Sage Journals: Discover world-class research

Abstract

Non-valvular atrial fibrillation and venous thromboembolism anticoagulation risk assessment tools have been increasingly utilized to guide implementation and duration of anticoagulant therapy. Anticoagulation significantly reduces stroke and recurrent venous thromboembolism risk, but comes at the cost of increased risk of major and clinically relevant non-major bleeding. The decision for anticoagulation in high-risk patients is complicated by the fact that many risk factors associated with increased thromboembolic risk are simultaneously associated with increased bleeding risk. Traditional risk assessment tools rely heavily on age, sex, and presence of cardiovascular comorbidities, with newer tools additionally taking into account changes in risk factors over time and novel biomarkers to facilitate more personalized risk assessment. These tools may help counsel and inform patients about the risks and benefits of starting or continuing anticoagulant therapy and can identify patients who may benefit from more careful management. Although the ability to predict anticoagulant-associated hemorrhagic risk is modest, ischemic and bleeding risk scores have been shown to add significant value to therapeutic management decisions. Ultimately, further work is needed to optimally implement accurate and actionable risk stratification into clinical practice.

Keywords

anticoagulation bleeding risk stratification stroke venous thromboembolism (VTE)atrial fibrillation

Introduction

Medical decision-making around anticoagulation (AC) has become increasingly complex, as new therapeutic agents have been introduced to populations with varying risk profiles and comorbidities. Risk assessment is an important first step in maximizing the beneficial effects of AC while mitigating bleeding events. This process requires knowledge of the indication for AC, the patient’s risk for major and clinically relevant non-major bleeding, anticipated duration of therapy, its expected adherence, and the intrinsic properties of the anticoagulant itself. This review specifically focuses on risk assessment models that are likely to be helpful in evaluating patients with non-valvular atrial fibrillation (NVAF) and/or venous thromboembolism (VTE).

The most commonly used oral anticoagulants include vitamin K antagonists (VKAs) and the direct oral anticoagulants (DOACs). Less often, patients may be treated with parenteral anticoagulation, including unfractionated heparin (UFH), low molecular weight heparin (LMWH), or the pentasaccharide, fondaparinux. Substantial data support long-term AC in patients with NVAF and/or VTE, particularly among those at higher thromboembolic risk.

Antithrombotic and anticoagulant therapies used for non-valvular atrial fibrillation (NVAF) and venous thromboembolism (VTE)

Antiplatelet therapy is less commonly used to reduce thrombotic risk in patients with NVAF or VTE. While aspirin is the most commonly used antithrombotic therapy worldwide and has been shown to reduce the risk of cardioembolic stroke in NVAF by roughly 20%,¹ the 2014 American Heart Association (AHA)/American College of Cardiology (ACC)/Heart Rhythm Society (HRS) atrial fibrillation guidelines limit its use to NVAF patients at lower thromboembolic risk.² The value of dual antiplatelet therapy in NVAF was evaluated in the ACTIVE A trial,³ where the addition of clopidogrel to aspirin reduced the risk of stroke by 27%. This benefit, however, was offset by an almost twofold increased risk of hemorrhage. As such, it is not a recommended therapy for NVAF.² Finally, while aspirin is not considered standard therapy for VTE, the American College of Chest Physicians (ACCP) endorses a grade 2B recommendation for its use in patients without contraindication to aspirin who are stopping AC.⁴

Substantial evidence exists to support the use of a VKA (e.g. warfarin) in NVAF, with an estimated stroke risk reduction of 68% compared to placebo.¹ Its use, however, is limited by a narrow therapeutic window, delayed pharmacodynamic response, numerous drug–food and drug–drug interactions, and the need for close monitoring. Adverse events with VKAs are also closely related to the quality of AC, with an inverse association between time in the therapeutic range (TTR) and risk of both major bleeding and mortality.^5,6

While LMWH represents an alternative to VKA, it is limited by dependence on renal excretion and subcutaneous administration. The DOACs, in contrast, are oral agents which act via direct inhibition of factors II or Xa and have proven efficacy in NVAF, VTE, and stable atherosclerotic cardiovascular disease (coronary artery disease and peripheral artery disease).^7–12 These agents are limited in part by cost and can be difficult to implement if monitoring of AC is required.

Given the diversity of agents to choose from, the decision for AC therapy has become increasingly complex. While a nuanced discussion of the risk–benefit differences among VKAs, DOACs, and parenteral anticoagulants is outside the scope of this review, it is important to consider the specific agents for which risk calculators were validated when making decisions about AC.

Variables implicated in prediction of thromboembolism and bleeding in NVAF and VTE

Assessment of thromboembolic risk for NVAF has historically been dependent on the presence of traditional pro-thrombotic risk factors, such as increased age (⩾ 75 years), female sex, and a variety of cardiovascular comorbidities. While risk prediction tools have been heavily weighted by these risk factors, novel markers have been increasingly incorporated to refine predictive accuracy.

Several biomarkers have been associated with NVAF thromboembolic risk, such as N-terminal pro-brain natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponin (hs-cTnT) that reflect high atrial stretch, and growth differentiating factor-15 (GDF-15) that reflects oxidative/myocardial stress.^13–15 Hs-cTnT and GDF-15 have additionally been associated with bleeding risk in NVAF patients, along with markers of anemia such as hemoglobin and markers of renal dysfunction such as cystatin C and estimated glomerular filtration rate (eGFR).^15–17 Several non-specific markers of inflammation, including D-dimer, C-reactive protein (CRP), and interleukin (IL)-6 have additionally been associated with higher bleeding risk,^18,19 but have been difficult to integrate into stratification algorithms. Echocardiographic findings such as increased left atrial volume index, abnormal left atrial longitudinal strain, and left atrial appendage pulse-wave velocity have been associated with increased thrombotic risk.²⁰ Finally, evidence of left atrial fibrosis by cardiac magnetic resonance imaging has also been found to be predictive.²¹

Similar to NVAF, there exists substantial interest in the identification of novel biomarkers to increase accuracy in the prediction of VTE and guide appropriateness for extended duration anticoagulant therapy. In addition to D-dimer, biomarkers including soluble P-selectin, coagulation factor VIII, thrombin, CRP, tumor necrosis factor, and interleukins have been evaluated.²² Some studies have shown niche value, such as the use of baseline P-selectin to evaluate risk of VTE recurrence in patients with cancer-associated VTE.²³ Moving forward, incorporation of biomarkers into VTE prediction models may allow for a more personalized approach to AC management.

It is also very important to recognize that many traditional risk factors and biomarkers that increase thromboembolic risk also simultaneously increase bleeding risk. In addition, risk factors may contribute differentially to risk. For example, the presence of vascular disease in NVAF does not contribute the same attendant risk as other traditional risk factors.²⁴ Collectively, these factors should be taken into consideration when evaluating risk prediction tools.

Prediction of stroke or systemic embolism in patients with NVAF

Atrial fibrillation is the most common sustained arrhythmia worldwide,²⁵ and is associated with an increased risk of thromboembolism. The treatment approach (rhythm- vs rate-control) has not been shown to meaningfully impact thromboembolic risk, as demonstrated in the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) and Rate Control Efficacy in Permanent Atrial Fibrillation (RACE) trials.^26,27 Less is known, however, about the relationship between the burden of NVAF and thromboembolic risk. In one study, a burden > 5.5 hours was associated with increased risk of thromboembolism.²⁸ Current major guidelines do not, however, specify how atrial fibrillation burden should impact decision making for AC.^2,29–31

Over time, a number of risk prediction tools have been developed to help identify patients at heightened risk for stroke or systemic embolism for which anticoagulant therapy would be beneficial. Consistent among these tools has been an attempt to divide patients into low-, moderate-, and high-risk strata based on a number of key traditional thrombotic risk factors. In addition, newer calculators have incorporated differential weighting of variables and biomarkers, as well as change in risk over time (Table 1). There has also been a shift to identify lower risk patients who can be successfully treated either with aspirin or no antithrombotic therapy at all.

Table 1.

NVAF stroke risk calculators.

Stroke risk calculator	Primary study	Study n	Validated disease state	Variables	Range of prediction	Validation studies
CHADS₂ score	Gage et al. (2001)³²	1733	NVAF, off AC	Heart failure, hypertension, age, diabetes mellitus, prior stroke/thromboembolism	Annual stroke risk: 0, low (1.9%); 1, intermediate (2.8%); ⩾ 2, high (4.0–18.2%)	Yes⁸⁴
CHA₂DS₂-VASc score	Lip et al. (2010)²⁴	1084	NVAF, off AC	Heart failure, hypertension, age, sex, diabetes mellitus, prior stroke/TIA/thromboembolism, vascular disease	Annual stroke risk: 0, low (0%); 1, low–moderate (1.3%); ⩾ 2, high (2.2–15.2%)	Yes^85–87
ATRIA stroke score	Singer et al. (2013)³⁵	10,927	NVAF, off AC	Stroke history, age, sex, diabetes mellitus, heart failure, hypertension, proteinuria, eGFR < 45 or ESRD	Annual stroke risk: 0–5, low (< 1%); 6, intermediate (1–1.9%); or > 6, high (⩾ 2%)	Yes^36,88,89
ABC stroke score	Hijazi et al. (2016)⁴⁰	14,701	NVAF, off or on AC	Prior stroke/TIA, age, troponin I, NT-proBNP	Annual stroke risk: low (< 1%); medium (1–2%); high (⩾ 2%)	Yes⁹⁰
Delta CHA₂DS₂-VASc score	Chao et al. (2018)³⁹	31,309	NVAF off antiplatelet/AC	Same as CHA₂DS₂-VASc	Annual stroke risk: 0.92–4.99%	No

AC, anticoagulation; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease; NT-proBNP, N-terminal pro-brain natriuretic peptide; NVAF, non-valvular atrial fibrillation; TIA, transient ischemic attack.

While the criteria used to define which patients with NVAF warrant AC has varied among international guidelines, there has been a tendency to rely upon either the CHADS₂ or CHA₂DS₂-VASc risk calculators.^2,29–31 Guide-lines such as the 2016 European Society of Cardiology/European Atherosclerosis Society (ESC/EAS) suggest that biomarker-based risk scores may provide additional prognostic information, but do not offer any specific recommendations for their incorporation into the treatment algorithm.³⁰

The first NVAF assessment tool to achieve widespread use was the CHADS₂ score, which was developed from a cohort of approximately 1,700 Medicare beneficiaries with non-rheumatic atrial fibrillation untreated with AC (warfarin) at hospital discharge.³² The score ranges from 0 to 6, with scores of zero, three, and six correlating with annual stroke risks of 1.9%, 5.9%, and 18.2%, respectively. A subsequent analysis by Fang et al. found the predicted annual stroke rate to be substantially lower: 0.5% for a CHADS₂ score of zero and > 3% for a CHADS₂ score ⩾ 3.³³ Although the score is simple to calculate and implement into daily practice, it is limited by inadequate refinement of risk at the low end of the stroke/systemic embolism spectrum, placing up to 50% of patients with a score of one in the intermediate-risk category.³⁴ This risk stratification tool is currently supported only by a limited number of guidelines.²⁹

The CHA₂DS₂-VASc score was created to refine risk stratification among patients with low to intermediate CHADS₂ scores by incorporating additional variables, including age of 65–74 years, vascular disease, and female sex.²⁴ It is superior in discriminating between patients at very low risk, and reclassifies many patients with a CHADS₂ score of zero into higher-risk categories. In the original study, patients characterized at lowest risk (CHADS₂ score of zero) had a stroke rate of 1.4% during the follow-up period, whereas no strokes were detected in those with a CHA₂DS₂-VASc score of zero.²⁴ Critics of both CHADS₂ and CHA₂DS₂-VASc scores note that they both have a less than optimal c-statistic (0.6–0.7).^35,36

The CHA₂DS₂-VASc score is the most commonly utilized score at present, with endorsement in several modern atrial fibrillation guidelines. According to the 2014 AHA/ACC/HRS atrial fibrillation guidelines, aspirin or no antithrombotic therapy is recommended for those with a CHA₂DS₂-VASc score of zero; no antithrombotic therapy, aspirin, or AC are all reasonable options for a score of one; and AC is recommended for those with a score ⩾ 2.² Similarly, the ESC/EAS 2016 guidelines give preference to the CHA₂DS₂-VASc score, recommending AC in men with a score ⩾ 1 and women with a score ⩾ 2.³⁰ In spite of this, several other stroke assessment tools have been proposed, with a goal of improving risk stratification afforded by the CHA₂DS₂-VASc score.

The Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) score expands upon the CHADS₂ score, incorporating elements such as age, prior stroke, female sex, eGFR < 45 mL/min, and proteinuria.³⁵ It also differentially weights risk factors depending on the presence of other co-factors. In contrast to the CHADS₂ and CHA₂DS₂-VASc scores, which assign a large number of patients into the intermediate stroke risk range, the ATRIA score has been shown to correctly reclassify more than 8.6% and 20% of CHA₂DS₂-VASc and CHADS₂ intermediate-risk patients, respectively, largely into lower risk categories.³⁷ These results are complicated by findings in higher-risk patient cohorts which found CHA₂DS₂-VASc to be superior to ATRIA in terms of accurately classifying patients as ‘low risk’³⁸; in these cohorts, patients marked as ‘low risk’ by ATRIA had higher risk on average than those marked as ‘low risk’ by CHA₂DS₂-VASc. To date, no major guidelines have endorsed the use of the ATRIA risk score for NVAF.

Alternative approaches to risk prediction in NVAF have been proposed that take into account temporal changes in risk factors and specific biomarkers. In patients with an initially low CHA₂DS₂-VASc score, recent evidence provides support for the delta CHA₂DS₂-VASc, which assesses change in risk factors over time to predict thromboembolic risk.³⁹ Most patients (90% of the cohort tested) will newly develop one or more comorbidities (most commonly hypertension) prior to developing an ischemic stroke, suggesting broad applicability. Reported hazard ratios for ischemic stroke for a delta CHA₂DS₂-VASc score of one, two, three, and four (calculated 1 year from baseline) are 1.56, 1.94, 2.39, and 2.88, respectively. The delta CHA₂DS₂-VASc score has been shown to be a better predictor of incident stroke than either the initial or subsequent absolute risk score.³⁹

A new biomarker-based risk tool called the ABC stroke risk score incorporates age, biomarkers, and a history of stroke/transient ischemic attack. The ABC stroke risk score has been shown to be superior to the CHA₂DS₂-VASc in both its derivation and external validation cohorts for predicting ischemic stroke.⁴⁰ The ABC stroke risk score uniquely utilizes two biomarkers, NT-proBNP and hs-cTnT, to improve predictive value. The addition of inflammatory (CRP, IL-6), oxidative stress (GDF-15), and coagulation (D-dimer) biomarkers to the risk model did not improve prognostic accuracy.⁴⁰

Modern NVAF risk calculators are subject to a number of limitations. While some calculators have begun to take a weighted approach to risk factors as reflected in epidemiologic data, the most commonly used scores are over-simplistic and binary in terms of how factors are weighted in the calculations. Co-morbidities associated with thromboembolic risk represent a range and do not equally contribute to stroke risk.⁴¹ Additionally, duration and frequency of NVAF episodes are not taken into consideration, which may have an impact on thromboembolic risk.⁴² Finally, these models take into account risk of thromboembolism, but not bleeding risk, so a comprehensive risk–benefit discussion requires incorporation of a separate bleeding risk assessment.

Prediction of progressive or recurrent thromboembolic complications in patients with VTE

Unlike NVAF, risk prediction for VTE is primarily used to define the duration of AC and whether long-term secondary prophylaxis with AC is warranted to reduce the risk of recurrent events. Generally, patients require a minimum of 3 months of AC for an acute, proximal, deep venous thrombosis (DVT) provoked by surgery or a proximal DVT provoked by a non-surgical, transient risk factor. In contrast, lifelong therapy is indicated for patients with a second unprovoked DVT or pulmonary embolism (PE).

According to the 2016 ACCP guidelines, decision-making about whether or not to extend AC treatment in a non-provoked VTE should be dictated by risk of bleeding.⁴ In general, after 3 months of therapy, AC is continued indefinitely in patients at low or moderate bleeding risk. In contrast, those at high bleeding risk will generally stop AC.

Currently, no risk assessment tools exist to acutely guide the duration of secondary prophylaxis in patients hospitalized for VTE. Instead, current risk assessment models are usually carried out after a defined period of AC treatment. Risk scores, however, do exist to predict risk of VTE among predisposed individuals hospitalized for other indications. In fact, hospitalization is a known risk factor for VTE, associated with an eightfold increased risk of VTE due, in part, to factors such as immobility.⁴³ The Padua prediction score for VTE was created to stratify patients into low- or high-risk groups for acute VTE, primarily to determine if prophylactic AC is indicated.⁴⁴ The score is based on presence of active cancer, previous VTE, reduced mobility, known thrombophilia, recent trauma/surgery, age, cardiac/respiratory failure, acute myocardial infarction/ischemic stroke, presence of acute infection, presence of a rheumatologic disorder, and obesity. For scores < 4, prophylactic doses of AC with subcutaneous heparin or LMWH should be considered on a case-by-case basis. For scores ⩾ 4, prophylactic AC is routinely recommended among non-pregnant patients without contraindications.

After the first VTE, risk of recurrence generally decreases and risk of AC-related bleeding increases over time.^45,46 Accordingly, the need for secondary prophylaxis should be serially reassessed based on individual risk factors at discrete follow-up intervals. Evidence from randomized controlled trials such as Rivaroxaban or Aspirin for Extended Treatment of Venous Thromboembolism (EINSTEIN CHOICE) and Apixaban for Extended Treatment of Venous Thromboembolism (AMPLIFY-EXT) suggest that extended-duration thromboprophylaxis with reduced dose rivaroxaban and apixaban are superior to aspirin in terms of preventing recurrent VTE.^47–49

A variety of risk calculators have been created to assess the need for therapeutic AC for secondary VTE prophylaxis after completion of acute VTE treatment, including the Vienna Prediction Model, DASH score, and HER DOO2 score (Table 2). The Vienna Prediction Model can assess recurrence of unprovoked VTE at any time 3 weeks to 15 months after AC is discontinued.^50,51 An update to the prediction model utilizes serial D-dimer levels at multiple random time points to calculate risk of recurrent VTE.⁵¹ The DASH prediction score utilizes similar risk factors, but includes hormonal use at VTE onset, which helps adjust for provoked VTE.⁵² This model indicates that in patients with low risk of recurrence (score ⩽ 1), long-term AC can be avoided, so that roughly 50% of patients with unprovoked VTE may obviate extended therapy. The HER DOO2 recurrent VTE assessment identifies patients at low risk for recurrence who can safely discontinue AC after 5–7 months of therapy.⁵³ The annual risk of recurrent VTE was higher among men and those with signs and symptoms of post-thrombotic syndrome. No men in the low-risk subgroup could be identified who could safely discontinue AC, but women with 0 or 1 risk factors were found to have a low annual risk of VTE recurrence (1.6%), thus allowing for discontinuation of AC. Although these three scores are not specifically endorsed for secondary VTE prophylaxis management in the most recent 2016 ACCP ‘Antithrombotic Therapy for VTE Disease’ guidelines, they have independent predictive ability, and may be used to quantify risk of recurrent VTE at various treatment time points to better facilitate AC risk–benefit assessment.⁴

Table 2.

NVAF bleeding risk calculators.

Bleeding risk calculator	Primary study	Study n	Variables	Range of prediction	Validation studies
HEMORR₂HAGES score	Gage et al. (2006)⁵⁹	3932	Hepatic/renal disease, ethanol abuse, history of malignancy, age, reduced platelet count/function, past bleeding episode, hypertension, anemia, excessive fall risk, stroke history	Annual bleeding risk: 0 (1.9%), 1 (2.5%), 2 (5.3%), 3 (8.4%), 4 (10.4%), and ≥ 5 (12.3%)	Yes⁹¹
HAS-BLED score	Pisters et al. (2010)⁵⁶	3456	Hypertension, renal disease, liver disease, stroke history, major bleeding/bleeding predisposition, labile INR, age, medication use predisposing to bleeding, alcohol use	Annual bleeding risk: 0.9% to > 10% (although upper limit is estimated as n too small)	Yes^61,92,93
ATRIA bleeding score	Fang et al. (2011)⁶⁰	6123	Anemia, severe renal disease (eGFR < 30 mL/min), age, prior hemorrhage, hypertension	Annual bleeding risk: low risk (0.76%) and high risk (5.8%)	Yes⁷¹
ORBIT bleeding score	O’Brien et al. (2015)⁶³	7411	Age, anemia, bleeding history, kidney insufficiency (eGFR < 60 mL/min), antiplatelet treatment	Annual bleeding risk: 0–2, low (1.7–2.9%); 3, medium (4.7%); 4–7, high (6.8–14.9%)	Yes⁶⁵
ABC bleeding score	Hijazi et al. (2016)¹⁶	14,537	Previous bleeding, age, troponin T, GDF-15, hemoglobin	Annual bleeding risk: low (0.6%), medium (1.67%), and high (4.87%)	Yes⁹⁴

eGFR, estimated glomerular filtration rate; GDF-15, growth differentiating factor-15; INR, international normalized ratio; NVAF, non-valvular atrial fibrillation.

Prediction of bleeding in patients with NVAF

When considering anticoagulant therapy for the treatment of NVAF, it is important to assess the risk of both major and minor bleeding. Outside of surgery, the former has been defined by the International Society on Thrombosis and Haemostasis as fatal bleeding, symptomatic bleeding in a critical area or organ, or bleeding causing a decrease in hemoglobin of ⩾ 2 g/dL or requiring transfusion of ⩾ 2 units of packed red blood cells^54,55 (Table 3). For patients with an absolute contraindication to anticoagulation, the decision to hold AC may be clear-cut; however, patients with relative contraindications due to high bleeding risk require a more nuanced approach. The 2014 Canadian Cardiovascular Society atrial fibrillation guidelines recommend implementation of bleeding risk scores (i.e. HAS-BLED) in addition to stroke risk assessment to inform decision-making in complex patients.²⁹ The 2018 ACCP Guideline and Expert Panel Report on antithrombotic therapy for atrial fibrillation further suggests that patients on a VKA at high bleeding risk (i.e. HAS-BLED score ⩾ 3) warrant more frequent follow-up and attention to modifiable bleeding risk factors.³¹ To this end, a variety of new stroke and bleeding risk prediction tools have been subsequently created to further improve risk estimation in clinically challenging patients (Table 2).

Table 3.

Brief summary of the 2010 International Society on Thrombosis and Haemostasis bleeding categories.^{54, 55}

Type of bleed	Criteria for classification
Major, non-surgical	1) Fatal bleeding AND/OR 2) Symptomatic bleeding in a critical area or organ (intracranial, intraocular, intraspinal, intra-articular, pericardial, retroperitoneal, or intramuscular with compartment syndrome) AND/OR 3) Bleeding causing a fall in hemoglobin (Hb) ⩾ 2 g/dL or leading to ⩾ 2 units of whole blood transfusion
Major, surgical	Patients with start of surgery within 5 half-lives of the target anticoagulant with: 1) Fatal bleeding AND/OR 2) Symptomatic bleeding in a critical area or organ AND/OR 3) Extrasurgical site bleeding causing a fall in Hb ⩾ 2 g/dL or leading to ⩾ 2 units transfusion AND/OR 4) Surgical site bleeding requiring a second intervention or hemarthrosis interfering with mobilization or wound healing AND/OR 5) Unexpected/prolonged surgical site bleeding causing hemodynamic instability within 24 hours of bleeding with a fall in Hb ⩾ 2 g/dL or ⩾ 2 units transfusion
Minor, clinically relevant	Does not meet criteria for major bleed but bleeding prompting: 1) Hospital admission for bleeding AND/OR 2) Physician-guided medical or surgical treatment for bleeding AND/OR 3) Change in antithrombotic therapy
Minor, non-clinically relevant	Does not meet criteria for major bleed AND bleeding does not prompt a clinical response

The first of these tools is the HAS-BLED score, which takes into account several bleeding factors to calculate yearly bleeding risk.⁵⁶ The HAS-BLED score draws attention to several modifiable bleeding risk factors such as uncontrolled hypertension, concomitant use of aspirin or non-steroidal anti-inflammatory drugs, and alcohol use. Furthermore, it takes into account the quality of AC control, reflected by TTR among VKA users, which can be easily assessed via the percentage of time having achieved a therapeutic international normalized ratio (INR) goal. A high HAS-BLED score (⩾ 3) is indicative of the need for regular clinical review and follow-up, but should not be used per se to stop AC.^57,58 Similar to other bleeding risk calculators, patients with NVAF and a high HAS-BLED score derive a higher net clinical benefit from AC when balancing the risks of ischemic stroke against intracranial bleeding. The HAS-BLED score has additionally been shown to have good predictive value for risk of intracranial bleeding and bleeding risk associated with bridging therapy.

Several other bleeding risk calculators, such as HEMORR₂HAGES and ATRIA, were designed concurrent with HAS-BLED but are less predictive of bleeding overall.^59,60 HAS-BLED was found to be superior to ATRIA and HEMORR₂HAGES upon retrospective analysis of VKA patients from the ‘Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation (AMADEUS)’ study,⁶¹ although all three risk scores were shown to have limited predictive accuracy overall (c-statistics < 0.70). In this comparison, the ATRIA score failed to demonstrate additional benefit over chance alone, whereas HEMORR₂HAGES was slightly better than random stratification. The HEMORR₂HAGES score had the worst mortality prediction, thought to be due to a larger number of bleeding-oriented factors used in its calculation. A subsequent study showed that HAS-BLED demonstrated superior predictive ability as compared to ATRIA amongst both VKA and non-VKA anticoagulant subgroups.⁶² Although the ATRIA tool may be of limited clinical utility, one unique aspect of the tool is that it was the first bleeding risk calculator to weigh risk factors differentially.

Each of the aforementioned bleeding risk scores was primarily designed and validated in those on VKAs. In an attempt to better predict bleeding risk among those on a VKA or DOAC, the ORBIT score was introduced in 2015, taking into account five easily calculable variables to predict bleeding risk: age, anemia, treatment with concurrent antiplatelet agent, renal insufficiency, and prior bleeding history.⁶³ The ORBIT score notably eliminated ‘labile INR’ as a criterion due to its difficulty to measure and poor applicability to patients on DOACs or those not on prior anticoagulation. The original study cohort included both patients on warfarin and dabigatran,⁶³ and was further validated in those on rivaroxaban.⁶⁴ The ORBIT score was subsequently shown to be superior to HAS-BLED, ATRIA, and HEMORR₂HAGES^64,65 among patients on VKA or dabigatran.

Subsequent analyses comparing ORBIT to HAS-BLED have demonstrated greater variability in predictive value, with several cohort studies showing HAS-BLED to be more predictive of bleeding risk than ORBIT for those on VKA or DOAC.^66–70 In these analyses, the ORBIT score tended to classify a higher percentage of patients as ‘low risk’⁷⁰ as compared to HAS-BLED, but was ultimately less predictive at identifying those at true ‘low risk’ of major bleeding among patients treated with a VKA.⁶⁷ It is possible that exclusion of labile INR or an alternative assessment of AC quality may have led the ORBIT score to underestimate risk, with one study showing significant improvement in predictive performance via the addition of ‘time in the therapeutic range’ as an additional variable.⁷¹ Ultimately, we believe that the ORBIT score is not demonstratively superior to HAS-BLED, and favor use of HAS-BLED, given that it is less likely to underestimate risk of bleeding in patients on a VKA or a DOAC.

Reflective of recent research into several new biomarkers correlated with AC-associated bleeding risk, the ABC bleeding score was created.¹⁶ This risk score incorporates hemoglobin, troponin, and GDF-15 in addition to demographic data to improve bleeding risk prediction. The ABC bleeding risk calculator, along with its counterpart ABC stroke risk calculator, were both derived from the 2011 ARISTOTLE study, with results translated into ‘low-’, ‘intermediate-’, and ‘high-’ risk of bleeding or stroke, respectively. The ABC bleeding score outperformed other bleeding scores, such as HAS-BLED and ORBIT, in its validation cohort, and continued to outperform these calculators when GDF-15 was exchanged for the more frequently assessed ‘creatinine clearance’.¹⁶

Use of biomarkers as part of the ABC score also allows for easier monitoring of changes in bleeding risk, which can be helpful in tailoring therapy over time.⁷² Prior to the ABC scores, pairing stroke and bleeding scores for overall clinical assessment was difficult given that scores were typically derived from different cohorts and utilized differing variables and clinical endpoints. In addition to being derived from similar cohorts, the ABC scores overlap traditional and biomarker variables (age and troponin) that contribute dually to thromboembolic and bleeding risk. Simultaneous use of these two risk tools may provide a future opportunity for improved tailoring of stroke prevention treatment in patients with atrial fibrillation due to their clinical overlap.

Ultimately, the decision whether or not to start AC for NVAF in high-risk patients can be aided by risk stratification tools that assist in quantifying estimated bleeding risk. Bleeding risk is not a static process, and patients require re-evaluation at frequent intervals over the course of their follow-up.⁶⁶ Importantly, modifiable risk factors for bleeding should be addressed to mitigate the risk of bleeding. Bleeding calculators do not take into account location or severity of bleeding, which are important factors for decision-making that may inform patient and physician preference. Finally, choice of AC may have an impact on risk stratification for these patients, as certain anticoagulants have been shown to have superior safety profiles as compared with VKAs, thus favoring treatment with AC in equivocal risk–benefit cases among those previously classified at lower ischemic risk.^10,12,73

Prediction of bleeding in patients with VTE

With regard to calculating bleeding risk in patients with VTE, options include use of NVAF bleeding tools (i.e. HAS-BLED, ATRIA, HEMORR₂HAGES) or those specific to VTE (Table 4). The 2010 International Medical Prevention Registry on Venous Thromboembolism (IMPROVE) score was created to assess risk of bleeding with VTE thromboprophylaxis in the inpatient setting, as a counterpart bleeding score to the Padua Prediction Score.^74,75 The Kuijer and RIETE scores were similarly derived and developed in VTE-specific populations to calculate the risk of hemorrhage with AC.^76,77 Disappointingly, these latter scores suffer from inferior predictive value within the first 3 months of treatment. They are also limited given their derivation from cohorts of patients using VKAs, and accordingly may not translate well to patients not treated with VKAs.

Table 4.

VTE risk calculators.

Bleeding risk calculator	Type of risk calculator	Primary study	Study n	Variables	Range of prediction	Validation studies
HER DOO2 score	VTE risk	Rodger et al. (2008)⁵³	646	Men continue. For women: hyperpigmentation (1 point), edema (1 point), redness (1 point), D-dimer ⩾ 250 µg/L during AC (1 point), BMI ⩾ 30 kg/m² (1 point), age ⩾ 65 years (1 point)	Annual recurrence risk: Men: 13.7%, Women: score of ⩽ 1 (1.6%), ⩾ 2 (14.1%)	Yes⁹⁵
Vienna risk calculator	VTE risk	Eichinger et al. (2010)⁵¹	929	Male > female, PE > proximal DVT > distal DVT, elevated D-dimer after AC	Annual recurrence risk: 2–15%, per normogram	Yes^50,96
DASH prediction score	VTE risk	Tosetto et al. (2012)⁵²	1,818	D-dimer abnormal after cessation of AC (2 points), age ⩽ 50 years (1 point), male sex (1 point), hormonal use at VTE onset (–2 points)	Annual recurrence risk: ⩽ 1 (3.1%), 2 (6.4%), ⩾ 3 (12.3%)	Yes⁹⁷
Modified Outpatient Bleeding Risk Index (mOBRI)	VTE and NVAF bleeding risk	Beyth et al. (1998)⁷⁸	565	One point for age ⩾ 65, history of GI bleeding, history of stroke AND one additional point for each of the following comorbidities: recent MI, anemia, renal impairment, and diabetes mellitus	48-month major bleeding risk: 0 low risk (3%), 1–2 moderate risk (12%), and 3–4 high risk (53%)	Yes^79,98
Kuijer calculator	VTE bleeding risk	Kuijer et al (1999)⁷⁶	241	Age > 60 years (1.6 points), female sex (1.3 points), presence of malignancy (2.2 points)	Annual bleeding risk: 0 points, low risk (0%); 1–3 points, intermediate risk (6%); and ⩾ 4 points, high risk (26%)	No
RIETE risk calculator	VTE bleeding risk	Ruíz-Giménez et al. (2008)⁷⁷	19,274	Recent major bleeding < 15 days prior to VTE (2 points), creatinine > 1.2 mg/dL (1.5 points), anemia (1.5 points), malignancy (1 point), PE (1 point), and age > 75 years (1 point)	Three-month bleeding risk: 0 points, low risk (0.3%); 1–4 points, intermediate risk (2.6%); > 4 points, high risk (7.3%)	Yes^99,100
IMPROVE score	VTE prophylaxis bleeding risk	Decousus et al. (2011)⁷⁴	15,156	Moderate or severe renal failure (1 or 2.5 points), sex (1 point), age (1 or 3.5 points), active malignancy (2 points), rheumatic disease (2 points), central venous catheter (2 points), intensive/cardiac care admission (2.5 points), hepatic failure, platelets < 50,000 cells/μL (4 points), bleeding 3 months prior to admission (4 points), active gastroduodenal ulcer (4.5 points)	Low-risk score < 7.0 points indicates low risk of bleeding, whereas high-risk score ⩾ 7.0 points indicates high risk of bleeding	Yes⁷⁵

AC, anticoagulation; BMI, body mass index; DVT, deep venous thrombosis; GI, gastrointestinal; MI, myocardial infarction; NVAF, non-valvular atrial fibrillation; PE, pulmonary embolism; VTE, venous thromboembolism.

From a pragmatic perspective, the NVAF bleeding scores were derived from separate cohorts and are not easily translated to data from counterpart VTE risk prediction scores. Even assuming one could simultaneously compare risk of subsequent VTE and bleeding from AC at a given time point, there exists no threshold to define what is considered an acceptable risk–benefit ratio or cut-off from which one could make an algorithmic decision in the indeterminate-risk patient.

Precision of bleeding in patients with NVAF and VTE

The first bleeding risk scores, the Outpatient Bleeding Risk Index (OBRI) and modified Outpatient Bleeding Risk Index (mOBRI), did not distinguish the indication for AC when measuring risk of hemorrhage with a VKA. These scores included patients with varied indications for AC (NVAF, VTE, and post-cardiac surgery), with calculated bleeding risk measured independently of etiology of AC.⁷⁸ mOBRI was initially developed based on patients discharged from the hospital on long-term warfarin therapy, but was later validated in outpatients initiated on warfarin after hospital discharge with up to a 7-year follow-up.^78,79

Subsequent bleeding risk calculators, as described in the sections above, have historically focused on either NVAF or VTE, with minimal clinical overlap. Given the relative paucity of VTE bleeding risk calculators, several of the bleeding risk scores developed in the NVAF population have been evaluated in patients with VTE. For example, HAS-BLED has been shown to be predictive in patients with VTE, such that a score ⩾ 3 is associated with an increased risk of major bleeding.⁸⁰

Currently, no large prospective studies exist that assess whether bleeding risk in patients with both atrial fibrillation and VTE varies compared to patients with only one indication for AC. Given that different DOAC dosing may be used for various AC indications, further investigation into the dose-dependent prediction of bleeding risk is needed.

Given a similar mechanism of hemorrhage for those receiving AC for NVAF and VTE, bleeding risk scores have significant potential for overlap. Ultimately, bleeding risk is dependent on the specific anticoagulant and dose, with the need to take these aspects into account to ensure fair comparison.

How to operationalize risk assessment tools in practice

Use of ischemic and bleeding risk scores in patients with either NVAF or VTE has been shown to outperform clinician judgment. Steinberg et al. investigated this question through a prospective, observational study of NVAF patients in the ORBIT-AF registry, assessing systemic embolism and major bleeding risk at 28 months with physician-assessment of risk in addition to CHADS₂, CHA₂DS₂-VASc, ATRIA, and HAS-BLED scores.⁸¹ Physician-based risk assessments for patients with NVAF showed risk stratification comparable to stroke scores, but these assessments were found to be inferior to bleeding scores. Furthermore, combining subjective and objective assessments did not add incremental accuracy as compared to empiric risk assessment alone.

Assessment of stroke risk for NVAF is relatively straightforward, in that stroke risk scores are easily actionable and supported by guidelines to direct therapeutic decisions. In contrast, assessment and implementation of bleeding risk is less straightforward. No prospective trial has demonstrated a benefit to withholding AC in patients identified as high bleeding risk by any criteria.⁸¹ Furthermore, some observational data have suggested that a net clinical benefit of AC persists even among patients at the highest risk of bleeding; thus, most major guidelines do not wholly endorse the use of bleeding score algorithms as the primary basis for treatment decision-making.^58,82 Ultimately, both stroke and bleeding assessments provide valuable risk stratification information in NVAF, and further work is needed to delineate maximally accurate and actionable risk stratification.

Management of extended duration AC for VTE is less well defined than NVAF given the limited predictive ability of current assessment tools and lack of firm incorporation of these tools into current guidelines.⁸³ A combined approach incorporating clinical VTE and bleeding risk calculators, subjective clinician assessment, and patient preference may be reasonable in the majority of cases.

Conclusions

A fair number of thromboembolism and bleeding risk prediction tools have been developed to guide AC in NVAF and VTE. Incorporation of these scores into practice has been challenging in spite of their ability to often improve decision-making. Biomarkers offer the potential promise of more personalized risk assessment over time, and choosing biomarkers that are widely available will be important to greater use. Finally, with increasingly greater use of DOACs in clinical practice, risk scores will need to be further refined.

Footnotes

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: TJG is a consultant for Boehringer-Ingelheim; VAP, RKK, KK, JWM, and RSB have no relevant disclosures or conflicts of interest to report.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Vincent A Pallazola

References

Risk factors for stroke and efficacy of antithrombotic therapy in atrial fibrillation. Analysis of pooled data from five randomized controlled trials. Arch Intern Med 1994; 154: 1449–1457.

January

Wann

Alpert

, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: Executive summary: A report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society. Circulation 2014; 130: 2071–2104.

ACTIVE Investigators Connolly

Pogue

, et al. Effect of clopidogrel added to aspirin in patients with atrial fibrillation. N Engl J Med 2009; 360: 2066–2078.

Kearon

Akl

Ornelas

, et al. Antithrombotic therapy for VTE disease: CHEST Guideline and Expert Panel Report. Chest 2016; 149: 315–352.

White

Gruber

Feyzi

, et al. Comparison of outcomes among patients randomized to warfarin therapy according to anticoagulant control: Results from SPORTIF III and V. Arch Intern Med 2007; 167: 239–245.

National Clinical Guideline Centre (UK). Atrial Fibrillation: The Management of Atrial Fibrillation, No. 180, http://www.ncbi.nlm.nih.gov/books/NBK248059/ (2014, accessed 27 August, 2018).

Eikelboom

Connolly

Bosch

, et al. Rivaroxaban with or without aspirin in stable cardiovascular disease. N Engl J Med 2017; 377: 1319–1330.

Connolly

Ezekowitz

Yusuf

, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med 2009; 361: 1139–1151.

Wallentin

Yusuf

Ezekowitz

, et al. Efficacy and safety of dabigatran compared with warfarin at different levels of international normalised ratio control for stroke prevention in atrial fibrillation: An analysis of the RE-LY trial. Lancet 2010; 376: 975–983.

10.

Patel

Mahaffey

Garg

, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 2011; 365: 883–891.

11.

Connolly

Eikelboom

Joyner

, et al. Apixaban in patients with atrial fibrillation. N Engl J Med 2011; 364: 806–817.

12.

Granger

Alexander

McMurray

JJV

, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 2011; 365: 981–992.

13.

McCarthy

Yousuf

Alonso

, et al. High-sensitivity troponin as a biomarker in heart rhythm disease. Am J Cardiol 2017; 119: 1407–1413.

14.

Hijazi

Siegbahn

Andersson

, et al. High-sensitivity troponin I for risk assessment in patients with atrial fibrillation: Insights from the Apixaban for Reduction in Stroke and other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial. Circulation 2014; 129: 625–634.

15.

Wallentin

Hijazi

Andersson

, et al. Growth differentiation factor 15, a marker of oxidative stress and inflammation, for risk assessment in patients with atrial fibrillation: Insights from the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial. Circulation 2014; 130: 1847–1858.

16.

Hijazi

Oldgren

Lindbäck

, et al. The novel biomarker-based ABC (age, biomarkers, clinical history)-bleeding risk score for patients with atrial fibrillation: A derivation and validation study. Lancet 2016; 387: 2302–2311.

17.

Hijazi

Oldgren

Siegbahn

, et al. Biomarkers in atrial fibrillation: A clinical review. Eur Heart J 2013; 34: 1475–1480.

18.

Aulin

Siegbahn

Hijazi

, et al. Interleukin-6 and C-reactive protein and risk for death and cardiovascular events in patients with atrial fibrillation. Am Heart J 2015; 170: 1151–1160.

19.

Christersson

Wallentin

Andersson

, et al. D-dimer and risk of thromboembolic and bleeding events in patients with atrial fibrillation—Observations from the ARISTOTLE trial. J Thromb Haemost 2014; 12: 1401–1412.

20.

Fukushima

Kato

, et al. Correlation between left atrial appendage morphology and flow velocity in patients with paroxysmal atrial fibrillation. Eur Heart J Cardiovasc Imaging 2016; 17: 59–66.

21.

Daccarett

Badger

Akoum

, et al. Association of left atrial fibrosis detected by delayed-enhancement magnetic resonance imaging and the risk of stroke in patients with atrial fibrillation. J Am Coll Cardiol 2011; 57: 831–838.

22.

Pabinger

Biomarkers and venous thromboembolism. Arterioscler Thromb Vasc Biol 2009; 29: 332–336.

23.

Van Es

Louzada

Carrier

, et al. Predicting the risk of recurrent venous thromboembolism in patients with cancer: A prospective cohort study. Thromb Res 2018; 163: 41–46.

24.

Lip

GYH

Nieuwlaat

Pisters

, et al. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: The Euro Heart Survey on Atrial Fibrillation. Chest 2010; 137: 263–272.

25.

Wyndham

CRC

. Atrial fibrillation: The most common arrhythmia. Tex Heart Inst J 2000; 27: 257–267.

26.

Wyse

Waldo

DiMarco

, et al. A comparison of rate control and rhythm control in patients with atrial fibrillation. N Engl J Med 2002; 347: 1825–1833.

27.

Hagens

Van Gelder

Crijns

HJGM

; RAte Control Versus Electrical Cardioversion Of Persistent Atrial Fibrillation (RACE) Study Group. The RACE study in perspective of randomized studies on management of persistent atrial fibrillation. Card Electrophysiol Rev 2003; 7: 118–121.

28.

Glotzer

Daoud

Wyse

, et al. The relationship between daily atrial tachyarrhythmia burden from implantable device diagnostics and stroke risk: The TRENDS study. Circ Arrhythm Electrophysiol 2009; 2: 474–480.

29.

Macle

Cairns

Andrade

, et al. The 2014 Atrial Fibrillation Guidelines Companion: A practical approach to the use of the Canadian Cardiovascular Society Guidelines. Can J Cardiol 2015; 31: 1207–1218.

30.

Kirchhof

Benussi

Kotecha

, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur J Cardiothorac Surg 2016; 50: e1–e88.

31.

Lip

GYH

Banerjee

Boriani

, et al. Antithrombotic therapy for atrial fibrillation: CHEST Guideline and Expert Panel Report. Chest 2018; 154: 1121–1201.

32.

Gage

Waterman

Shannon

, et al. Validation of clinical classification schemes for predicting stroke: Results from the National Registry of Atrial Fibrillation. JAMA 2001; 285: 2864–2870.

33.

Fang

Chang

, et al. Comparison of risk stratification schemes to predict thromboembolism in people with nonvalvular atrial fibrillation. J Am Coll Cardiol 2008; 51: 810–815.

34.

Conen

Chae

Glynn

, et al. Risk of death and cardiovascular events in initially healthy women with new-onset atrial fibrillation. JAMA 2011; 305: 2080–2087.

35.

Singer

Chang

Borowsky

, et al. A new risk scheme to predict ischemic stroke and other thromboembolism in atrial fibrillation: The ATRIA Study Stroke Risk Score. J Am Heart Assoc 2013; 2: e000250.

36.

Van den Ham

Klungel

Singer

, et al. Comparative performance of ATRIA, CHADS2, and CHA2DS2-VASc risk scores predicting stroke in patients with atrial fibrillation: Results from a national primary care database. J Am Coll Cardiol 2015; 66: 1851–1859.

37.

Eikelboom

Connolly

Brueckmann

, et al. Dabigatran versus warfarin in patients with mechanical heart valves. N Engl J Med 2013; 369: 1206–1214.

38.

Chao

T-F

Liu

C-J

Wang

K-L

, et al. Using the CHA2DS2-VASc score for refining stroke risk stratification in ‘low-risk’ Asian patients with atrial fibrillation. J Am Coll Cardiol 2014; 64: 1658–1665.

39.

Chao

T-F

Lip

GYH

Liu

C-J

, et al. Relationship of aging and incident comorbidities to stroke risk in patients with atrial fibrillation. J Am Coll Cardiol 2018; 71: 122–132.

40.

Hijazi

Lindbäck

Alexander

, et al. The ABC (age, biomarkers, clinical history) stroke risk score: A biomarker-based risk score for predicting stroke in atrial fibrillation. Eur Heart J 2016; 37: 1582–1590.

41.

Lip

GYH

Nielsen

PB.

Should patients with atrial fibrillation and 1 stroke risk factor (CHA2DS2-VASc Score 1 in men, 2 in women) be anticoagulated? Yes: even 1 stroke risk factor confers a real risk of stroke. Circulation 2016; 133: 1498–1503; discussion 1503.

42.

Hunter

McCready

Diab

, et al. Maintenance of sinus rhythm with an ablation strategy in patients with atrial fibrillation is associated with a lower risk of stroke and death. Heart 2012; 98: 48–53.

43.

Anderson

Spencer

FA.

Risk factors for venous thromboembolism. Circulation 2003; 107(23 Suppl 1): I9–16.

44.

Barbar

Noventa

Rossetto

, et al. A risk assessment model for the identification of hospitalized medical patients at risk for venous thromboembolism: The Padua Prediction Score. J Thromb Haemost 2010; 8: 2450–2457.

45.

Christiansen

Cannegieter

Koster

, et al. Thrombophilia, clinical factors, and recurrent venous thrombotic events. JAMA 2005; 293: 2352–2361.

46.

Heit

Mohr

Silverstein

, et al. Predictors of recurrence after deep vein thrombosis and pulmonary embolism: A population-based cohort study. Arch Intern Med 2000; 160: 761–768.

47.

Weitz

Lensing

AWA

Prins

, et al. Rivaroxaban or aspirin for extended treatment of venous thromboembolism. N Engl J Med 2017; 376: 1211–1222.

48.

Agnelli

Buller

Cohen

, et al. Apixaban for extended treatment of venous thromboembolism. N Engl J Med 2013; 368: 699–708.

49.

Vasanthamohan

Boonyawat

Chai-Adisaksopha

, et al. Reduced-dose direct oral anticoagulants in the exten-ded treatment of venous thromboembolism: A systematic review and meta-analysis. J Thromb Haemost 2018; 16: 1288–1295.

50.

Eichinger

Heinze

Kyrle

PA.

D-dimer levels over time and the risk of recurrent venous thromboembolism: An update of the Vienna Prediction Model. J Am Heart Assoc 2014; 3: e000467.

51.

Eichinger

Heinze

Jandeck

, et al. Risk assessment of recurrence in patients with unprovoked deep vein thrombosis or pulmonary embolism: The Vienna Prediction Model. Circulation 2010; 121: 1630–1636.

52.

Tosetto

Iorio

Marcucci

, et al. Predicting disease recurrence in patients with previous unprovoked venous thromboembolism: A proposed prediction score (DASH). J Thromb Haemost 2012; 10: 1019–1025.

53.

Rodger

Kahn

Wells

, et al. Identifying unprovoked thromboembolism patients at low risk for recurrence who can discontinue anticoagulant therapy. CMAJ 2008; 179: 417–426.

54.

Schulman

Kearon

; Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005; 3: 692–694.

55.

Schulman

Angerås

Bergqvist

, et al. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in surgical patients. J Thromb Haemost 2010; 8: 202–204.

56.

Pisters

Lane

Nieuwlaat

, et al. 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.

57.

Olesen

Lip

GYH

Lindhardsen

, 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: 739–749.

58.

Friberg

Rosenqvist

Lip

GYH

. Net clinical benefit of warfarin in patients with atrial fibrillation: A report from the Swedish atrial fibrillation cohort study. Circulation 2012; 125: 2298–2307.

59.

Gage

Yan

Milligan

, et al. Clinical classification schemes for predicting hemorrhage: Results from the National Registry of Atrial Fibrillation (NRAF). Am Heart J 2006; 151: 713–719.

60.

Fang

Chang

, 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.

61.

Apostolakis

Lane

Guo

, et al. Performance of the HEMORR₂HAGES, 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: 861–867.

62.

Roldán

Marín

Fernández

, 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: 179–184.

63.

O’Brien

Simon

Thomas

, et al. The ORBIT bleeding score: A simple bedside score to assess bleeding risk in atrial fibrillation. Eur Heart J 2015; 36: 3258–3264.

64.

Lip

GYH

Lane

DA.

ACP Journal Club. The 5-factor ORBIT bleeding score predicted major bleeding at 2 years in patients with atrial fibrillation. Ann Intern Med 2016; 164: JC11.

65.

Proietti

Hijazi

Andersson

, et al. Comparison of bleeding risk scores in patients with atrial fibrillation: Insights from the RE-LY trial. J Intern Med 2018; 283: 282–292.

66.

Lip

GYH

Lane

DA.

Bleeding risk assessment in atrial fibrillation: Observations on the use and misuse of bleeding risk scores. J Thromb Haemost 2016; 14: 1711–1714.

67.

Proietti

Senoo

Lane

, et al. Major bleeding in patients with non-valvular atrial fibrillation: Impact of time in therapeutic range on contemporary bleeding risk scores. Sci Rep 2016; 6: 24376.

68.

Senoo

Lip

GYH

. Predictive abilities of the HAS-BLED and ORBIT bleeding risk scores in non-warfarin anticoagulated atrial fibrillation patients: An ancillary analysis from the AMADEUS trial. Int J Cardiol 2016; 221: 379–382.

69.

Guo

Zhang

Shi

, et al. Assessing bleeding risk in 4824 Asian patients with atrial fibrillation: The Beijing PLA Hospital Atrial Fibrillation Project. Sci Rep 2016; 6. 31755.

70.

Wang

Zhu

, et al. Comparing the ORBIT and HAS-BLED bleeding risk scores in anticoagulated atrial fibrillation patients: A systematic review and meta-analysis. Oncotarget 2017; 8: 109703–109711.

71.

Senoo

Proietti

Lane

, et al. Evaluation of the HAS-BLED, ATRIA, and ORBIT bleeding risk scores in patients with atrial fibrillation taking warfarin. Am J Med 2016; 129: 600–607.

72.

Hijazi

Oldgren

Andersson

, et al. Importance of persistent elevation of cardiac biomarkers in atrial fibrillation: A RE-LY substudy. Heart 2014; 100: 1193–1200.

73.

Giugliano

Ruff

Braunwald

, et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med 2013; 369: 2093–2104.

74.

Decousus

Tapson

Bergmann

J-F

, et al. Factors at admission associated with bleeding risk in medical patients: Findings from the IMPROVE investigators. Chest 2011; 139: 69–79.

75.

Rosenberg

Press

Fishbein

, et al. External validation of the IMPROVE Bleeding Risk Assessment Model in medical patients. Thromb Haemost 2016; 116: 530–536.

76.

Kuijer

Hutten

Prins

, et al. Prediction of the risk of bleeding during anticoagulant treatment for venous thromboembolism. Arch Intern Med 1999; 159: 457–460.

77.

Ruíz-Giménez

Suárez

González

, et al. Predictive variables for major bleeding events in patients presenting with documented acute venous thromboembolism. Findings from the RIETE Registry. Thromb Haemost 2008; 100: 26–31.

78.

Beyth

Quinn

Landefeld

CS.

Prospective evaluation of an index for predicting the risk of major bleeding in outpatients treated with warfarin. Am J Med 1998; 105: 91–99.

79.

Aspinall

DeSanzo

Trilli

, et al. Bleeding Risk Index in an anticoagulation clinic. Assessment by indication and implications for care. J Gen Intern Med 2005; 20: 1008–1013.

80.

Kooiman

van Hagen

Iglesias del Sol

, et al. The HAS-BLED score identifies patients with acute venous thromboembolism at high risk of major bleeding complications during the first six months of anticoagulant treatment. PLoS One 2015; 10(4): e0122520.

81.

Steinberg

Shrader

Kim

, et al. How well does physician risk assessment predict stroke and bleeding in atrial fibrillation? Results from the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF). Am Heart J 2016; 181: 145–152.

82.

Singer

Chang

Fang

, et al. The net clinical benefit of warfarin anticoagulation in atrial fibrillation. Ann Intern Med 2009; 151: 297–305.

83.

Barnes

Kanthi

Froehlich

JB.

Venous thromboembolism: Predicting recurrence and the need for extended anticoagulation. Vasc Med 2015; 20: 143–152.

84.

Gage

van Walraven

Pearce

, et al. Selecting patients with atrial fibrillation for anticoagulation: Stroke risk stratification in patients taking aspirin. Circulation 2004; 110: 2287–2292.

85.

Ntaios

Lip

GYH

Makaritsis

, et al. CHADS₂, CHA₂S₂DS₂-VASc, and long-term stroke outcome in patients without atrial fibrillation. Neurology 2013; 80: 1009–1017.

86.

Friberg

Rosenqvist

Lip

GYH

. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: The Swedish Atrial Fibrillation cohort study. Eur Heart J 2012; 33: 1500–1510.

87.

Okumura

Inoue

Atarashi

, et al. Validation of CHA₂DS₂-VASc and HAS-BLED scores in Japanese patients with nonvalvular atrial fibrillation: An analysis of the J-RHYTHM Registry. Circ J 2014; 78: 1593–1599.

88.

Aspberg

Chang

Atterman

, et al. Comparison of the ATRIA, CHADS2, and CHA2DS2-VASc stroke risk scores in predicting ischaemic stroke in a large Swedish cohort of patients with atrial fibrillation. Eur Heart J 2016; 37: 3203–3210.

89.

Zhu

Ding

, et al. Meta-analysis of ATRIA versus CHA2DS2-VASc for predicting stroke and thromboembolism in patients with atrial fibrillation. Int J Cardiol 2017; 227: 436–442.

90.

Oldgren

Hijazi

Lindbäck

, et al. Performance and validation of a novel biomarker-based stroke risk score for atrial fibrillation. Circulation 2016; 134: 1697–1707.

91.

Caldeira

Costa

Fernandes

, et al. Performance of the HAS-BLED high bleeding-risk category, compared to ATRIA and HEMORR2HAGES in patients with atrial fibrillation: A systematic review and meta-analysis. J Interv Card Electrophysiol 2014; 40: 277–284.

92.

Lip

GYH

Frison

Halperin

, et al. Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: The HAS-BLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol 2011; 57: 173–180.

93.

Fauchier

Chaize

Gaudin

A-F

, et al. Predictive ability of HAS-BLED, HEMORR2HAGES, and ATRIA bleeding risk scores in patients with atrial fibrillation. A French nationwide cross-sectional study. Int J Cardiol 2016; 217: 85–91.

94.

Zulkifly

Lip

GYH

Lane

DA.

Bleeding risk scores in atrial fibrillation and venous thromboembolism. Am J Cardiol 2017; 120: 1139–1145.

95.

Rodger

Le Gal

Anderson

, et al. Validating the HERDOO2 rule to guide treatment duration for women with unprovoked venous thrombosis: Multinational prospective cohort management study. BMJ 2017; 356: j1065.

96.

Ensor

Riley

Moore

, et al. Systematic review of prognostic models for recurrent venous thromboembolism (VTE) post-treatment of first unprovoked VTE. BMJ Open 2016; 6: e011190.

97.

Tosetto

Testa

Martinelli

, et al. External validation of the DASH prediction rule: A retrospective cohort study. J Thromb Haemost 2017; 15: 1963–1970.

98.

Wells

Forgie

Simms

, et al. The outpatient bleeding risk index: Validation of a tool for predicting bleeding rates in patients treated for deep venous thrombosis and pulmonary embolism. Arch Intern Med 2003; 163: 917–920.

99.

Guijarro

Montes

Sanromán

, et al.; RIETE Investigators. Venous thromboembolism in Spain. Comparison between an administrative database and the RIETE registry. Eur J Intern Med 2008; 19: 443–446.

100.

Nieto

Solano

Trapero Iglesias

, et al. Validation of a score for predicting fatal bleeding in patients receiving anticoagulation for venous thromboembolism. Thromb Res 2013; 132: 175–179.

Anticoagulation risk assessment for patients with non-valvular atrial fibrillation and venous thromboembolism: A clinical review

Abstract

Keywords

Introduction

Antithrombotic and anticoagulant therapies used for non-valvular atrial fibrillation (NVAF) and venous thromboembolism (VTE)

Variables implicated in prediction of thromboembolism and bleeding in NVAF and VTE

Prediction of stroke or systemic embolism in patients with NVAF

Prediction of progressive or recurrent thromboembolic complications in patients with VTE

Prediction of bleeding in patients with NVAF

Prediction of bleeding in patients with VTE

Precision of bleeding in patients with NVAF and VTE

How to operationalize risk assessment tools in practice

Conclusions

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References