding event at 1 year, as well as to generate the individual bleeding risks at 1 year for the various bleeding risk scores. For the dichotomous categorical bleeding risk analysis, multiple analyses were performed. Based on the accepted low/moderate/high risk cut offs, we compared low/moderate vs high and low vs moderate/high. We also used the median bleeding risk score of all patients to create low and high risk bleeding groups.
Table 1 Bleeding Risk Scores. Risk Elements
CHADS2
CHA2DS2-VASc
HEMORR2HAGES
HAS-BLED
ATRIA
CHF HTN Age
1 point 1 point 1 point (≥75)
1 point 1 point (N75)
1 point 1 point (N65)
1 point 2 points (≥75)
DM Prior Stroke or TIA CAD, PAD or Aortic plaque Female Gender Chronic Liver Disease or Cirrhosis Chronic Renal Insufficiency Heavy Alcohol Use Malignancy Thrombocytopenia or Antiplatelet Use Prior Bleeding Event Anemia History of Falls Genetic Factors Prior Bleeding Event or Anemia TTR b 60% Use of ASA, clopidogrel, prasugrel, ticagrelor or NSAIDs Low Intermediate High
1 point 2 points 0-1 2 3+
1 point 1 point 2 points (≥75) 1 point (65-74) 1 point 2 points 1 point 1 point 0-1 2 3+
1 point 1 point 1 point 1 point 1 point 1 point 2 point 1 point 1 point 1 point 0-1 2-3 4+
1 point 1 point 1 point 1 point 1 point 1 point 1 point 0 1-2 3+
3 points 1 point 3 points 0-3 4 5-10
CHF – congestive heart failure, HTN – hypertension (N140/90 or use of antihypertensive medications), DM – diabetes mellitus, TIA – transient ischemic attack, CAD – coronary artery disease, PAD – peripheral artery disease, ASA – aspirin, TTR – time in therapeutic range, NSAIDs – non-steroidal anti-inflammatory drugs.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
3
Table 2 Demographic and Comorbidity Characteristics.
Patient-years (mean ± SD) Age (mean ± SD) Female BMI (mean ± SD) HTN DM CAD CHF Cancer Current Tobacco Use Renal Disease Liver Disease Stroke Bleeding Diathesis Age N 65 Concurrent drugs Drugs/EtOH HAS-BLED score (mean ± SD) ATRIA score (mean ± SD) HEMORR2HAGES score (mean ± SD) CHADS2 score (mean ± SD) CHA2DS2-VASc score (mean ± SD) TTR (mean ± SD)
Total N = 2600
Patients with major bleeding event N = 110
Patients without major bleeding event N = 2490
P-value
1.0 ± 0.8 70.1 ± 12.8 1085 (41.7) 30.2 ± 6.8 1937 (74.5) 655 (25.2) 847 (32.6) 629 (24.2) 412 (15.8) 154 (5.9) 310 (11.9) 17 (0.7) 300 (11.5) 815 (31.3) 1748 (67.2) 1466 (56.4) 126 (4.8) 2.6 ± 1.3 3.0 ± 2.4 2.6 ± 1.4 1.9 ± 1.2 3.4 ± 1.8 59.3 ± 22.7
1.0 ± 0.8 73.6 ± 10.9 51 (46.4) 29.8 ± 6.5 90 (81.8) 39 (35.5) 47 (42.7) 34 (30.9) 27 (24.5) 6 (5.5) 33 (30.0) 1 (0.9) 15 (13.6) 71 (54.5) 87 (79.1) 78 (70.9) 11 (10.0) 3.5 ± 1.2 4.7 ± 2.6 3.5 ± 1.5 2.2 ± 1.2 4.0 ± 1.7 54.6 ± 19.6
1.0 ± 0.8 70.0 ± 12.9 1034 (41.5) 30.3 ± 6.9 1847 (74.2) 616 (24.7) 800 (32.1) 595 (23.9) 385 (15.5) 148 (5.9) 281 (11.3) 16 (0.6) 285 (11.4) 744 (29.9) 1661 (66.7) 1388 (55.7) 115 (4.6) 2.5 ± 1.2 3.0 ± 2.4 2.6 ± 1.4 1.9 ± 1.2 3.4 ± 1.8 59.5 ± 22.8
0.78 0.085 0.65 0.51 0.27 0.018 0.022 0.09 0.025 0.92 b0.001 0.65 0.75 b0.001 0.0012 0.0012 0.016 b0.001 b0.001 b0.001 0.020 0.0026 b0.0001
BMI – Body-mass index, TTR – Time in therapeutic range, HTN – hypertension, DM – Diabetes mellitus, CAD – coronary artery disease, CHF – congestive heart failure.
Time-dependent receiver operating characteristic (ROC) curves were generated and the area under the ROC curve (C statistic) was used to compare the predictability of both non-categorical (using the individual scores) and category-based (dichotomous) risk schemes. Net reclassification improvement (NRI) analysis on censored data was performed for two-category based risk groups [25]. A 2-sided P b 0.05 was considered statistically significant for all analyses. All analyses were performed with statistical software SAS and R.
The formal bleeding risk scores (HAS-BLED, ATRIA and HEMORR2HAGES) had a modest predictive value for first major bleeding events (C statistic 0.66-0.69), performing better than CHADS2 and CHA2DS2VASc scores (Table 4). The CHADS2 score had a C statistic of 0.53 (95% CI 0.47-0.60) and the CHA2DS2-VASc score had a C statistic of 0.56 (95% CI 0.49-0.62). The HAS-BLED score had a C statistic that was similar to the ATRIA and HEMORR2HAGES scores, 0.018 (95% CI -0.038–0.074) and 0.032 (95% CI -0.021–0.067) in difference, respectively (Table 4).
Results
Clinical Risk Group
110 individual patients experienced major bleeding events during 2581.6 patient-years of follow-up (4.5%/year). Mean follow up was 1.0 ± 0.8 years for patients with and without bleeding events. Patients with bleeding events had higher CHADS 2, CHA2DS2-VASc, HEMORR2 HAGES, HAS-BLED and ATRIA scores (Table 2). Further demographic and comorbidity characteristics are summarized in Table 2. Patients with bleeding events had mean TTR 54.6 ± 19.6 while patients without bleeding events had a mean TTR 59.5 ± 22.8 (p b 0.0001). Median scores for all patients are listed in Table 3. Event rates for major bleeding are shown in Table 3. The predictive abilities of the five risk scores (non-categorical) are presented in Fig. 1.
NRI analysis was performed using dichotomous risk groups to compare the ability to better classify patients with bleeding events into the higher risk group and patients without bleeding events into the lower risk group. The NRI was highest when risk groups were defined based on the median scores (Appendix A). NRI demonstrated improvement of the HAS-BLED, ATRIA and HEMORR2HAGES scores over the CHADS2 score and a 36-54% improvement over the CHA2DS2VASc score (Table 4). The HAS-BLED score also demonstrated a 26% and 31% improvement over the ATRIA and HEMORR2HAGES scores, respectively (Table 4). NRI did not demonstrate superiority of the HASBLED score compared to the other bleeding risk scores if the risk groups
Table 3 Major bleeding event rates. Bleeding Risk
HAS-BLED
ATRIA
HEMORR2HAGES
CHADS2
Median Score
3
3
2
2
3
Low Moderate High Low High C statistic at 12 Months (non-categorical)
0 1.3 (0.03–7.3) 1-2 2.0 (1.3–3.1) 3+ 6.6 (5.3–8.0) 0-3 2.7 (2.0–3.5) 4+ 10.4 (8.0–13.4) 0.69 (0.63–0.75)
0-3 2.3 (1.7–3.2) 4 7.4 (4.6–11.3) 5+ 9.1 (6.8–11.8) 0-3 2.3 (1.7-3.2) 4+ 8.5 (6.7–10.7) 0.67 (0.61-0.74)
0-1 1.7 (0.7–3.3) 2 3.6 (2.7-4.8) 3+ 8.5 (6.4–11.0) 0-2 2.1 (1.4-3.0) 3+ 6.8 (5.4–8.3) 0.66 (0.61-0.74)
0-1 3.3 (2.2-4.7) 2 4.8 (3.5-6.4) 3+ 5.6 (4.0-7.6) 0-2 4.1 (3.2-5.1) 3+ 5.6 (4.0-7.6) 0.53 (0.47-0.60)
0-1 3.4 (1.6-6.4) 2 3.3 (1.8-5.7) 3+ 4.9 (3.9-6.0) 0-3 3.3 (2.3-4.4) 3+ 5.6 (4.4-7.0) 0.56 (0.49 – 0.62)
CHA2DS2-VASc
Number of major bleeding events per 100-patient-years (95% CI).
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
4
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
Fig. 1. Receiver Operating Characteristic (ROC) Curves and C Statistics for the Risk Scores (non-categorical) at 12 Months.
were defined by actual scores and when compared as low/moderate vs. high or low vs. moderate/high (Appendix A).
Discussion In this study of a “real-world” inception cohort of non-valvular AF patients treated with warfarin for stroke prevention, our data show that the three bleeding risk scores performed better than the CHADS2 [4] and CHA2DS2-VASc [5] scores for the prediction of a first major bleeding event. By NRI analysis, the HAS-BLED [7] score was superior to the other two established bleeding risk scores (ATRIA [8] and HEMORR2HAGES [6] for the prediction of a first major bleeding event. Consistent with prior reports, all three of the formal bleeding risk models had only moderate predictive ability [9,12,13,22,26,27]. Despite the knowledge that the bleeding risk increased with the CHADS2 score [17–19], we believe that this is the first report to perform a complete comparison of the CHADS2 and CHA2DS2-VASc scores to the three established bleeding risk models (HAS-BLED, ATRIA and HEMORR2HAGES) in a diverse, multi-centered dataset. While patients experiencing major bleeding events had higher CHADS2 and CHA2DS2VASc scores, these stroke risks scores are not as predictive as bleeding risk scores, consistent with other recent reports [26,27]. Therefore, a formalized bleeding risk score should be used when clinicians estimate the bleeding risk for patients with non-valvular AF when considering anticoagulation treatment. Clinically, this highlights the importance of using a stroke risk score (e.g. CHADS 2 or CHA 2 DS 2 -VASc) for predicting thromboembolism risk while using a bleeding risk score (e.g. HAS-BLED) to predict bleeding risk separately.
Although easy to use clinically, all available stroke and bleeding risk models have been limited by their ability to predict clinically relevant events. Prior analyses of the CHADS2 and CHA2DS2-VASc scores have demonstrated C statistics between 0.51-0.82 and 0.53-0.888, respectively, for the prediction of stroke in non-valvular AF [4,5,26,28–30]. This represents modest to good predictive ability. However, the bleeding risk models have continuously demonstrated only modest predictive abilities with C statistics between 0.57-0.795, 0.61-0.69 and 0.58-0.771 for the HAS-BLED, ATRIA and HEMORR2HAGES scores, respectively [6–8,13,26,27,31,32]. Our findings of C statistics of 0.66-0.69 for the three established bleeding risk models are in line with those previously published validation studies. Despite the increasing risk of bleeding with increasing CHADS2 score, a large Danish population registry analysis previously suggested a net clinical benefit for all patients receiving warfarin therapy, irrespective of their CHADS2 and HAS-BLED score profiles [33]. Additionally, there have been numerous studies highlighting the underuse of warfarin in eligible AF patient and a perceived risk of bleeding is regularly given as a common reason not to anticoagulate patients [34–36]. Although the bleeding risk is increased in patients at higher CHADS 2 and CHA2 DS2 -VASc scores in this study, this should not be a hindrance to their receiving the benefits of stroke prophylaxis. Therefore, it seems reasonable to focus on the underuse of warfarin in all patients with AF who are likely to derive benefit from stroke risk reduction instead of on the perceived bleeding risk, as predicted by provider intuition or bleeding risk scores. This study has a number of strengths. This is an inception cohort of “real world” patients from a diverse United States population, including both academic and private-practice centers in urban and suburban locations. Warfarin control in this study, as measured by the TTR, is on par with those of other real-world inception cohorts [37] but likely differs from those that preselected patients with stable INRs for a minimum of 6 months [12,27] or were analyzed from randomized trial populations [26]. We also abstracted and audited patient chart data, instead of relying on ICD9 billing data, to verify comorbidity elements and outcomes. This is a contemporary data set, with all patients initiating in warfarin since 2009. Still, there are a few limitations. First, since data are collected only at participating centers, bleeding events that occurred at non-participating centers may not have been captured by the ACSs participating in MAQI2. However, since all patients are contacted either face-to-face or over the telephone by nurses and/or pharmacists at the ACS, it is reasonable to assume that clinical events occurring at other hospital centers would be mentioned in the patient’s ACS chart and therefore captured in this dataset. We did not include cytochrome 2C9 or vitamin K oxide reductase genentics (HEMORR2HAGES) or labile INR measurement (HAS-BLED) scores, which may have effective their predictive accuracy. Also, since this is an observational inception cohort, our follow up time is variable based on when the patients initiated warfarin therapy. To correctly calculate the NRI in this setting, we generated
Table 4 Comparison of Risk Scores for First Major Bleeding Event at 12 Months.
HAS-BLED vs. ATRIA HAS-BLED vs. HEMORR2HAGES ATRIA vs. HEMORR2HAGES HAS-BLED vs. CHADS2 ATRIA vs. CHADS2 HEMORR2HAGES vs. CHADS2 HAS-BLED vs. CHA2DS2-VASc ATRIA vs. CHA2DS2-VASc HEMORR2HAGES vs. CHA2DS2-VASc CHADS2 vs CHA2DS2-VASc
C Statistic difference
95% CI
p-value
NRI
IQR
p-value
0.018 0.032 0.014 0.16 0.14 0.12 0.13 0.11 0.10 -0.024
-0.038–0.074 -0.0021–0.067 -0.034–0.062 0.096–0.22 0.071–0.21 0.065–0.18 0.073–0.19 0.047–0.18 0.039–0.16 -0.057–0.0084
0.53 0.065 0.56 b0.001 b0.001 0.03 b0.001 b0.001 0.031 0.15
0.26 0.31 0.34 0.58 0.59 0.54 0.36 0.40 0.54 -0.071
-0.40-0.42 -0.12-0.52 -0.06-0.50 0.018-0.79 0.32-0.80 0.26-0.74 0.18-0.71 0.22-0.80 0.014-0.73 -0.41-0.14
0.006 0.001 0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 0.25
C statistic difference calculated using ROC curve analysis for non-categorical scores. Net Reclassification Improvement (NRI) calculated using dichotomous risk group analysis based on median scores. IQR – Interquartile Range.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
one-year bleeding rates using a Kaplan-Meier estimator [38]. Lastly, our rate of major bleeding is slightly higher than those of other recently published studies (4.5%/year vs 1.0-3.2%/year) [12,26,27]. This likely corresponds to the real-world, American population with a median TTR that is slightly lower than commonly seen in European systems or randomized trial settings of patients previously treated with oral anticoagulants instead of an inception cohort. Further refinement of bleeding risk prediction would be helpful, but is likely challenging given the transient nature of many possible bleeding triggers, e.g. drug-drug interactions, non-compliance, trauma. While European and Canadian guidelines have favored the use of the HASBLED score for estimation of bleeding risk, American-based groups have not yet endorsed the use of any bleeding risk tool, citing their modest predictive capacity [3,14–16,39–40]. Additionally, further consideration about the integration of both bleeding and stroke risk prediction tools into a single tool for patient education and discussion of risks and benefits warrants further development and clinical testing. In summary, the three formal bleeding risk scores performed better than the CHADS2 and CHA2DS2-VASc scores for predicting major bleeding in non-valvular AF patients treated with warfarin. We recommend using a formal bleeding risk score instead of a stroke risk score when counseling patients about their risk of major bleeding when considering warfarin therapy for stroke prevention in AF, particularly in the highest risk patients. Authorship Details G Barnes conceived the study idea, designed the study and drafted the initial manuscript. X Gu performed all statistical analysis. B Haymart, E Kline-Rogers, S Almany, J Kozlowski, D Besley, G Grol and J Froehlich provided critical analysis and substantial edits to the study design and manuscript revisions. S Kaatz provided direct oversight of the study design, statistical analysis and manuscript editing with critical and substantial revisions. Conflict of Interest Statements Geoff Barnes – No conflicts to report Xiaokui Gu – No conflicts to report Brian Haymart – No conflicts to report Eva Kline-Rogers – 45% of salary is funded by BCBSM Steve Almany – No conflicts to report Jay Kozlowski – No conflicts to report Dennis Besley – No conflicts to report Gregory Krol – No conflicts to report James Froehlich – Consultant for: Sanofi-Aventis, Ortho-McNeil, Merck. Research grants from Sanofi-Aventis, Blue Cross/Blue Shield of Michigan, Mardigian Foundation, Fibromuscular Disease Society of America. • Scott Kaatz – Grant support • Boehringer-Ingelheim • Bristol Myer Squibb • Bayer/Jansen/Johnson and Johnson • Eisai • Iverson Genetics Diagnostics/Medicare • National Institute of Health • Canadian Institute of Health Research • Blue Cross/Blue Shield of Michigan • Speaker honorarium • Jansen/Johnson and Johnson • Boehringer-Ingelheim • Bristol Myer Squibb/Pfizer • Consultant • Boehringer Ingelheim • Bristol Myer Squibb/Pfizer
5
• Jansen/Johnson and Johnson • Daiichi Sankyo • Board membership (non-profit) • AC Forum • National Certification Board of Anticoagulation Providers • National Blood Clot Alliance Medical and Scientific Advisory Board Acknowledgement Blue Cross-Blue Shield of Michigan provided funding for data collection and analysis, but was not involved in the interpretation of the data, the decision to publish or any revisions of the manuscript.
Appendix A Category-based NRI at 12 Months
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Low vs. Mod/High NRI (IQR) -0.55 (-0.76 – -0.33) -0.11 (-0.17 – -0.072) -0.042 (-0.31 – 0.051) -0.018 (-0.094 – 0.051) 0.46 (0.15 – 0.65) 0.46 (0.28 – 0.77) 0.53 (0.29 – 0.76) 0.15 (-0.057 – 0.28) 0.14 (-0.0025 – 0.22) 0.035 (-0.14 – 0.19)
p-value b0.001 0.001 0.51 0.61 b0.001 b0.001 b0.001 0.018 0.002 0.50
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Low/Mod vs. High NRI (IQR) 0.097 (-0.36 – 0.31) 0.13 (-0.083 – 0.48) 0.55 (0.28 – 0.72) 0.55 (0.79 – 0.20) 0.20 (-0.19 – 0.65) 0.47 (0.18 – 0.69) 0.38 (0.12 – 0.62) 0.39 (0.084 – 0.61) 0.28 (-0.011 – 0.56) -0.16 (-0.26 – 0.22)
p-value 0.39 0.23 b0.001 b0.001 0.017 b0.001 b0.001 0.001 0.006 0.89
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Median as cut-off NRI (IQR) 0.26 (-0.40 – 0.42) 0.31 (-0.12 – 0.52) 0.58 (0.18 – 0.79) 0.36 (0.18 – 0.71) 0.34 (-0.061 – 0.50) 0.59 (0.32 – 0.80) 0.40 (0.22 – 0.80) 0.54 (0.26 – 0.74) 0.54 (0.014 – 0.73) -0.071 (-0.41 – 0.14)
p-value 0.006 0.001 b0.001 b0.001 0.001 b0.001 b0.001 b0.001 b0.001 0.25
References [1] Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med 2007;146:857–67. [2] Palareti G, Cosmi B. Bleeding with anticoagulation therapy - who is at risk, and how best to identify such patients. Thromb Haemost 2009;102:268–78. [3] You JJ, Singer DE, Howard PA, Lane DA, Eckman MH, Fang MC, et al. Antithrombotic therapy for atrial fibrillation: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012;141:e531S–75S. [4] Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA 2001;285:2864–70.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
6
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
[5] Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factorbased approach: the euro heart survey on atrial fibrillation. Chest 2010;137:263–72. [6] Gage BF, Yan Y, Milligan PE, Waterman AD, Culverhouse R, Rich MW, et al. Clinical classification schemes for predicting hemorrhage: results from the National Registry of Atrial Fibrillation (NRAF). Am Heart J 2006;151:713–9. [7] Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, 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;138:1093–100. [8] Fang MC, Go AS, Chang Y, Borowsky LH, Pomernacki NK, Udaltsova N, et al. A new risk scheme to predict warfarin-associated hemorrhage: The ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) Study. J Am Coll Cardiol 2011;58:395–401. [9] Apostolakis S, Lane DA, Guo Y, Buller H, Lip GY. Performance of the HEMORR(2) HAGES, ATRIA, and HAS-BLED bleeding risk-prediction scores in patients with atrial fibrillation undergoing anticoagulation: the AMADEUS (evaluating the use of SR34006 compared to warfarin or acenocoumarol in patients with atrial fibrillation) study. J Am Coll Cardiol 2012;60:861–7. [10] Apostolakis S, Lane DA, Guo Y, Buller H, Lip GY. Performance of the HEMORR(2) HAGES, ATRIA, and HAS-BLED Bleeding Risk-Prediction Scores in Nonwarfarin Anticoagulated Atrial Fibrillation Patients. J Am Coll Cardiol 2013;61:386–7. [11] Lip GY, Lin HJ, Hsu HC, Su TC, Chen MF, Lee YT, et al. Comparative assessment of the HAS-BLED score with other published bleeding risk scoring schemes, for intracranial haemorrhage risk in a non-atrial fibrillation population: The Chin-Shan Community Cohort Study. Int J Cardiol 2013;57:173–80. [12] Roldán V, Marín F, Fernández H, Manzano-Fernandez S, Gallego P, Valdés M, et al. Predictive value of the HAS-BLED and ATRIA bleeding scores for the risk of serious bleeding in a "real-world" population with atrial fibrillation receiving anticoagulant therapy. Chest 2013;143:179–84. [13] Donzé J, Rodondi N, Waeber G, Monney P, Cornuz J, Aujesky D. Scores to predict major bleeding risk during oral anticoagulation therapy: a prospective validation study. Am J Med 2012;125:1095–102. [14] Cairns JA, Connolly S, McMurtry S, Stephenson M, Talajic M, Committee CAFG. Canadian Cardiovascular Society atrial fibrillation guidelines 2010: prevention of stroke and systemic thromboembolism in atrial fibrillation and flutter. Can J Cardiol 2011;27:74–90. [15] Lip GY, Andreotti F, Fauchier L, Huber K, Hylek E, Knight E, et al. Bleeding risk assessment and management in atrial fibrillation patients: a position document from the European Heart Rhythm Association, endorsed by the European Society of Cardiology Working Group on Thrombosis. Europace 2011;13:723–46. [16] Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Europace 2010;12:1360–420. [17] Suh DC, Choi JC, Schein J, Kim S, Nelson WW. Factors associated with warfarin discontinuation, including bleeding patterns, in atrial fibrillation patients. Curr Med Res Opin 2013;29:761–71. [18] Lopes RD, Al-Khatib SM, Wallentin L, Yang H, Ansell J, Bahit MC, et al. Efficacy and safety of apixaban compared with warfarin according to patient risk of stroke and of bleeding in atrial fibrillation: a secondary analysis of a randomised controlled trial. Lancet 2012;380:1749–58. [19] Oldgren J, Alings M, Darius H, Diener HC, Eikelboom J, Ezekowitz MD, et al. Risks for stroke, bleeding, and death in patients with atrial fibrillation receiving dabigatran or warfarin in relation to the CHADS2 score: a subgroup analysis of the RE-LY trial. Ann Intern Med 2011;155:660–7 [W204]. [20] Barnes GD, Kaatz S, Golgotiu V, Gu X, Leidal A, Kobeissy A, et al. Use of warfarin for venous thromboembolism prophylaxis following knee and hip arthroplasty: results of the Michigan Anticoagulation Quality Improvement Initiative (MAQI(2)). J Thromb Thrombolysis 2013;35:10–4. [21] Barnes GD, Kaatz S, Winfield J, Gu X, Haymart B, Kline-Rogers E, et al. Warfarin use in atrial fibrillation patients at low risk for stroke: analysis of the Michigan Anticoagulation Quality Improvement Initiative (MAQI(2)). J Thromb Thrombolysis 2014;37:171–6. [22] Lip GY, Banerjee A, Lagrenade I, Lane DA, Taillandier S, Fauchier L. Assessing the risk of bleeding in patients with atrial fibrillation: the Loire Valley Atrial Fibrillation project. Circ Arrhythm Electrophysiol 2012;5:941–8.
[23] Rosendaal FR, Cannegieter SC, van der Meer FJ, Briet E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69:236–9. [24] Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005;3:692–4. [25] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72 [discussion 207–12]. [26] Apostolakis S, Lane DA, Buller H, Lip GY. Comparison of the CHADS2, CHA2DS2 -VASc and HAS-BLED scores for the prediction of clinically relevant bleeding in anticoagulated patients with atrial fibrillation: The AMADEUS trial. Thromb Haemost 2013;110. [27] Roldan V, Marin F, Manzano-Fernandez S, Gallego P, Vilchez JA, Valdes M, et al. The HAS-BLED score has better prediction accuracy for major bleeding than the CHADS or CHADS-VASc scores In anticoagulated patients with atrial fibrillation. J Am Coll Cardiol 2013;62:2199–204. [28] Rietbrock S, Heeley E, Plumb J, van Staa T. Chronic atrial fibrillation: Incidence, prevalence, and prediction of stroke using the Congestive heart failure, Hypertension, Age N75, Diabetes mellitus, and prior Stroke or transient ischemic attack (CHADS2) risk stratification scheme. Am Heart J 2008;156:57–64. [29] Piccini JP, Stevens SR, Chang Y, Singer DE, Lokhnygina Y, Go AS, et al. Renal Dysfunction as a Predictor of Stroke and Systemic Embolism in Patients With Nonvalvular Atrial Fibrillation: Validation of the R2CHADS2 Index in the ROCKET AF (Rivaroxaban Oncedaily, oral, direct factor Xa inhibition Compared with vitamin K antagonism for prevention of stroke and Embolism Trial in Atrial Fibrillation) and ATRIA (AnTicoagulation and Risk factors In Atrial fibrillation) Study Cohorts. Circulation 2013;127:224–32. [30] Olesen JB, Lip GY, Hansen ML, Hansen PR, Tolstrup JS, Lindhardsen J, et al. Validation of risk stratification schemes for predicting stroke and thromboembolism in patients with atrial fibrillation: nationwide cohort study. BMJ 2011;342:d124. [31] Olesen JB, Lip GY, Hansen PR, Lindhardsen J, Ahlehoff O, Andersson C, et al. Bleeding risk in 'real world' patients with atrial fibrillation: comparison of two established bleeding prediction schemes in a nationwide cohort. J Thromb Haemost 2011;9:1460–7. [32] Lip GY, Frison L, Halperin JL, Lane DA. Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: the HASBLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol 2011;57:173–80. [33] Olesen JB, Lip GY, Lindhardsen J, Lane DA, Ahlehoff O, Hansen ML, et al. Risks of thromboembolism and bleeding with thromboprophylaxis in patients with atrial fibrillation: A net clinical benefit analysis using a 'real world' nationwide cohort study. Thromb Haemost 2011;106:739–49. [34] Wilke T, Groth A, Mueller S, Pfannkuche M, Verheyen F, Linder R, et al. Oral anticoagulation use by patients with atrial fibrillation in Germany. Adherence to guidelines, causes of anticoagulation under-use and its clinical outcomes, based on claims-data of 183,448 patients. Thromb Haemost 2012;107:1053–65. [35] Gussoni G, Di Pasquale G, Vescovo G, Gulizia M, Mathieu G, Scherillo M, et al. Decision making for oral anticoagulants in atrial fibrillation: The ATA-AF study. Eur J Intern Med 2013;24:324–32. [36] Rosenman MB, Baker L, Jing Y, Makenbaeva D, Meissner B, Simon TA, et al. Why is warfarin underused for stroke prevention in atrial fibrillation? A detailed review of electronic medical records. Curr Med Res Opin 2012;28:1407–14. [37] Gallego P, Roldan V, Marin F, Romera M, Valdes M, Vicente V, et al. Cessation of oral anticoagulation in relation to mortality and the risk of thrombotic events in patients with atrial fibrillation. Thromb Haemost 2013;110:1189–98. [38] Steyerberg EW, Pencina MJ. Reclassification calculations for persons with incomplete follow-up. Ann Intern Med 2010;152:195–6 [author reply 6-7]. [39] Fuster V, Rydén LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, et al. 2011 ACCF/ AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines. Circulation 2011;123:e269–367. [40] Furie KL, Goldstein LB, Albers GW, Khatri P, Neyens R, Turakhia MP, et al. Oral antithrombotic agents for the prevention of stroke in nonvalvular atrial fibrillation: a science advisory for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke 2012;43:3442–53.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
Contents lists available at ScienceDirect
Thrombosis Research journal homepage: www.elsevier.com/locate/thromres
Regular Article
The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience☆ Geoffrey D. Barnes a,⁎, Xiaokui Gu a, Brian Haymart a, Eva Kline-Rogers a, Steve Almany b, Jay Kozlowski c, Dennis Besley d, Gregory D. Krol e, James B. Froehlich a, Scott Kaatz f a
Cardiovascular Center, University of Michigan Health System, Ann Arbor, MI William Beaumont Hospital, Troy, MI Huron Valley Sinai Hospital, Commerce, MI d West Michigan Heart, Grand Rapids, MI e Henry Ford Hospital, Detroit, MI f Hurley Medical Center, Flint, MI b c
a r t i c l e
i n f o
Article history: Received 11 December 2013 Received in revised form 22 April 2014 Accepted 20 May 2014 Available online xxxx Keywords: Anticoagulants Atrial Fibrillation Risk Factors Stroke Warfarin
a b s t r a c t Introduction: Guidelines recommend the assessment of stroke and bleeding risk before initiating warfarin anticoagulation in patients with atrial fibrillation. Many of the elements used to predict stroke also overlap with bleeding risk in atrial fibrillation patients and it is tempting to use stroke risk scores to efficiently estimate bleeding risk. Comparison of stroke risk scores to bleeding risk scores to predict bleeding has not been thoroughly assessed. Methods: 2600 patients followed at seven anticoagulation clinics were followed from October 2009-May 2013. Five risk models (CHADS2, CHA2DS2-VASc, HEMORR2HAGES, HAS-BLED and ATRIA) were retrospectively applied to each patient. The primary outcome was the first major bleeding event. Area under the ROC curves were compared with C statistic and net reclassification improvement (NRI) analysis was performed. Results: 110 patients experienced a major bleeding event in 2581.6 patient-years (4.5%/year). Mean follow up was 1.0 ± 0.8 years. All of the formal bleeding risk scores had a modest predictive value for first major bleeding events (C statistic 0.66-0.69), performing better than CHADS2 and CHA2DS2-VASc scores (C statistic difference 0.10 - 0.16). NRI analysis demonstrated a 52-69% and 47-64% improvement of the formal bleeding risk scores over the CHADS2 score and CHA2DS2-VASc score, respectively. Conclusions: The CHADS2 and CHA2DS2-VASc scores did not perform as well as formal bleeding risk scores for prediction of major bleeding in non-valvular atrial fibrillation patients treated with warfarin. All three bleeding risk scores (HAS-BLED, ATRIA and HEMORR2HAGES) performed moderately well. © 2014 Elsevier Ltd. All rights reserved.
Introduction Atrial fibrillation (AF) is known to significantly increase the risk of stroke and thromboembolism, leading to significant morbidity and mortality. For over 40 years, warfarin has been used to reduce the risk
Abbreviations: ACS, Anticoagulation clinics; AF, Atrial Fibrillation; BCBSM/BCN, Blue Cross Blue Shield of Michigan/Blue Care Network; MAQI2, Michigan Anticoagulation Quality Improvement Initiative; NRI, Net Reclassification Improvement; ROC, Receiver Operator Curve; TTR, Time in Therapeutic Range. ☆ Part of this data was presented as an oral abstract at the XXIV Congress of the International Society of Thrombosis and Haemostasis on July 1, 2013 in Amsterdam, Netherlands. ⁎ Corresponding author at: University of Michigan Cardiovascular Center, 1500 E Medical Center Dr – SPC 5853, Ann Arbor, MI 48109-5853. Tel.: + 1 734 998 5909; fax: + 1 734 998 9939. E-mail address: [email protected] (G.D. Barnes).
by up to 60% [1]. However, the beneficial effects of warfarin must be balanced with the increased risk of major bleeding [2]. Guidelines recommend taking into account both thromboembolic and bleeding risk when considering stroke prevention therapy for AF [3]. Prediction of thromboembolic risk is predominately performed using the CHADS2 or CHA2DS2-VASc risk scoring systems [4,5]. Multiple scoring systems have been proposed to predict the risk of major bleeding in AF populations, including the HEMORR2HAGES, HAS-BLED and ATRIA scores [6–8]. Choosing the most effective scoring system has been the topic of debate in the literature [9–13]. Additionally, use of bleeding risk assessment tools, especially the HAS-BLED score, has been endorsed by several guidelines for management of AF [14–16]. With overlap of some of the risks factors for stoke or bleeding in the risk prediction rules, the CHADS2 scoring system has been associated with an increased risk of both stroke and bleeding in patients with AF [17–19]. In this study, we retrospectively compared the CHADS2 and
http://dx.doi.org/10.1016/j.thromres.2014.05.034 0049-3848/© 2014 Elsevier Ltd. All rights reserved.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
2
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
CHA2DS2-VASc stroke scores to the HEMORR2HAGES, HAS-BLED and ATRIA bleeding scoring systems for predicting major bleeding in a contemporary, “real-world” population of non-valvular AF patients treated with warfarin.
Methods MAQI2 Collaborative The Michigan Anticoagulation Quality Improvement Initiative (MAQI 2 ) is a Blue Cross Blue Shield of Michigan/Blue Care Network (BCBSM/BCN) sponsored continuous quality improvement consortium. Full details of the MAQI2 consortium have previously been published [20,21]. Briefly, the aim of the MAQI2 collaborative is to describe anticoagulation care across the state, identify best practices and procedures associated with best outcomes, and to partner in quality improvement projects. The collaborative was formed in 2008 with initial patient enrollment starting in October 2009. As a BCBSM/BCN-funded quality improvement collaborative, selected anticoagulation clinics (ACS) in the state of Michigan were invited to participate. Participating clinics are provided reimbursement by BCBSM/BCN to cover the costs of participation in MAQI2, including data abstraction and associated quality improvement efforts. All data abstractors undergo training and each center undergoes regular audits to ensure high quality data collection. Data collected is verified by random audits to ensure agreement with pre-defined data element definitions, including the primary outcomes for this study.
Patient Selection Between October 2009 and December 2012, 2600 new patients with non-valvular AF were enrolled in MAQI2. Patients were identified at the time of ACS enrollment in this inception cohort and clinical data was abstracted from the individual ACS database and hospital or group medical records.
Risk Stratification The CHADS2, CHA2DS2-VASc, HEMORR2HAGES, HAS-BLED and ATRIA scoring systems were retrospectively calculated and patients were grouped into low, intermediate and high risk as noted in Table 1 [4–8,10,19,22]. Presence of an aortic plaque was not available and not included in the calculation of vascular disease for the CHA2DS2-VASc score. Genetic factors associated with increased risk for bleeding were not available and not included in the calculation of the HEMORR2HAGES score. Labile INR was not included in the HAS-BLED score as this data was not available at the time of warfarin initiation. Time in the therapeutic range (TTR) was calculated using the Rosendaal method [23]. Study Endpoints The primary endpoint was occurrence of first major bleeding event. Major bleeding was defined according to the International Society of Thrombosis and Haemostasis consensus, which includes fatal bleeding, bleeding into a critical organ, overt bleeding requiring transfusion of 2 + units of red blood cells or an overt bleed causing a hemoglobin fall of 2+ g/dL [24]. Statistical Analysis Because of censored data, the association of patient demographics and comorbidities with the bleeding outcome was evaluated by using Cox regression analysis. We determined the cumulative incidence of first major bleeding in all patients. Patients were stratified according to the various risk scoring systems. Kaplan-Meier analysis was performed to determine the likelihood of a major bleeding event at 1 year, as well as to generate the individual bleeding risks at 1 year for the various bleeding risk scores. For the dichotomous categorical bleeding risk analysis, multiple analyses were performed. Based on the accepted low/moderate/high risk cut offs, we compared low/moderate vs high and low vs moderate/high. We also used the median bleeding risk score of all patients to create low and high risk bleeding groups.
Table 1 Bleeding Risk Scores. Risk Elements
CHADS2
CHA2DS2-VASc
HEMORR2HAGES
HAS-BLED
ATRIA
CHF HTN Age
1 point 1 point 1 point (≥75)
1 point 1 point (N75)
1 point 1 point (N65)
1 point 2 points (≥75)
DM Prior Stroke or TIA CAD, PAD or Aortic plaque Female Gender Chronic Liver Disease or Cirrhosis Chronic Renal Insufficiency Heavy Alcohol Use Malignancy Thrombocytopenia or Antiplatelet Use Prior Bleeding Event Anemia History of Falls Genetic Factors Prior Bleeding Event or Anemia TTR b 60% Use of ASA, clopidogrel, prasugrel, ticagrelor or NSAIDs Low Intermediate High
1 point 2 points 0-1 2 3+
1 point 1 point 2 points (≥75) 1 point (65-74) 1 point 2 points 1 point 1 point 0-1 2 3+
1 point 1 point 1 point 1 point 1 point 1 point 2 point 1 point 1 point 1 point 0-1 2-3 4+
1 point 1 point 1 point 1 point 1 point 1 point 1 point 0 1-2 3+
3 points 1 point 3 points 0-3 4 5-10
CHF – congestive heart failure, HTN – hypertension (N140/90 or use of antihypertensive medications), DM – diabetes mellitus, TIA – transient ischemic attack, CAD – coronary artery disease, PAD – peripheral artery disease, ASA – aspirin, TTR – time in therapeutic range, NSAIDs – non-steroidal anti-inflammatory drugs.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
3
Table 2 Demographic and Comorbidity Characteristics.
Patient-years (mean ± SD) Age (mean ± SD) Female BMI (mean ± SD) HTN DM CAD CHF Cancer Current Tobacco Use Renal Disease Liver Disease Stroke Bleeding Diathesis Age N 65 Concurrent drugs Drugs/EtOH HAS-BLED score (mean ± SD) ATRIA score (mean ± SD) HEMORR2HAGES score (mean ± SD) CHADS2 score (mean ± SD) CHA2DS2-VASc score (mean ± SD) TTR (mean ± SD)
Total N = 2600
Patients with major bleeding event N = 110
Patients without major bleeding event N = 2490
P-value
1.0 ± 0.8 70.1 ± 12.8 1085 (41.7) 30.2 ± 6.8 1937 (74.5) 655 (25.2) 847 (32.6) 629 (24.2) 412 (15.8) 154 (5.9) 310 (11.9) 17 (0.7) 300 (11.5) 815 (31.3) 1748 (67.2) 1466 (56.4) 126 (4.8) 2.6 ± 1.3 3.0 ± 2.4 2.6 ± 1.4 1.9 ± 1.2 3.4 ± 1.8 59.3 ± 22.7
1.0 ± 0.8 73.6 ± 10.9 51 (46.4) 29.8 ± 6.5 90 (81.8) 39 (35.5) 47 (42.7) 34 (30.9) 27 (24.5) 6 (5.5) 33 (30.0) 1 (0.9) 15 (13.6) 71 (54.5) 87 (79.1) 78 (70.9) 11 (10.0) 3.5 ± 1.2 4.7 ± 2.6 3.5 ± 1.5 2.2 ± 1.2 4.0 ± 1.7 54.6 ± 19.6
1.0 ± 0.8 70.0 ± 12.9 1034 (41.5) 30.3 ± 6.9 1847 (74.2) 616 (24.7) 800 (32.1) 595 (23.9) 385 (15.5) 148 (5.9) 281 (11.3) 16 (0.6) 285 (11.4) 744 (29.9) 1661 (66.7) 1388 (55.7) 115 (4.6) 2.5 ± 1.2 3.0 ± 2.4 2.6 ± 1.4 1.9 ± 1.2 3.4 ± 1.8 59.5 ± 22.8
0.78 0.085 0.65 0.51 0.27 0.018 0.022 0.09 0.025 0.92 b0.001 0.65 0.75 b0.001 0.0012 0.0012 0.016 b0.001 b0.001 b0.001 0.020 0.0026 b0.0001
BMI – Body-mass index, TTR – Time in therapeutic range, HTN – hypertension, DM – Diabetes mellitus, CAD – coronary artery disease, CHF – congestive heart failure.
Time-dependent receiver operating characteristic (ROC) curves were generated and the area under the ROC curve (C statistic) was used to compare the predictability of both non-categorical (using the individual scores) and category-based (dichotomous) risk schemes. Net reclassification improvement (NRI) analysis on censored data was performed for two-category based risk groups [25]. A 2-sided P b 0.05 was considered statistically significant for all analyses. All analyses were performed with statistical software SAS and R.
The formal bleeding risk scores (HAS-BLED, ATRIA and HEMORR2HAGES) had a modest predictive value for first major bleeding events (C statistic 0.66-0.69), performing better than CHADS2 and CHA2DS2VASc scores (Table 4). The CHADS2 score had a C statistic of 0.53 (95% CI 0.47-0.60) and the CHA2DS2-VASc score had a C statistic of 0.56 (95% CI 0.49-0.62). The HAS-BLED score had a C statistic that was similar to the ATRIA and HEMORR2HAGES scores, 0.018 (95% CI -0.038–0.074) and 0.032 (95% CI -0.021–0.067) in difference, respectively (Table 4).
Results
Clinical Risk Group
110 individual patients experienced major bleeding events during 2581.6 patient-years of follow-up (4.5%/year). Mean follow up was 1.0 ± 0.8 years for patients with and without bleeding events. Patients with bleeding events had higher CHADS 2, CHA2DS2-VASc, HEMORR2 HAGES, HAS-BLED and ATRIA scores (Table 2). Further demographic and comorbidity characteristics are summarized in Table 2. Patients with bleeding events had mean TTR 54.6 ± 19.6 while patients without bleeding events had a mean TTR 59.5 ± 22.8 (p b 0.0001). Median scores for all patients are listed in Table 3. Event rates for major bleeding are shown in Table 3. The predictive abilities of the five risk scores (non-categorical) are presented in Fig. 1.
NRI analysis was performed using dichotomous risk groups to compare the ability to better classify patients with bleeding events into the higher risk group and patients without bleeding events into the lower risk group. The NRI was highest when risk groups were defined based on the median scores (Appendix A). NRI demonstrated improvement of the HAS-BLED, ATRIA and HEMORR2HAGES scores over the CHADS2 score and a 36-54% improvement over the CHA2DS2VASc score (Table 4). The HAS-BLED score also demonstrated a 26% and 31% improvement over the ATRIA and HEMORR2HAGES scores, respectively (Table 4). NRI did not demonstrate superiority of the HASBLED score compared to the other bleeding risk scores if the risk groups
Table 3 Major bleeding event rates. Bleeding Risk
HAS-BLED
ATRIA
HEMORR2HAGES
CHADS2
Median Score
3
3
2
2
3
Low Moderate High Low High C statistic at 12 Months (non-categorical)
0 1.3 (0.03–7.3) 1-2 2.0 (1.3–3.1) 3+ 6.6 (5.3–8.0) 0-3 2.7 (2.0–3.5) 4+ 10.4 (8.0–13.4) 0.69 (0.63–0.75)
0-3 2.3 (1.7–3.2) 4 7.4 (4.6–11.3) 5+ 9.1 (6.8–11.8) 0-3 2.3 (1.7-3.2) 4+ 8.5 (6.7–10.7) 0.67 (0.61-0.74)
0-1 1.7 (0.7–3.3) 2 3.6 (2.7-4.8) 3+ 8.5 (6.4–11.0) 0-2 2.1 (1.4-3.0) 3+ 6.8 (5.4–8.3) 0.66 (0.61-0.74)
0-1 3.3 (2.2-4.7) 2 4.8 (3.5-6.4) 3+ 5.6 (4.0-7.6) 0-2 4.1 (3.2-5.1) 3+ 5.6 (4.0-7.6) 0.53 (0.47-0.60)
0-1 3.4 (1.6-6.4) 2 3.3 (1.8-5.7) 3+ 4.9 (3.9-6.0) 0-3 3.3 (2.3-4.4) 3+ 5.6 (4.4-7.0) 0.56 (0.49 – 0.62)
CHA2DS2-VASc
Number of major bleeding events per 100-patient-years (95% CI).
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
4
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
Fig. 1. Receiver Operating Characteristic (ROC) Curves and C Statistics for the Risk Scores (non-categorical) at 12 Months.
were defined by actual scores and when compared as low/moderate vs. high or low vs. moderate/high (Appendix A).
Discussion In this study of a “real-world” inception cohort of non-valvular AF patients treated with warfarin for stroke prevention, our data show that the three bleeding risk scores performed better than the CHADS2 [4] and CHA2DS2-VASc [5] scores for the prediction of a first major bleeding event. By NRI analysis, the HAS-BLED [7] score was superior to the other two established bleeding risk scores (ATRIA [8] and HEMORR2HAGES [6] for the prediction of a first major bleeding event. Consistent with prior reports, all three of the formal bleeding risk models had only moderate predictive ability [9,12,13,22,26,27]. Despite the knowledge that the bleeding risk increased with the CHADS2 score [17–19], we believe that this is the first report to perform a complete comparison of the CHADS2 and CHA2DS2-VASc scores to the three established bleeding risk models (HAS-BLED, ATRIA and HEMORR2HAGES) in a diverse, multi-centered dataset. While patients experiencing major bleeding events had higher CHADS2 and CHA2DS2VASc scores, these stroke risks scores are not as predictive as bleeding risk scores, consistent with other recent reports [26,27]. Therefore, a formalized bleeding risk score should be used when clinicians estimate the bleeding risk for patients with non-valvular AF when considering anticoagulation treatment. Clinically, this highlights the importance of using a stroke risk score (e.g. CHADS 2 or CHA 2 DS 2 -VASc) for predicting thromboembolism risk while using a bleeding risk score (e.g. HAS-BLED) to predict bleeding risk separately.
Although easy to use clinically, all available stroke and bleeding risk models have been limited by their ability to predict clinically relevant events. Prior analyses of the CHADS2 and CHA2DS2-VASc scores have demonstrated C statistics between 0.51-0.82 and 0.53-0.888, respectively, for the prediction of stroke in non-valvular AF [4,5,26,28–30]. This represents modest to good predictive ability. However, the bleeding risk models have continuously demonstrated only modest predictive abilities with C statistics between 0.57-0.795, 0.61-0.69 and 0.58-0.771 for the HAS-BLED, ATRIA and HEMORR2HAGES scores, respectively [6–8,13,26,27,31,32]. Our findings of C statistics of 0.66-0.69 for the three established bleeding risk models are in line with those previously published validation studies. Despite the increasing risk of bleeding with increasing CHADS2 score, a large Danish population registry analysis previously suggested a net clinical benefit for all patients receiving warfarin therapy, irrespective of their CHADS2 and HAS-BLED score profiles [33]. Additionally, there have been numerous studies highlighting the underuse of warfarin in eligible AF patient and a perceived risk of bleeding is regularly given as a common reason not to anticoagulate patients [34–36]. Although the bleeding risk is increased in patients at higher CHADS 2 and CHA2 DS2 -VASc scores in this study, this should not be a hindrance to their receiving the benefits of stroke prophylaxis. Therefore, it seems reasonable to focus on the underuse of warfarin in all patients with AF who are likely to derive benefit from stroke risk reduction instead of on the perceived bleeding risk, as predicted by provider intuition or bleeding risk scores. This study has a number of strengths. This is an inception cohort of “real world” patients from a diverse United States population, including both academic and private-practice centers in urban and suburban locations. Warfarin control in this study, as measured by the TTR, is on par with those of other real-world inception cohorts [37] but likely differs from those that preselected patients with stable INRs for a minimum of 6 months [12,27] or were analyzed from randomized trial populations [26]. We also abstracted and audited patient chart data, instead of relying on ICD9 billing data, to verify comorbidity elements and outcomes. This is a contemporary data set, with all patients initiating in warfarin since 2009. Still, there are a few limitations. First, since data are collected only at participating centers, bleeding events that occurred at non-participating centers may not have been captured by the ACSs participating in MAQI2. However, since all patients are contacted either face-to-face or over the telephone by nurses and/or pharmacists at the ACS, it is reasonable to assume that clinical events occurring at other hospital centers would be mentioned in the patient’s ACS chart and therefore captured in this dataset. We did not include cytochrome 2C9 or vitamin K oxide reductase genentics (HEMORR2HAGES) or labile INR measurement (HAS-BLED) scores, which may have effective their predictive accuracy. Also, since this is an observational inception cohort, our follow up time is variable based on when the patients initiated warfarin therapy. To correctly calculate the NRI in this setting, we generated
Table 4 Comparison of Risk Scores for First Major Bleeding Event at 12 Months.
HAS-BLED vs. ATRIA HAS-BLED vs. HEMORR2HAGES ATRIA vs. HEMORR2HAGES HAS-BLED vs. CHADS2 ATRIA vs. CHADS2 HEMORR2HAGES vs. CHADS2 HAS-BLED vs. CHA2DS2-VASc ATRIA vs. CHA2DS2-VASc HEMORR2HAGES vs. CHA2DS2-VASc CHADS2 vs CHA2DS2-VASc
C Statistic difference
95% CI
p-value
NRI
IQR
p-value
0.018 0.032 0.014 0.16 0.14 0.12 0.13 0.11 0.10 -0.024
-0.038–0.074 -0.0021–0.067 -0.034–0.062 0.096–0.22 0.071–0.21 0.065–0.18 0.073–0.19 0.047–0.18 0.039–0.16 -0.057–0.0084
0.53 0.065 0.56 b0.001 b0.001 0.03 b0.001 b0.001 0.031 0.15
0.26 0.31 0.34 0.58 0.59 0.54 0.36 0.40 0.54 -0.071
-0.40-0.42 -0.12-0.52 -0.06-0.50 0.018-0.79 0.32-0.80 0.26-0.74 0.18-0.71 0.22-0.80 0.014-0.73 -0.41-0.14
0.006 0.001 0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 0.25
C statistic difference calculated using ROC curve analysis for non-categorical scores. Net Reclassification Improvement (NRI) calculated using dichotomous risk group analysis based on median scores. IQR – Interquartile Range.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
one-year bleeding rates using a Kaplan-Meier estimator [38]. Lastly, our rate of major bleeding is slightly higher than those of other recently published studies (4.5%/year vs 1.0-3.2%/year) [12,26,27]. This likely corresponds to the real-world, American population with a median TTR that is slightly lower than commonly seen in European systems or randomized trial settings of patients previously treated with oral anticoagulants instead of an inception cohort. Further refinement of bleeding risk prediction would be helpful, but is likely challenging given the transient nature of many possible bleeding triggers, e.g. drug-drug interactions, non-compliance, trauma. While European and Canadian guidelines have favored the use of the HASBLED score for estimation of bleeding risk, American-based groups have not yet endorsed the use of any bleeding risk tool, citing their modest predictive capacity [3,14–16,39–40]. Additionally, further consideration about the integration of both bleeding and stroke risk prediction tools into a single tool for patient education and discussion of risks and benefits warrants further development and clinical testing. In summary, the three formal bleeding risk scores performed better than the CHADS2 and CHA2DS2-VASc scores for predicting major bleeding in non-valvular AF patients treated with warfarin. We recommend using a formal bleeding risk score instead of a stroke risk score when counseling patients about their risk of major bleeding when considering warfarin therapy for stroke prevention in AF, particularly in the highest risk patients. Authorship Details G Barnes conceived the study idea, designed the study and drafted the initial manuscript. X Gu performed all statistical analysis. B Haymart, E Kline-Rogers, S Almany, J Kozlowski, D Besley, G Grol and J Froehlich provided critical analysis and substantial edits to the study design and manuscript revisions. S Kaatz provided direct oversight of the study design, statistical analysis and manuscript editing with critical and substantial revisions. Conflict of Interest Statements Geoff Barnes – No conflicts to report Xiaokui Gu – No conflicts to report Brian Haymart – No conflicts to report Eva Kline-Rogers – 45% of salary is funded by BCBSM Steve Almany – No conflicts to report Jay Kozlowski – No conflicts to report Dennis Besley – No conflicts to report Gregory Krol – No conflicts to report James Froehlich – Consultant for: Sanofi-Aventis, Ortho-McNeil, Merck. Research grants from Sanofi-Aventis, Blue Cross/Blue Shield of Michigan, Mardigian Foundation, Fibromuscular Disease Society of America. • Scott Kaatz – Grant support • Boehringer-Ingelheim • Bristol Myer Squibb • Bayer/Jansen/Johnson and Johnson • Eisai • Iverson Genetics Diagnostics/Medicare • National Institute of Health • Canadian Institute of Health Research • Blue Cross/Blue Shield of Michigan • Speaker honorarium • Jansen/Johnson and Johnson • Boehringer-Ingelheim • Bristol Myer Squibb/Pfizer • Consultant • Boehringer Ingelheim • Bristol Myer Squibb/Pfizer
5
• Jansen/Johnson and Johnson • Daiichi Sankyo • Board membership (non-profit) • AC Forum • National Certification Board of Anticoagulation Providers • National Blood Clot Alliance Medical and Scientific Advisory Board Acknowledgement Blue Cross-Blue Shield of Michigan provided funding for data collection and analysis, but was not involved in the interpretation of the data, the decision to publish or any revisions of the manuscript.
Appendix A Category-based NRI at 12 Months
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
HAS-BLED HAS-BLED HAS-BLED HAS-BLED ATRIA ATRIA ATRIA HEMORR2HAGES HEMORR2HAGES CHADS2
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Low vs. Mod/High NRI (IQR) -0.55 (-0.76 – -0.33) -0.11 (-0.17 – -0.072) -0.042 (-0.31 – 0.051) -0.018 (-0.094 – 0.051) 0.46 (0.15 – 0.65) 0.46 (0.28 – 0.77) 0.53 (0.29 – 0.76) 0.15 (-0.057 – 0.28) 0.14 (-0.0025 – 0.22) 0.035 (-0.14 – 0.19)
p-value b0.001 0.001 0.51 0.61 b0.001 b0.001 b0.001 0.018 0.002 0.50
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Low/Mod vs. High NRI (IQR) 0.097 (-0.36 – 0.31) 0.13 (-0.083 – 0.48) 0.55 (0.28 – 0.72) 0.55 (0.79 – 0.20) 0.20 (-0.19 – 0.65) 0.47 (0.18 – 0.69) 0.38 (0.12 – 0.62) 0.39 (0.084 – 0.61) 0.28 (-0.011 – 0.56) -0.16 (-0.26 – 0.22)
p-value 0.39 0.23 b0.001 b0.001 0.017 b0.001 b0.001 0.001 0.006 0.89
ATRIA HEMORR2HAGES CHADS2 CHADS2-VASc HEMORR2HAGES CHADS2 CHADS2-VASc CHADS2 CHADS2-VASc CHADS2-VASc
Median as cut-off NRI (IQR) 0.26 (-0.40 – 0.42) 0.31 (-0.12 – 0.52) 0.58 (0.18 – 0.79) 0.36 (0.18 – 0.71) 0.34 (-0.061 – 0.50) 0.59 (0.32 – 0.80) 0.40 (0.22 – 0.80) 0.54 (0.26 – 0.74) 0.54 (0.014 – 0.73) -0.071 (-0.41 – 0.14)
p-value 0.006 0.001 b0.001 b0.001 0.001 b0.001 b0.001 b0.001 b0.001 0.25
References [1] Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med 2007;146:857–67. [2] Palareti G, Cosmi B. Bleeding with anticoagulation therapy - who is at risk, and how best to identify such patients. Thromb Haemost 2009;102:268–78. [3] You JJ, Singer DE, Howard PA, Lane DA, Eckman MH, Fang MC, et al. Antithrombotic therapy for atrial fibrillation: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012;141:e531S–75S. [4] Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA 2001;285:2864–70.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
6
G.D. Barnes et al. / Thrombosis Research xxx (2014) xxx–xxx
[5] Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factorbased approach: the euro heart survey on atrial fibrillation. Chest 2010;137:263–72. [6] Gage BF, Yan Y, Milligan PE, Waterman AD, Culverhouse R, Rich MW, et al. Clinical classification schemes for predicting hemorrhage: results from the National Registry of Atrial Fibrillation (NRAF). Am Heart J 2006;151:713–9. [7] Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, 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;138:1093–100. [8] Fang MC, Go AS, Chang Y, Borowsky LH, Pomernacki NK, Udaltsova N, et al. A new risk scheme to predict warfarin-associated hemorrhage: The ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) Study. J Am Coll Cardiol 2011;58:395–401. [9] Apostolakis S, Lane DA, Guo Y, Buller H, Lip GY. Performance of the HEMORR(2) HAGES, ATRIA, and HAS-BLED bleeding risk-prediction scores in patients with atrial fibrillation undergoing anticoagulation: the AMADEUS (evaluating the use of SR34006 compared to warfarin or acenocoumarol in patients with atrial fibrillation) study. J Am Coll Cardiol 2012;60:861–7. [10] Apostolakis S, Lane DA, Guo Y, Buller H, Lip GY. Performance of the HEMORR(2) HAGES, ATRIA, and HAS-BLED Bleeding Risk-Prediction Scores in Nonwarfarin Anticoagulated Atrial Fibrillation Patients. J Am Coll Cardiol 2013;61:386–7. [11] Lip GY, Lin HJ, Hsu HC, Su TC, Chen MF, Lee YT, et al. Comparative assessment of the HAS-BLED score with other published bleeding risk scoring schemes, for intracranial haemorrhage risk in a non-atrial fibrillation population: The Chin-Shan Community Cohort Study. Int J Cardiol 2013;57:173–80. [12] Roldán V, Marín F, Fernández H, Manzano-Fernandez S, Gallego P, Valdés M, et al. Predictive value of the HAS-BLED and ATRIA bleeding scores for the risk of serious bleeding in a "real-world" population with atrial fibrillation receiving anticoagulant therapy. Chest 2013;143:179–84. [13] Donzé J, Rodondi N, Waeber G, Monney P, Cornuz J, Aujesky D. Scores to predict major bleeding risk during oral anticoagulation therapy: a prospective validation study. Am J Med 2012;125:1095–102. [14] Cairns JA, Connolly S, McMurtry S, Stephenson M, Talajic M, Committee CAFG. Canadian Cardiovascular Society atrial fibrillation guidelines 2010: prevention of stroke and systemic thromboembolism in atrial fibrillation and flutter. Can J Cardiol 2011;27:74–90. [15] Lip GY, Andreotti F, Fauchier L, Huber K, Hylek E, Knight E, et al. Bleeding risk assessment and management in atrial fibrillation patients: a position document from the European Heart Rhythm Association, endorsed by the European Society of Cardiology Working Group on Thrombosis. Europace 2011;13:723–46. [16] Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Europace 2010;12:1360–420. [17] Suh DC, Choi JC, Schein J, Kim S, Nelson WW. Factors associated with warfarin discontinuation, including bleeding patterns, in atrial fibrillation patients. Curr Med Res Opin 2013;29:761–71. [18] Lopes RD, Al-Khatib SM, Wallentin L, Yang H, Ansell J, Bahit MC, et al. Efficacy and safety of apixaban compared with warfarin according to patient risk of stroke and of bleeding in atrial fibrillation: a secondary analysis of a randomised controlled trial. Lancet 2012;380:1749–58. [19] Oldgren J, Alings M, Darius H, Diener HC, Eikelboom J, Ezekowitz MD, et al. Risks for stroke, bleeding, and death in patients with atrial fibrillation receiving dabigatran or warfarin in relation to the CHADS2 score: a subgroup analysis of the RE-LY trial. Ann Intern Med 2011;155:660–7 [W204]. [20] Barnes GD, Kaatz S, Golgotiu V, Gu X, Leidal A, Kobeissy A, et al. Use of warfarin for venous thromboembolism prophylaxis following knee and hip arthroplasty: results of the Michigan Anticoagulation Quality Improvement Initiative (MAQI(2)). J Thromb Thrombolysis 2013;35:10–4. [21] Barnes GD, Kaatz S, Winfield J, Gu X, Haymart B, Kline-Rogers E, et al. Warfarin use in atrial fibrillation patients at low risk for stroke: analysis of the Michigan Anticoagulation Quality Improvement Initiative (MAQI(2)). J Thromb Thrombolysis 2014;37:171–6. [22] Lip GY, Banerjee A, Lagrenade I, Lane DA, Taillandier S, Fauchier L. Assessing the risk of bleeding in patients with atrial fibrillation: the Loire Valley Atrial Fibrillation project. Circ Arrhythm Electrophysiol 2012;5:941–8.
[23] Rosendaal FR, Cannegieter SC, van der Meer FJ, Briet E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69:236–9. [24] Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005;3:692–4. [25] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72 [discussion 207–12]. [26] Apostolakis S, Lane DA, Buller H, Lip GY. Comparison of the CHADS2, CHA2DS2 -VASc and HAS-BLED scores for the prediction of clinically relevant bleeding in anticoagulated patients with atrial fibrillation: The AMADEUS trial. Thromb Haemost 2013;110. [27] Roldan V, Marin F, Manzano-Fernandez S, Gallego P, Vilchez JA, Valdes M, et al. The HAS-BLED score has better prediction accuracy for major bleeding than the CHADS or CHADS-VASc scores In anticoagulated patients with atrial fibrillation. J Am Coll Cardiol 2013;62:2199–204. [28] Rietbrock S, Heeley E, Plumb J, van Staa T. Chronic atrial fibrillation: Incidence, prevalence, and prediction of stroke using the Congestive heart failure, Hypertension, Age N75, Diabetes mellitus, and prior Stroke or transient ischemic attack (CHADS2) risk stratification scheme. Am Heart J 2008;156:57–64. [29] Piccini JP, Stevens SR, Chang Y, Singer DE, Lokhnygina Y, Go AS, et al. Renal Dysfunction as a Predictor of Stroke and Systemic Embolism in Patients With Nonvalvular Atrial Fibrillation: Validation of the R2CHADS2 Index in the ROCKET AF (Rivaroxaban Oncedaily, oral, direct factor Xa inhibition Compared with vitamin K antagonism for prevention of stroke and Embolism Trial in Atrial Fibrillation) and ATRIA (AnTicoagulation and Risk factors In Atrial fibrillation) Study Cohorts. Circulation 2013;127:224–32. [30] Olesen JB, Lip GY, Hansen ML, Hansen PR, Tolstrup JS, Lindhardsen J, et al. Validation of risk stratification schemes for predicting stroke and thromboembolism in patients with atrial fibrillation: nationwide cohort study. BMJ 2011;342:d124. [31] Olesen JB, Lip GY, Hansen PR, Lindhardsen J, Ahlehoff O, Andersson C, et al. Bleeding risk in 'real world' patients with atrial fibrillation: comparison of two established bleeding prediction schemes in a nationwide cohort. J Thromb Haemost 2011;9:1460–7. [32] Lip GY, Frison L, Halperin JL, Lane DA. Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: the HASBLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol 2011;57:173–80. [33] Olesen JB, Lip GY, Lindhardsen J, Lane DA, Ahlehoff O, Hansen ML, et al. Risks of thromboembolism and bleeding with thromboprophylaxis in patients with atrial fibrillation: A net clinical benefit analysis using a 'real world' nationwide cohort study. Thromb Haemost 2011;106:739–49. [34] Wilke T, Groth A, Mueller S, Pfannkuche M, Verheyen F, Linder R, et al. Oral anticoagulation use by patients with atrial fibrillation in Germany. Adherence to guidelines, causes of anticoagulation under-use and its clinical outcomes, based on claims-data of 183,448 patients. Thromb Haemost 2012;107:1053–65. [35] Gussoni G, Di Pasquale G, Vescovo G, Gulizia M, Mathieu G, Scherillo M, et al. Decision making for oral anticoagulants in atrial fibrillation: The ATA-AF study. Eur J Intern Med 2013;24:324–32. [36] Rosenman MB, Baker L, Jing Y, Makenbaeva D, Meissner B, Simon TA, et al. Why is warfarin underused for stroke prevention in atrial fibrillation? A detailed review of electronic medical records. Curr Med Res Opin 2012;28:1407–14. [37] Gallego P, Roldan V, Marin F, Romera M, Valdes M, Vicente V, et al. Cessation of oral anticoagulation in relation to mortality and the risk of thrombotic events in patients with atrial fibrillation. Thromb Haemost 2013;110:1189–98. [38] Steyerberg EW, Pencina MJ. Reclassification calculations for persons with incomplete follow-up. Ann Intern Med 2010;152:195–6 [author reply 6-7]. [39] Fuster V, Rydén LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, et al. 2011 ACCF/ AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines. Circulation 2011;123:e269–367. [40] Furie KL, Goldstein LB, Albers GW, Khatri P, Neyens R, Turakhia MP, et al. Oral antithrombotic agents for the prevention of stroke in nonvalvular atrial fibrillation: a science advisory for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke 2012;43:3442–53.
Please cite this article as: Barnes GD, et al, The Predictive Ability of the CHADS2 and CHA2DS2-VASc Scores for Bleeding Risk in Atrial Fibrillation: The MAQI2 Experience..., Thromb Res (2014), http://dx.doi.org/10.1016/j.thromres.2014.05.034
