0
Original Research: Cardiovascular Disease |

Factors Affecting Quality of Anticoagulation Control Among Patients With Atrial Fibrillation on WarfarinAnticoagulation Control in Atrial Fibrillation: The SAMe-TT2R2 Score FREE TO VIEW

Stavros Apostolakis, MD, PhD; Renee M. Sullivan, MD; Brian Olshansky, MD; Gregory Y. H. Lip, MD
Author and Funding Information

From the University of Birmingham Centre for Cardiovascular Sciences (Drs Apostolakis and Lip), City Hospital, Birmingham, England; and the Division of Cardiovascular Medicine (Drs Sullivan and Olshansky), University of Iowa Hospitals and Clinics, Iowa City, IA.

Correspondence to: Gregory Y. H. Lip, MD, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Dudley Rd, Birmingham, B18 7QH, England; e-mail: g.y.h.lip@bham.ac.uk


For editorial comment see page 1437

Funding/Support: The authors have reported to CHEST that no funding was received for this study.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details.


Chest. 2013;144(5):1555-1563. doi:10.1378/chest.13-0054
Text Size: A A A
Published online

Background:  When oral anticoagulation with adjusted-dose vitamin K antagonist (VKA) is used, the quality of anticoagulation control (as reflected by the time in therapeutic range [TTR] of the international normalized ratio [INR]) is an important determinant of thromboembolism and bleeding. Our objective was to derive a validated scheme using patient-related clinical parameters to assess the likelihood of poor INR control among patients with atrial fibrillation (AF) on VKA therapy.

Methods:  The Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial population was randomly divided into derivation and internal validation cohorts using a 1:1 ratio. We used linear regression analysis to detect the clinical factors associated with TTR and binary logistic regression to evaluate the predictive performance of a model incorporating these factors for different cutoff values of TTR. The derived model was validated externally in a cohort of patients receiving anticoagulant therapy who were recruited prospectively.

Results:  In the linear regression model, nine variables emerged as independent predictors of TTR: female sex (P < .0001), age < 50 years (P < .0001), age 50 to 60 years (P = .02), ethnic minority status (P < .0001), smoking (P = .03), more than two comorbidities (P < .0001), and being treated with a β-blocker (P = .02), verapamil (P = .02), or, inversely, with amiodarone (P = .05). We incorporated these factors into a simple clinical prediction scheme with the acronym SAMe-TT2R (sex female, age < 60 years, medical history [more than two comorbidities], treatment [interacting drugs, eg, amiodarone for rhythm control], tobacco use [doubled], race [doubled]). The score demonstrated good discrimination performance in both the internal and external validation cohorts (c-index, 0.72; 95% CI, 0.64-0.795; and c-index, 0.7; 95% CI, 0.57-0.82, respectively).

Conclusions:  Common clinical and demographic factors can influence the quality of oral anticoagulation. We incorporated these factors into a simple score (SAMe-TT2R2) that can predict poor INR control and aid decision-making by identifying those patients with AF who would do well on VKA (SAMe-TT2R2 score = 0-1), or conversely, those who require additional interventions to achieve acceptable anticoagulation control (SAMe-TT2R2 score ≥ 2).

Figures in this Article

Vitamin K antagonists (VKAs) remain the main therapeutic agents used to prevent thromboembolism.1 Their efficiency and safety have been proved in multiple disease settings.13 But VKAs have several limitations, including a narrow therapeutic window and numerous drug and dietary interactions.46 Maintaining therapeutic range in patients treated with VKAs has always been challenging and the potential consequences of deviating from the therapeutic range are deleterious.6 Patients who are undertreated remain at risk of thromboembolic events, and patients who receive too much anticoagulant therapy are exposed to unnecessary bleeding risk.5,6

Several indices of anticoagulation quality have been proposed, with the time in therapeutic range (TTR) being the most widely used.7 Both major bleeding and mortality rates have been reported to be significantly higher in patients with TTR < 60% compared with those with TTR > 75%.8

The quality of oral anticoagulation is strongly related to the quality of the coagulation services. Patient-related factors may also affect the quality of oral anticoagulation, but limited data are currently available.8,9 Factors such as female sex, nonwhite race, paroxysmal as opposed to permanent atrial fibrillation (AF), and being new to VKAs have been associated with poor anticoagulation control.814 However, it is unclear how these data can be applied in everyday clinical practice, especially in the era of the new oral anticoagulants. Indeed, being able to single out those patients most likely to present with good anticoagulation control would be beneficial because these patients would gain the same prognostic benefit from VKAs and the new oral anticoagulants, whereas the health-care system would benefit from the lower treatment cost of VKAs.

In the current analysis, we used data from the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial with the objective of developing a validated scheme using patient-related clinical parameters for assessing the likelihood of poor international normalized ratio (INR) control among patients with AF.

Study Population

The AFFIRM trial, which investigated clinical outcomes in participants undergoing rate control vs rhythm control for AF management, has been described in detail before.15 Warfarin anticoagulation was recommended throughout the duration of the study but was left to physician discretion and could be discontinued for participants in the rhythm control arm if they remained in sinus rhythm for (preferably) 12 weeks.

The current analysis considered all participants enrolled in the AFFIRM study (rate or rhythm control arms) treated with oral anticoagulation in whom sufficient data were available to estimate the quality of the oral anticoagulation. Patients enrolled in the study had at least a moderate risk of stroke or death by virtue of age or clinical risk factors. The mean duration of follow-up was 3.5 years, with a maximum of 6 years. This analysis of the AFFIRM database was approved by the University of Iowa institutional review board; the database was obtained from the National Institutes of Health.

A cohort of patients with AF was used for the external validation of the retrieved results. This cohort was recruited prospectively during attendance at the outpatient cardiology clinic or the anticoagulation clinic in Sandwell and West Birmingham Hospitals. Informed consent was obtained from all the participants, and access to their medical records was approved by the local ethics committee in Sandwell and West Birmingham Hospitals.

Quality of Oral Anticoagulation

Not all AFFIRM participants received anticoagulation therapy throughout the study follow-up period. In the absence of consecutive INR measurements, the AFFIRM population was stratified with respect to quality of oral anticoagulation by the fraction of INR values between 2 and 3. Patients with > 12 months of uninterrupted VKA treatment and more than eight INR values available were included in the analysis; these eight INR values provided the best combination of sample size and individual INR variance.

For the external validation cohort, TTR was calculated (between 0 and 1) using the Rosendaal method,16 which uses linear interpolation to assign an INR value to each day between two successive observed INR values. Patients with more than eight consecutive INR values, < 6 weeks apart, within a 12-month follow-up period were included in the cohort. After interpolation, the percentage of time during which the interpolated INR values lay between 2 and 3 was calculated.

Statistical Analysis

The AFFIRM population was divided randomly into a split-sample derivation and an internal validation cohort using a 1:1 ratio. Means, SDs, and 95% CIs of the means were calculated for continuous variables. Frequencies and percentages were calculated for categorical variables. Continuous variables were analyzed using one-way analysis of variance or independent-sample Student t test. Categorical variables were analyzed using χ2 tests.

Covariates associated with TTR at α levels of < 0.05 in univariate analysis were incorporated into a multivariate stepwise linear regression model. Determinants with a prevalence rate below 5% were excluded from the model. The retrieved factors were further tested in binary logistic regression models using different percentiles of TTR as cutoff points.

Model calibration was tested by the Hosmer-Lemeshow (HL) goodness-of-fit test. The test assesses whether the observed event rates match expected event rates in subgroups of the model population. Explanatory power was tested using the pseudo-R2 statistic according to the “Nagelkerke R2” to assess the degree to which the model explained the variance of the binary outcome.

Variables from the final binary logistic regression model were converted to a risk score, with points assigned to each predictor approximately proportional to the magnitude of the regression coefficients (using the 25th percentile as the TTR cutoff point). The score was validated in the internal validation and external validation cohorts by c-index calculated for different TTR cutoff levels.

P values < .05 were considered statistically significant. Analysis was performed with SPSS, version 17.0 (IBM) and the R statistical package (www.r-project.org).

Clinical characteristics of the derivation population are summarized in Table 1. The derivation cohort included 1,061 patients, with 59.3% men and a mean age of 69 years (SD, 8). TTR was calculated from a median of 12 INR values per patient (range, nine to 18). Mean TTR in the cohort was 0.642 (SD, 0.18). The distribution of the cohort population in TTR cutoffs is summarized in Table 2.

Table Graphic Jump Location
Table 1 —Clinical Characteristics of Patients From the Three Cohorts

TTR = time in therapeutic range.

Table Graphic Jump Location
Table 2 —Quality of Anticoagulation in the Three Cohorts

See Table 1 legend for expansion of abbreviation.

a 

TTR calculated as the fraction of measurement within the therapeutic range.

b 

TTR calculated using the Rosendaal formula.

Derivation Analysis

Most of the comorbidities were associated with lower TTR (Fig 1). Eighteen variables were correlated with TTR beyond the level of statistical significance in univariate analysis (Table 3).

Figure Jump LinkFigure 1. Mean time in therapeutic range (TTR) and 95% CI of mean in subgroups of the derivation cohort. Dotted line represents the mean TTR in the cohort (0.642).Grahic Jump Location
Table Graphic Jump Location
Table 3 —Factors Associated With Anticoagulation Control at a Level Below 0.05

See Table 1 legend for expansion of abbreviation.

a 

P values for independent-sample Student t test.

b 

Defined as more than one or two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease.

In the linear regression model, nine variables emerged as independent predictors of TTR: female sex (unstandardized coefficient β = −0.04, P < .0001), age < 50 years (β = −0.14, P < .0001), age 50 to 60 years (β = −0.04, P = .02), ethnic minority status (β = −0.09, P < .0001), smoking (β = −0.04, P = .03), and more than two comorbidities (β = −0.04, P < .0001). Treatment with a β-blocker (β = 0.03, P = .02) or verapamil (β = 0.05, P = .02) favored INR control, whereas treatment with amiodarone (β = −0.03, P = .05) was associated with low TTR (Table 4). Thus, treatment with verapamil and β-blockers (the main rate control drugs in AFFIRM) were associated with better TTR, whereas treatment with amiodarone (the main rhythm control drug in AFFIRM) was associated with lower TTR. These observations were grouped as “rhythm control strategy” (ie, not treated with β-blockers or verapamil, treated with amiodarone) for development of our score (see later discussion).

Table Graphic Jump Location
Table 4 —Linear Logistic Regression Analysis: Factors Affecting TTR

See Table 1 legend for expansion of abbreviation.

a 

Defined as nonwhite.

b 

Defined as more than two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease.

Overall, the regression model demonstrated a significant F statistic (F = 14.3; P < .0001), indicating that using the model was better than guessing the mean. Patient-related factors, as a whole, explained 10% of the variation in TTR (R2, 0.098).

We tested the ability of the nine-factor model to predict the quality of oral anticoagulation. In binary logistic regression analysis, we used three different cutoff points of TTR corresponding to the fifth, 10th, and 25th percentiles of the cohort (Table 2). The model showed good calibration by the HL goodness-of-fit test for the detection of patients with TTR lower than the fifth, 10th, and 25th percentile (HL P values = .38, .37, and .85, respectively). Similarly, the nine-factor model demonstrated good discriminatory ability, as reflected by c-indexes of 0.73 (95% CI, 0.65-0.8), 0.69 (95% CI, 0.64-0.75), and 0.64 (95% CI, 0.6-0.68) for the fifth, 10th, and 25th percentile of TTR, respectively.

Based on the regression coefficients of the final model, a clinical prediction score was developed. We merged the drug treatment variables into a single variable (rhythm control strategy). Similarly, the two age variables were merged into a single variable (age < 60 years). Minority status and smoking were assigned two points each; age < 60 years, more than two comorbidities, female sex, and rhythm control strategy were assigned one point each. Despite our efforts to further simplify the scheme by using unweighed variables or by omitting weaker variables, the simpler scores were associated with significantly reduced performance.

The acronym SAMe-TT2R2 (sex female, Age < 60 years, medical history [more than two comorbidities], treatment [interacting drugs, eg, amiodarone for rhythm control], tobacco use [doubled], race [doubled]) was used for the score (Table 5). In the derivation cohort, the simpler SAME-TT2R2 demonstrated discriminatory performance similar to that of the nine-factor model (c-indexes, 0.71 [95% CI, 0.62-0.79]; 0.66 [95% CI, 0.61-0.72]; 0.62 [95% CI, 0.58-0.66] for the cohorts of the fifth, 10th, and 25th percentile, respectively).

Table Graphic Jump Location
Table 5 —Acronym and Definition of the SAMe-TT2R2 Score
a 

Defined as more than two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease. SAMe-TT2R2 = sex female, age < 60 years, medical history (more than two comorbidities), treatment (interacting drugs, eg, amiodarone for rhythm control), tobacco use (doubled), race (doubled).

Internal and External Validation

Clinical characteristics of the internal and external validation cohorts are summarized in Table 1. The internal validation cohort included 1,019 patients (63% men; mean age, 69 years (SD, 8 years). The external validation cohort included 286 patients (45% men; mean age, 74 years [SD, 10 years]). Table 6 presents the population of the derivation and the internal and external validation cohorts stratified by the SAMe-TT2R2 score.

Table Graphic Jump Location
Table 6 —Mean TTR in the Development and Validation Cohorts Stratified by the SAMe-TT2R2 Score

See Table 1 and 5 legends for expansion of abbreviations.

In binary logistic regression analysis, the SAMe-TT2R2 model demonstrated good calibration for goodness of fit in both the internal and external validation cohorts for all the tested TTR percentiles (fifth, 10th, and 25th). Nagelkerke R2, HL goodness-of-fit test, P values, and c-indexes for the two cohorts are summarized in Table 7. Discrimination performance, as reflected by the c-index of the SAMe-TT2R2 score for different percentiles of TTR in the internal and external validation cohorts, are illustrated in Figure 2. The score demonstrated good discrimination performance for patients in the fifth TTR percentile (c-index ≥ 0.7) and better-than-chance discrimination for the 10th and 25th TTR percentiles (Fig 2). Using a mean TTR of approximately 0.65 as a cutoff, the score could aid decision-making by identifying those patients with AF who would do well on VKA (SAMe-TT2R2 score = 0-1) or, conversely, those (ie, SAMe-TT2R2 score ≥ 2) who would be at risk of suboptimal anticoagulation control.

Table Graphic Jump Location
Table 7 —Model Summary and Calibration for Different Percentiles of TTR

See Table 1 legend for expansion of abbreviation.

Figure Jump LinkFigure 2. Receiver operating characteristic curves demonstrating discrimination performance of SAMe-TT2R2 (sex female, age < 60 y, medical history [more than two comorbidities], treatment [interacting drugs, eg, amiodarone for rhythm control], tobacco use [doubled], race [doubled]) score for different percentiles of TTR in the internal and external validation cohorts: fifth TTR percentile (dark gray line), 10th TTR percentile (light gray line), 25th TTR percentile (gray dotted line). See Figure 1 legend for expansion of abbreviation.Grahic Jump Location

In this analysis we have shown that common clinical and demographic factors can influence the quality of oral anticoagulation, making it feasible to identify patients who are less likely to keep within the target INR range. Second, we have incorporated these factors into a simple score (SAMe-TT2R2) that can predict poor INR control and could potentially aid decision-making in the management of patients with AF.

Because the most important determinant of anticoagulation control is the (local) available anticoagulation services, we did not use specific TTR cutoff points but rather percentiles of TTR values. This gives the opportunity for every anticoagulation unit to detect its own “outliers” irrespective of the center’s achieved mean TTR.

Despite the multiple risk factors that have been shown to correlate with TTR in previous studies, our analysis developed a model (with the acronym SAMe-TT2R2) based on six simple clinical variables with good discriminatory performance, especially for extreme TTR outliers (below the fifth percentile). The scheme was developed and validated in the AFFIRM cohort and further validated externally from “real world practice” in a small registry of patients receiving anticoagulant therapy.

Previous studies have explored only a few possible predictors of TTR, including inception status, race, and sex.10,12 Other studies have extensively examined the effect of individual predictors such as cancer.11 In our analysis, we evaluated the impact of multiple clinical and demographic factors. Our results in patients are in agreement with previous large “general” cohorts (patients with and without AF), including the Veterans Affairs Study to Improve Anticoagulation (VARIA) registry.17 As in our study, the VARIA investigators concluded that female sex, minority status, and multiple comorbidities negatively affected TTR. Hospitalizations and alcohol abuse further emerged as important predictors in the VARIA registry.

It is beyond the scope of the current analysis to identify the pathophysiology underlying the association between clinical factors and quality of anticoagulation. The observation that women have lower TTR than men is a common finding in every study that investigated TTR predictors, although the precise reason(s) remain unclear. Also, women are known to be at higher risk of AF-related stroke regardless of warfarin use, which may be related to poorer anticoagulation control in women, and thus, TTR may be a focus for intervention.18 In our cohort, younger patients experienced worse TTR, perhaps as a result of compliance parameters associated with the more active lifestyle of young patients. We can also assume that the number of comorbidities affects compliance.

Nonetheless, how race affects TTR is unclear, because it may reflect differences in genetic background related to warfarin metabolism or differences in socioeconomic status that can affect compliance. In a study of 16,000 Medicare beneficiaries, Birman-Deych et al13 reported that blacks and Hispanics did not benefit from warfarin anticoagulation, partly because of less adequate warfarin treatment and anticoagulation monitoring. Smoking is known to affect warfarin metabolism19,20 but our observation may simply reflect the lower compliance profile of smokers in general.

Finally, when considering various rate control vs rhythm control approaches, given how common drug interactions are with VKA use, treatment strategy (rhythm control) may affect TTR because of different drug-drug interactions.21 We found that treatment with verapamil and β-blockers (the main rate control drugs in AFFIRM) were associated with better TTR, whereas treatment with amiodarone (the main rhythm control drug in AFFIRM) was associated with lower TTR; however, our analysis cannot fully address all the numerous drug therapy permutations in patients with AF or the many potential drug interactions with VKAs.

The SAMe-TT2R2 score was developed and validated in patients anticoagulated for a homogeneous indication: stroke and systemic embolism prevention in AF. Thus, the bias and residual confounding related to the different indications for anticoagulation are avoided. In various prediction schemes, clinical definitions may affect reproducibility, but with SAMe-TT2R2, five out of six variables are clearly defined (age, sex, race, smoking status, and treatment strategy), and the clinical variables are only included numerically. Moreover, the SAMe-TT2R2 scheme is simple and can be calculated easily without the use of special equipment or software.

How can SAMe-TT2R2 be used? The focus here is on predicting those who would do well on VKA therapy using a newly validated simple score based on simple clinical variables. Thus, the SAMe-TT2R2 score can aid physicians in identifying the patients likely to present with poor anticoagulation control before starting VKA therapy. Such information may be useful in determining which patients may require more aggressive clinical interventions to help achieve an acceptable level of control (eg, regular review/follow up and educational interventions). Because all patients on warfarin treatment need regular observation and consistent advice anyway, risk assessment is a dynamic process; during regular review and follow-up, assessment of the SAMe-TT2R2 score allows the physician to make regular informed assessments of the individual patient to see if he/she would continue to do well on warfarin (rather than use guesswork).

Even in the era of novel oral anticoagulants for AF, VKAs are still very much used in patients with AF (and for many indications for which the novel oral anticoagulants are still not licensed) and in health economies where cost is an issue. The availability of an easily applied clinical tool that can identify patients unlikely (or likely) to benefit from VKAs may have several applications. In health systems with limited resources, patients with SAME-TT2R2 scores of 0 to 1 could be treated initially with VKAs. Based on our validation data, these patients have an almost 50% lower risk of presenting with TTR less than the 25th percentile of the center’s average and a 75% lower risk of presenting with TTR less than the 10th percentile of the center’s average compared with patients with SAMe-TT2R2 score > 1. This new score could identify the patients who could potentially have such poor predicted control that they may not be suitable candidates for VKAs and would be better off on novel oral anticoagulants.

Limitations

The current study has the advantage of incorporating derivation and internal and external validation data, but it also has several limitations. Most importantly, the derivation and validation data originate in a clinical trial and may not reflect “real-world” practice. Moreover, AFFIRM was not designed to assess the impact of anticoagulation on outcome. The AFFIRM dataset did not include all the consecutive INR measurements. Thus, the most valid formula to assess TTR was to calculate the fractions of the available measurements within therapeutic range. Nevertheless, we tried to accurately estimate the quality of anticoagulation by including only that fraction of patients with at least 12 months of uninterrupted anticoagulation and more than eight INR values available. Finally, the data set did not include information on other important variables that may reflect TTR, such as distance from a coumadin clinic, level of education, employment history, the multitude of drugs that interact with VKAs, and CYP2C9 or VKORC1 genotype. Moreover, some established factors that could affect coagulation control (alcohol abuse or severe liver or renal impairment) were missing from the AFFIRM study cohort. Nevertheless, we believe that these factors are less common and by themselves constitute relative contraindications to VKA treatment. The factors included in the SAMe-TT2R2 score do not individually affect decision-making for oral anticoagulation Indeed, although the parameters in the SAMe-TT2R2 score have been identified as associated with poor anticoagulation control, we have, for the first time, derived and validated a simple score that can be used to identify those patients who would do well on VKA

Finally, external validation was performed in a single-center small cohort of patients with AF; however, our data included a detailed assessment of the quality of anticoagulation with 12 months of uninterrupted anticoagulation and more than eight INR values per patient per year recorded.

In conclusion, we have identified common clinical and demographic factors that can influence the quality of oral anticoagulation. We incorporated those factors into a simple score (SAMe-TT2R2) that can predict poor INR control and could potentially aid decision-making in the management of patients with AF, which was internally and externally validated with good discrimination performance, especially for patients at the lowest percentiles of the center’s average. The SAMe-TT2R2 score can aid decision-making by identifying those patients with AF who would do well on VKA (SAMe-TT2R2 score = 0-1) or, conversely, those (ie, SAMe-TT2R2 score ≥ 2) who may require some intervention(s) to help them achieve acceptable anticoagulation control.

Author contributions: Drs Apostolakis and Lip are guarantors of the manuscript.

Dr Apostolakis: contributed to the study hypothesis, concept of the SAMe-TT2R2 score (“the Birmingham Atrial Fibrillation anticoagulation TTR prediction score”), data analysis and interpretation, and drafting of the paper.

Dr Sullivan: contributed to the data interpretation and drafting of the paper.

Dr Olshansky: contributed to the data interpretation and drafting of the paper.

Dr Lip: contributed to the study hypothesis, concept of the SAMe-TT2R2 score (“the Birmingham Atrial Fibrillation anticoagulation TTR prediction score”), data analysis and interpretation, and drafting of paper.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Lip has served as a consultant for Bayer; Astellas Pharma; Merck & Co; Sanofi; Bristol-Myers Squibb/Pfizer Inc; Daiichi Sankyo, Inc; BIOTRONIK; Portola Pharmaceuticals, Inc; and Boehringer-Ingelheim GmbH and has been on the speakers’ bureau of Bayer, Bristol-Myers Squibb/Pfizer Inc, Boehringer-Ingelheim GmbH, and Sanofi-Aventis. Drs Apostolakis, Sullivan, and Olshansky have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

AF

atrial fibrillation

AFFIRM

Atrial Fibrillation Follow-up Investigation of Rhythm Management

HL

Hosmer-Lemeshow

INR

international normalized ratio

SAMe-TT2R2

sex female, age < 60 years, medical history (more than two comorbidities), treatment (interacting drugs, eg, amiodarone for rhythm control), tobacco use (doubled), race (doubled)

TTR

time in therapeutic range

VARIA

Veterans Affairs Study to Improve Anticoagulation

VKA

vitamin K antagonist

Lip GY, Tse HF, Lane DA. Atrial fibrillation. Lancet. 2012;379(9816):648-661. [CrossRef] [PubMed]
 
Singer DE, Chang Y, Fang MC, et al. The net clinical benefit of warfarin anticoagulation in atrial fibrillation. Ann Intern Med. 2009;151(5):297-305. [CrossRef] [PubMed]
 
Mant J, Hobbs FD, Fletcher K, et al; BAFTA investigators; Midland Research Practices Network (MidReC). Warfarin versus aspirin for stroke prevention in an elderly community population with atrial fibrillation (the Birmingham Atrial Fibrillation Treatment of the Aged Study, BAFTA): a randomised controlled trial. Lancet. 2007;370(9586):493-503. [CrossRef] [PubMed]
 
Olesen JB, Lip GY, Lindhardsen J, 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(4):739-749. [CrossRef] [PubMed]
 
Flaherty ML, Kissela B, Woo D, et al. The increasing incidence of anticoagulant-associated intracerebral hemorrhage. Neurology. 2007;68(2):116-121. [CrossRef] [PubMed]
 
Lip GY, Andreotti F, Fauchier L, et al; Document reviewers. 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(5):723-746. [CrossRef] [PubMed]
 
Schmitt L, Speckman J, Ansell J. Quality assessment of anticoagulation dose management: comparative evaluation of measures of time-in-therapeutic range. J Thromb Thrombolysis. 2003;15(3):213-216. [CrossRef] [PubMed]
 
White HD, Gruber M, Feyzi J, 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(3):239-245. [CrossRef] [PubMed]
 
Nieuwlaat R, Connolly BJ, Hubers LM, et al. Quality of individual INR control and the risk of stroke and bleeding events in atrial fibrillation patients: a nested case control analysis of the ACTIVE W study. Thromb Res. 2012;129(6):715-719. [CrossRef] [PubMed]
 
Rose AJ, Ozonoff A, Henault LE, Hylek EM. Warfarin for atrial fibrillation in community-based practise. J Thromb Haemost. 2008;6(10):1647-1654. [CrossRef] [PubMed]
 
Rose AJ, Sharman JP, Ozonoff A, Henault LE, Hylek EM. Effectiveness of warfarin among patients with cancer. J Gen Intern Med. 2007;22(7):997-1002. [CrossRef] [PubMed]
 
Bhandari VK, Wang F, Bindman AB, Schillinger D. Quality of anticoagulation control: do race and language matter? J Health Care Poor Underserved. 2008;19(1):41-55. [CrossRef] [PubMed]
 
Birman-Deych E, Radford MJ, Nilasena DS, Gage BF. Use and effectiveness of warfarin in Medicare beneficiaries with atrial fibrillation. Stroke. 2006;37(4):1070-1074. [CrossRef] [PubMed]
 
Shen AY, Yao JF, Brar SS, Jorgensen MB, Wang X, Chen W. Racial/ethnic differences in ischemic stroke rates and the efficacy of warfarin among patients with atrial fibrillation. Stroke. 2008;39(10):2736-2743. [CrossRef] [PubMed]
 
Wyse DG, Waldo AL, DiMarco JP, et al; Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) Investigators. A comparison of rate control and rhythm control in patients with atrial fibrillation. N Engl J Med. 2002;347(23):1825-1833. [CrossRef] [PubMed]
 
Rosendaal FR, Cannegieter SC, van der Meer FJ, Briët E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69(3):236-239. [PubMed]
 
Rose AJ, Hylek EM, Ozonoff A, Ash AS, Reisman JI, Berlowitz DR. Patient characteristics associated with oral anticoagulation control: results of the Veterans AffaiRs Study to Improve Anticoagulation (VARIA). J Thromb Haemost. 2010;8(10):2182-2191. [CrossRef] [PubMed]
 
Avgil Tsadok M, Jackevicius CA, Rahme E, Humphries KH, Behlouli H, Pilote L. Sex differences in stroke risk among older patients with recently diagnosed atrial fibrillation. JAMA. 2012;307(18):1952-1958. [CrossRef] [PubMed]
 
Mungall DR, Ludden TM, Marshall J, Hawkins W, Talbert RL, Crawford MH. Population pharmacokinetics of racemic warfarin in adult patients. J Pharmacokinet Biopharm. 1985;13(3):213-227. [CrossRef] [PubMed]
 
Lenzini PA, Grice GR, Milligan PE, et al. Laboratory and clinical outcomes of pharmacogenetic vs. clinical protocols for warfarin initiation in orthopedic patients. J Thromb Haemost. 2008;6(10):1655-1662. [CrossRef] [PubMed]
 
Ageno W, Gallus AS, Wittkowsky A, Crowther M, Hylek EM, Palareti G; American College of Chest Physicians. Oral anticoagulant therapy: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2012;141(2_suppl):e44S-e88S. [CrossRef] [PubMed]
 

Figures

Figure Jump LinkFigure 1. Mean time in therapeutic range (TTR) and 95% CI of mean in subgroups of the derivation cohort. Dotted line represents the mean TTR in the cohort (0.642).Grahic Jump Location
Figure Jump LinkFigure 2. Receiver operating characteristic curves demonstrating discrimination performance of SAMe-TT2R2 (sex female, age < 60 y, medical history [more than two comorbidities], treatment [interacting drugs, eg, amiodarone for rhythm control], tobacco use [doubled], race [doubled]) score for different percentiles of TTR in the internal and external validation cohorts: fifth TTR percentile (dark gray line), 10th TTR percentile (light gray line), 25th TTR percentile (gray dotted line). See Figure 1 legend for expansion of abbreviation.Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1 —Clinical Characteristics of Patients From the Three Cohorts

TTR = time in therapeutic range.

Table Graphic Jump Location
Table 2 —Quality of Anticoagulation in the Three Cohorts

See Table 1 legend for expansion of abbreviation.

a 

TTR calculated as the fraction of measurement within the therapeutic range.

b 

TTR calculated using the Rosendaal formula.

Table Graphic Jump Location
Table 3 —Factors Associated With Anticoagulation Control at a Level Below 0.05

See Table 1 legend for expansion of abbreviation.

a 

P values for independent-sample Student t test.

b 

Defined as more than one or two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease.

Table Graphic Jump Location
Table 4 —Linear Logistic Regression Analysis: Factors Affecting TTR

See Table 1 legend for expansion of abbreviation.

a 

Defined as nonwhite.

b 

Defined as more than two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease.

Table Graphic Jump Location
Table 5 —Acronym and Definition of the SAMe-TT2R2 Score
a 

Defined as more than two of the following: hypertension, diabetes, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease. SAMe-TT2R2 = sex female, age < 60 years, medical history (more than two comorbidities), treatment (interacting drugs, eg, amiodarone for rhythm control), tobacco use (doubled), race (doubled).

Table Graphic Jump Location
Table 6 —Mean TTR in the Development and Validation Cohorts Stratified by the SAMe-TT2R2 Score

See Table 1 and 5 legends for expansion of abbreviations.

Table Graphic Jump Location
Table 7 —Model Summary and Calibration for Different Percentiles of TTR

See Table 1 legend for expansion of abbreviation.

References

Lip GY, Tse HF, Lane DA. Atrial fibrillation. Lancet. 2012;379(9816):648-661. [CrossRef] [PubMed]
 
Singer DE, Chang Y, Fang MC, et al. The net clinical benefit of warfarin anticoagulation in atrial fibrillation. Ann Intern Med. 2009;151(5):297-305. [CrossRef] [PubMed]
 
Mant J, Hobbs FD, Fletcher K, et al; BAFTA investigators; Midland Research Practices Network (MidReC). Warfarin versus aspirin for stroke prevention in an elderly community population with atrial fibrillation (the Birmingham Atrial Fibrillation Treatment of the Aged Study, BAFTA): a randomised controlled trial. Lancet. 2007;370(9586):493-503. [CrossRef] [PubMed]
 
Olesen JB, Lip GY, Lindhardsen J, 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(4):739-749. [CrossRef] [PubMed]
 
Flaherty ML, Kissela B, Woo D, et al. The increasing incidence of anticoagulant-associated intracerebral hemorrhage. Neurology. 2007;68(2):116-121. [CrossRef] [PubMed]
 
Lip GY, Andreotti F, Fauchier L, et al; Document reviewers. 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(5):723-746. [CrossRef] [PubMed]
 
Schmitt L, Speckman J, Ansell J. Quality assessment of anticoagulation dose management: comparative evaluation of measures of time-in-therapeutic range. J Thromb Thrombolysis. 2003;15(3):213-216. [CrossRef] [PubMed]
 
White HD, Gruber M, Feyzi J, 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(3):239-245. [CrossRef] [PubMed]
 
Nieuwlaat R, Connolly BJ, Hubers LM, et al. Quality of individual INR control and the risk of stroke and bleeding events in atrial fibrillation patients: a nested case control analysis of the ACTIVE W study. Thromb Res. 2012;129(6):715-719. [CrossRef] [PubMed]
 
Rose AJ, Ozonoff A, Henault LE, Hylek EM. Warfarin for atrial fibrillation in community-based practise. J Thromb Haemost. 2008;6(10):1647-1654. [CrossRef] [PubMed]
 
Rose AJ, Sharman JP, Ozonoff A, Henault LE, Hylek EM. Effectiveness of warfarin among patients with cancer. J Gen Intern Med. 2007;22(7):997-1002. [CrossRef] [PubMed]
 
Bhandari VK, Wang F, Bindman AB, Schillinger D. Quality of anticoagulation control: do race and language matter? J Health Care Poor Underserved. 2008;19(1):41-55. [CrossRef] [PubMed]
 
Birman-Deych E, Radford MJ, Nilasena DS, Gage BF. Use and effectiveness of warfarin in Medicare beneficiaries with atrial fibrillation. Stroke. 2006;37(4):1070-1074. [CrossRef] [PubMed]
 
Shen AY, Yao JF, Brar SS, Jorgensen MB, Wang X, Chen W. Racial/ethnic differences in ischemic stroke rates and the efficacy of warfarin among patients with atrial fibrillation. Stroke. 2008;39(10):2736-2743. [CrossRef] [PubMed]
 
Wyse DG, Waldo AL, DiMarco JP, et al; Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) Investigators. A comparison of rate control and rhythm control in patients with atrial fibrillation. N Engl J Med. 2002;347(23):1825-1833. [CrossRef] [PubMed]
 
Rosendaal FR, Cannegieter SC, van der Meer FJ, Briët E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69(3):236-239. [PubMed]
 
Rose AJ, Hylek EM, Ozonoff A, Ash AS, Reisman JI, Berlowitz DR. Patient characteristics associated with oral anticoagulation control: results of the Veterans AffaiRs Study to Improve Anticoagulation (VARIA). J Thromb Haemost. 2010;8(10):2182-2191. [CrossRef] [PubMed]
 
Avgil Tsadok M, Jackevicius CA, Rahme E, Humphries KH, Behlouli H, Pilote L. Sex differences in stroke risk among older patients with recently diagnosed atrial fibrillation. JAMA. 2012;307(18):1952-1958. [CrossRef] [PubMed]
 
Mungall DR, Ludden TM, Marshall J, Hawkins W, Talbert RL, Crawford MH. Population pharmacokinetics of racemic warfarin in adult patients. J Pharmacokinet Biopharm. 1985;13(3):213-227. [CrossRef] [PubMed]
 
Lenzini PA, Grice GR, Milligan PE, et al. Laboratory and clinical outcomes of pharmacogenetic vs. clinical protocols for warfarin initiation in orthopedic patients. J Thromb Haemost. 2008;6(10):1655-1662. [CrossRef] [PubMed]
 
Ageno W, Gallus AS, Wittkowsky A, Crowther M, Hylek EM, Palareti G; American College of Chest Physicians. Oral anticoagulant therapy: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2012;141(2_suppl):e44S-e88S. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
Antithrombotic Therapy in Atrial Fibrillation*: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition)
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543