Author + information
- Received June 29, 2016
- Revision received August 25, 2016
- Accepted September 13, 2016
- Published online December 13, 2016.
- Antonio Gómez-Outes, MD, PhDa,∗ (, )
- Julián Lagunar-Ruíz, MDb,
- Ana-Isabel Terleira-Fernández, MD, PhDb,c,
- Gonzalo Calvo-Rojas, MD, PhDd,
- Maria Luisa Suárez-Gea, PharmD, PhDa and
- Emilio Vargas-Castrillón, MD, PhDb,c
- aDivision of Pharmacology and Clinical Drug Evaluation, Spanish Agency for Medicines and Medical Devices (AEMPS), Madrid, Spain
- bDepartment of Clinical Pharmacology, Hospital Clínico San Carlos, Madrid, Spain
- cDepartment of Pharmacology, Universidad Complutense, Madrid, Spain
- dDepartment of Clinical Pharmacology, Hospital Clinic, Barcelona, Spain
- ↵∗Reprint requests and correspondence:
Dr. Antonio Gómez-Outes, Division of Pharmacology and Clinical Drug Evaluation, Agencia Española de Medicamentos y Productos Sanitarios (AEMPS), Campezo 1, 28022 Madrid, Spain.
Background Oral anticoagulation reduces the risk of mortality in atrial fibrillation (AF), but examination of the causes of death is essential to design new strategies to further reduce the high mortality rates observed in this population.
Objectives The authors sought to analyze and compare causes of death in patients receiving direct oral anticoagulants (DOAC) or warfarin for prevention of stroke and systemic embolism (SE) in AF.
Methods The authors systematically searched for randomized trials of DOAC versus warfarin for prevention of stroke/SE in AF. The main outcome was mortality and independently adjudicated specific causes of death. The authors used the random effects model of meta-analysis to combine the studies.
Results 71,683 patients from 4 trials were included (134,046 patient-years of follow-up). A total of 6,206 patients (9%) died during follow-up. Adjusted mortality rate was 4.72%/year (95% confidence interval [CI]: 4.19 to 5.28). Cardiac deaths accounted for 46% of all deaths, whereas nonhemorrhagic stroke/SE and hemorrhage-related deaths represented 5.7% and 5.6% of the total mortality, respectively. Compared with patients who were alive, those who died had more frequent history of heart failure (odds ratio [OR]: 1.75; 95% CI: 1.25 to 2.44), permanent/persistent AF (OR: 1.38; 95% CI: 1.25 to 1.52) and diabetes (OR: 1.37; 95% CI: 1.11 to 1.68); were more frequently male (OR: 1.24; 95% CI: 1.13 to 1.37) and older (mean difference 3.2 years; 95% CI: 1.6 to 4.8); and had a lower creatinine clearance (−9.9 ml/min; 95% CI: −11.3 to −8.4). There was a small, but significant, reduction in all-cause mortality with the DOAC versus warfarin (difference −0.42%/year; 95% CI: −0.66 to −0.18), mainly driven by a reduction in fatal bleedings.
Conclusions In contemporary AF trials, most deaths were cardiac-related, whereas stroke and bleeding represented only a small subset of deaths. Interventions beyond anticoagulation are needed to further reduce mortality in AF.
Atrial fibrillation (AF) is the most common arrhythmia in the developed world, being associated with a 5-fold increased risk of stroke and higher mortality (1). Oral anticoagulation with warfarin and other vitamin K antagonists (VKA) are highly effective treatments in reducing the risk of stroke and mortality in AF (2), but their management remains problematic due to their narrow therapeutic index and variability in drug exposure necessitating routine coagulation monitoring (international normalized ratio), clinical surveillance, and continuous patient education (3).
In recent years, several direct oral anticoagulants (DOACs) (dabigatran, rivaroxaban, apixaban, and edoxaban) (3), also referred in the published reports as “novel anticoagulants” or “non–vitamin K antagonist oral anticoagulants” and “target-specific oral anticoagulants,” have been developed to overcome some of these limitations. Contemporary studies of anticoagulation in AF have shown a significant mortality rate (≈8%) (4). Examination of the causes of death during these studies is essential to understand the specific impact of the new antithrombotic agents compared with warfarin in the risk of death, and to design new strategies, beyond anticoagulation, to further reduce the high mortality rates observed in this indication.
We systematically reviewed and meta-analyzed data from randomized controlled trials of the DOACs for prophylaxis against stroke and systemic embolism (SE) in patients with AF with the aim of describing the specific causes of death reported during these studies and exploring potential differences in cause-specific mortality between the DOACs and warfarin.
We considered randomized controlled trials comparing any of the approved new oral anticoagulants (dabigatran, rivaroxaban, apixaban, and edoxaban) with warfarin in patients with AF at risk of stroke and ≥1-year follow-up. We included all doses tested in the experimental arms of pivotal trials of the DOACs in this indication in the base-case (primary) analysis (3).
Trial identification and data collection
We searched Medline and CENTRAL (up to May 2016), websites of regulatory agencies, clinical trial registries, and relevant conference proceedings (Online Appendix 1). No language restrictions were applied. Two investigators (A.G.-O. and J.L.-R.) separately assessed trials for eligibility and extracted data. If a trial was covered in >1 report, we used the following hierarchy of data sources: peer-reviewed articles, public reports from regulatory authorities, reports from the web-based repository for results of clinical studies, and other sources. Finally, we contacted the main investigators by e-mail to retrieve unpublished data from clinical trials (differential demographic characteristics in patients who died and those alive after follow-up, as well as disaggregated classification of causes of death when they were reported as a composite, e.g., death due to stroke, death due to hemorrhage). In case of no response, we sent a reminder to the main investigator after 1 week, with copy to a sponsor's representative (e.g., coauthor/s being employee/s of the sponsor).
Study characteristics and quality assessment
We collected data on patient characteristics, numbers of patients evaluable for efficacy and safety, dosage used in the experimental and control groups, duration of treatment and follow-up, as well as inclusion and exclusion criteria and methodology for the adjudication of deaths. We assessed study quality using the Cochrane Collaboration’s tool for assessing risk of bias in randomized studies (5).
The pre-specified outcome was all-cause death and adjudicated specific causes of death. They had to correspond to the principal cause of death as assessed by the adjudication committees of the studies. Therefore, each death was attributed to one cause exclusively. Fatal hemorrhagic stroke was counted among fatal bleedings, but not among fatal strokes, in order to avoid double counting.
Quantitative data synthesis
We conducted this meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations (6). We used the intention-to-treat population whenever available. Firstly, we conducted an incidence rate meta-analysis to describe the average rate of deaths (total and specific causes) in each treatment group and overall population, for each study and pooled studies. We calculated exposure-adjusted incidence rate as the number of patients who died (numerator) by the person-time at risk, expressed as patient-years (denominator), in order to standardize different follow-up durations across studies. We calculated patient-years by multiplying the number of patients in each treatment group by the mean years that patients in each group were in the study. Secondly, we conducted a proportion meta-analysis of the relative risk (RR) of death with the DOACs (all doses combined) and warfarin. Thirdly, we conducted an incidence-rate difference meta-analysis of the absolute risk of death with the DOACs (all doses combined) and warfarin, expressed as difference in events per 10,000 patients/year. We also compared baseline characteristics of patients who died versus those alive during follow-up using the odds ratio (OR) as measure of association. Heterogeneity between studies was assessed using the Cochran Q test and the Higgins I2 index (percentage of variation across studies that is due to heterogeneity rather than chance). A Cochran’s Q p < 0.05 and I2 >50% indicate significant heterogeneity. We used the random effects model described by Der-Simonian and Laird (7) for the main analysis.
We also conducted ancillary analyses for high-dose (HD) and low-dose (LD) DOAC regimens, in order to explore potential differences in rates and specific causes of death depending on the DOAC dose tested. The HD DOAC regimen was defined as the use of dabigatran 150 mg twice daily, edoxaban 60 mg once daily, or standard doses of rivaroxaban 20 mg once daily and apixaban 5 mg twice daily. The LD DOAC regimen was defined as the use of dabigatran 110 mg twice daily and edoxaban 30 mg once daily.
We conducted sensitivity analyses taking into account different methodological issues that could influence the results of the meta-analysis: 1) statistical model (fixed effects instead of the random effects model); 2) type of measure (OR instead of RR); 3) type of analysis (incidence rate meta-analysis instead of proportion meta-analysis); and 4) study quality (including only studies at low risk of bias).
Analyses of incidence rates were done using StatsDirect software, version 2.8.0 (StatsDirect, Cheshire, United Kingdom), whereas analyses of proportions were done using RevMan statistical software, version 5.1 (Nordic Cochrane Center, Copenhagen, Denmark).
Study selection, design, and methodology
The literature search identified 4,712 articles. Of these, 47 reports were related to clinical trials including mortality data in AF, and were selected for checking as full text. An additional hand search identified 8 reports from regulatory authorities related to assessment of the DOACs in AF (Figure 1). Of these, 37 reports were excluded due to several causes detailed in Figure 1. Finally, 18 reports (8–25) including mortality data from 4 unique randomized clinical trials with dabigatran (RE-LY [Randomized Evaluation of Long-Term Anticoagulation Therapy]) (8–13), rivaroxaban (ROCKET AF [An Efficacy and Safety Study of Rivaroxaban With Warfarin for the Prevention of Stroke and Non-Central Nervous System Systemic Embolism in Patients With Non-Valvular Atrial Fibrillation]) (14–18), apixaban (ARISTOTLE [Apixaban for the Prevention of Stroke in Subjects With Atrial Fibrillation]) (19–21), and edoxaban (ENGAGE AF–TIMI 48 [Effective Anticoagulation With Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis In Myocardial Infarction 48]) (22–25), in patients with AF were included in the quantitative meta-analysis.
The 4 studies comprised 71,683 AF patients (134,046 patient-years of exposure) (Table 1), who were randomized to receive a DOAC or warfarin. The risk of bias was low in 3 studies (14,19,22) and unclear in the RE-LY trial due to lack of double-blinding (8) (Table 1, Online Appendix 2, Online Table 1). An independent adjudicating committee blindly adjudicated deaths in all 4 studies. Adjudication of deaths was made in the intention-to-treat population in 3 studies (9,15,20) and in randomized patients in the remaining study (23). The time-point for adjudication of deaths was up to end of study in 3 studies (8,14,22) and at the end of the intended treatment period (expected date of attainment of the study’s event target) in the remaining study (19). The rate of patients lost to follow-up for vital status at end of study ranged from 0.1% to 2% across studies, without significant differences between treatment groups (Table 1). Each death was attributed to 1 cause exclusively in all studies (principal cause of death). For more detailed information on the adjudication process of causes of deaths by trial and assumptions made in the meta-analysis for pooling cause-specific deaths, see Online Appendix 3 and Online Table 2.
Mean or median age ranged between 70 and 73 years (Table 1). There was a predominance of men (range 60% to 65%) and permanent/persistent AF (range 67% to 83%). Mean thromboembolic risk (CHADS2) was the highest in the ROCKET AF and ENGAGE AF trials (3.5 and 2.8, respectively) and the lowest in the RE-LY and ARISTOTLE trials (2.1 each). Consistently, the rate of patients aged >75 years and history of heart failure, hypertension, diabetes, and prior stroke was higher in the ROCKET-AF and ENGAGE AF trials than in the RE-LY and ARISTOTLE trials. Demographic characteristics by vital status were available from 2 studies (Online Table 3). Patients who died during the studies were significantly older (difference 3.2 years; 95% confidence interval [CI]: 1.6 to 4.8), had a lower creatinine clearance (difference −9.9 ml/min; 95% CI: −11.3 to −8.4), were more frequently male (OR: 1.24; 95% CI: 1.13 to 1.37), and had more frequent history of congestive heart failure (OR: 1.75; 95% CI: 1.25 to 2.44), diabetes (OR: 1.37; 95% CI: 1.11 to 1.68), permanent/persistent AF (OR: 1.38; 95% CI: 1.25 to 1.52), and less history of VKA use (OR: 0.88; 95% CI: 0.78 to 0.98) than patients alive at the end of follow-up.
Descriptive analysis of causes of death
A total of 6,206 deaths were adjudicated in the 4 studies. Vascular deaths accounted for 64% of all deaths, whereas 30% of deaths were of nonvascular cause, and 6% of deaths were due to unknown causes (Table 2). The more frequent causes of death were cardiac death, in 46% of cases (28% sudden cardiac deaths/dysrhythmia, 15% heart failure, and 3% myocardial infarction), followed by malignancies (11%), infections (9%), ischemic stroke/SE (6%), and bleeding (6%).
The average annual mortality rate in the study populations was 4.63%/year (95% CI: 3.99 to 5.32) (Table 3), but there was significant heterogeneity across studies (p < 0.0001; I2 = 95.5%). Heterogeneity was due to a higher death rate in the ROCKET AF and ENGAGE AF trials than in the other studies, which is consistent with a higher rate of elderly patients and significant comorbidities in the ROCKET AF and ENGAGE AF trials than in the other 2 studies (Table 1). Average annual incidences for the more frequent specific categories were cardiac death 2.07%/year (95% CI: 1.55% to 2.68%/year), nonvascular death 1.37%/year (1.18% to 1.57%/year), heart failure 0.67%/year (0.53% to 0.82%/year), malignancies 0.52%/year (0.42% to 0.63%/year), infections 0.37%/year (0.23% to 0.55%/year), bleeding 0.27%/year (0.19% to 0.35%/year), and ischemic stroke/SE 0.26%/year (0.20% to 0.32%/year) (Table 3).
The average annual mortality rate by anticoagulant type was 4.46%/year (95% CI: 3.85% to 5.12%/year) with the DOACs and 4.87%/year (4.15% to 5.64%/year) with warfarin (Online Tables 4 and 5, respectively). The main difference in annual rates of specific causes of death between the DOACs and warfarin was in the fatal bleeding subcategory (DOACs 0.19%/year [0.13% to 0.26%/year] and warfarin 0.38%/year [0.28% to 0.49%/year]) (Online Tables 4 and 5).
Differences in specific causes of death between DOACs and warfarin
There were significant relative risk differences in favor of the DOACs in comparison with warfarin with respect to all-cause mortality (RR: 0.90; 95% CI: 0.86 to 0.95) (Figure 2A), as well as for cardiovascular (CV) deaths (RR: 0.89; 95% CI: 0.84 to 0.95) and unknown/undetermined deaths (RR: 0.80; 95% CI: 0.66 to 0.98) (Figure 2B), with homogeneity between trials. The differences in CV mortality were mainly driven by a reduction in fatal bleedings (hemorrhagic stroke, other fatal intracranial and extracranial bleedings) (Figure 3). There was a significant difference in the effect on bleeding and nonbleeding CV mortality (heterogeneity p < 0.00001; I2 = 97%) (Figure 4). Finally, no differences in specific causes of non-CV mortality were apparent between treatments, although the DOACs were also associated to a nonsignificant trend toward a lower risk of some specific types of deaths than warfarin, such as those due to trauma/accident (RR: 0.62; 95% CI: 0.39 to 1.01) (Figure 5).
Absolute risk differences between DOACs and warfarin for specific causes of deaths per 10,000 patients treated per year were homogeneous across trials and were generally in favor of the DOACs (Table 4). For all-cause death, the DOACs avoided 42 additional deaths per 10,000 patients treated per year (95% CI: −66 to −18) in the pool of the 4 studies analyzed, with approximately 238 patients needed to be treated with the DOAC instead of warfarin to prevent 1 additional death per year (95% CI: 152 to 548).
Ancillary analyses by DOAC dose
Compared with warfarin, all-cause mortality was significantly reduced with both the HD DOAC regimens (RR: 0.91; 95% CI: 0.86 to 0.96) and LD DOAC regimens (RR: 0.89; 95% CI: 0.83 to 0.96) (Online Figure 1). The same applied for vascular death (Online Figure 2A) and fatal bleedings, including hemorrhagic stroke (Online Figures 3 and 4). However, the HD DOAC regimes tended to decrease the risk of fatal ischemic stroke/SE compared with warfarin (RR: 0.83; 95% CI: 0.65 to 1.05), whereas the LD DOAC regimens tended to increase the risk of fatal ischemic stroke/SE (RR: 1.12; 95% CI: 0.82 to 1.54) (p value for subgroup differences = 0.13; I2 = 57.5%) (Online Figure 5). No differences were found between HD and LD DOAC regimens versus warfarin for overall nonvascular deaths (Online Figure 2B) and subtypes (Online Figures 6 and 7).
The sensitivity analyses (Online Table 6) were consistent with the base-case analysis. There were significant differences between the DOACs and warfarin in all-cause mortality and vascular mortality in all cases, which were mainly driven by a halved risk of fatal bleedings. The results were robust regardless of statistical model, effect measure, use of incidence rates instead of proportions, and exclusion of studies at uncertain risk of bias (the RE-LY trial) (Online Table 6).
Selective outcome reporting, dissemination bias, and missing data
Adjudication of causes of death were pre-specified in the protocols and reported in the publications or regulatory reviews of the large multicenter studies included in this meta-analysis. No statistical methods were applied to assess publication bias due to the low number of included studies.
The use of oral anticoagulation reduces all-cause mortality in AF patients (2), but a better understanding of causes of death among anticoagulated AF patients is necessary to identify additional interventions to improve care for these patients (15). This systematic review and meta-analysis indicates that the main cause of death in anticoagulated patients with AF, about 46%, is heart-related mortality, including sudden cardiac death, heart failure, and myocardial infarction, whereas ischemic stroke and fatal bleedings only account for approximately 6% of all deaths each (Central Illustration).
We found heart failure, decreased creatinine clearance, diabetes, advanced age, and male sex to be significantly associated with a higher risk of death. These findings are consistent with published multivariate analyses of risk factors of death in the RE-LY (9) and ROCKET AF (15) trials, which showed that history of heart failure, previous myocardial infarction, low creatinine clearance, and diabetes mellitus were among the strongest independent predictors of cardiac death in this population, whereas the presence of nonfatal major bleeding during follow-up and the history of stroke, heart failure, and older age were strong predictors of stroke-related death. In addition, statin use and optimal treatment of heart failure and coronary artery disease were independently associated with a lower mortality (9). Therefore, from a global health care perspective, further measures to reduce mortality in this population, beyond an appropriate anticoagulation not incurring in excessive increase in major bleeding, have to include an improvement in the management of relevant comorbidities, mainly heart failure, coronary artery disease, and diabetes, together with proven global CV risk-reduction measures and healthy lifestyle changes (26,27).
We also found that, in comparison with warfarin, the DOAC reduced all-cause mortality and CV mortality, which was mainly driven by a halved risk of fatal bleeding (Central Illustration). Of all fatal bleedings, the main component accounting for the differences was fatal intracerebral bleeding (hemorrhagic stroke). This trend for causes of death (halved risk of fatal bleeding and similar risk of fatal ischemic stroke/SE) is consistent with previous meta-analyses (4,28), in which the DOACs were associated to a lower bleeding tendency, including a lower risk of intracranial bleeding, but they were not more effective than warfarin in reducing the risk of ischemic stroke/SE. The DOACs also appeared to reduce deaths due to unknown/undetermined causes to a significant extent, but the results are difficult to interpret due to lack of data. It might suggest that these unknown causes could be directly or indirectly related to bleeding, either due to a higher rate of sudden fatal bleedings with warfarin or due to a higher withdrawal rate due to bleeding in the warfarin group and subsequent fatal thromboembolism. However, information bias cannot be ruled out (e.g., sufficient follow-up information for adjudication of deaths potentially obtained more frequently when patients had been treated with the investigational drug than with warfarin). Finally, the DOACs were associated with a trend toward a lower risk of death due to trauma or accident than warfarin. Because uncontrolled post-traumatic bleeding is the leading cause of potentially preventable death among trauma patients (29), our findings might reflect a lower risk of trauma-associated severe bleeding in patients anticoagulated with the DOAC than in those anticoagulated with VKA, which may be particularly relevant in patients at risk of falling (30).
We also explored potential differences in rates and causes of death depending on the DOAC dose tested. Similar to the HD regimens, the LD regimes significantly reduced all-cause death and CV death in comparison with warfarin, mainly at the expense of a reduced risk of fatal bleedings, including hemorrhagic stroke. However, compared with warfarin, the HD DOAC regimens tended to decrease the risk of fatal ischemic stroke, whereas LD DOAC regimens tended to increase the risk of fatal ischemic stroke/SE. Balancing the effects on death due to ischemic and hemorrhagic stroke, it seems reasonable to give preference to the use of the HD DOAC regimens unless there are clinical factors for increased plasma concentrations of the novel anticoagulant that would warrant dose reduction, in accordance with approved label information (package inserts).
Our review has several strengths that include the large sample size and analysis of studies in which deaths were adjudicated using specific definitions by a blinded committee of expert physicians. To our knowledge, this is the first systematic review that explores the specific causes of death in the pivotal trials conducted with the DOACs for the prevention of stroke/SE in patients with AF. A previous relevant meta-analysis reviewed a pool of trials that compared the DOAC versus warfarin in different indications (31). The authors found a reduced risk of fatal bleeding, CV mortality, and all-cause mortality with the DOACs compared with warfarin, but did not analyze specific causes of CV mortality (beyond aggregated fatal bleedings) or non-CV mortality, and did not provide a calculation of absolute risks. We have shown that the effect in the reduction of different types of mortality (total, CV) reflects a reduction in fatal (mainly intracerebral) bleedings. We have to bear in mind that this benefit may be highly dependent on the quality of anticoagulation that could be achieved with warfarin, as measured by the time within therapeutic range (TTR) (32,33). The quality of anticoagulation with VKA during pivotal studies with the DOAC greatly differed across trials and regions, with the highest TTR reported in Western Europe, and the lowest TTR in centers from Asia and Latin America (34,35). Therefore, the expected absolute benefit in fatal bleedings and all-cause mortality versus warfarin may differ across centers and regions (36).
First, there was heterogeneity across trials in absolute death rates, which are likely due to differences in baseline characteristics. Both the ROCKET AF and ENGAGE AF trials included patients at high to very high risk of stroke (CHADS2 score ≥2), and therefore included a higher number of patients with a history of heart failure, advanced age, hypertension, diabetes, and prior stroke/SE than those patients enrolled in the RE-LY and ARISTOTLE trials, which recruited a broader population with a lower thromboembolic risk (approximately corresponding to a CHADS2 score ≥1). However, because baseline characteristics were balanced between treatment groups, the RR of death was homogeneous across trials.
Second, there were methodological differences across trials in the determination of cause-specific mortality by the adjudicating committees (e.g., particularly in the subclassification of cardiac deaths or deaths due to stroke or hemorrhage). These differences may have led to under- or overestimation of specific causes of death. Anyway, as the subclassification of deaths made by the adjudicating committees of the studies was blinded, any potential differences are likely to be randomly distributed between treatment groups. In addition, there were some differences in the quality of follow-up for vital status across studies, thus potentially leading to underestimation of the numbers of deaths to some extent. However, the rate of patients lost to follow-up for vital status at end of study was at most 2% (the ARISTOTLE trial), without significant imbalances between treatment groups, thus making detection bias unlikely.
Third, our meta-analysis is based on data from selected populations from clinical trials, and therefore, the extrapolation to general practice should be done cautiously. The populations included in contemporary AF trials with the DOAC included a high proportion of VKA-experienced patients with demonstrated tolerance to warfarin (37), whereas patients with anticipated poor adherence to treatment or conditions associated with a high risk of bleeding and mortality were excluded, such as those with severe renal impairment (38). In fact, <50% of AF patients in real-life cohorts would have met the strict inclusion/exclusion criteria applied in these studies (39). The main gap seems to be an under-representation of women and elderly patients in contemporary AF trials compared with those from registries (40). Some reports have shown that despite these baseline differences, mortality rates in real-life cohorts of patients with AF are consistent with those reported in clinical trials (41,42), but caution in the interpretation of the data is still necessary.
Almost one-half of deaths in contemporary AF anticoagulation trials are cardiac deaths (sudden cardiac death, heart failure, and myocardial infarction), whereas stroke and bleeding only account for approximately 6% of all deaths each as the principal cause. The introduction of the DOAC may confer some benefit in the reduction of fatal bleedings compared with VKA, but there is still a need to improve current management of comorbidities and associated risk factors in order to reduce mortality rates in AF beyond anticoagulation.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: In contemporary trials of anticoagulation, most deaths among patients with nonvalvular AF were related to underlying cardiovascular disease, whereas stroke and bleeding accounted for a relatively small proportion of deaths.
TRANSLATIONAL OUTLOOK: Further studies are necessary to investigate new interventions, beyond anticoagulation, to reduce mortality in patients with AF.
The authors thank Prof. Dr. Martina Brückmann (Boehringer Ingelheim Pharma GmbH & Co. KG) and Prof. Dr. Robert P. Giugliano (Division of Cardiovascular Medicine, Brigham and Women’s Hospital, Boston, Massachusetts) for answering the authors’ request for supplementary information related to the RE-LY and ENGAGE AF studies, respectively.
For an expanded Methods section as well as supplemental tables and figures, please see the online version of this paper.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- confidence interval
- direct oral anticoagulant
- odds ratio
- relative risk
- systemic embolism
- time within therapeutic range
- vitamin K antagonist
- Received June 29, 2016.
- Revision received August 25, 2016.
- Accepted September 13, 2016.
- American College of Cardiology Foundation
- Fuster V.,
- Rydén L.E.,
- Cannom D.S.,
- et al.
- Gómez-Outes A.,
- Suárez-Gea M.L.,
- Lecumberri R.,
- et al.
- Higgins J.P.,
- Altman D.G.,
- Gøtzsche P.C.,
- et al.
- Liberati A.,
- Altman D.G.,
- Tetzlaff J.,
- et al.
- Marijon E.,
- Le Heuzey J.Y.,
- Connolly S.,
- et al.
- Hart R.G.,
- Diener H.C.,
- Yang S.,
- et al.
- Hohnloser S.H.,
- Oldgren J.,
- Yang S.,
- et al.
- Beasley N, Thompson A. Questions: dabigatran. FDA Draft Briefing Document for the Cardiovascular and Renal Drugs Advisory Committee. September 20, 2010. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/CardiovascularandRenalDrugsAdvisoryCommittee/UCM247244.pdf. Accessed May 9, 2016.
- Committee for Medicinal Products for Human Use (CHMP). Pradaxa (dabigatran). Assessment report No.: EMEA/H/C/000829/X/13/G. August 23, 2011. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Assessment_Report_-_Variation/human/000829/WC500110875.pdf. Accessed May 10, 2016.
- Pokorney S.D.,
- Piccini J.P.,
- Stevens S.R.,
- et al.
- Hankey G.J.,
- Stevens S.R.,
- Piccini J.P.,
- et al.
- Beasley N, Dunnmon P, Rose M. Xarelto (rivaroxaban). FDA Draft Briefing Document for the Cardiovascular and Renal Drugs Advisory Committee (CRDAC). September 8, 2011. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/drugs/CardiovascularandRenalDrugsAdvisoryCommittee/ucm270796.pdf. Accessed May 9, 2016.
- Committee for Medicinal Products for Human Use (CHMP). Xarelto (rivaroxaban). Assessment report No.: EMEA/H/C/000944/II/0012. September 22, 2011. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Assessment_Report_-_Variation/human/000944/WC500120735.pdf. Accessed May 10, 2016.
- Center for Drug Evaluation and Research. Eliquis (apixaban). Clinical Review Addendum. December 17, 2012. Available at: http://www.accessdata.fda.gov/drugsatfda_docs/nda/2012/202155Orig1s000MedR.pdf. Accessed May 9, 2016.
- Committee for Medicinal Products for Human Use (CHMP). Eliquis (apixaban). Assessment report No.: EMEA/H/C/002148/X/04/G. September 20, 2012. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Assessment_Report_-_Variation/human/002148/WC500136575.pdf. Accessed May 10, 2016.
- Giugliano R.P.,
- Ruff C.T.,
- Wiviott S.D.,
- et al.
- Blank M, McDowell TY, Rose M. Savaysa (edoxaban). FDA Draft Briefing Document for the Cardiovascular and Renal Drugs Advisory Committee (CRDAC). October 30, 2014. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/CardiovascularandRenalDrugsAdvisoryCommittee/UCM420704.pdf. Accessed May 9, 2016.
- Committee for Medicinal Products for Human Use (CHMP). Lixiana (edoxaban). Assessment report No.: EMEA/H/C/002629/0000. April 23, 2015. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Public_assessment_report/human/002629/WC500189047.pdf. Accessed May 10, 2016.
- Piepoli M.F.,
- Hoes A.W.,
- Agewall S.,
- et al.
- Fox C.S.,
- Golden S.H.,
- Anderson C.,
- et al.
- Steffel J.,
- Giugliano R.P.,
- Braunwald E.,
- et al.
- Connolly S.J.,
- Pogue J.,
- Eikelboom J.,
- et al.
- Singer D.E.,
- Hellkamp A.S.,
- Piccini J.P.,
- et al.
- Gómez-Outes A.,
- Terleira-Fernández A.I.,
- Calvo-Rojas G.,
- et al.
- Wang S.V.,
- Franklin J.M.,
- Glynn R.J.,
- et al.
- Sarnak M.J.,
- Levey A.S.,
- Schoolwerth A.C.,
- et al.
- Desmaele S.,
- Steurbaut S.,
- Cornu P.,
- et al.
- Blin P.,
- Dureau-Pournin C.,
- Lassalle R.,
- et al.
- Larsen T.B.,
- Skjøth F.,
- Nielsen P.B.,
- et al.